Datasets:

urialon

/

gov_report_validation

like 0

gao_GAO-19-196

gao_GAO-19-196_0

Background The Consumer Reporting Process Information on consumers is exchanged through a consumer reporting process that includes consumers, CRAs, furnishers, and users of that information (see fig.1). Consumers are individuals whose information is collected and shared to make eligibility decisions, such as for credit, insurance, or employment. CRAs are companies that assemble or evaluate consumer information for the purpose of furnishing consumer reports to third parties who use the reports to determine consumer eligibility for employment, or products and services such as credit and insurance. Furnishers are entities such as banks or credit card companies that provide CRAs with consumer information, such as account openings, bill payments, or delinquency information. CRAs use this information, along with other information, including from public records such as bankruptcies, to compile consumer reports. Users are banks, credit card companies, employers, or other entities that use consumer reports to make eligibility decisions for individual consumers. Users vary in the specific information they request from CRAs and how they interpret the data. Some institutions, such as banks, may act as both furnishers and users. During the consumer reporting process, a consumer would not necessarily interact with the CRA; however, if the consumer discovered inaccurate information on their credit report as a result of, for example, being denied credit, the consumer could file a dispute with the CRA or the furnisher. Consumers may also request copies of their consumer reports from CRAs directly, and CRAs may provide consumers with disclosures about how their information is being shared. Oversight Agencies FTC and, most recently, CFPB, are the federal agencies primarily responsible for overseeing CRAs. FTC has authority to investigate most organizations that maintain consumer data and to bring enforcement actions for violations of statutes and regulations that concern the security of data and consumer information. CFPB, created in 2010 by the Dodd- Frank Wall Street Reform and Consumer Protection Act (Dodd-Frank Act), has enforcement authority over all CRAs for violations of certain consumer financial protection laws. In general, it also has the authority to issue regulations and guidance for those laws. CFPB has supervisory authority over larger market participants in the consumer reporting market. In 2012, CFPB defined larger market participant CRAs as those with more than $7 million in annual receipts from consumer reporting. CFPB’s supervision of these companies includes monitoring, inspecting, and examining them for compliance with the requirements of certain federal consumer financial laws and regulations. As discussed below, these laws include most provisions of the Fair Credit Reporting Act (FCRA); several provisions of the Gramm-Leach-Bliley Act (GLBA); and provisions of the Dodd-Frank Act concerning unfair, deceptive, or abusive acts or practices. Data Breaches and the Equifax Breach Although there is no commonly agreed-upon definition of “data breach,” the term generally refers to an unauthorized or unintentional exposure, disclosure, or loss of sensitive information. This information can include personally identifiable information such as Social Security numbers, or financial information such as credit card numbers. A data breach can be inadvertent, such as from the loss of an electronic device; or deliberate, such as from the theft of a device. A breach can also occur as a result of a cyber-based attack by individuals or groups, including organizations’ own employees, foreign nationals, or terrorists. Data breaches have occurred at all types of organizations, including private, nonprofit, and federal and state entities. In the Equifax data breach, Equifax system administrators discovered on July 29, 2017, that intruders had gained unauthorized access via the Internet to a server housing the company’s online dispute portal. The breach compromised the personally identifiable information of at least 145.5 million individuals, and included names, addresses, and birth dates; and credit card, driver’s license, and Social Security numbers. Equifax’s investigation of the breach identified factors that led to the breach: software vulnerabilities, failure to detect malicious traffic, failure to isolate databases from each other, and inadequately limiting access to sensitive information such as usernames and passwords. Equifax’s public filings after the breach noted that the company took steps to improve security and notify individuals about the breach. Our August 2018 report provides more information on the breach and Equifax’s response. While data breaches do not always result in measurable harm, intruders may retain or resell stolen information to commit identity theft, which can include existing-account fraud and new-account fraud. In existing-account fraud, identity thieves use financial account identifiers, such as credit card or debit card numbers, to take over an individual’s existing accounts to make unauthorized charges or withdraw money. In new-account fraud, identity thieves use an individual’s identifying data, such as Social Security and driver’s license numbers, to open new financial accounts and incur charges and obtain credit in an individual’s name without that person’s knowledge. In addition, identity thieves may commit synthetic identity fraud, where they combine real and/or fictitious information to create identities with which they may defraud financial institutions, government agencies, or individuals. Several Federal Laws Govern the Collection, Use, and Protection of Consumer Information FCRA Governs the Accuracy, Use, and Sharing of Consumer Information, and CRAs Reported Taking Actions to Comply FCRA, enacted in 1970, is one of the primary federal laws governing the personal information that CRAs hold. It governs the accuracy of this information and gives consumers rights to view, correct, or opt out of the sharing or use of certain aspects of their personal information among affiliates. FCRA also applies to how CRAs can use and share the information. Accuracy of collected information. FCRA requires that when preparing a consumer report, CRAs follow reasonable procedures to assure “maximum possible accuracy” of the information concerning the individual about whom the report relates. Companies that furnish information to CRAs also must take steps regarding the accuracy of information they report, as required by FCRA and its implementing regulation, Regulation V. A 2012 CFPB report cited steps that nationwide CRAs take to help ensure that information they collect from furnishers is legitimate and accurate. The report notes that initial screening of furnishers generally includes an inspection of the companies’ physical headquarters, phone numbers, websites, business licenses, and company records such as annual reports. In addition, these CRAs may hire third-party investigation services to screen for illegal or unethical business practices. They may also conduct additional inspections in response to consumer complaints, variations in data reporting, or changes in a furnisher’s ownership. To conduct quality checks on data submitted by furnishers, CFPB reported that the nationwide CRAs check for blank fields or logical inconsistencies. Representatives of CRAs we spoke with provided examples of the quality assurance steps they take. For example, one representative told us that they look for violations of logical patterns, such as a loan going from 30 days past due to 90 days past due over the course of one month. CFPB reported that when inaccuracies are identified, the CRAs can reject the information. These steps may improve the quality of the information received from furnishers, but they cannot ensure the accuracy of such data. Use and sharing of information. FCRA permits CRAs to provide users with consumer reports only if the user has a “permissible purpose,” such as to process a credit application, screen a job applicant, or underwrite an insurance policy, subject to limitations where the credit or insurance transaction is not initiated by the consumer. FCRA also prohibits the use of a consumer report for any purpose other than that specified to the CRA when the user obtained the report. It also requires that CRAs take steps to validate the legitimacy of users and their requests for consumer report information and apply FCRA requirements to the sharing of information within their companies. Validating the legitimacy of users and their requests for consumer report information. Representatives of CRAs told us they take several steps to validate the legitimacy of users and their requests, including verifying credit transactions, periodically evaluating user agreements, and validating users’ identities. For example, representatives of one CRA said they sometimes conduct on-site visits to verify the existence of an entity and the business it conducts. In addition, they said they randomly select 6,000 to 8,000 consumer files each year and ask users associated with those files to show proof that the consumers engaged in the credit transactions contained in those files. However, several CRAs told us that these steps cannot guarantee that the users and requests are valid. For example, representatives of one CRA noted that once a user has the information, a CRA would find it difficult to prevent that user from retaining and reusing it for purposes other than the original permissible purpose. Applying FCRA requirements to sharing information internally. As amended by the Fair and Accurate Credit Transactions Act of 2003, FCRA limits the ability of affiliated companies to market products or services to consumers using shared consumer data. Affiliates may use consumer report information for product or service marketing only if they clearly and conspicuously disclose to the consumer that the information may be shared for such solicitations, the consumer is provided a simple method to opt out of such solicitations, and the consumer does not opt out. Representatives of CRAs told us that they apply the same FCRA protections when they share consumer reporting data among their departments or subsidiaries, which may use the data for other purposes. For example, one nationwide CRA said that one of its internal groups seeks to ensure that the company implements appropriate legal protections when it shares data for other uses within the company. Staff from state Attorneys General offices we spoke with told us that their states also have laws pertaining to consumer reporting, which have similar requirements to those in FCRA. In addition, they noted that while there is no federal data breach notification law, all 50 states have laws requiring companies to notify consumers in the event of a data breach. According to the National Conference of State Legislatures, those laws have varying requirements, such as the timing or method of notification, and who must be notified. GLBA and Other Laws Govern the Protection of Consumer Information Congress enacted GLBA in part to protect the privacy and security of nonpublic personal information that individuals provide to financial institutions. According to FTC staff, CRAs may be considered financial institutions under GLBA if they collect, maintain, and report on consumer information. As with FCRA, GLBA restricts financial institutions from sharing consumers’ private information, but GLBA restricts sharing with nonaffiliated third parties specifically, and those parties face similar restrictions in how they may further share or use the information. In addition, unlike FCRA, GLBA includes a provision directing FTC and certain federal regulators (not including CFPB) to establish standards specifically with respect to protection against any anticipated threats or hazards to the security of customer records. Specifically, under GLBA, these federal regulators are directed to establish appropriate standards for financial institutions under their jurisdiction to ensure the security and confidentiality of customer records and information; protect against any anticipated threats or hazards to the security or integrity of such records; and protect against unauthorized access to or use of such records or information that could result in substantial harm or inconvenience to any customer. To implement these standards for CRAs, and other entities that fall under its jurisdiction, FTC adopted its Safeguards Rule, which requires, among other things, that financial institutions have a written information security program, assess the risks to customer information, and evaluate and adjust the information security program in light of foreseeable risks. FTC staff told us that because GLBA applies to information about a consumer with a customer relationship with a financial institution, the Safeguards Rule may not apply in all cases where a CRA holds personal information on individuals. For example, they said that GLBA would more clearly apply if the consumer had purchased credit monitoring or other products or services directly from the CRA, or if the CRA obtained customer information from another financial institution, such as a bank. Representatives of the three nationwide CRAs told us that for purposes of protecting information, they do not distinguish between consumers with whom they have a direct customer relationship and those with whom they do not. CRAs we spoke with provided examples of how they protect consumer information and meet GLBA requirements to maintain administrative, technical, and physical safeguards. For example, with respect to administrative safeguards, representatives of one CRA said they enforce contractual requirements for data access and data security. Representatives of another CRA said that the technical safeguards they use include firewalls, anti-virus software, and malware protection. Examples of physical safeguards from another CRA included monitoring data centers by video and restricting access to secure data rooms. To address data protection more generally, representatives of CRAs we spoke with told us they routinely conduct internal audits of their data security systems, and that the financial institutions they work with frequently conduct audits of their risk management practices, including CRAs’ data security controls. Provisions related to unfair or deceptive acts or practices also may apply to CRAs’ protection of consumer data. Specifically, under FTC’s authority, section 5 of the Federal Trade Commission Act (FTC Act) prohibits “unfair or deceptive acts or practices” in or affecting commerce. In the context of privacy and security, these provisions require companies to truthfully represent practices to consumers. For example, FTC has found companies that alleged that they were following certain security protections, but did not in fact have such security features, to have engaged in unfair or deceptive practices. Similarly, the Dodd-Frank Act prohibits providers of consumer financial products or services from engaging in “unfair, deceptive, or abusive acts or practices,” and CFPB has authority to enforce and supervise for compliance with this provision. CFPB has alleged that claims to consumers that transactions are safe and secure while simultaneously lacking basic security practices can constitute unfair, deceptive, or abusive acts or practices. FTC and CFPB officials said that in the case of data breaches, they would examine each case individually to determine whether the institution violated these provisions in connection with the breach. Some states also have laws that protect consumer information, including laws that generally govern data security. For example, staff from the Massachusetts Attorney General’s office told us that their state has a data security law similar to FTC’s Safeguards Rule but with more specific requirements, including those for malware detection and firewalls. According to the National Consumer Law Center, all 50 states have consumer protection laws that prohibit unfair or deceptive practices. Staff from state Attorneys General offices told us that they can prosecute entities for potential violations of these provisions, including data breaches. They told us that following the Equifax breach, several states’ Attorneys General launched a joint investigation into whether Equifax violated state laws, including prohibitions of unfair or deceptive practices. According to staff from one state Attorney General office, as of February 2019, this investigation was ongoing. In addition, Equifax reported that individual states have also filed legal action or have ongoing investigations. For example, Massachusetts and West Virginia have filed civil enforcement actions against Equifax that seek various remedies, including civil penalties. FTC Has Taken Enforcement Measures against CRAs but Lacks Civil Penalty Authority for GLBA Data Protection Provisions FTC Enforces CRA Compliance with Consumer Protection Laws FTC enforces compliance with consumer protection laws under authorities provided in FCRA, GLBA, and the FTC Act. FCRA authorizes FTC to enforce compliance for nearly all companies not supervised by either a federal banking regulator or certain other federal agencies. GLBA authorizes FTC to issue certain rules and enforce compliance for all nonbank financial institutions and other entities not under the jurisdiction of a federal banking regulator, the National Credit Union Administration, Securities and Exchange Commission, or state insurance regulators. The FTC Act authorizes FTC to investigate and take administrative and civil enforcement actions against companies under its jurisdiction that engage in unfair or deceptive acts or practices in or affecting commerce. According to FTC, in the last 10 years, it has brought 34 enforcement actions for FCRA violations, including 17 against CRAs. In addition, FTC said that it had taken a total of 66 actions against companies (not just in the last 10 years), including CRAs, that allegedly engaged in unfair or deceptive practices relating to data protection. If FTC has reason to believe that a company has violated laws under its jurisdiction, it may initiate an investigation to determine whether to take enforcement action. FTC staff said that in determining whether to take on a case related to privacy and data security matters, they consider factors such as the company’s size and the sensitivity of the data in the company’s network. For example, FTC may choose not to investigate a data breach of a small company that affects few people; however, it may investigate a potential data security violation of a large company, even without evidence of a breach. Under its statutory authority, FTC can ask or compel companies to produce documents, testimony, and other materials to assist in its investigations. In June 2018, FTC notified Equifax that it was considering legal action against the company as a result of its 2017 data breach, including seeking civil penalties. If FTC finds that a company violated consumer law, the agency may take several different actions depending on its legal authority and what it considers to be the most appropriate response. For example, FTC may, in administrative proceedings, issue cease-and-desist orders for unfair or deceptive acts or practices. Further, FTC generally may seek a range of remedies from the U.S. district courts, including injunctions, damages to compensate consumers for their actual losses, and disgorgement of ill- gotten funds. In limited circumstances, FTC also may seek civil money penalties, which are monetary fines imposed for a violation of a statute or regulation. Examples of FTC enforcement actions related to consumer reporting include: In May 2016, FTC settled with a furnisher that allegedly violated FCRA requirements to have adequate policies and procedures for reporting accurate credit information to CRAs. FTC alleged that a debt collector acting as a furnisher did not have a written policy regarding the accuracy and integrity of information it furnished, and in numerous instances failed to inform consumers about these outcomes. In 2011, FTC brought enforcement actions against three CRAs that merge, and then sell, information from the three nationwide CRAs. FTC alleged that these companies did not meet GLBA standards and violated unfair or deceptive practices prohibitions by not providing reasonable and appropriate security for consumers’ personal information. These violations included not developing and disseminating information security policies, and not addressing risks by, for example, evaluating the security of end users’ computer networks. In 2006, FTC settled with ChoicePoint—a CRA—and imposed a $10 million civil penalty for violations of FCRA stemming from a 2005 data breach. In 2009, FTC obtained an additional $275,000 in equitable monetary relief due to ChoicePoint’s violation of the order after an additional data breach occurred in 2008. FTC’s Lack of Civil Penalty Authority for GLBA May Hinder Its Effectiveness in Enforcing Data Security Provisions As previously discussed, in some circumstances, FTC enforcement authority can include civil money penalties. This includes cases of knowing violations of FCRA. For example, in a 2014 settlement, FTC levied $525,000 in civil penalties against a CRA after alleging that the company did not comply with FCRA provisions to ensure the accurate and permissible use of its reports. FTC does not have civil penalty authority for initial violations of the FTC Act but may obtain civil penalties from companies for violations of FTC Act orders. FTC’s civil penalty authority does not extend to initial violations of GLBA’s privacy and safeguarding provisions, which require administrative, physical, and technical safeguards with an emphasis on protection against anticipated threats and unauthorized access to customer records. For violations of GLBA provisions, which are enforced pursuant to FTC Act authority, FTC may seek an injunction to stop a company from violating these provisions and may seek redress (damages to compensate consumers for losses) or disgorgement. However, determining the appropriate amount of consumer compensation requires FTC to identify the consumers affected and the amount of monetary harm they suffered. In cases involving security or privacy violations resulting from data breaches, assessing monetary harm can be difficult. Consumers may not be aware that their identities have been stolen as a result of a breach and or identity theft, and related harm may occur years in the future. In addition, it can be difficult to trace instances of identity theft to specific data breaches. According to FTC staff, these factors can make it difficult for the agency to identify which individuals were victimized as a result of a particular breach and to what extent they were harmed and then obtain related redress or disgorgement. Having civil penalty authority for GLBA provisions would allow FTC to fine a company for a violation such as a data breach without needing to prove the monetary harm to individual consumers. FTC staff told us and testified before Congress that civil penalties are often the most appropriate remedy for a data breach, and that such penalties serve as an effective deterrent in cases involving weak data privacy and security policies and practices. FTC staff noted that in the case of a data breach, each consumer record exposed could constitute a violation; as a result, a data breach that involved a large number of consumer records could result in substantial fines. Unlike FTC, other regulators have civil penalty authority to punish entities that violate provisions of GLBA. For example, the Office of the Comptroller of the Currency has said that it can enforce GLBA privacy and safeguard provisions with civil money penalties against any insured depository institution or institution-affiliated party subject to its supervision. In our 2009 report on modernizing the financial regulatory framework, we stated that financial regulators should have the authority to carry out and enforce their statutory missions. In the case of FTC, this includes having the tools necessary to meet its mission of protecting consumers from harm, including the harm caused by misuse of personal information, by having the range of authorities to punish entities for violations of the statutes and regulations the agency enforces. In 2006, we suggested that Congress consider providing FTC with civil penalty authority for its enforcement of GLBA’s privacy and safeguarding provisions. We noted that providing this authority would give FTC a practical enforcement tool to more effectively enforce provisions related to security of data and consumer information. Following the 2008 financial crisis, Congress introduced several bills related to data protection and identity theft, which included giving FTC civil penalty authority for its enforcement of GLBA. However, in the final adoption of these laws, Congress did not provide FTC with this authority. Since that time, data breaches at Equifax and other large organizations have highlighted the need to better protect sensitive personal information. Accordingly, we continue to believe FTC and consumers would benefit if FTC had such authority. CFPB Enforces and Examines CRAs for Compliance with Consumer Protection Laws but Does Not Fully Consider Data Security in Prioritizing Examinations CFPB Enforces and Examines CRA Compliance with Consumer Protection Laws CFPB enforces compliance with most provisions of FCRA; several provisions of GLBA; and the prohibition of unfair, deceptive, or abusive acts or practices under the Dodd-Frank Act. According to CFPB staff, CFPB cannot enforce data security standards under these statutory provisions or the FTC’s implementing rules because CFPB does not have authority to supervise for or enforce compliance with the GLBA’s safeguards provision or FCRA’s red flags or records disposal provisions. Since 2015, CFPB has had five public settlements with CRAs. Four of these settlements included alleged violations of FCRA and three included alleged violations of unfair, deceptive, or abusive practices provisions. For example, in March 2017, CFPB settled with Experian for $3 million in civil penalties for an alleged violation of FCRA and alleged deceptive acts or practices. Experian marketed to consumers an “educational credit score” that the company claimed lenders used to make credit decisions. CFPB alleged that lenders did not use these “educational credit scores” for this purpose, and that Experian violated FCRA’s implementing regulation by requiring consumers to view Experian advertisements before obtaining a free credit report. In December 2015, CFPB levied a fine of $8 million against another CRA—Clarity Services, Inc.—for obtaining consumer reports without a permissible purpose in violation of FCRA and failing to investigate consumer disputes. CFPB is also continuing its investigation of Equifax’s data breach. CFPB supervises the larger market participant CRAs (those with more than $7 million in annual receipts from consumer reporting, as defined by CFPB) and has the authority to examine these CRAs for compliance with federal consumer financial protection laws. From 2015 through 2017, CFPB examined several CRAs. Some of these examinations resulted in findings of deficiencies related to data accuracy and dispute processes, and follow-up examinations were conducted as necessary. As part of its supervisory role, CFPB also periodically monitors the nationwide CRAs by requesting information on their activities and identifying any changes in risk to consumers and the market. CFPB uses this information to learn of changes to a CRA’s compliance, personnel, issues raised by the CRA’s internal audits, or other developments that might affect CFPB’s strategy for supervising the CRA. CFPB May Not Be Identifying All CRAs under Its Supervisory Authority CFPB has examined several larger market participant CRAs, but may not be identifying all CRAs that meet the $7 million threshold. CFPB staff told us that as of October 2018, they were tracking between 10 and 15 CRAs that might qualify as larger market participants (as defined by CFPB). CFPB staff told us that they believe the CRA market is highly concentrated and there were not likely to be many larger market participants beyond the 10 to 15 they are tracking. However, CFPB staff said that the 10 to 15 CRAs may not comprise the entirety of larger market participants because whether CRAs meet the threshold may vary from year to year and CFPB has limited data to determine whether CRAs meet the threshold. Specifically, CFPB staff said that identifying additional larger market participant CRAs can be challenging. For example, the Securities and Exchange Commission does not require nonpublicly traded CRAs to file financial and other information that CFPB could otherwise use to identify these CRAs, which are generally not widely known to the public. In addition, CFPB staff said they do not ask CRAs to provide their annual receipts, with the exception of the specific CRAs being considered for examination in a given year, because CFPB staff said calculating these receipts could create an additional cost to the companies. Our January 2009 report on reforming the U.S. financial regulatory structure noted that regulators should be able to identify institutions and products that pose risks to the financial system, and monitor similar institutions consistently. One method for identifying institutions for oversight, particularly where data are limited, is to require companies to register with the relevant regulator. For example, among other requirements, insured depository institutions must obtain a charter to operate, and money services businesses generally must register with the Financial Crimes Enforcement Network. Similarly, CFPB could identify CRAs that meet the larger market participant threshold by requiring such businesses to register with them, subject to a rulemaking process and cost-benefit analysis of the burden it could impose on the industry. Another method CFPB could use to identify CRAs and inform its oversight activities would be to leverage information collected by states. Stakeholders we spoke with cited New York and Maine as examples where CRAs are required to register with the state. Implementing strategies such as registration or leveraging existing information could be a cost-effective way for CFPB to identify all CRAs under its authority. Identifying additional sources of information on the population of larger market participant CRAs—including those that are lesser-known, possibly unknown to CFPB, and possibly in possession of large amounts of sensitive consumer information—could help ensure that CFPB has more comprehensive information for carrying out its supervisory responsibilities. CFPB’s Prioritization of CRA Examinations Does Not Specifically Account for Data Security Risk To determine which product lines, institutions, and compliance issues to examine, CFPB determines the institutions (for example, banks, credit unions, non-bank mortgage servicers, and CRAs) and the consumer product lines that pose the greatest risk to consumers, and prioritizes these for examinations annually (see fig.2). CFPB segments the consumer product market into institution product lines, or specific institutions’ offerings of consumer product lines. CFPB then assesses each institution product line’s risk to consumers at the market level and institutional level. To assess risk at the market level, CFPB considers market size and other factors that contribute to market risk. Market size includes a consideration of a product’s market size relative to other consumer finance product markets. Other market risk factors include the potential risk to consumers from new or existing products offered in the market as well as emerging risks and trends in consumer financial products. For example, CFPB noted that a market may be considered higher risk if consumers cannot choose the provider of a financial product or service in that market, or if the transactions occur between two businesses rather than between a business and consumers. Because they do not face the same risk of losing customers as companies in other markets, companies in higher-risk markets may not have the same financial incentives to protect the interests of consumers. To assess risk at the institution level, CFPB considers an institution’s market share within a product line, as well as field and market intelligence. An institution’s market share correlates with the number of consumers who could be affected by that institution’s practices; therefore, CFPB generally places a higher priority on larger providers of products. Field and market intelligence includes quantitative and qualitative information on an institution’s operations for a given product line, including the strength of its compliance management systems, the number of regulatory actions directed at the institution, findings from prior CFPB examinations, information obtained from CFPB’s quarterly monitoring of institutions, public reports, and the number and severity of consumer complaints CFPB has received about the institution. Field and market intelligence can also include information about an institution’s fair lending practices and its ability to provide fair, equitable, and nondiscriminatory access to credit. Taking market and institutional considerations together, CFPB places institution product lines into tiers based on its determination of their relative risk to consumers. These risk tiers range from 1 to 5, with 1 being the lowest risk and 5 being the highest risk. Risk tiers then feed into CFPB’s development of its supervision strategy, which includes other information, including information from subject matter experts and recent legal and policy decisions that could affect examinations, and consultations with internal stakeholders. CFPB uses both the risk tiers and information from its supervision strategy to identify potential institutions for examination. Following this process, CFPB has regularly determined CRAs’ consumer reporting to be a high priority for examination since it began supervising them in 2012. After identifying institution product lines to examine, CFPB determines specific areas of compliance to assess. These determinations are made by considering sources such as consumer complaints, public filings and reports, and past examination findings related to the same or similar products or institutions. Most recently, CFPB examinations of CRA’s consumer reporting have focused on issues such as data accuracy, dispute processes, compliance management, and permissible purposes. Although CFPB’s examination prioritization incorporates several important factors and sources, the process does not routinely include assessments of data security risk, such as how institutions detect and respond to cyber threats. According to CFPB staff, the agency’s process for determining risk tiers incorporates the risk factors specifically cited in the Dodd-Frank Act, including those related to the size of a product market. The Act also states that CFPB should consider other factors it determines to be relevant; as such, CFPB staff noted that certain elements of data protection have been included in the scope of some of its past CRA examinations. For example, CFPB staff said that in assessing compliance with FCRA’s permissible purposes provision, the examination scope would include ensuring that data are not improperly shared. CFPB staff noted that the bureau cannot examine for compliance with or enforce the data security standards in provisions of GLBA and FCRA or FTC’s implementing rules, even at larger participant CRAs. After the Equifax breach, however, CFPB used its existing supervisory authority to develop internal guidelines for examining data security, and conduct some CRA data security examinations. CFPB staff said that they do not routinely consider data security risks during their examination prioritization process and have not reassessed the process to determine how to incorporate such risks going forward. The Dodd-Frank Act requires CFPB, when implementing its risk-based supervision program, to consider risks posed to consumers in the relevant product and geographic markets. In addition, federal internal control standards state that agencies should identify, analyze, and respond to risks related to achieving defined objectives. This can entail considering all significant internal and external factors to identify risks and their significance, including magnitude of impact, likelihood of occurrence, nature of the risk, and appropriate response. In light of the Equifax breach, as well as CFPB’s acknowledgment of the CRA market as a higher-risk market for consumers, it is important for CFPB to routinely consider factors that could inform the extent of CRA data security risk such as the number of consumers that could be affected by a data security incident and the nature of potential harm resulting from the loss or exposure of information. CFPB’s reliance primarily on consumer complaints, information from public filings, and information and findings from past examination for prioritizing examinations may not fully detect data security risks that CRAs pose. Data accuracy and dispute resolution feature prominently in consumer complaints, according to CFPB staff, because consumers mostly interact with CRAs in these contexts. But consumers likely did not know, for example, about Equifax’s data security challenges prior to its breach, so that vulnerability was not a focus of complaints. While the three nationwide CRAs acknowledged the risk of data breaches in recent public filings, other larger participant CRAs may not be publicly traded and therefore may not have public filings. Further, if CFPB’s past examinations have not addressed data security, the agency cannot use those past examination findings to target current risks. The Equifax breach demonstrated the vulnerability that CRAs may face with regard to data security. We have noted that advancements in technology, combined with the increasing sophistication of hackers and others with malicious intent, have increased the risk of sensitive personal information being exposed and compromised. We have also reported that rapid developments in new technologies will continue to pose new threats to security, privacy, and safety. In recent years, insured depository institutions—which, like CRAs, maintain large amounts of sensitive consumer data—have been subject to regular information technology examinations, which, according to one regulator, may include a cybersecurity component. Banking regulators have noted that unauthorized access to the information and systems that support these institutions can affect operations, pose risk to consumers through exposure of private information, and undermine consumer confidence. The risks may be similar for CRAs—companies that by definition also maintain extensive amounts of sensitive consumer information. By including routine consideration of data security risks into its process for prioritizing CRA examinations, CFPB can better ensure that its supervision of CRAs proactively detects such risks and helps prevent the further exposure or compromise of consumer information. Regulators Inform Consumers about Protections Available and Consumers Can Take Some Actions after a CRA Data Breach FTC and CFPB Provide Consumers with Information on How to Address Identity Theft Risk FTC and CFPB provide educational information for consumers on ways to mitigate the risk of identity theft. For example, FTC has a dedicated website (IdentityTheft.gov) that allows consumers to report suspected identity theft to FTC and develop and implement a recovery plan. In addition, FTC offers businesses guidance on steps to take in the event of a data breach, including notification of relevant parties and a model notification letter. CFPB’s website offers consumers tips on how to protect their information and spot identity theft. CFPB also publishes a consumer guide that lists CRAs and their websites, and ways to obtain free credit reports. After a breach, FTC and CFPB publish information specific to that breach. For example, shortly after Equifax’s announcement of the breach, FTC published information on when the breach occurred, the types of data compromised, and links to additional information on Equifax’s website. Similarly, CFPB released three blog posts and several social media posts shortly after Equifax’s public announcement of the breach. These included information on ways that consumers could protect themselves in the wake of the breach and special protections and actions for service members. Consumers Have Options to Mitigate Identity Theft Risk and Respond to Breaches At any time, consumers can take actions to help mitigate identity theft risk. For example, consumers can implement a credit freeze free of charge, which can help prevent new-account fraud by restricting potential creditors from accessing the consumer’s credit report. Similarly, implementing a free fraud alert with a credit bureau can help prevent fraud because it requires a business to verify a consumer’s identity before issuing credit. Consumers also can monitor their credit report for suspicious activity, either through self-review or by using a free or paid credit monitoring service. FTC and others recommend that consumers regularly review their credit card and bank statements to detect fraudulent charges. Consumers whose data have been compromised in any data breach can file a complaint with FTC or CFPB. FTC has an online “complaint assistant,” and FTC staff told us they use consumer complaints to help inform their investigatory and enforcement activity. CFPB staff told us that they use consumer complaints to help prioritize examinations and inform enforcement activity. In the 6 months following Equifax’s announcement of its data breach, CFPB received more than 20,000 consumer complaints about the impact of the breach or Equifax’s response. However, consumers are limited in the direct actions they can take against a CRA in the event of a data breach, for two primary reasons. First, consumers generally cannot trace the source of the data used to commit identity theft to a particular breached entity. As a result, it can be difficult to link a breach by a CRA (or any other entity) to the harm a consumer suffers from a particular incidence of identity theft, which may make it challenging to prevail in a legal action. Second, unlike with many other products and services, consumers generally cannot exercise choice if they are dissatisfied with a CRA’s privacy or security practices. Specifically, consumers cannot choose which CRAs maintain information about them. In addition, consumers do not have a legal right to delete their records with CRAs, according to CFPB staff, and therefore cannot choose to remove themselves entirely from the CRA market. FTC and CFPB have noted that the level of consumer protection required can depend on the consumer’s ability to exercise choice in a marketplace. For example, when determining whether a practice constitutes an unfair practice, FTC considers whether the practice is one that consumers could choose to avoid. Similarly, according to CFPB staff, the consumer reporting market may pose higher risk to consumers because consumers cannot choose whether or which CRAs possess and sell their information. Conclusions The 2017 data breach of Equifax highlighted the data security risks associated with CRAs. While companies in many industries have experienced data breaches, CRAs may present heightened risks because of the scope of sensitive information they possess and because consumers have very limited control over what information CRAs hold and how they protect it. These challenges underscore the importance of appropriate federal oversight of CRAs’ data security. While FTC has taken significant enforcement actions against CRAs that have violated federal privacy or data security laws, it is important that the agency have all of the appropriate enforcement options to fulfill its mission of protecting consumers. However, GLBA, one of the key laws governing the security of consumer information, does not provide FTC with civil penalty authority. The remedies that FTC does have available under GLBA—such as disgorgement and consumer redress—may be less practical enforcement tools for violations involving breaches of mass consumer data. Accordingly, providing FTC with civil penalty authority can enable it to more effectively or efficiently enforce GLBA’s privacy and safeguarding provisions. Although CFPB is responsible for overseeing larger market participant CRAs, it lacks the data to identify with certainty all the CRAs under its supervision, in part because the sources it is using, such as public filings, are not comprehensive. Using additional methods to obtain information, such as requiring larger market participant CRAs to register with the agency or leveraging state registration information, would help CFPB ensure it is tracking all CRAs under its supervision and is providing appropriate oversight. CFPB considers a number of market and institutional factors in prioritizing which CRAs to examine, but data security has not routinely been among these factors. Given the nature and amount of consumer information CRAs hold, as well as increasing threats from hackers and others with malicious intent, vulnerabilities in these companies’ data security can pose significant risk to a vast number of consumers. By ensuring that its process for determining the scope of CRA examinations routinely includes factors that would detect data security risks, CFPB can better ensure the effectiveness of its supervision and help prevent further exposure or compromise of consumer information. Matter for Congressional Consideration Congress should consider providing the Federal Trade Commission with civil penalty authority for the privacy and safeguarding provisions of the Gramm-Leach-Bliley Act to help ensure that the agency has the tools it needs to most effectively act against data privacy and security violations. (Matter for Consideration 1) Recommendations for Executive Action We are making two recommendations to CFPB: The Director of CFPB should identify additional sources of information, such as through registering CRAs or leveraging state information, that would help ensure the agency is tracking all CRAs that meet the larger participant threshold. (Recommendation 1) The Director of CFPB should assess whether its process for prioritizing CRA examinations sufficiently incorporates the data security risks CRAs pose to consumers, and take any needed steps identified by the assessment to more sufficiently incorporate these risks. (Recommendation 2) Agency Comments and Our Evaluation We provided a draft of this report to CFPB, the Board of Governors of the Federal Reserve System, the Federal Deposit Insurance Corporation, FTC, and the Office of the Comptroller of the Currency. All of the agencies provided technical edits, which we incorporated as appropriate. In addition, we received written comments from CFPB, which are reprinted in appendix II. CFPB neither agreed nor disagreed with our recommendations. Regarding our recommendation that it identify additional sources of information that would help ensure that it is tracking all CRAs that meet the larger market participant threshold, CFPB noted that it cannot require CRAs to register with the bureau without first undertaking a rulemaking. While we acknowledge the challenges of tracking larger participant CRAs, we maintain that CFPB should be able to identify and monitor them consistently. In its letter, CFPB stated that this may be feasible. The agency noted that, short of rulemaking, there may be cost-effective ways to better ensure that it is appropriately tracking larger participant CRAs and added that they intend to track these CRAs by exploring ways to leverage state registration information. These actions, if fully implemented, would meet the intent of our recommendation. With respect to the recommendation that CFPB assess whether its process for prioritizing CRA examinations sufficiently incorporates data security risks, CFPB said it will continue to evaluate risks to consumers, including data security risks, as part of its prioritization process. CFPB also said it will assess whether that process should incorporate data security risks CRAs pose to consumers. However, CFPB expressed concern with the scope of its statutory authority, such as its lack of authority to supervise for compliance with GLBA safeguard provisions. CFPB noted that we did not adequately consider or discuss its limited statutory authority in the area of data security. Specifically, CFPB stated that it does not have authority to supervise for, enforce compliance with, or write regulations implementing GLBA’s safeguards provisions or FCRA’s records disposal provision. In response, we added language in the report to clarify CFPB’s lack of certain authorities over these data security provisions. Nonetheless, as we discuss in the report, CFPB has conducted data security examinations of some CRAs under its existing authority, including its authority to assess compliance with the requirements of federal consumer financial law. We continue to believe that effective supervision of CRAs and the protection of consumer information require that CFPB consider data security risks in its prioritization of CRA examinations. As agreed with your offices, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies to the appropriate congressional committees, CFPB, the Board of Governors of the Federal Reserve System, the Federal Deposit Insurance Corporation, FTC, and the Office of the Comptroller of the Currency. In addition, the report will be available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact Michael Clements at (202) 512-8678 or clementsm@gao.gov, or Nick Marinos at (202) 512-9342 or marinosn@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made major contributions to this report are listed in appendix III. Appendix I: Objectives, Scope, and Methodology Our objectives were to examine (1) the federal laws and regulations governing consumer reporting agencies’ (CRA) collection, use, and protection of consumer information; (2) measures the Federal Trade Commission (FTC) has taken to enforce CRA compliance with requirements to protect consumer information; (3) measures the Consumer Financial Protection Bureau(CFPB) has taken to ensure that CRAs protect consumer information; and (4) FTC’s and CFPB’s roles in assisting consumers following a data breach and actions consumers can take following a data breach of a CRA. To examine the laws governing CRAs, we identified the relevant laws, including the Fair Credit Reporting Act, the Gramm-Leach-Bliley Act, and statutes related to unfair or deceptive acts or practices. We reviewed these laws for their application to CRAs and their collection, use, and protection of consumer information. We interviewed representatives of relevant federal agencies, including CFPB and FTC, about these laws and regulations and how they apply to CRAs. We also reviewed documents from and interviewed federal banking regulators on their role in overseeing financial institutions’ management of third-party risk, including those of CRAs. We selected four states with existing or proposed information protection laws or regulations that vary from federal requirements (California, Illinois, Massachusetts, and New York); reviewed related documentation; and interviewed Attorneys General from these states about their enforcement of state laws. In addition, we interviewed and reviewed documentation from the three nationwide CRAs and interviewed three other CRAs that produce or compile consumer reporting information. We selected these CRAs because they are not sector-specific and hold information on a broad segment of the population. We conducted a site visit to Equifax’s Alpharetta, Georgia data center to learn more about steps the company takes to comply with relevant consumer protection laws. We also interviewed representatives of furnishers and users of CRA consumer information—the American Bankers Association, the Property Casualty Insurance Association of America, and the National Retail Federation—about their roles in the collection, use, and protection of consumer data, and steps their members take to comply with relevant laws. To assess FTC’s and CFPB’s measures to enforce information protection provisions and to ensure CRAs’ proper collection, use, and protection of consumer information, we reviewed agency documentation and interviewed agency officials on their oversight activities. We reviewed the types of enforcement actions available to FTC and CFPB for violations of laws related to consumer reporting, as well as specific enforcement actions these agencies have brought against CRAs, data furnishers, and users of consumer reports. We also interviewed agency staff about FTC enforcement actions against CRAs and how it determines when to pursue such actions. We reviewed CFPB documentation on the scope of its supervisory examinations of larger market participant CRAs since 2015, as well as findings from recent CRA examinations. In addition, we reviewed CFPB examination guidance for supervising these CRAs, including CFPB’s internal guidelines for conducting data security examinations. We also reviewed documents related to CFPB’s process for prioritizing which institutions and which product lines (specific product offerings) should receive supervisory examination, and we interviewed CFPB staff about this process. Finally, we interviewed representatives of industry, consumer, and privacy groups for their views on the supervision of CRAs. These included the three nationwide CRAs, three other CRAs, the Consumer Data Industry Association, National Consumer Law Center, Consumer Federation of America, Consumers Union, World Privacy Forum, ID Theft Resource Center, and Consumer Action. To assess FTC and CFPB roles in assisting consumers, and actions consumers can take following a data breach of a CRA, we reviewed the two agencies’ efforts to inform and educate consumers following breaches. Specifically, we reviewed consumer education materials on FTC’s and CFPB’s websites related to data breaches and identify theft in general, as well as specific information posted after the Equifax data breach. We also interviewed staff from these agencies about their roles in assisting consumers following a breach. To identify actions consumers can take following a data breach, we reviewed our prior related reports and spoke with representatives of the industry and consumer representatives noted above. We conducted this performance audit from November 2017 to February 2019 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Comments from the Bureau of Consumer Financial Protection Appendix III: GAO Contacts and Staff Acknowledgments GAO Contacts GAO Staff Acknowledgments: In addition to the individuals named above, John de Ferrari and John Forrester (Assistant Directors); Winnie Tsen (Analyst-in-Charge), Bethany Benitez, Kavita Daitnarayan, Farrah Graham, Andrea Harvey, Thomas Johnson, Tovah Rom, Rachel Siegel, Jena Sinkfield, and Tina Torabi made key contributions to this report.

Why GAO Did This Study CRAs collect, maintain, and sell to third parties large amounts of sensitive data about consumers, including Social Security numbers and credit card numbers. Businesses and other entities commonly use these data to determine eligibility for credit, employment, and insurance. In 2017, Equifax, one of the largest CRAs, experienced a breach that compromised the records of at least 145.5 million consumers. GAO was asked to examine issues related to federal oversight of CRAs. Among other things, this report discusses (1) measures FTC has taken to enforce CRA compliance with requirements to protect consumer information, (2) measures CFPB has taken to ensure CRA protection of consumer information, and (3) actions consumers can take after a breach. GAO reviewed relevant laws, documentation related to CRA examinations, and policies and practices of selected CRAs; and interviewed representatives of regulatory agencies, CRAs, consumer and industry groups, and Attorneys General from four states with consumer reporting requirements. What GAO Found Since 2008, the Federal Trade Commission (FTC) has settled 34 enforcement actions against various entities related to consumer reporting violations of the Fair Credit Reporting Act (FCRA), including 17 actions against consumer reporting agencies (CRA). Some of these settlements included civil penalties—fines for wrongdoing that do not require proof of harm—for FCRA violations or violations of consent orders. However, FTC does not have civil penalty authority for violations of requirements under the Gramm-Leach-Bliley Act (GLBA), which, unlike FCRA, includes a provision directing federal regulators and FTC to establish standards for financial institutions to protect against any anticipated threats or hazards to the security of customer records. To obtain monetary redress for these violations, FTC must identify affected consumers and any monetary harm they may have experienced. However, harm resulting from privacy and security violations can be difficult to measure and can occur years in the future, making it difficult to trace a particular harm to a specific breach. As a result, FTC lacks a practical enforcement tool for imposing civil money penalties that could help to deter companies, including CRAs, from violating data security provisions of GLBA and its implementing regulations. Since 2015, the Consumer Financial Protection Bureau (CFPB) has had five public settlements with CRAs. Four of these settlements included alleged violations of FCRA; and three included alleged violations of unfair, deceptive, or abusive practices provisions. CFPB is also responsible for supervising larger CRAs (those with more than $7 million in annual receipts from consumer reporting) but lacks the data needed to ensure identification of all CRAs that meet this threshold. Identifying additional sources of information on these CRAs, such as by requiring them to register with the agency through a rulemaking or leveraging state registration information, could help CFPB ensure that it can comprehensively carry out its supervisory responsibilities. According to CFPB staff, the bureau does not have authority to examine for or enforce the GLBA’s safeguards provisions. After the Equifax breach, however, CFPB used its existing supervisory authority to examine the data security of certain CRAs. CFPB’s process for prioritizing which CRAs to examine does not routinely include an assessment of companies’ data security risks, but doing so could help CFPB better detect such risks and prevent the further exposure or compromise of consumer information. If a CRA experiences a data breach, affected consumers can take actions to mitigate the risk of identity theft—such as implementing a fraud alert or credit freeze—and can file a complaint with FTC or CFPB. However, consumers are limited in the direct actions they can take against the CRA. Consumers generally cannot exercise choice in the consumer reporting market—such as by choosing which CRAs maintain their information—if they are dissatisfied with a CRA’s privacy or security practices. In addition, according to CFPB, consumers cannot remove themselves from the consumer reporting market entirely because they do not have a legal right to delete their records with CRAs. This limited control by consumers, coupled with the large amount and sensitive nature of the information CRAs possess, underscores the importance of appropriate federal oversight of CRAs’ data security. What GAO Recommends GAO recommends that Congress consider giving FTC civil penalty authority to enforce GLBA’s safeguarding provisions. GAO also recommends that CFPB (1) identify additional sources of information on larger CRAs, and (2) reassess its prioritization of examinations to address CRA data security. CFPB neither agreed nor disagreed with GAO’s recommendations.

gao_GAO-18-190

gao_GAO-18-190_0

Background DOD guidance states that the Air Force and other services are responsible for providing trained and ready forces to fulfill the current and future operational requirements of the combatant commands. The Air Force is specifically responsible for gaining and maintaining air superiority. The Air Force Strategic Master Plan states that the Air Force must focus clearly on the capabilities that will allow freedom of maneuver and decisive action in highly contested spaces, including high-end air capabilities. Fifth generation fighter capabilities and ready and trained Airmen who are properly equipped for their missions are central components of the Air Force’s ability to provide air superiority in contested environments. The F-22 is the Air Force’s fifth generation, air superiority fighter that incorporates a stealthy and highly maneuverable airframe, advanced integrated avionics, and engines capable of sustained supersonic flight. The F-22 is optimized for air-to-air combat, able to carry up to eight air-to- air missiles, and equipped with a 20-millimeter cannon. After development began, the Air Force also added air-to-ground capabilities to the F-22. Air Force officials emphasized the synergistic benefits of the F-22 to the joint force. Specifically, the F-22’s individual capabilities, like its stealth and sensors, help it to coordinate and improve the performance of other aircraft during operations, including fourth generation fighters. The Air Force views the F-22 and the F-35—its other fifth generation fighter—as complementary platforms with some overlapping capabilities. For example, the F-22 is focused on, and more capable in, air-to-air missions and the F-35 is focused on, and more capable in, air-to-ground missions. The Air Force announced in its fiscal year 2018 budget request that it now intends to retain the F-22 until 2060. It has also begun an effort to define and develop the next generation of air superiority capabilities that it plans to field in 2030 and beyond. Figure 1 shows a picture of an F-22. The F-22 and the F-15C are the two operational fighters in the Air Force’s Air Superiority Core Function. The Air Force assigns two primary (air-to- air focused) missions and one secondary (air-to-ground focused) mission to the F-22. These missions are described in table 1. The Air Force requires its pilots to be proficient in their primary missions and familiar with their secondary missions. The size of the current F-22 fleet is smaller than the Air Force originally planned. The Air Force F-22 acquisition program began in 1991 with an intended development period of 12 years and a planned quantity of 648 aircraft. The Air Force had intended to station 40 percent of the operational fleet outside of the United States. However, schedule delays, cost increases, and changes to threats, missions, and requirements led DOD to reduce the number of F-22s it eventually purchased. The Air Force identified a requirement for 381 F-22s in 2002, but ended aircraft production in 2012 with approximately half of that number. As of May 2018, the Air Force had a total of 186 F-22s, as shown in figure 2. The total aircraft inventory includes primary mission aircraft in each community—those authorized to perform combat—as well as aircraft that are designated for other purposes. The operational portion of the F-22 fleet is organized into 6 operational squadrons at four locations. According to Air Force officials, the small number of F-22s provides a less than ideal fifth generation fighter capacity until F-35 numbers grow. However, in a June 2017 report to Congress, the Air Force stated that it would not make economic or operational sense to reopen the F-22 production line, and reported that it would cost approximately $50 billion to procure an additional 194 F-22s. The Air Force is continuing to fund programs to modernize the F-22 and make reliability improvements. Figure 3 shows the basing locations of the F-22 fleet, and the numbers of aircraft at each base. F-22 Unit Size and Organizational Structure Constrain Aircraft Availability and Have Not Been Reviewed by the Air Force since 2010 The size and structure of F-22 units diminishes the Air Force’s ability to maximize the number of F-22s available for operations and have not been reviewed since 2010. The F-22 has sustainment issues due to the fleet’s maintenance and supply challenges. These challenges have affected aircraft availability rates, which have remained below Air Force standards. The small size of F-22 squadrons and wings has contributed to low aircraft availability rates. Further, the Air Force practice of deploying a small portion of a squadron makes it difficult for F-22 squadrons, as currently organized, to make aircraft available for their missions at home station. The Air Force would also face difficulties generating aircraft to support DOD’s concepts for using distributed operations in high threat environments with its current F-22 squadron organization. Although in 2016 it assessed its future air superiority capability needs, the Air Force has not comprehensively assessed whether the current F-22 organizational structure is optimized to support combatant commander needs. Sustainment Issues Limit F-22 Aircraft Availability The F-22 has sustainment issues due to the fleet’s maintenance and supply challenges that have affected aircraft availability rates. In fiscal year 2016, this resulted in the fleet having an average of 80 F-22s available for operations, as shown in figure 4. According to the Air Force, from fiscal year 2012 through 2016, the F-22 fleet availability rate was below the Air Force’s annual F-22 availability standard by 4 to 19 percent. First, the F-22 has some unique maintenance challenges, which have affected aircraft availability rates. The maintenance demands of the F- 22’s Low Observable (LO) coating, a critical component of its stealth characteristic, reduces aircraft availability. Without the LO maintenance issues, availability would have been significantly closer to meeting the annual availability standard, according to the Air Force officials. Fourth generation fighters do not have to contend with this maintenance issue. The F-22’s LO coating is actually a series of coatings that require diligent and time-consuming application and curing, which results in extended periods of time when the aircraft are not available, according to Air Force officials. The F-22’s LO coating is also beginning to reach the end of its service live, requiring maintenance actions that further reduce aircraft availability. The Air Force has begun to address these maintenance issues by using a more durable coating and standing up additional repair facilities. Second, the F-22 faces a number of supply challenges that have contributed to reduced and unpredictable aircraft availability. Officials from all four operational locations expressed concerns over low supply levels and difficulties with obtaining needed parts. The F-22 fleet’s small size and resulting low demand for parts contributes to this problem. Obtaining missing parts can be a time-consuming and costly process because some original manufacturers no longer make the parts or are completely out of business, according to Air Force officials. When this is the case, the Air Force may need to find the original aircraft and parts design plans, and obtain a new contractor to produce a small number of parts. Officials at one operational location said a simple wiring harness required a 30-week lead time. Appendix I contains additional information on F-22 maintenance and supply issues. Small F-22 Squadrons and Wings Exacerbate Aircraft Availability Challenges With 18 to 21 primary mission aircraft per F-22 squadron, and 1 or 2 F-22 squadrons per wing, the Air Force has been unable to gain the maintenance and supply efficiencies associated with its larger traditional squadrons and wings, and this has contributed to low aircraft availability rates. According to service officials, the Air Force has traditionally structured its fighter wings to have 3 squadrons with 24 primary mission aircraft per squadron to optimize maintenance efficiency and combat power. The Air Force is planning to organize its active duty F-35 fleet into traditional sized squadrons with 2 or 3 squadrons per wing. A RAND study also concluded that larger squadrons and multiple squadrons per wing create efficiencies. Larger squadrons and wings create efficiencies because people, equipment, and parts can be shared, according to Air Force officials. Having a multi-squadron wing is also beneficial when one squadron deploys a portion of its aircraft, pilots, and maintenance personnel and leaves another portion of the squadron at the squadron’s home station. In these cases, collocated squadron(s) can help backfill shortfalls for the portion of the squadron that remained at home station. The Air Force recognizes that smaller F-22 operational squadrons and wings face sustainment challenges due to their size. Facing cuts in the total number of aircraft purchased, the Air Force decided in 2006 to organize its F-22s into 7 operational squadrons, each with 18 primary mission aircraft. However, in 2010, the Air Force found that this plan was unsustainable because operational squadrons were not able to produce adequate sorties. The Air Force then decided to eliminate 1 squadron and used some of the aircraft from that squadron to increase the number of primary mission aircraft to 21 in its 5 remaining active duty squadrons. The Air Force left its one F-22 National Guard squadron with only 18 primary mission aircraft. The Air Force’s intent with this restructuring was to increase fleet sustainability while retaining enough squadrons for force projection needs. F-22 aircraft availability metrics have fluctuated, but have generally been better for operational locations with more aircraft per squadron and more squadrons per wing. For example, table 2 shows that the operational locations in Alaska and Virginia—locations with 2 operational squadrons—have higher aircraft availability rates than the locations with only 1 operational squadron. Although Air Force maintenance data shows that the Florida operational squadron had a lower availability rate than the locations with 2 operational squadrons in fiscal years 2014, 2015, and 2016, Air Force officials noted that this squadron should be able to leverage the maintenance benefits of having the F-22 training squadron on base. However, a major maintenance backlog for the training squadron currently limits that benefit, according to the officials. The F-22 units in Alaska and Virginia are also generally able to produce more sorties per month. Further, F-22 squadron officials in Hawaii stated that increasing their squadron—the smallest in the fleet—by 4 additional aircraft would allow the squadron to generate 32 percent more sorties. Air Force officials cautioned that there are many factors that influence maintenance metrics for the F-22, including the age of the aircraft, climate and leadership. However, they agreed that larger squadrons and wings increase maintenance performance. Partial Unit Deployments Hinder the F-22 Squadrons’ Abilities to Maximize the Aircraft Available for Operations Further, the Air Force practice of deploying a small portion of a squadron forward makes it difficult for F-22 squadrons as currently organized to make aircraft available for their missions at home station, according to officials from all four operational locations. The Air Force organizes its F- 22 squadrons and other fighter squadrons based on a model where a squadron deploys to a single forward location, according to Air Force officials. In order to facilitate deployments, the Air Force has for approximately the last two decades organized squadrons into smaller deployable pieces called Unit Type Codes (UTCs). However, the UTCs are not the same size. For example, one of the F-22’s UTCs is designed to have only 6 of a squadron’s 21 aircraft but contains almost 50 percent of the squadron’s equipment, approximately 40 percent of the squadron’s maintenance personnel and 60 percent of its operational personnel. This organizational approach therefore creates a disproportionate split among UTCs in terms of equipment and personnel, making it more difficult for the underserved portions of the squadron to maintain readiness or generate sorties. Furthermore, different UTCs will not only have unequal amounts of equipment and personnel, but will also tend to unevenly apportion their best aircraft, more experienced personnel, and critical parts, according to Air Force officials. The officials noted that during “split operations,” the portion of the squadron remaining at home struggles to keep aircraft available for missions. According to Air Force officials, traditional fighter squadrons have larger UTCs, which provides a better balance in equipment and personnel that lessens the strain of split operations. Organizing for New Operational Concepts Poses Additional Aircraft Availability Challenges With its current F-22 squadron organization, the Air Force would also face difficulties generating aircraft to support DOD’s concepts for using distributed operations in high threat environments. According to DOD, potential adversaries are increasingly capable of challenging U.S. access to operational areas by, for example, developing cruise and ballistic missiles that are able to reach U.S. forward air bases. In its Air Superiority 2030 Flight Plan, the Air Force states that the ability to deploy and operate forces in non-permissive environments is essential to air superiority. One approach for doing this is to use distributed bases. Instead of operating from well-developed and vulnerable forward air bases, squadrons would break up into smaller units and operate independently from multiple locations, moving around so as to complicate enemy targeting. The Air Force is drafting an adaptive basing concept and implementation plan to help guide its efforts in this area. Sustaining and maintaining multiple independent deployable units so that they have operational aircraft available for the combatant commander is not possible with the current F-22 squadron structure and would require significant investment, according to Air Force officials. F-22 squadrons have made a number of short exercise deployments, with small numbers of aircraft to provide forward presence and examine the units’ abilities to conduct distributed operations. These deployments showed that rapidly deploying small numbers of F-22s for short durations is possible. The deployments also identified a number of challenges the Air Force needs to address if it implements a distributed operations concept, including maintenance, logistics, spare parts, and tanker support challenges, according to after-action reports and service officials. Furthermore, according to the Commander of U.S. Pacific Command, distributed operations requires a dynamic logistics system that is more responsive, agile, and flexible than DOD is used to employing. Air Force officials told us that the Air Force is early in the process of examining the implications of distributed operations and has not determined the extent to which F-22 squadron organization should be adjusted to support distributed operations. The Air Force Has Not Comprehensively Assessed F-22 Organizational Structure since 2010 While the Air Force reviews F-22 operations and sustainment needs as part of the annual programming and budgeting process within DOD, the Air Force has not comprehensively assessed whether the current F-22 organizational structure is the optimal structure to support combatant commander needs since 2010, according to Air Force officials. As previously discussed, the Air Force found in 2010 that operational squadrons were not able to produce adequate sorties and so eliminated 1 squadron and used some of the aircraft from that squadron to increase the number of primary mission aircraft in its 5 remaining active duty squadrons DOD’s Joint Publication 3-0, Joint Operations, states that risk management is the process to identify, assess, and control hazards arising from operational factors and make decisions that balance risk and cost with mission benefits. It assists organizations and individuals in making informed decisions to reduce or offset risk, thereby increasing operational effectiveness and the probability of mission success. Furthermore, Standards for Internal Control in the Federal Government states that management should periodically evaluate the organizational structure so that it meets the entity’s objectives and has adapted to any new objectives for the entity. Furthermore, management should periodically review policies, procedures, and related control activities for continued relevance and effectiveness in achieving the entity’s objectives or addressing related risks. However, the Air Force has not conducted a comprehensive assessment of the F-22 organizational structure since 2010, according to Air Force officials. As previously discussed, while the larger squadrons and wings created after the 2010 restructuring have generally had higher availability rates than smaller ones, fleet aircraft availability rates remain below the Air Force standard for what is needed. Further, the F-22’s role has also evolved since 2010. For example, F-22s have begun participating in combat operations in Iraq and Syria. Additionally, potential adversaries are increasingly able to challenge U.S. air superiority, according to the Air Force. In 2016, the Air Force examined its future air superiority capability needs in its Air Superiority 2030 Flight Plan, but that review did not include an assessment of organizational structure, according to officials involved with the review. Such an assessment could consider a number of alternatives, such as consolidating the F-22 fleet into larger squadrons and/or wings to improve aircraft availability or revising the design of the deployable units in squadrons to better support current deployment practices and future operational concepts, as well as any risks associated with those alternatives. Without conducting a comprehensive assessment of the F-22 organizational structure that identifies and assesses alternative approaches to organizing F-22 squadrons, the Air Force may be forgoing opportunities to improve the availability of its small yet critical F-22 fleet to support current and future combatant commander high-end air superiority needs. Air Force Utilization of F-22s Reduces Pilot High-End Air Superiority Training Opportunities The Air Force’s utilization of its F-22 fleet has limited its pilots’ opportunities to train for their high-end air superiority missions, and contributed to F-22 pilots not meeting their training requirements. F-22 pilots need extensive training in order to be prepared to execute their high-end air superiority missions. However, Air Force utilization of F-22 units for exercises, the low supply of adversary air training capabilities, and the use of F-22s to meet combatant commander needs, including the alert mission, affects the ability of pilots to meet those requirements. F-22 Pilots Are Not Meeting Minimum Training Requirements Necessary for Their Air Superiority Missions F-22 pilots are not meeting their minimum yearly training requirements for the air superiority missions, according to Air Force training reports and service officials. F-22 pilots need extensive training for F-22 units to fulfill their air superiority role. Air Force guidance notes that a key to maintaining air superiority is trained and ready Airmen that must possess a well-honed combat edge so that they are ready to prevail even against the most advanced opponents. The Air Force strategy also notes that the training of Airmen must be relevant and responsive if they are to maintain superior agility in the future. Through its Ready Aircrew Program, the Air Force establishes annual continuation training requirements for F-22 pilots. These requirements focus on the training needed to accomplish the core missions of F-22 units. They define the minimum required mix of annual sorties, simulator missions, and training events aircrews must accomplish to sustain combat mission readiness. Air Force officials emphasized that the requirements outlined in the Ready Aircrew Program are minimums and noted that some pilots may need additional sorties to achieve proficiency. In 2016, GAO reported that combat fighter squadrons were unable to meet annual training requirements across the full range of core missions. Further, an Air Force analysis conducted in 2016 determined that, based on current aircraft availability rates, pilots in an F-22 squadron with 21 primary mission aircraft need 270 days of home station training each year to meet their minimum annual continuation training requirements. However, F-22 pilots are generally not meeting those minimums, according to the officials, and F-22 operational squadrons have reported numerous shortfalls. For example, one squadron identified training shortfalls in its primary missions for four consecutive years in its annual training reports. Another squadron identified training shortfalls in one of its primary missions, offensive counter-air, in three of the last four annual training reports. F-22 Pilots Benefit from Exercises, but Exercise Quality and Quantity Can Detract from Training Needs Although participation in exercises is an important component of F-22 pilot training, multiple exercises can interrupt pilot training cycles and restrictions in some exercises can detract from F-22 pilot training for the high threat environment. Exercises provide pilots an opportunity to train in a more realistic setting. At the same time, frequent participation in exercises can take time away from the home station training that is required to maintain combat mission readiness for high-end air superiority missions. Although high demand for exercise participation is causing stress across the Air Force, the problem is particularly acute for F-22 pilots, according to a 2016 Air Force analysis. While F-22 pilots require 270 days at home each year, they are getting only 191 days on average, according to the analysis. Pilots from other fighter aircraft, such as the F- 16 and F-15E, are also experiencing home station training shortfalls, but not as great as those faced by F-22 pilots, according to the analysis. Furthermore, F-22 units are often directed to participate in exercises as part of Air Force efforts to build relationships with partners. However, due to security concerns regarding exposing the F-22’s unique capabilities, F- 22 pilots may be restricted from flying the aircraft the way they would in combat, according to Air Force officials. As a result, the value of the training is reduced and these types of exercises can result in the F-22 pilots developing bad habits that must be corrected in future training, according to Air Force officials. The Air Force recognizes that exercise demands on F-22 units and other fighter units make it difficult for pilots to complete their required training. Based on its analysis, the Air Force is planning to increase the time pilots have available to conduct home station training, including by establishing a goal of no more than 1 day on travel for every 5 days at home station. As a result, the Air Force will be reducing total exercise participation and thereby increasing the number of days F-22 pilots are at home station in fiscal year 2018 by 8 days. However, the Air Force projects that F-22 pilots in fiscal year 2018 will still fall 71 days short of the 270 days they need to meet their yearly training requirement, based on current aircraft availability rates. Without exploring further reductions in exercise events that do not contribute to high-end air superiority training, at current aircraft availability rates F-22 pilots may not be fully prepared to effectively support combatant commander needs against the most advanced threats. Adversary Air Demands for F-22s Detract from the Ability of Pilots to Meet Training Requirements F-22 pilot training requires flying against aircraft playing the role of adversaries, but high demand and low supply of adversary air resources have resulted in training shortfalls. Due to the F-22’s unique air superiority role and high-end capabilities, the Air Force expects F-22 pilots to face and defeat numerically superior adversaries. This results in an annual demand of between 145 and 171 adversary air sorties for every operational F-22 pilot. The adversary air demand for fourth generation fighters is much lower. For example, continuation training for the Air Force’s other air superiority fighter—the F-15C—results in an annual demand of between 45 and 73 adversary air sorties. To support F-22 training requirements, the Air Force has provided two of the four operational locations (Virginia and Florida) with a squadron of T-38s to provide dedicated adversary air support for use in training. In Alaska, an adversary air squadron is located at a nearby base that is able to provide some support for F-22 training, according to officials. The F-22s in Hawaii have no adversary air support on base or nearby. Figure 5 shows F-22 operational locations and their adversary air support. All F-22 operational locations report that insufficient adversary air caused pilots to have shortfalls in their training. For example, the operational F-22 squadron in Florida, which shares an adversary air squadron with a collocated F-22 training unit, reported that F-22 pilot training deficiencies in fiscal year 2016 were caused in part by limited adversary air support. Specifically, adversary air shortfalls negatively impacted the training of 83 percent of the squadron’s pilots for the offensive counter-air mission and 54 percent of the pilots for the defensive counter-air mission. Operational squadrons at other locations reported similar negative effects on training caused in part by the limited adversary air. Moreover, the limited supply of dedicated adversary aircraft means that often F-22 pilots must fly their aircraft in an adversary aircraft role to support the training of the squadron’s other F-22 pilots. For example, according to a 2017 Air Force memo, 55 percent of all sorties generated by F-22s based in Hawaii were dedicated to adversary air. The F-22 squadron in Hawaii reported that this practice negatively affected the combat readiness of all of the squadron’s pilots. The Air Force categorizes adversary air sorties as useful only for maintaining basic flying proficiency. Officials from the Virginia unit explained that F-22 pilots flying adversary air do not fly like they would during combat missions and so these sorties are wasteful, having no or negative training value. An official representing the Hawaii unit indicated that the high percentage of sorties dedicated to adversary air leads to wasteful training and declines in readiness against potential threats. Air Force officials expect competing demands for limited adversary air to grow as the Air Force stands up more F-35 squadrons. The Air Force recognizes and is attempting to mitigate adversary air shortfalls. For example, the Air Force has hired contractors to address Air Force adversary air shortfalls at exercises, as we previously reported. In addition, the Air Force has outlined a plan to provide additional adversary air support for its fighter units, including contract adversary air support for the F-22 training squadron in Florida and the operational squadrons at two of the four operational locations (Virginia and Hawaii) in the 2019 timeframe. However, the Air Force must first complete additional analysis and finalize funding before additional adversary support is provided to these locations, according to an August 2017 Air Force briefing on the plan. Until the Air Force explores additional alternatives for increasing external adversary air training support at all of the operational locations, F-22 pilots will likely continue to face training shortfalls and use limited sorties on flying adversary air themselves. Furthermore, this may result in the F-22 squadrons not being fully prepared to execute the high-end air superiority missions. Current Operations Reduce F-22 Pilot High- End Air Superiority Training Opportunities The Air Force is providing F-22s in support of current combatant commander needs, including alert missions and operational deployments, but the alert mission and these operational deployments take time away from air superiority training. Although these missions are important, they take no or limited advantage of the unique capabilities provided by the F- 22, as figure 6 illustrates. Classified details regarding the current and projected operational requirements for the F-22 are included in the classified version of this report. DOD has an established risk-informed process to distribute the service’s operational forces to the combatant commanders. Air Force officials told us that combatant commanders can request a general fighter capability or a very specific capability that only an F-22 can provide. Air Force officials also said the Air Force does not set aside F-22 units for only the most advanced threat missions, and it does not set aside any other fighter units for unique missions. The Air Force provides F-22 units to the combatant commands when those units address the combatant commander’s capability requirement and are available, according to service officials. Use of F-22s for Alert Missions Diminishes Pilots’ Ability to Train for High-End Air Superiority Missions F-22 support for ongoing air sovereignty alert missions further reduces F- 22 pilots’ abilities to train for the high-end air superiority mission. The alert mission supports homeland defense, DOD’s top priority. This mission requires certain air bases have fully fueled, fully armed jets ready at all times to respond to threats from civil or military aviation. Two F-22 operational locations have full time alert mission responsibilities (Alaska and Hawaii) and one location (Virginia) performs alert missions on an as- needed basis. According to Air Force officials, the alert mission does not require the high-end capabilities provided by the F-22 and currently F- 15C and F-16 squadrons are filling alert mission requirements in other parts of the United States. F-35s could also conduct this mission if they were assigned it, according to Air Force officials. The Air Force plans to start fielding 2 F-35 squadrons in Alaska beginning in 2020. However, there are currently no plans to use F-35s for the alert mission, according to U.S. Northern Command. With no other operational Air Force fighter squadrons currently based in Hawaii and Alaska, the alert mission falls to the F-22 units. Dedicating F-22s to the alert mission reduces the ability of F-22 pilots to train for their primary missions. Operational squadrons in Alaska and Hawaii have F-22 pilots sitting alert in order to address the 24-hour per day alert commitment. During this time they are not able to train for their high-end air superiority missions. Further, the squadrons must dedicate a number of mission-capable aircraft to this mission, which is more challenging for squadrons with a smaller number of aircraft. Squadron officials from one location estimated that they could generate hundreds of additional training sorties on an annual basis if they could use the aircraft that are currently dedicated to the alert mission. Operational Deployments Diminish F-22 Pilots’ Ability to Train for High-End Air Superiority Missions The Air Force also deploys F-22s outside of the United States to address combatant commander requirements and these deployments also reduce the time available for F-22 pilots to conduct home station training for their high-end air superiority missions. Since 2007, the Air Force has deployed F-22s to a number of combatant commands to address a variety of needs, including providing assurance to friends and allies and deterring potential adversaries. F-22s deployed to U.S. Central Command have also been supporting ongoing operations against ISIS in Iraq and Syria. F-22 pilots can gain valuable experience from deployments but their ability to train for the high-end air superiority mission can suffer. For example, F-22 involvement in current combat operations against ISIS provides pilots with experience deploying for combat, integrating with coalition forces, and conducting air-to-ground attack operations, according to Air Force officials. Although its high-end capabilities provide some benefits in current operations against ISIS, F-22s have primarily been used for close air support (CAS) missions in operations against ISIS, according to Air Force officials. However, CAS is not a primary or secondary mission for the F-22. As such, F-22 pilot air superiority skills degrade while on deployment because they are conducting CAS missions and not able to train for their air superiority missions, according to Air Force officials. Conclusions The F-22’s current availability and pilot training challenges will likely become more significant as fourth generation fighters become less survivable and the Air Force’s reliance on its small fleet of F-22s to execute the air superiority mission grows. Limitations on F-22 availability are due in part to maintenance challenges inherent to the F-22, including maintaining its LO coating. It is also due in part to Air Force decisions to organize the F-22 fleet into small wings and squadrons, resulting in lost efficiencies that come with larger organizations. Further, F-22 squadrons, designed to operate from one location, face challenges generating available aircraft when they are split, as current Air Force practices and future concepts require. The Air Force also requires well-trained pilots in order to take full advantage of the F-22’s high-end capabilities. However, F-22 pilots’ ability to train for its air superiority missions and meet associated training requirements is constrained by factors including adversary air limitations and participation in exercises with limited training value. Operational use of the F-22 for missions that have no or limited need for the F-22’s unique capabilities, such as the alert mission, further limit the ability of pilots to prepare for the high-end air superiority challenges the nation increasingly faces. One option for addressing these challenges would be to purchase more F-22s. However, the Air Force’s determination that it does not make economic or operational sense to restart F-22 production means that the Air Force has to find other ways to improve its F-22 fleet’s ability to address high-end air superiority challenges. Air Force efforts to improve F-22 capabilities and maintainability and wider efforts to address high exercise demand and adversary air shortfalls are examples of positive steps the service is taking. The Air Force has also shown a prior willingness to consolidate its F-22 fleet. Further, the Air Force outlined its commitment to addressing high-end air superiority challenges in its Air Superiority 2030 Flight Plan. This effort, along with the planned fielding of a large number of F-35s provides the Air Force with the opportunity to more comprehensively review and, if necessary, transform how it should best organize, posture, train, and utilize its fifth generation assets, including the F-22. However, unless the Air Force takes steps to assess and make necessary adjustments to the current organization and use of its F-22s, F-22 units are likely to continue to experience aircraft availability and pilot training rates that are below what they could be. As a result, the Air Force may be incurring increased risks in future operations in high threat areas. Recommendations for Executive Action We are making the following two recommendations to the Air Force: The Secretary of the Air Force should conduct a comprehensive assessment of the F-22 organizational structure that identifies and assesses alternative approaches to organizing F-22 squadrons. The assessment could at a minimum assess the following two alternatives: consolidating the fleet into larger squadrons and/or wings in order to improve aircraft availability, and revising the design of the deployable units in squadrons to better support current deployment practices and future operational concepts. (Recommendation 1) The Secretary of the Air Force should identify and assess actions to increase F-22 pilot training opportunities for the high-end air superiority missions. This effort could consider alternatives such as: reducing exercise events that do not contribute to F-22 pilot high-end air superiority training, increasing external adversary air support so all F-22 pilots can use their available limited sorties to conduct high-end air superiority training rather than having a significant portion of the F-22 pilots providing training support, and finding alternatives to using F-22 units for alert missions, and other missions that do not require the jet’s unique capabilities or prepare F-22 pilots for their primary missions. (Recommendation 2) Agency Comments and Our Evaluation We provided a draft of the classified version of the report to DOD for review and comment. That draft contained the same recommendations as this unclassified version. In written comments (reproduced in appendix II), DOD concurred with our recommendations and noted planned actions to address each recommendation. DOD also provided technical comments, which we incorporated as appropriate. We are sending copies of this report to the Senate Armed Services Committee and the House Armed Services Committee and the Secretary of Defense; the Chairman of the Joint Chiefs of Staff; and the Secretary of the Air Force, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staffs have any questions about this report, please contact me at (202) 512-3489 or pendletonj@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix III. Appendix I: F-22 Maintenance and Supply Challenges Limit Aircraft Availability Maintenance demands of the F-22’s unique Low Observable (LO) coating, along with supply challenges exacerbated by the fleet’s small size, limit the number of aircraft available for missions. In part because of these challenges, the Air Force had an average of 80 F-22s available for operations during fiscal year 2016. Maintenance of the F-22’s Unique Low Observable Coating Reduces Aircraft Availability Maintenance demands of the F-22’s unique LO coating, a critical component that gives the F-22 its stealth characteristics, reduces aircraft availability. Without the LO maintenance issues, availability would have been significantly closer to meeting the annual availability standard, according to Air Force officials. Fourth generation fighters, lacking a full LO coating, do not have to contend with this maintenance issue. The LO coating is actually a series of coatings that require diligent and time- consuming application and curing, resulting in extended periods of time during which aircraft are not available, according to Air Force officials. Further, the LO coating for each F-22 requires regular and thorough inspection to ensure that any damaged or degraded areas are identified and repaired. If damage to the LO coating exceeds a threshold, the F-22 is considered not capable of conducting its mission. An Air Force report summarizing fiscal year 2016 maintenance issues reported that LO maintenance was the primary reason F-22s were not considered mission capable due to maintenance. Maintaining the integrity of the LO coating complicates other F-22 maintenance actions because the LO coating must be removed and then restored. According to Air Force officials, removing and replacing a part on an F-22 and a fourth generation fighter, like an F-15C, could take a similar amount of time. However, the F-22 would require additional time at the beginning and end of the maintenance action to gain access to the part through the LO coating and then restore the integrity of the coating, significantly increasing the time aircraft would be unavailable due to maintenance. The Air Force is taking steps to reduce the impact of LO maintenance by, for example, creating panels that can be removed without requiring a full recoating procedure and by developing a more durable coating. Additionally, Air Force officials told us that the LO coating for its other fifth generation fighter—the F-35—uses different materials and processes and should be easier to maintain than the F-22’s LO. The F-22’s LO coating is also beginning to reach the end of its service life, requiring maintenance actions that further reduce aircraft availability. According to Air Force officials, the LO coating has an 8-to-10 year life span, but environmental factors such as high temperatures, humidity, and salinity can reduce that span by 2 to 3 years. Further, the Air Force does not house its F-22s in climate-controlled hangars at 3 of the 4 operational locations (Florida, Hawaii, and Virginia), thus exposing them to these LO- degrading environmental factors. The Air Force has taken action to address maintenance challenges by using a more durable coating and standing up additional repair facilities. The Air Force also plans to use more durable materials to make long-term corrective repairs beginning in calendar year 2019, but this will constitute a costly long-term effort, according to the Air Force. The F-22’s Small Fleet Size Exacerbates Spare Part Supply Challenges As a result of the F-22 fleet’s small size and resulting low demand for parts, the F-22 faces a number of supply challenges that have contributed to reduced and unpredictable aircraft availability. Officials from all four operational locations identified low supply levels and difficulty obtaining needed parts as a concern. Obtaining parts can be a time-consuming and costly process because some original manufacturers no longer make the parts or are completely out of business, according to Air Force officials. When this is the case, the Air Force may need to find the original aircraft and parts design plans, and obtain a new contractor to produce a small number of parts. Officials at one operational location said a simple wiring harness required a 30-week lead time. Air Force maintenance statistics for fiscal year 2016 show that 14 percent of the F-22 fleet was not mission capable, and therefore not available, due to supply issues. According to Air Force officials, the F-22 fleet’s small size and resulting low demand for parts makes this problem more acute. F-22 squadrons face an unenviable choice when necessary parts are not available, according to Air Force officials: they can make the aircraft unavailable until the spare part arrives and can be installed or they can take the part from another aircraft that may be broken for a different reason. The second option, called cannibalization, is an inefficient way to conduct maintenance because it doubles the work. A good part needs to be removed from one aircraft and put into another. Once the replacement part arrives, it needs to be installed on the cannibalized aircraft. There is also a chance that the cannibalized part could get damaged in the process or just not work. Further, cannibalization could result in additional LO repairs on the donor aircraft. An Air Force report summarizing fiscal year 2016 maintenance issues reported that F-22 cannibalization rates have grown by 6 percent between fiscal years 2012 and 2016. The F-22’s small fleet size also exacerbates supply challenges it is facing with its engines, potentially falling below minimum spare part requirements for multiple calendar years. Further, an increase in flying hours in 2014 resulted in engines requiring overhauls sooner than previously anticipated. It is taking time for the engine maintenance contractor to build up enough capacity to deal with this increased demand. The officials said that the F-22’s small fleet size contributed to this problem because, as was the case with other parts issues, low early demand meant that many of the vendors that built parts for those engines no longer build the parts or are not in business. Additionally, it takes time to find vendors and skilled people to build those parts again. The Air Force is implementing a mitigation plan that includes increasing production, overflying the standard engine maintenance interval, and borrowing engines from aircraft in long-term maintenance. According to Air Force officials, this kind of engine issue is not unique to the F-22. They noted that there was a time when B-1s, another small fleet, had a major engine shortfall that resulted in aircraft parked without engines in them. An Air Force forecast shows that mitigation efforts will avoid that problem, barring unanticipated increases in demand or maintenance problems. Appendix II: Comments from the Department of Defense Appendix III: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact name above, Michael Ferren, Assistant Director; Vincent Buquicchio; Nicolaas Cornelisse, Analyst-in-Charge; Patricia Donahue; Amie Lesser; Tamiya Lunsford; Matthew Jacobs; Travis Masters; Richard Powelson; Walter Vance; and Nicole Volchko made key contributions to this report.

Why GAO Did This Study The F-22 was designed and fielded as the Air Force's premier air-to-air fighter. The small fleet of 186 F-22s is central to the Air Force's ability to accomplish its air superiority mission in high threat areas. While the Air Force has focused on other missions over the last 15 years of conflict, it is now trying to refocus on overcoming advanced threats, even as it continues to support ongoing operations. Though the recent introduction of the F-35 gives the Air Force another advanced fighter, the F-35 is primarily designed for the air-to-ground missions and so is intended to complement but not replace the F-22. Senate Report 114-255 included a provision for GAO to review a variety of issues related to Air Force F-22 fighter squadrons. This report examines the extent to which the Air Force's (1) organization of its F-22 fleet maximizes availability of aircraft and (2) utilization of its F-22 fleet affects pilot air superiority training. GAO reviewed Department of Defense (DOD) guidance, analyzed maintenance data and training information for the F-22, evaluated the use of F-22s during deployments, and interviewed agency officials. This is a public version of a classified report issued in April 2018. Information DOD deemed classified or sensitive has been omitted. What GAO Found The Air Force's organization of its small F-22 fleet has not maximized the availability of these 186 aircraft. Availability is constrained by maintenance challenges and unit organization. For example, stealth is a central feature of the F-22 and, according to Air Force officials, maintaining the stealth coating on the outside of the aircraft is time consuming and significantly reduces the time F-22s are available for missions. Maintenance availability challenges are exacerbated by the Air Force's decision to organize the F-22 fleet into small units—18 or 21 primary mission aircraft per squadron and one or two squadrons per wing. Traditional fighter wings have three squadrons per wing with 24 aircraft in each squadron, which creates maintenance efficiencies because people, equipment, and parts can be shared, according to Air Force officials. Moreover, the Air Force organized F-22 squadrons to operate from a single location. However, it generally deploys only a part of a squadron, and the remaining part struggles to keep aircraft available for missions at home. Larger, traditional Air Force squadrons and deployable units provide a better balance of equipment and personnel, according to service officials. The Air Force has not reassessed the structure of its F-22 fleet since 2010. Without conducting a comprehensive assessment to identify and assess F-22 organization, the Air Force may be foregoing opportunities to improve the availability of its small yet critical F-22 fleet, and support combatant commander air superiority needs in high threat environments. The Air Force's utilization of its F-22 fleet has limited pilot opportunities to train for air superiority missions in high threat environments. To complete the annual training requirements for air superiority missions, F-22 pilots must train almost the entire year. However, F-22 pilots are not meeting their minimum yearly training requirements for the air superiority missions, according to Air Force training reports and service officials. Moreover, the utilization of F-22s for exercises and operational missions that do not require the F-22's unique capabilities interrupt pilot training and lead to reduced proficiency. For example, F-22 units are often directed to participate in partnership building exercises. However, during these exercises, F-22 pilots may be restricted from flying the F-22 the way they would fly it in combat—due to security concerns about exposing the F-22's unique capabilities. These restrictions limit the value of the exercises and can result in pilots developing bad habits, according to Air Force officials. The Air Force also uses F-22s to support alert missions—a mission that requires certain bases to have jets ready at all times to respond to threats from civil or military aviation. The alert mission does not require the advanced capabilities of the F-22, but there are no other operational Air Force fighter squadrons currently based at the F-22 locations in Alaska and Hawaii, so the alert mission falls to the F-22 units. Pilots and aircraft assigned to the alert mission cannot be used for any other purposes, including training. This limits opportunities for pilots to enhance air superiority skills. Without examining and implementing options to improve F-22 pilot training opportunities, the Air Force may be foregoing opportunities to improve its capability to address the high-end air superiority challenges it expects to face. What GAO Recommends GAO recommends that the Air Force reassess its F-22 organizational structure to determine alternative approaches to organizing F-22 squadrons, and identify ways to increase F-22 pilot training opportunities for high-end air superiority missions. DOD concurred with GAO's recommendations.

gao_GAO-18-457

gao_GAO-18-457_0

Background History of Conflict in the DRC and the Region The DRC is a vast, mineral-rich nation with an estimated population of about 83 million people and an area that is roughly one-quarter the size of the United States, according to the United Nations. Figure 1 shows the DRC’s provinces and adjoining countries. Since gaining its independence from Belgium in 1960, the DRC has undergone political upheaval and armed conflict. From 1998 to 2003, the DRC and eight other African countries were involved in what has become known as “Africa’s World War,” which resulted in a death toll of an estimated 5 million people in the DRC, according to the U.S. Department of State (State). The eastern DRC has continued to be plagued by violence, often perpetrated against civilians by illegal armed groups and some members of the Congolese national military. Notably, in 2012, an illegal armed group occupied the city of Goma and other cities in the eastern DRC and clashed with the Congolese national army. During this time, the United Nations reported numerous cases of sexual violence against civilians, including women and children, which were perpetrated by armed groups and some members of the Congolese national military. In 2017, the United Nations reported that serious violations of human rights remain widespread in the DRC, including continued acts of sexual violence by government security forces as well as nonstate armed groups. Uses of Conflict Minerals Various industries, particularly manufacturing industries, use the four conflict minerals specifically named in the Dodd-Frank Act—tin, tungsten, tantalum, and gold—in a wide variety of products. For example, tin is used to solder metal pieces and is also found in food packaging, steel coatings on automobile parts, and some plastics. Tungsten is used in automobile manufacturing, drill bits and cutting tools, and other industrial manufacturing tools and is the primary component of filaments in incandescent light bulbs. Most tantalum is used to manufacture capacitors that enable energy storage in electronic products such as cell phones and computers or to produce alloy additives used in turbines in jet engines. Gold is held as bullion by central bank reserves and used in making jewelry and also in the electronics industry, for example, to manufacture cell phones and laptops. SEC Conflict Minerals Disclosure Rule In August 2012, SEC adopted its conflict minerals disclosure rule in response to Section 1502(b) of the Dodd-Frank Act. The act required that SEC promulgate disclosure and reporting regulations regarding the use of conflict minerals originating from the DRC and covered countries. In the summary section of the adopting release for the rule, SEC noted that to accomplish the goal of helping to end the human rights abuses in the DRC caused by the conflict, Congress chose to use the Dodd-Frank Act’s disclosure requirements to bring greater public awareness of the sources of companies’ conflict minerals and to promote the exercise of due diligence on conflict mineral supply chains. The SEC disclosure rule addresses the four conflict minerals named in the Dodd-Frank Act from the DRC and covered countries. The rule outlines a process for companies to follow, as applicable, to comply with the rule (see app. II). The process broadly requires a company to 1. determine whether it manufactures, or contracts to be manufactured, products with “necessary” conflict minerals; 2. conduct a reasonable country-of-origin inquiry (RCOI) concerning the origin of those conflict minerals; and 3. exercise due diligence, if appropriate, to determine the source and chain of custody of those conflict minerals, adhering to a nationally or internationally recognized due diligence framework, if such a framework is available for these necessary conflict minerals. If companies choose to disclose that their products are “DRC conflict free,” the SEC disclosure rule requires companies to obtain an independent private-sector audit (IPSA). Conflict Minerals Disclosures Filed in 2017 Were Similar to Those Filed in Prior Years, and SEC Updated Guidance on Enforcement of Due Diligence Requirements Our review of companies’ conflict minerals disclosures filed with the SEC in 2017 found that, in general, they were similar to disclosures filed in the prior 2 years. In 2017, a similar number of companies filed conflict minerals disclosures as in 2015 and 2016. Based on our review of a generalizable sample, we found that almost all companies that filed conflict minerals disclosures in 2017 reported performing inquiries about their conflict minerals’ country of origin, similar to the results we previously reported for 2016 and 2015. In their 2017 disclosure reports, many companies described actions they took to improve data collection processes, and most companies indicated challenges in determining the country of origin. Our review of company filings found that almost all companies required to conduct due diligence, as a result of their country- of-origin inquires, reported performing it. SEC issued revised guidance in April 2017, indicating that the SEC’s Division of Corporation Finance would not recommend enforcement action if companies did not report on specified due diligence disclosure requirements. About as Many Companies Filed Conflict Minerals Disclosures in 2017 as in Each of the Prior 2 Years In 2017, 1,165 companies filed conflict minerals disclosures—almost as many companies as filed in 2016 and 2015 (1,230 and 1,281 respectively). Our analysis of a generalizable sample of filings found that an estimated 90 percent of the companies that filed in 2017 were domestic and an estimated 10 percent were foreign companies, similar to the domestic-to-foreign ratio we found in 2016 and 2015. While not all companies reported the minerals used, of those that disclosed this information, an estimated 69 percent reported using tin, 54 percent reported using tantalum, 59 percent reported using tungsten, and 63 percent reported using gold, figures that are similar to the percentages reported in 2016. Percentages of Companies Reporting Minerals’ Country-of-Origin in 2017 Were Similar to Prior 2 Years Our analysis of a generalizable sample of 2017 filings found that, as in 2016 and 2015, almost all companies that filed conflict minerals disclosures indicated that they performed country-of-origin inquiries. Specifically, an estimated 100 percent of the companies reported that they performed such an inquiry, similar to the percentages that we estimated reported doing so in 2016 and 2015. As a result of the inquiries they conducted, an estimated 53 percent of companies reported in 2017 whether the conflict minerals in their products came from covered countries—similar to the estimate of 49 percent in 2016 and in 2015 but significantly higher than the estimate of 30 percent in 2014 (see fig. 2). Some Companies Reported Improvements in Supply Chain Data Collection Efforts, but Challenges Remain In the filings we reviewed, many companies indicated they had taken actions to improve their data collection processes, such as gathering missing information about their supply chains and working with suppliers to encourage conflict-free sourcing. In interviews, representatives of selected companies that filed conflict minerals disclosures in 2017 and other industry participants noted that (1) awareness among suppliers about the use of conflict minerals had continued to increase and (2) the process for collecting data on supply chains had become more routine and standardized. However, as in prior years, our review of filings found that most companies reported challenges in determining the country of origin of conflict minerals, in part due to lack of access to suppliers and complex supply chains involving many suppliers and processing facilities. Almost All Companies Required to Conduct Due Diligence Reported Performing It Our review of company filings found that almost all companies that were required, as a result of their country-of-origin inquires, to conduct due diligence on the source and chain of custody of the conflict minerals in their products reported doing so. In 2017, an estimated 96 percent reported conducting due diligence, compared with 96 and 97 percent in 2016 and 2015 respectively. An estimated 87 percent of companies in 2017 reported using a due diligence framework prescribed by the Organization for Economic Co-operation and Development (OECD) guidance for conducting due diligence on the source and chain of custody of the conflict minerals in their products. That result is comparable to an estimated 92 percent that we reported in 2016 and an estimated 95 percent that we reported in 2015. The remaining 13 percent of the companies that reported conducting due diligence in 2017 did not specify a framework for their due diligence activities. After conducting due diligence, an estimated 37 percent of the companies reported in 2017 that they were able to determine that their conflict minerals came from covered countries or from scrap or recycled sources, compared with an estimated 39 and 23 percent in 2016 and 2015, respectively. An estimated 47 percent of the companies in 2017 reported that they could not definitively confirm the source of the conflict minerals in their products, compared with an estimated 55 and 67 percent in 2016 and 2015, respectively. However, as in previous years, almost all of the companies that reported conducting due diligence in 2017 reported that they could not determine whether the conflict minerals financed or benefited armed groups. Four companies in our sample reported determining that the minerals in their products did not finance or benefit armed groups in covered countries, and declared some products “DRC- conflict free.” Three of these companies included the required independent private-sector audit (IPSA) report, and one company did not include an IPSA report. Overall, a total of 16 companies filed an IPSA report in 2017, compared with 19 in 2016. SEC’s Division of Corporation Finance Issued Updated Guidance Indicating It Would Not Recommend Enforcement Action on Due Diligence Disclosure Requirements In April 2017, the SEC’s Division of Corporation Finance issued revised guidance indicating that it would not recommend enforcement action to the Commission if companies did not report on specified due diligence disclosure requirements. The SEC disclosure rule requires companies, if applicable, to report on their due diligence in a conflict minerals report (see app. II for additional detail on the disclosure requirements). SEC’s Division of Corporation Finance staff told us that they received inquiries from a small number of companies about the filing process for 2017. In response to these inquiries, these staff noted that they advised companies that the companies had the flexibility to determine whether or not to report on their due diligence and to report their country-of-origin inquiry findings in either the Form SD or in a conflict minerals report. However, the Division of Corporation Finance staff also told us that, regardless of the division’s revised guidance, the SEC could still initiate enforcement action if companies do not report on their due diligence, as required by the SEC disclosure rule. In our sample, three companies cited the updated guidance and other statements issued by the SEC in their filings as a rationale for not reporting on due diligence activities. In interviews, representatives of some companies and other industry participants told us that even though the revised guidance and other statements made by the SEC raised some uncertainty about the filing process, generally, companies plan to continue to report similar conflict minerals disclosure information. New Survey Data Are Available on the Rate of Sexual Violence in Burundi and Uganda We identified two new population-based surveys since our last report related to sexual violence in Burundi and Uganda published in 2018; the most recent information for eastern DRC and Rwanda is from 2016. We also identified some new case-file data on sexual violence in the DRC; however, as we reported previously, case-file data on sexual violence are not suitable for estimating an overall rate of sexual violence. Results of Two New Demographic and Health Surveys Conducted in Uganda and Burundi Were Published in 2018 We identified two new population-based surveys related to sexual violence that were conducted in Uganda and Burundi in 2016 and 2017, respectively, and whose results were published in 2018. The Uganda Demographic and Health Survey was conducted from June to December 2016 by the Uganda Bureau of Statistics with technical assistance from ICF International. The survey estimated that 12.7 percent of women nationwide, ages 15-49, reported they had experienced sexual violence in the 12-month period preceding the survey, while 21.9 percent reported they had experienced sexual violence at some point in their lifetime. In addition, 4 percent of men nationwide, ages 15-49, reported they had experienced sexual violence in the 12-month period preceding the survey, while 8.3 percent reported they had experienced sexual violence at some point in their lifetime. The Burundi Demographic and Health Survey was conducted from October of 2016 to March of 2017 by the Burundi Institute of Statistics and Economic Studies with technical assistance from ICF International. The survey estimated that 12.7 percent of women nationwide, ages 15- 49, reported they had experienced sexual violence in the 12-month period preceding the survey, while 23.1 percent reported they had experienced sexual violence at some point in their lifetime. In addition, 1.9 percent of men nationwide, ages 15-49, reported they had experienced sexual violence in the 12-month period preceding the survey, while 6.1 percent reported they had experienced sexual violence at some point in their lifetime. The most recent information on the rate of sexual violence for eastern DRC and Rwanda is from 2016 and is discussed in our previous reports. Figure 3 shows the publication dates for the population-based surveys with data on rates of sexual violence in the eastern DRC, Rwanda, Uganda, and Burundi that have been published since 2007. Some Additional Case-File Information about Sexual Violence Has Become Available State and United Nations entities have provided additional case-file information about instances of sexual violence in the DRC and adjoining countries. State’s annual country reports on human rights practices provided the following case-file data pertaining to sexual violence in the DRC and Burundi: DRC. In 2017, the UN documented 267 adult victims and 171 child victims, including two boys, of sexual violence in conflict. This violence was perpetrated by illegal armed groups as well as state security forces and civilians and was concentrated in North Kivu Province and in the Kasai region, according to State. Burundi. One government organization—Humara Center— responsible for investigating cases of sexual violence and rape received 197 cases of sexual and gender-based violence through early December 2017, according to State. Observers stated many women were reluctant to report rape, in part due to fear of reprisal, according to State. In addition, UN entities reported the following case-file data about sexual violence in the DRC: DRC. In 2017, the United Nations Organization Stabilization Mission in the DRC verified 195 cases of conflict-related sexual violence, with illegal armed groups responsible for 80 percent of the cases and DRC security forces responsible for the remaining 20 percent. United Nations officials we interviewed raised concerns about a resurgence of sexual violence in certain regions in the DRC due to a variety of factors, including political instability arising from the government’s postponement of the presidential election originally scheduled to take place in November 2016. Agency Comments We provided a draft of this report to the SEC, State, and the U.S. Agency for International Development for comment. SEC provided technical comments, which we incorporated as appropriate. State and USAID did not provide comments. We are sending copies of this report to appropriate congressional committees and to the Chairman of the Securities and Exchange Commission, the Secretary of State, and the Administrator of the U.S. Agency for International Development. The report is also available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-8612 or gianopoulosk@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix III. Appendix I: Objectives, Scope, and Methodology In this report, we provide information about (1) companies’ disclosures filed with the U.S. Securities and Exchange Commission (SEC) in 2017 compared with disclosures filed in the prior 2 years and (2) the rate of sexual violence in the eastern Democratic Republic of the Congo and neighboring countries published in 2017 and early 2018. To examine the fourth annual company disclosures filed with the SEC in 2017 in response to the SEC disclosure rule, we downloaded the specialized disclosure reports (Form SD) and conflict minerals reports (CMR) from SEC’s publically available Electronic Data Gathering, Analysis, and Retrieval (EDGAR) database in September 2017. We downloaded 1,165 filings identified as Form SDs and the CMRs included in EDGAR. To review the completeness and accuracy of the EDGAR database, we reviewed relevant documentation, interviewed knowledgeable SEC officials, and reviewed prior GAO reports on internal controls related to SEC’s financial systems. We determined that the EDGAR database was sufficiently reliable for identifying the universe of SD filings. We reviewed the conflict minerals section of the Dodd-Frank Wall Street Reform and Consumer Protection Act (Dodd-Frank Act) and the requirements of the SEC disclosure rule to develop a data collection instrument (DCI) that guided our analysis of Form SDs and CMRs that contain the information disclosed by the filing companies. Our DCI was not a compliance review of the Form SDs and CMRs. The questions were written in both yes-no and multiple-choice formats. An analyst reviewed the Form SDs and CMRs and recorded responses to the DCI for all of the companies in the sample. A second analyst also reviewed the Form SDs and CMRs and verified the responses recorded by the first analyst. Analysts met to discuss and resolve any discrepancies. We randomly sampled 100 Form SDs from a population of 1,165 to create estimates generalizable to the population of all companies that filed. All estimates based on our sample have a margin of error of plus or minus 10 percentage points or less at the 95-percent confidence level. Because we followed a probability procedure based on random selections, our sample is only one of a large number of samples that we might have drawn. Since each sample could have provided different estimates, we express our confidence in the precision of our particular sample’s results as a 95-percent confidence interval. This is the interval that would contain the actual population value for 95 percent of the samples we could have drawn. We also attended an industry conference on conflict minerals and spoke with company representatives to provide additional perspective. To address our second objective, we identified and assessed any information on sexual violence in the eastern DRC and the three adjoining countries—Rwanda, Uganda, and Burundi—that had been published or otherwise had become available in 2017 and early 2018. We discussed the collection of sexual violence–related data in the DRC and adjoining countries, including population-based survey data and case-file data, during interviews with U.S. Department of State and U.S. Agency for International Development officials and with representatives of nongovernmental organizations and researchers whom we interviewed for our prior review of sexual violence rates in the eastern DRC and adjoining countries. We also interviewed officials from the United Nations Population Fund and the United Nations Special Representative of the Secretary-General in New York on Sexual Violence in Conflict. In addition, we conducted Internet searches to identify new academic articles containing any additional information on sexual violence published in 2017 and early 2018. We conducted this performance audit from September 2017 to June 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: U.S. Securities and Exchange Commission (SEC) Summary of the Conflict Minerals Disclosure Rule The U.S. Securities and Exchange Commission (SEC) conflict minerals disclosure rule requires certain companies to file a specialized disclosure report, known as the Form SD, if the company manufactures, or contracts to have manufactured, a product or products containing conflict minerals that are necessary to the functionality or the production of those products. The rule also requires each company, as applicable, to provide a description of the measures the company took to exercise due diligence in determining the source and chain of custody of the conflict minerals, the facilities used to process them, their country of origin, and the efforts made to determine the mine or location of origin with the greatest possible specificity. Form SD provides general instructions for filing conflict minerals disclosures and specifies the information that companies must provide. Companies were required to file under the rule for the first time by June 2, 2014, and annually thereafter on May 31. Figure 4 shows the SEC’s flowchart summarizing the conflict minerals disclosure rule. Appendix III: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the individual named above, Godwin Agbara (Assistant Director), Farahnaaz Khakoo-Mausel (Analyst-in-Charge), Diana Blumenfeld, Andrew Kurtzman, Justin Fisher, Grace Lui, David Dayton, Christopher Keblitis, and Michael McAtee made key contributions to this report. Related GAO Products Conflict Minerals: Information on Artisanal Mined Gold and Efforts to Encourage Responsible Sourcing in the Democratic Republic of the Congo. GAO-17-733. Washington, D.C.: August 23, 2017. SEC Conflict Minerals Rule: 2017 Review of Company Disclosures in Response to the U.S. Securities and Exchange Commission Rule. GAO-17-517R. Washington, D.C.: April 26, 2017. Conflict Minerals: Insights from Company Disclosures and Agency Actions. GAO-17-544T. Washington, D.C.: April 5, 2017. SEC Conflict Minerals Rule: Companies Face Continuing Challenges in Determining Whether Their Conflict Minerals Benefit Armed Groups. GAO-16-805. Washington, D.C.: August 25, 2016. SEC Conflict Minerals Rule: Insights from Companies’ Initial Disclosures and State and USAID Actions in the Democratic Republic of the Congo Region. GAO-16-200T. Washington, D.C.: November 17, 2015. SEC Conflict Minerals Rule: Initial Disclosures Indicate Most Companies Were Unable to Determine the Source of Their Conflict Minerals. GAO-15-561. Washington, D.C.: August 18, 2015. Conflict Minerals: Stakeholder Options for Responsible Sourcing Are Expanding, but More Information on Smelters Is Needed. GAO-14-575. Washington, D.C.: June 26, 2014. SEC Conflict Minerals Rule: Information on Responsible Sourcing and Companies Affected. GAO-13-689. Washington D.C.: July 18, 2013. Conflict Minerals Disclosure Rule: SEC’s Actions and Stakeholder- Developed Initiatives. GAO-12-763. Washington, D.C.: July 16, 2012. The Democratic Republic of Congo: Information on the Rate of Sexual Violence in War-Torn Eastern DRC and Adjoining Countries. GAO-11-702. Washington, D.C.: July 13, 2011. The Democratic Republic of the Congo: U.S. Agencies Should Take Further Actions to Contribute to the Effective Regulation and Control of the Minerals Trade in Eastern Democratic Republic of the Congo. GAO-10-1030. Washington, D.C.: September 30, 2010.

Why GAO Did This Study Over the past decade, the United States and the international community have sought to improve security in the DRC. In the eastern DRC, armed groups have committed severe human rights abuses, including sexual violence, and reportedly profit from the exploitation of “conflict minerals”— in particular, tin, tungsten, tantalum, and gold, according to the United Nations. Congress included a provision in the 2010 Dodd-Frank Wall Street Reform and Consumer Protection Act that, among other things, required the SEC to promulgate regulations regarding the use of conflict minerals from the DRC and adjoining countries. The SEC adopted these regulations in 2012. The act also included a provision for GAO to annually assess the SEC regulations' effectiveness in promoting peace and security and report on the rate of sexual violence in the DRC and adjoining countries. In this report, GAO provides information about (1) companies' conflict minerals disclosures filed with the SEC in 2017 compared with disclosures filed in the prior 2 years and (2) the rate of sexual violence in the eastern DRC and adjoining countries published in 2017 and early 2018. GAO analyzed a generalizable random sample of SEC filings and interviewed relevant officials. GAO reviewed U.S., United Nations, and international organizations' reports; interviewed DRC officials, and other stakeholders; and conducted fieldwork in New York at the United Nations headquarters. GAO is not making any recommendations. What GAO Found GAO's review of companies' conflict minerals disclosures filed with the U.S. Securities and Exchange Commission (SEC) in 2017 found that, in general, they were similar to disclosures filed in the prior 2 years. In 2017, 1,165 companies filed conflict minerals disclosures—about the same as in 2016 and 2015. Percentages of companies reporting country-of-origin inquiries in 2017 were also similar to the percentages from those 2 prior years. As a result of the inquiries they conducted, an estimated 53 percent of companies reported in 2017 whether the conflict minerals in their products came from the Democratic Republic of the Congo (DRC) and adjoining countries—similar to the estimated 49 percent in 2016 and 2015 but significantly higher than the estimate of 30 percent in 2014 (see figure). In their 2017 disclosure reports, many companies described actions they took to improve data collection processes, and most companies indicated some challenges in determining the country of origin. Similar to the prior 2 years, almost all companies required to conduct due diligence, as a result of their country-of-origin inquiries, reported doing so. After conducting due diligence to determine the source and chain of custody of any conflict minerals used, an estimated 37 percent of these companies reported in 2017 that they were able to determine that their conflict minerals came from covered countries or from scrap or recycled sources, compared with 39 and 23 percent in 2016 and 2015, respectively. Four companies in GAO's sample declared their products “DRC conflict-free,” and of those, three included the required Independent Private Sector Audit report (IPSA), and one did not. In 2017, 16 companies filed an IPSA; 19 did so in 2016. GAO found information on the rate of sexual violence in the 2017 Uganda and Burundi Demographic and Health Surveys. For Uganda, 22 percent of women and 9 percent of men reported they had experienced sexual violence at least once in their lifetime. For Burundi, 23 percent of women and 6 percent of men reported they had experienced sexual violence at least once in their lifetime. The most recent information on the rate of sexual violence for eastern DRC and Rwanda is from 2016 and is discussed in our previous GAO reports.

gao_GAO-18-696T

gao_GAO-18-696T_0

Background Historically, patient health information has been scattered across paper records kept by many different caregivers in many different locations, making it difficult for a clinician to access all of a patient’s health information at the time of care. Lacking access to these critical data, a clinician may be challenged in making the most informed decisions on treatment options, potentially putting the patient’s health at risk. The use of technology to electronically collect, store, retrieve, and transfer clinical, administrative, and financial health information has the potential to improve the quality and efficiency of health care. Electronic health records are particularly crucial for optimizing the health care provided to military personnel and veterans. While in active military status and later as veterans, many DOD and VA personnel, along with their family members, tend to be highly mobile and may have health records residing at multiple medical facilities within and outside the United States. VA and DOD operate separate electronic health record systems that they rely on to create and manage patient health information. In particular, VA currently uses its integrated medical information system—VistA—which was developed in-house by the department’s clinicians and IT personnel and has been in operation since the early 1980s. Over the last several decades, VistA has evolved into a technically complex system comprised of about 170 modules that support health care delivery at 170 VA Medical Centers and over 1,200 outpatient sites. In addition, customization of VistA, such as changes to the modules by the various medical facilities, has resulted in about 130 versions of the system—referred to as instances. For its part, DOD relies on its Armed Forces Health Longitudinal Technology Application (AHLTA), which comprises multiple legacy medical information systems that were developed from commercial software products and customized for specific uses. For example, the Composite Health Care System (CHCS), which was formerly DOD’s primary health information system, is used to capture information related to pharmacy, radiology, and laboratory order management. In addition, the department uses Essentris (also called the Clinical Information System), a commercial health information system customized to support inpatient treatment at military medical facilities. In July 2015, DOD awarded a contract for a new commercial electronic health record system to be developed by the Cerner Corporation. Known as MHS GENESIS, this system is intended to replace DOD’s existing AHLTA system. The transition to MHS GENESIS began in February 2017 and implementation is expected to be complete throughout the department in 2022. Interoperability: An Overview The sharing of health information among organizations is especially important because the health care system is highly fragmented, with care and services provided in multiple settings, such as physician offices and hospitals, that may not be able to coordinate patient medical care records. Thus, a means for sharing information among providers, such as between DOD’s and VA’s health care systems, is by achieving interoperability. The Office of the National Coordinator for Health IT, within the Department of Health and Human Services, has issued guidance, describing interoperability as: the ability of systems to exchange electronic health information and the ability to use the electronic health information that has been exchanged from other systems without special effort on the part of the user. Similarly, the National Defense Authorization Act for Fiscal Year 2014 defines interoperability, per its use in the provision governing VA’s and DOD’s electronic health records, as “the ability of different electronic health records systems or software to meaningfully exchange information in real time and provide useful results to one or more systems.” Thus, in these contexts, interoperability allows patients’ electronic health information to be available from provider to provider, regardless of where the information originated. Achieving interoperability depends on, among other things, the use of agreed-upon health data standards to ensure that information can be shared and used. If electronic health records conform to interoperability standards, they potentially can be created, managed, and consulted by authorized clinicians and staff across more than one health care organization, thus providing patients and their caregivers the information needed for optimal care. Information that is electronically exchanged from one provider to another must adhere to the same standards in order to be interpreted and used in electronic health records, thereby permitting interoperability. In the health IT field, standards may govern areas ranging from technical issues, such as file types and interchange systems, to content issues, such as medical terminology. On a national level, the Office of the National Coordinator has been assigned responsibility for identifying health data standards and technical specifications for electronic health record technology and overseeing the certification of this technology. In addition to exchanging the information, systems must be able to use the information that is exchanged. Thus, if used in a way that improves providers’ and patients’ access to critical information, electronic health record technology has the potential to improve the quality of care that patients receive and to reduce health care costs. For example, with interoperability, medical providers have the ability to query data from other sources while managing chronically ill patients, regardless of geography or the network on which the data reside. VA and DOD Have a Long History of Efforts to Achieve Electronic Health Record Interoperability Since 1998, DOD and VA have relied on a patchwork of initiatives involving their health information systems to exchange information and increase electronic health record interoperability. These have included initiatives to share viewable data in existing (legacy) systems; link and share computable data between the departments’ updated health data repositories; develop a virtual lifetime electronic health record to enable private sector interoperability; implement IT capabilities for the first joint federal health care center; and jointly develop a single integrated system. Table 1 provides a brief description of the history of these various initiatives. In addition to the initiatives mentioned in table 1, DOD and VA previously responded to provisions in the National Defense Authorization Act for Fiscal Year 2008 directing the departments to jointly develop and implement fully interoperable electronic health record systems or capabilities in 2009. The act also called for the departments to set up the Interagency Program Office to be a single point of accountability for their efforts to implement these systems or capabilities by the September 30, 2009, deadline. The Interagency Program Office Has Not Functioned as the Single Point of Accountability for VA and DOD’s Efforts to Increase Electronic Health Record Interoperability The Interagency Program Office has been involved in the various approaches taken by VA and DOD to increase health information interoperability and modernize their respective electronic health record systems. These approaches have included development of the Virtual Lifetime Electronic Record (VLER) and a new, common integrated electronic health record (iEHR) system. However, although the Interagency Program Office has led efforts to identify data standards that are critical to interoperability between systems, the office has not been effectively positioned to be the single point of accountability as called for in the National Defense Authorization Act for Fiscal Year 2008. Moreover, the future role of the office with respect to VA’s current electronic health record modernization program is uncertain. The Interagency Program Office Became Operational, but Was Not Positioned to Be the Single Point of Accountability for Achieving Interoperability Although VA and DOD took steps to set up the Interagency Program Office, the office was not positioned to be the single point of accountability for the departments’ efforts to achieve electronic health record interoperability by September 30, 2009. When we first reported in July 2008 on its establishment, VA and DOD’s efforts to set up the office were still in their early stages. Leadership positions in the office were not yet permanently filled, staffing was not complete, and facilities to house the office had not been designated. Further, the implementation plan for setting up the office was in draft and, although the plan included schedules and milestones, the dates for several activities (such as implementing a capability to share immunization records) had not yet been determined, even though all capabilities were to be achieved by September 2009. We concluded that without a fully established program office and a finalized implementation plan with set milestones, the departments could be challenged in meeting the required date for achieving interoperability. Accordingly, we recommended that the departments give priority to fully establishing the office by putting in place permanent leadership and staff, as well as finalizing the draft implementation plan. Both departments agreed with this recommendation. We later reported in January 2009 that VA and DOD had continued to take steps to set up the Interagency Program Office. For example, the departments had developed descriptions for key positions within the office. In addition, the departments had developed a document that depicted the Interagency Program Office’s organizational structure; they also had approved a program office charter to describe, among other things, the mission and functions of the office. However, we pointed out that VA and DOD had not yet fully executed their plan to set up the office. For example, among other activities, they had not filled key positions for the Director and Deputy Director, or for 22 of 30 other positions identified for the office. Our report stressed that, in the continued absence of a fully established Interagency Program Office, the departments would remain ineffectively positioned to assure that interoperable electronic health records and capabilities would be achieved by the required date. Thus, we recommended that the departments develop results-oriented performance goals and measures to be used as the basis for reporting interoperability progress. VA and DOD agreed with our recommendation. Nevertheless, in a subsequent July 2009 report, we noted that the Interagency Program Office was not effectively positioned to function as a single point of accountability for the implementation of fully interoperable electronic health record systems or capabilities between VA and DOD. While the departments had made progress in setting up the office by hiring additional staff, they continued to fill key leadership positions on an interim basis. Further, while the office had begun to demonstrate responsibilities outlined in its charter, it was not yet fulfilling key IT management responsibilities in the areas of performance measurement (as we previously recommended), project planning, and scheduling, which were essential to establishing the office as a single point of accountability for the departments’ interoperability efforts. Thus, we recommended that the departments improve the management of their interoperability efforts by developing a project plan and a complete and detailed integrated master schedule. VA and DOD stated that they agreed with this recommendation. In our January 2010 final report in response to the National Defense Authorization Act for Fiscal Year 2008, we noted that VA and DOD officials believed they had satisfied the act’s September 30, 2009, requirement for full interoperability by meeting specific interoperability- related objectives that the departments had established. These objectives included: refine social history data, share physical exam data, and demonstrate initial document scanning between the departments. Additionally, the departments had made progress in setting up their Interagency Program Office by hiring additional staff, including a permanent director. In addition, consistent with our recommendations in the three previously mentioned reports, the office had begun to demonstrate responsibilities outlined in its charter in the areas of scheduling, planning, and performance measurement. Nevertheless, the office’s efforts in these areas did not fully satisfy the recommendations and were incomplete. Specifically, the office did not have a schedule that included information about tasks, resource needs, or relationships between tasks associated with ongoing activities to increase interoperability. Also, key IT management responsibilities in the areas of planning and performance measurement remained incomplete. We reiterated that, by not having fulfilled key management responsibilities, as we had previously recommended, the Interagency Program Office continued to not be positioned to function as a single point of accountability for the delivery of the future interoperable capabilities that the departments were planning. The Interagency Program Office Was to Be the Single Point of Accountability for Establishing a Lifetime Electronic Record for Servicemembers and Veterans, but VA and DOD Did Not Develop Complete Plans for the Effort Although the Interagency Program Office charter named the office as the single point of accountability for the initiative, the office did not have key plans to define and guide the effort. In April 2009, the President announced that VA and DOD would work together to define and build VLER to streamline the transition of electronic medical, benefits, and administrative information between the two departments. VLER was intended to enable access to all electronic records for service members as they transition from military to veteran status, and throughout their lives. Further, the initiative was to expand the departments’ health information sharing capabilities by enabling access to private sector health data. Shortly after the April 2009 announcement, VA, DOD, and the Interagency Program Office began working to define and plan for the VLER initiative. Further, the office was rechartered in September 2009 and named as the single point of accountability for the coordination and oversight of jointly approved IT projects, data, and information sharing activities, including VLER. In our February 2011 report on the departments’ efforts to address their common health IT needs, we noted that, among other things, the Interagency Program Office had not developed an approved integrated master schedule, master program plan, or performance metrics for the VLER initiative, as outlined in the office’s charter. We noted that if the departments did not address these issues, their ability to effectively deliver capabilities to support their joint health IT needs would be uncertain. Thus, we recommended that the Secretaries of VA and DOD strengthen their efforts to establish VLER by developing plans that would include scope definition, cost and schedule estimation, and project plan documentation and approval. Although the departments stated they agreed with this recommendation, they did not implement it. The Interagency Program Office Was Responsible for the Development of a Joint Electronic Health Record System for VA and DOD, but the Office Was Not Positioned for Effective Collaboration The Interagency Program Office was assigned responsibility for the development of an electronic health record system that VA and DOD were to share. However, the departments did not provide the office with control over the resources (i.e., funds and staff) it needed to facilitate effective collaboration. In March 2011, the Secretaries of VA and DOD committed the two departments to developing the iEHR system, and in May 2012 announced their goal of implementing it across the departments by 2017. To oversee this new effort, in October 2011, VA and DOD re-chartered the Interagency Program Office to give it increased authority, expanded responsibilities, and increased staffing levels for leading the integrated system effort. The new charter also gave the office responsibility for program planning and budgeting, acquisition and development, and implementation of clinical capabilities. However, in February 2013, the Secretaries of VA and DOD announced that they would not continue with their joint development of a single electronic health record system. In February 2014, we reported on the departments’ decision to abandon their plans for iEHR. Specifically, we reported that VA and DOD had not addressed management barriers to effective collaboration on their joint health IT efforts. For example, the Interagency Program Office was intended to better position the departments to collaborate, but the departments had not implemented the office in a manner consistent with effective collaboration. Specifically, the Interagency Program Office lacked effective control over essential resources such as funding and staffing. In addition, decisions by the departments had diffused responsibility for achieving integrated health records, potentially undermining the office’s intended role as the single point of accountability. We concluded that providing the Interagency Program Office with control over essential resources and clearer lines of authority would better position it for effective collaboration. Further, we recommended that VA and DOD better position the office to function as the single point of accountability for achieving interoperability between the departments’ electronic health record systems by ensuring that the office has authority (1) over dedicated resources (e.g., budget and staff), (2) to develop interagency processes, and (3) to make decisions over the departments’ interoperability efforts. Although VA and DOD stated that they agreed with this recommendation, they did not implement it. The Interagency Program Office Subsequently Took Steps to Improve Interoperability Measurement and Additional Actions Are Planned In light of the departments’ not having implemented a solution that allowed for seamless electronic sharing of medical health care data, the National Defense Authorization Act for Fiscal Year 2014 included requirements pertaining to the implementation, design, and planning for interoperability between VA and DOD’s separate electronic health record systems. Among other things, the departments were each directed to (1) ensure that all health care data contained in VA’s VistA and DOD’s AHLTA systems complied with national standards and were computable in real time by October 1, 2014, and (2) deploy modernized electronic health record software to support clinicians while ensuring full standards- based interoperability by December 31, 2016. In August 2015, we reported that VA and DOD, with guidance from the Interagency Program Office, had taken actions to increase interoperability between their electronic health record systems. Among other things, the departments had initiated work focused on near-term objectives, including standardizing their existing health data and making them viewable by both departments’ clinicians in an integrated format. The departments also developed longer-term plans to modernize their respective electronic health record systems. For its part, the Interagency Program Office issued guidance outlining the technical approach for achieving interoperability between the departments’ systems. However, even with the actions taken, VA and DOD did not certify by the October 1, 2014, deadline established in the National Defense Authorization Act for Fiscal Year 2014 for compliance with national data standards that all health care data in their systems complied with national standards and were computable in real time. We also reported that the departments’ system modernization plans identified a number of key activities to be implemented beyond December 31, 2016—the deadline established in the act for the two departments to deploy modernized electronic health record software to support clinicians while ensuring full standards-based interoperability. Specifically, DOD had issued plans and announced the contract award for acquiring a modernized system to include interoperability capabilities across military operations. VA had issued plans describing an incremental approach to modernizing its existing electronic health records system. These plans—if implemented as described—indicated that deployment of the new systems with interoperability capabilities would not be completed across the departments until after 2018. With regard to its role, the Interagency Program Office had taken steps to develop process metrics intended to monitor progress related to the data standardization and exchange of health information consistent with its responsibilities. For example, it had issued guidance that calls for tracking metrics, such as the percentage of data domains within the departments’ current health information systems that are mapped to national standards. However, the office had not yet specified outcome-oriented metrics and established related goals that are important to gauging the impact that interoperability capabilities have on improving health care services for shared patients. As a result, we recommended that VA and DOD, working with the Interagency Program Office, take actions to establish a time frame for identifying outcome-oriented metrics, define goals to provide a basis for assessing and reporting on the status of interoperability-related activities and the extent to which interoperability is being achieved by the departments’ modernized electronic health record systems, and update Interagency Program Office guidance to reflect the metrics and goals identified. Subsequently, we reported that VA and DOD had certified in April 2016 that all health care data in their systems complied with national standards and were computable in real time. However, VA acknowledged that it did not expect to complete a number of key activities related to its electronic health record system until sometime after the December 31, 2016, statutory deadline for deploying modernized electronic health record software with interoperability. Further, in following up on implementation of the recommendations in our August 2015 report, we found that VA, DOD, and the Interagency Program Office had addressed the recommendations in full by updating guidance to include goals and objectives and an approach to developing metrics that would improve the departments’ ability to report on the status of interoperability activities. The Interagency Program Office’s Role in Governing VA’s New Electronic Health Record System Acquisition Is Uncertain In June 2017, the former VA Secretary announced a significant shift in the department’s approach to modernizing the department’s electronic health record system. Specifically, rather than continue to use VistA, the Secretary stated that the department planned to acquire the same Cerner electronic health record system that DOD has been acquiring. Accordingly, the department awarded a contract to Cerner in May 2018 for a maximum of $10 billion over 10 years. Cerner is to replace VistA with a commercial electronic health record system. This new system is to support a broad range of health care functions that include, for example, acute care, clinical decision support, dental care, and emergency medicine. When implemented, the new system will be expected to provide access to authoritative clinical data sources and become the authoritative source of clinical data to support improved health, patient safety, and quality of care provided by VA. Deployment of the new electronic health record system at three initial sites is planned for within 18 months of October 1, 2018, with a phased implementation of the remaining sites over the next decade. Each VA medical facility is expected to continue using VistA until the new system has been deployed at that location. As we testified in June 2018, VA has taken steps to establish a program management office and has drafted a structure for technology, functional, and joint governance of the electronic health record implementation. Specifically, in January 2018, the former VA Secretary established the Electronic Health Record Modernization (EHRM) program office that reports directly to the VA Deputy Secretary. Further, VA has drafted a memorandum that describes the role of governance bodies within VA, as well as governance intended to facilitate coordination between the department and DOD. According to EHRM program documentation, VA is in the process of establishing a Functional Governance Board, a Technical Governance Board, and a Governance Integration Board comprised of program officials intended to provide guidance and coordinate with DOD, as appropriate. Further, a joint governance structure between VA and DOD has been proposed that would be expected to leverage existing joint governance facilitated by the Interagency Program Office. Nevertheless, while VA’s plans for governance of the EHRM program provide a framework for high-level oversight for program decisions moving forward, EHRM officials have noted that the governance bodies will not be finalized until October 2018. Accordingly, the officials have not yet indicated what role, if any, the Interagency Program Office is to have in the governance process. Conclusions The responsibilities of the Interagency Program Office have been intended to support the numerous approaches taken by VA and DOD to increase health information interoperability and modernize their respective electronic health record systems. Yet, while the office has led key efforts to identify data standards that are critical to interoperability between systems, the office has not been effectively positioned to be the single point of accountability originally described in the National Defense Authorization Act for Fiscal Year 2008. Further, the future role of the Interagency Program Office remains unclear despite the continuing need for VA and DOD to share the electronic health records of servicemembers and veterans. In particular, what role, if any, that the office is to have in VA’s acquisition of the same electronic health record system that DOD is currently acquiring is uncertain. Recommendation for Executive Action We are making the following recommendation to VA: The Secretary of Veterans Affairs should ensure that the role and responsibilities of the Interagency Program Office are clearly defined within the governance plans for acquisition of the department’s new electronic health record system. (Recommendation 1) Chairman Banks, Ranking Member Lamb, and Members of the Subcommittee, this completes my prepared statement. I would be pleased to respond to any questions that you may have at this time. GAO Contact and Staff Acknowledgments If you or your staffs have any questions about this testimony, please contact Carol C. Harris, Director, Information Technology Management Issues, at (202) 512-4456 or harrisc@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this testimony statement. GAO staff who made key contributions to this testimony are Mark Bird (Assistant Director), Jennifer Stavros-Turner (Analyst in Charge), Rebecca Eyler, Jacqueline Mai, Scott Pettis, and Charles Youman. This is a work of the U.S. government and is not subject to copyright protection in the United States. The published product may be reproduced and distributed in its entirety without further permission from GAO. However, because this work may contain copyrighted images or other material, permission from the copyright holder may be necessary if you wish to reproduce this material separately.

Why GAO Did This Study The National Defense Authorization Act for Fiscal Year 2008 included provisions that VA and DOD jointly develop and implement electronic health record systems or capabilities and accelerate the exchange of health care information. The act also required that these systems be compliant with applicable interoperability standards. Further, the act established a joint Interagency Program Office to act as a single point of accountability for the efforts, with the function of implementing, by September 30, 2009, electronic health record systems that allow for full interoperability. This testimony discusses GAO's previously reported findings on the establishment and evolution of the Interagency Program Office over the last decade. In developing this testimony, GAO summarized findings from its reports issued in 2008 through 2018, and information on the departments' actions in response to GAO's recommendations. What GAO Found Since its establishment in 2008, the Department of Defense (DOD) and Department of Veterans Affairs (VA) Interagency Program Office has been involved in various approaches to increase health information interoperability. However, the office has not been effectively positioned to function as the single point of accountability for the departments' electronic health record system interoperability efforts. For example, Between July 2008 and January 2010, GAO issued reports on VA's and DOD's efforts to set up the office, which highlighted steps the departments had taken, but also identified deficiencies, such as vacant leadership positions and a lack of necessary plans. GAO recommended that the departments improve management of their interoperability efforts by developing a project plan and results-oriented performance goals and measures. In April 2009, the Interagency Program Office was assigned responsibility for establishing a lifetime electronic record for servicemembers and veterans, called the Virtual Lifetime Electronic Record. GAO reported in February 2011 that, among other things, the office had not developed and approved an integrated master schedule, a master program plan, or performance metrics for the initiative, as outlined in the office's charter. Accordingly, GAO recommended that the departments correct these deficiencies to strengthen their efforts to establish the Virtual Lifetime Electronic Record. In March 2011, VA and DOD committed to jointly developing a new, common integrated electronic health record system and empowered the Interagency Program Office with increased authority, expanded responsibilities, and increased staffing levels for leading the integrated system effort. However, in February 2013, the departments abandoned their plan to develop the integrated system and stated that they would again pursue separate modernization efforts. In February 2014, GAO reported on this decision and recommended that VA and DOD take steps to better position the office to function as the single point of accountability for achieving interoperability between the departments' electronic health record systems. VA and DOD stated that they agreed with the above GAO recommendations. However, in several cases the departments' subsequent actions were incomplete and did not fully address all recommendations. In June 2017 VA announced that it planned to acquire the same electronic health record system that DOD has been acquiring. GAO testified in June 2018 that a governance structure had been proposed that would be expected to leverage existing joint governance facilitated by the Interagency Program Office. At that time, VA's program officials had stated that the department's governance plans for the new program were expected to be finalized in October 2018. However, the officials have not yet indicated what role, if any, the Interagency Program Office is to have in the governance process. Ensuring that the role and responsibilities of the office are clearly defined within these governance plans is essential to VA successfully acquiring and implementing the same system as DOD. What GAO Recommends GAO recommends that VA clearly define the role and responsibilities of the Interagency Program Office in the governance plans for acquisition of the department's new electronic health record syst

gao_GAO-18-19

gao_GAO-18-19_0

Background Employee Retirement Income Security Act of 1974 (ERISA) and the Internal Revenue Code (IRC) The Employee Retirement Income Security Act of 1974 (ERISA) contains various provisions intended to protect the interests of plan participants and beneficiaries in workplace retirement plans. These protections include requirements related to reporting and disclosure, participation, vesting, and benefit accrual, as well as plan funding. For example, ERISA requires plans to provide plan participants with a summary plan description, including information on their rights under ERISA, periodic benefit statements, and upon request, a copy of the annual report including a financial statement, according to DOL. ERISA sets fiduciary standards that generally require workplace retirement plan funds to be handled prudently and in the sole interest of participants. ERISA also establishes certain requirements related to plan termination. ERISA does not require employers to provide workplace retirement plans, but those that do must comply with applicable requirements and standards. The Internal Revenue Code (IRC) provides favorable tax treatment for workplace retirement plans that meet certain qualification requirements set out in the IRC. For example, employees are generally not taxed on contributions made on their behalf but instead are taxed on benefits received. Federal Agencies’ Roles with Respect to U.S. Workplace Retirement Plans Department of Labor’s Employee Benefits Security Administration (EBSA) Several federal agencies play a role with respect to U.S. workplace retirement plans. Responsibility for enforcing ERISA is shared by DOL, Treasury, and PBGC. Treasury, through IRS, is primarily responsible for enforcing the IRC. The mission of DOL’s Employee Benefits Security Administration (EBSA) is to assure the security of retirement, health, and other workplace-related benefits of U.S. workers and their families. DOL administers Title I of ERISA, which includes the fiduciary standards and disclosure and reporting requirements. To carry out its responsibilities, EBSA issues regulations in these and other areas, and conducts programs and initiatives to assist and educate workers, plan sponsors, fiduciaries, and service providers on their rights and obligations under ERISA. EBSA also issues guidance and field assistance bulletins to assist plan sponsors and plan fiduciaries with managing retirement plans. For instance, in 2014, EBSA issued Field Assistance Bulletin (FAB) 2014-01 to assist fiduciaries of terminating DC plans in fulfilling their obligations under ERISA to locate missing participants and properly distribute their account balances. EBSA also maintains an outreach program employing approximately 100 benefits advisors throughout the country in 13 field offices. The program offers services to educate U.S. workers, beneficiaries, and plan sponsors about their rights and obligations under federal employee benefit laws, and helps individuals obtain retirement benefits that have been improperly denied. Internal Revenue Service (IRS) Under Title II of ERISA and subsequent amendments to the IRC, IRS issues and enforces rules that plans must meet to be qualified for preferential tax treatment. IRS also enforces certain provisions in Title I of ERISA regarding participation, vesting, benefit accrual, and minimum funding. IRS’ mission is to help U.S. taxpayers understand and meet their tax responsibilities and to enforce the law with integrity and fairness. To help achieve its mission, IRS issues tax regulations and other guidance to help taxpayers comply with the IRC. IRS guidance provides detailed and technical explanations of tax laws for professional tax preparers as well as taxpayers. IRS also manages a number of initiatives, programs, and systems to enforce federal tax law and assist taxpayers that are related to our review of unclaimed retirement accounts. For example, to assist taxpayers, IRS adopted a Taxpayer Bill of Rights in 2014 to provide a better understanding of taxpayers’ rights under the IRC. IRS also periodically publishes a strategic plan for a given period that outlines how it will improve service to taxpayers and enforce the law. To help ensure that taxpayers are paying the correct amount of tax due and to identify discrepancies, IRS’ Automated Underreporter Program matches taxpayer income and deductions submitted on information returns by third parties against amounts reported by taxpayers on their individual income tax returns. IRS also assists taxpayers and payors with information about federal tax withholding obligations. To assist taxpayers with foreign accounts, since 2003 IRS has offered an Offshore Voluntary Disclosure Program. This program provides a way for taxpayers with previously undisclosed income and undisclosed offshore accounts that need to be reported to contact IRS and resolve their tax matters. IRS also assists sponsors that administer qualified retirement plans through a number of systems and programs. For example, IRS offers assistance to plan sponsors through the Employee Plans Compliance Resolution System, which helps sponsors of qualified plans remedy operational and form mistakes made in the course of administering a retirement plan and avoid plan disqualification. IRS also forwards letters to missing individuals on behalf of private individuals or government agencies for a “humane purpose” when there is no other way to relay the information to the individual. Between 1994 and 2012, IRS forwarded letters through a letter forwarding program on behalf of entities that control assets that may be due a taxpayer, such as from sponsors of qualified plans that are attempting to locate missing participants. Pension Benefit Guaranty Corporation (PBGC) Title IV of ERISA created PBGC as a U.S. government corporation to provide plan termination insurance for certain DB plans that are unable to pay promised benefits. For example, when a PBGC-insured single- employer DB plan fails, PBGC trustees the plan and pays benefits up to statutory limits. PBGC also oversees the voluntary (“standard”) termination of fully funded PBGC-insured single-employer DB plans to ensure participants will receive the benefits to which they are entitled. As part of the standard termination process, PBGC’s Missing Participants Program connects participants—missing when the plan closes out—to their retirement benefits, in part by maintaining a centralized, online database the public can use to find lost retirement benefits. Social Security Administration (SSA) SSA provides retirement benefits to eligible individuals under the federal Social Security Old Age and Survivors’ Insurance program (Social Security). Although SSA does not oversee workplace retirement plans, SSA maintains data that are reported to IRS by plans using Form 8955- SSA on separated participants with vested but undistributed benefits. When individuals claim Social Security benefits, SSA may provide them with a “Potential Private Pension Benefit Information” notice that indicates they may be entitled to a retirement benefit through a past employer. Ongoing and Terminated Workplace Retirement Plans DOL reported that in 2014 there were just over 639,000 DC plans and nearly 43,500 DB plans in the United States. These plans were sponsored by individual employers (i.e., private single-employer plans) and provided benefits to nearly 118 million participants. When a qualified plan terminates—whether it is a DB or DC plan—federal law requires plan participants to immediately be 100 percent vested in all accrued benefits (to the extent funded in the case of a DB plan) regardless of the vesting schedule in the plan document, according to IRS. A plan sponsor is required to distribute assets from a terminated plan as soon as administratively feasible, but generally within 1 year after plan termination. For terminated DC plans, such as 401(k) plans, participants generally receive the full amount of their vested account balance upon plan termination, according to IRS. Transferring Savings of Missing Participants in U.S. Defined Contribution Retirement Plans When an employee separates from an employer but still has vested savings in a qualified defined contribution retirement plan, the plan can, under certain conditions and without the participant’s consent, transfer accounts out of the plan—commonly referred to as a “forced transfer.” Before the Economic Growth and Tax Relief Reconciliation Act of 2001 was enacted, ongoing DC plans could, in the absence of participant instructions, distribute balances of $5,000 or less by paying them directly to the participant, referred to as a “cash-out.” This law sought to protect participants’ retirement savings by requiring ongoing plans that have a cash-out limit that exceeds $1,000 (up to $5,000), in the absence of participant instructions and subject to certain notice requirements, to transfer balances that exceed $1,000 (up to $5,000) to an individual retirement account (IRA), preserving their tax-preferred status. Terminating plans are subject to different requirements. Fiduciaries of terminating plans are obligated to search for missing participants, to notify them of the termination and pending distribution of benefits before transferring participants’ unclaimed accounts to an IRA or elsewhere, according to DOL guidance. The guidance further provides that fiduciaries of terminating plans who are unable to locate missing participants may also be permitted to transfer accounts belonging to missing participants, without consent, to a federally-insured bank account or to a state’s unclaimed property fund. This occurs if the plan fiduciary cannot find an IRA provider to accept a direct rollover distribution for a missing participant or otherwise determines not to roll over the distribution to an IRA, for some other compelling reason. For tax reporting purposes, transfers made to a bank or a state unclaimed property fund are generally subject to income taxation, according to the guidance. This contrasts with rollovers to IRAs in which transferred retirement savings remain tax- favored. Plan sponsors are generally required to withhold 20 percent of the account balance on transfers to a bank or state unclaimed property fund and will send the withheld amount to Treasury to be used toward any potential taxes due on the distribution. Tax Treatment of U.S. and Foreign Workplace Retirement Plans in which U.S. Individuals Participate In the United States, employee and employer contributions and investment earnings in a qualified retirement plan are generally not taxed as income until the employee receives the benefit. For example, employees participating in a 401(k) plan can generally elect to have their employer contribute a portion of their compensation to their account on a pretax basis. This deferred compensation (commonly referred to as a pre- tax elective contribution by IRS) is not subject to income tax withholding, and employees are not required to report it as wages on their individual U.S. tax returns at the time of the contribution. In addition, employers can provide matching or non-elective contributions to an employee’s 401(k) account; these matching or non-elective contributions are generally tax- deductible by the employer and employees also are not required to report these contributions as wages on their U.S. tax returns or pay income tax on these contributions at the time the contributions are made. Distributions from a qualified DC plan in the United States made to participants, including those who have separated from their employer, may be treated differently for tax purposes, depending on the nature and timing of the distribution. For example, a direct rollover, in which money is transferred directly from one qualified workplace retirement plan or IRA to another eligible retirement plan or IRA, is not taxable at the time of the rollover but should be reported on the participant’s federal tax return. By contrast, a distribution that is not rolled over is generally taxable income in the year in which it is received by the participant, according to IRS. Foreign workplace retirement plans are generally not tax-qualified under the IRC or covered by ERISA, according to IRS officials and tax experts with whom we spoke. They are, however, generally subject to the regulatory structure in place in the country where the retirement plan exists. Foreign workplace retirement plans that cover U.S. individuals may be subject to certain provisions of the IRC and other federal laws governing reporting and taxation of these retirement assets, as well as any applicable income tax treaties between the United States and the foreign country (see more about these treaties below). The extent to which U.S. individuals are subject to U.S. income tax on the contributions and earnings accruing in their foreign workplace retirement account depends on the specific characteristics of the plan. For example, according to tax experts with whom we spoke, many foreign workplace retirement plans qualify as employees’ trusts, and the taxation of contributions and earnings from these plans are governed by section 402(b) of the IRC. According to IRS, as long as the foreign retirement plan is determined to be an employees’ trust, the U.S. individual must include on their U.S. tax return contributions to the trust if the contributions are not subject to a substantial risk of forfeiture (vested). In addition, IRS officials said contributions that become vested after the year of contribution are taxable in the year of vesting, and earnings are taxable when distributed. Some foreign workplace retirement plans may include investments in a Passive Foreign Investment Company (PFIC), which, according to one tax preparer with whom we spoke, are investments in foreign mutual funds, hedge funds, or other kinds of pooled investments not incorporated in the United States. A U.S. individual who is a shareholder of a PFIC may be subject to annual reporting requirements and a high income tax rate on certain distributions. U.S. individuals who participate in a foreign workplace retirement plan also may be subject to income tax on any distribution they receive from their plan during the current tax year. Depending on the circumstances, U.S. individuals also may be subject to income tax on certain distributions they have not actually received, such as transfers of assets between or within foreign workplace retirement plans, if they are in “constructive receipt” of (or otherwise have income inclusion with respect to) the funds. In addition, U.S. individuals who pay foreign income taxes on distributions from their foreign workplace retirement plans may be eligible to claim a foreign tax credit on their U.S. tax return. Income Tax Treaties U.S. Income Tax Treaties with Other Countries One objective of tax treaties is to provide taxpayers some relief from having to pay taxes in both the United States and a foreign country on the same income—referred to as “double taxation”—without creating opportunities for tax evasion or avoidance. Treaty provisions generally apply to both countries that have signed the treaty. A U.S. resident who receives income from a treaty country may be entitled to certain treaty benefits—credits, deductions, exemptions, or reductions in the rate of tax—on the taxes owed to that foreign country. Similarly, residents of the foreign country may be entitled to treaty benefits on their U.S. taxes on income from U.S. sources. However, with certain exceptions, tax treaties generally do not reduce the U.S. tax liability of U.S. residents. As of October 2017, the United States had a network of 57 comprehensive income tax treaties covering 66 countries, according to Treasury. U.S. individuals are subject to U.S. income tax on their worldwide income, and this could include contributions and earnings within and distributions from a foreign workplace retirement plan. However, tax treaty provisions may reduce foreign income taxes owed by U.S. individuals who receive income sourced from a treaty country; for example, through the use of credits, deductions, exemptions, or tax rate reductions. (See appendix I for more information on how IRS recommends taxpayers review tax treaties.) According to IRS and Treasury, almost all U.S. tax treaties also contain what is known as a “saving clause,” which IRS describes as a way to preserve or “save” the right of each country to tax its own residents (and in the case of the United States, its citizens) as if no tax treaty were in effect. As a result, these treaties do not generally reduce the U.S. income tax for U.S. individuals, unless an exception applies. In February 2016, Treasury issued a revised U.S. Model Income Tax Convention (i.e., model treaty), which is the baseline text the agency uses when it negotiates tax treaties. Depending on the outcome of the treaty negotiations, the final treaty with a particular foreign country may or may not include language from the model treaty. Reporting Foreign Accounts and Foreign Financial Assets According to IRS, the Foreign Account Tax Compliance Act (FATCA) and IRC section 6038D are important developments in U.S. efforts to combat tax evasion by U.S. individuals holding accounts and other financial assets offshore, which may have implications for U.S. individuals who have foreign retirement accounts. FATCA generally requires foreign financial institutions to provide information to IRS regarding foreign financial accounts held by U.S. taxpayers. IRC section 6038D generally requires U.S. individuals to report to IRS their foreign financial assets that exceed a certain threshold. Beginning in July 2014, U.S. entities were required to withhold 30 percent on certain payments to a foreign financial institution unless the institution has entered into an agreement with IRS regarding FATCA reporting or is in a jurisdiction that is treated as having an Intergovernmental Agreement (IGA) in effect. However, FATCA regulations exempt foreign financial institutions from reporting on retirement accounts that meet certain requirements. Treasury has entered into IGAs with other countries to assist with implementing FATCA that may also provide an exemption for foreign financial institutions reporting of certain retirement accounts. This exemption does not exist under IRC section 6038D, which requires individuals, including U.S. citizens, to report their foreign retirement accounts on IRS Form 8938 if they meet certain regulatory thresholds. For example, unmarried U.S. individuals living abroad must file if the total value of their specified foreign financial assets is more than $200,000 on the last day of the tax year or more than $300,000 at any time during the tax year. Estimated Number and Financial Profile of U.S. Citizens Living and Working Abroad The U.S. Department of State (State) estimates that as of April 2015, 8.7 million U.S. citizens lived abroad. (See appendix II for other estimates.) Income data published by IRS for 2011 suggest that a majority of U.S. taxpayers who earned income from foreign sources likely owed little federal income tax because their reported adjusted gross income was relatively low due to tax credits and exemptions available to taxpayers on foreign-earned income. IRS estimated that in 2011 over 449,000 returns were filed by taxpayers reporting foreign-earned income and just over 445,000 of these returns reported using the foreign-earned income exclusion. We previously reported that for tax year 2011, taxpayers claiming the foreign-earned income exclusion had higher average income ($163,450) than the average Form 1040 filer ($58,706), and about 45 percent of those taxpayers had an adjusted gross income of less than $10,000. These data reflect that some taxpayers were able to exclude all or most of their foreign-earned income in calculating their adjusted gross income. We also reported that taxpayers claiming the foreign- earned income exclusion had lower average U.S. tax rates than all Form 1040 filers. Participants Face Challenges Managing Unclaimed Retirement Accounts and Agencies Have Not Provided Sufficient Guidance and Information to Assist Them or Plan Sponsors Participants Have Challenges Managing Unclaimed Retirement Accounts Participants in U.S. workplace retirement plans face challenges managing unclaimed accounts accumulated over the course of their careers. We previously reported that some 401(k) plan participants find it difficult to keep track of their savings, particularly when they change jobs, because of challenges with consolidation, communication, and information. First, we found that individuals who accrue multiple accounts over the course of a career may be unable to consolidate their accounts by rolling over savings from one employer’s plan to the next. Second, maintaining communication with a former employer’s plan can be challenging if companies are restructured and plans are terminated or merged and renamed. Third, key information on lost accounts may be held by different plans, service providers, or government agencies, and participants may not know where to turn for assistance. As one witness testified to the ERISA Advisory Council in 2013, it is not uncommon for former employees to have difficulty locating a previous employer. Existing reporting and disclosure requirements directed at plan sponsors can provide participants who separate from their employer information about their accounts via multiple disclosures. However, plan sponsors have no automatic way to keep participants’ contact information up to date, nor do they have ways to ensure that separated participants will respond to their communications. Many participants rarely read the notices they receive. We conducted a review of private sector pension plan notices in 2013, and found that participants were interested in information about their individual benefits, which could reasonably include information about a pending distribution of their unclaimed account. Due to the large number of participant notices, we found participants struggled with what they must or should read. When participant notices are ineffective, accounts can become lost or unclaimed and eventually shrink or disappear entirely, diminishing a source of income in retirement. For example, accounts with a balance of $1,000 or less can be cashed out of a plan without participant consent; account balances can be reduced by tax withholding and early distribution taxes, or conditionally forfeited by the plan sponsor until the participant emerges to make a claim. Accounts with balances under $5,000, and sometimes those with larger balances, can be forcibly transferred to an IRA, where the account balances may decrease over time as the fees outpace low investment returns, as we reported in our prior work. In 13 of the 19 forced-transfer IRA scenarios we considered in 2014, a $1,000 account balance was reduced to zero within 30 years. DOL has also uncovered tens of thousands of participants of retirement age with unclaimed accounts that remained in their plans who were not receiving the retirement income they were due. DOL Has Not Provided Guidance to Ongoing Plan Sponsors for Locating Missing Participants Although DOL has provided guidance to plan sponsors of terminated DC plans about locating missing participants and unclaimed accounts, DOL has not provided similar guidance to ongoing plans. DOL officials told us that they are conducting investigations of steps taken by ongoing plans to find missing participants under their authority to oversee compliance with ERISA’s fiduciary requirement that plans be administered for the exclusive purpose of providing benefits. Plan sponsors are required to send notices to participants in a variety of circumstances, such as to obtain direction before making a distribution. However, the communication is not always successful, and may result in a mailing to an out-of-date address. With the absence of guidance, it is not clear to sponsors of ongoing DC plans how they should satisfy requirements to notify participants when participant addresses are out of date. Undeliverable mail is the main indicator for identifying a participant as missing, according to third-party administrators (TPA), who help manage missing participant issues for plan sponsors. However, DOL officials told us a recent pilot investigation found that some ongoing plans send notices that were returned undeliverable but then fail to follow-up with any search process. In contrast, if participants in a terminated plan do not respond to a notice, plan sponsors need to take certain steps, at a minimum, to locate the participant or a beneficiary. According to our analysis of stakeholder interviews, in some circumstances plan sponsors may be considering a participant to have been “notified,” even when the mail used to notify them was returned undeliverable. Executives at one firm that conducts missing participant searches told us that for an average client, 7 to 10 percent of mail will be returned undeliverable, which means communication was unsuccessful, potentially leaving participants without notification of changes to the plan or potential distributions or transfers. It also is not clear how ongoing plan sponsors should arrange for paying to obtain updated addresses of participants with unclaimed accounts. Because search costs are not all paid from plan assets, finding missing participants can be an additional business expense for plan sponsors. Once an account is force-transferred out of the plan to an IRA, the account may be charged a $65 annual search fee by the IRA provider, as we reported one provider did in 2014. Plan sponsors are permitted to pay only reasonable plan administration expenses, although they may charge expenses associated with a specific participant to that participant’s account. To reduce costs for its plan sponsor clients, representatives at one TPA told us that it will generally try to cash out accounts in ongoing plans under $1,000 immediately, before an address becomes obsolete. DOL audit findings also show that ongoing plans have challenges staying in touch with missing participants and paying them their benefits when due. DOL officials told us that in a recent DOL pilot investigation of 50 large DB plans, they found tens of thousands of separated participants who were entitled to benefits but were not receiving them. They told us that between 1 and 7 percent of all participants could be missing and not receiving letters from the plan, depending on the industry. They said their investigations found databases with missing names, addresses, and Social Security Numbers, and data they suspected were unreliable, such as participants named “Jane Doe” or with birth dates listed as “1/1/1900.” DOL enforces the fiduciary standards of ERISA, which require plan fiduciaries to act solely in the interest of plan participants and their beneficiaries, for the exclusive purpose of providing benefits to them, among other things. After plan termination, plan fiduciaries must distribute all plan assets as soon as administratively feasible, which could create an urgent need for plan sponsors to find participants. DOL officials said that part of their enforcement role is examining how plans are maintaining good records and what plans are doing to find and communicate with participants—officials are aware that additional guidance indicating what is expected of plan fiduciaries would be helpful. PBGC has recently published a final rule which expands its Missing Participants Program to cover most terminated DC plans, and DOL intends to revisit its guidance within that context. At that time, DOL will have an opportunity to also provide guidance to ongoing DC plan sponsors on their obligations under ERISA to prevent, search for, and pay costs associated with missing participants. By doing so, DOL can provide plan sponsors with better tools to manage unclaimed accounts and help ensure that future DOL investigations do not also uncover ongoing DC plans with substantial numbers of participants not receiving benefits to which they are entitled. IRS Has Not Issued Guidance Clarifying Tax Withholding Requirements for Cashed-Out Unclaimed Retirement Accounts Based in part on our discussions with IRS and our review of ERISA Advisory Council documentation, when a plan sponsor cashes out an unclaimed account and sends the money to the participant address it has on file, the address may be obsolete. As a result, the participant may not include the distribution in his or her taxable income for the year because the participant may not have received the payment from the plan sponsor or be aware of the transfer. According to an IRS publication on tax withholding for plan sponsors, a 20 percent income tax withholding generally is mandatory on amounts distributed from the plan that are not rolled over directly into another qualified plan or an IRA. However, our findings that some participants may not actually receive these distributions raise questions about whether withholding should be required in situations when it is reasonable to believe distributions will not be received by the participants. Misconceptions exist regarding how and when IRS will credit tax withholding toward a taxpayer’s tax liability. For example, two TPAs told us they believed that IRS will credit tax withholding on cashed-out accounts to the tax liabilities of missing participants. One industry representative we interviewed in 2013 told us that he withheld taxes when he could not find a participant because he believed the withholding would cause IRS to make the participant aware of the account. According to DOL bulletins issued in 2004 and 2014, some plan sponsors were using 100 percent withholding—in effect transferring the entire account to IRS— under the assumption that the withheld amounts would be matched and applied to a participant’s tax liabilities. DOL bulletins clarified it was not an appropriate distribution option for plan sponsors. Table 1 shows a variety of approaches to tax withholding. However, according to IRS, none of the tax withholding strategies automatically reduces the tax liability of the account holder. IRS officials told us that the agency does not routinely credit federal tax withholding to a taxpayer’s current federal tax liability unless the taxpayer has made a claim. Retirement accounts with small balances are most vulnerable to the tax consequences of tax withholding by plan sponsors. We previously reported that in the absence of participant instructions, accounts with a balance of $1,000 or less can be cashed out of the tax-deferred plan environment by plan fiduciaries without the separated participant’s consent. From 2004 to 2013, separated participants left more than 13 million accounts of $1,000 or less in workplace retirement plans with an aggregate value of $1.2 billion, according to SSA. Withholding taxes on balances of $1,000 or less at the time of distribution may result in participants paying taxes twice on the account. IRS told us that missing participants generally have up to 3 years to become aware of and claim the withheld amounts for them to be credited towards their tax liability. However, missing participants who claim their account after 3 years may again pay federal income tax on the account balance, although IRS officials said they thought such a scenario would be rare. (See fig. 1). IRS has not issued specific guidance clarifying the withholding requirements that apply to distributions from unclaimed accounts in situations in which the participant may be unlikely to receive the distribution. By reviewing the issue of distributions to participants with unclaimed accounts, including reviewing the IRC in this context, IRS may be able to issue guidance on applicable tax withholding and other tax requirements with respect to such accounts. The Taxpayer Bill of Rights states that taxpayers are entitled to clear explanations of the tax laws in IRS publications and notices, and federal internal control standards require agencies to communicate effectively with external stakeholders to help achieve agency goals. U.S. participants already facing the challenge of finding a small account transferred without their consent may discover, when the account is located, 20 percent of their account eliminated by taxes. Without an IRS review of this issue and subsequent guidance, questions may remain about withholding from distributions in situations where the participant may be missing. Plan Sponsors Can No Longer Use IRS’ Letter Forwarding Program to Locate Missing Participants Under IRS’ letter forwarding program, between 1994 and 2012 plan sponsors could ask IRS to use IRS’ most current address on file to forward a letter with information about an account to a missing plan participant. However, in 2012 IRS modified the service and no longer forwards letters on behalf of qualified retirement plan sponsors attempting to locate plan participants. According to the 2013 Report of the ERISA Advisory Council on Locating Missing and Lost Participants, the letter forwarding program was a popular alternative for plan sponsors when email and U.S. mail proved ineffective at contacting separated participants. Executives at one large record keeper told us the letter forwarding program provided very important assistance for locating missing participants, noting that few individuals are going to ignore correspondence from IRS. Although the letter forwarding program never notified the plan sponsor as to whether or not the letter reached the intended recipient, executives at one TPA characterized the program as effective. In addition, they said the fact that it was sponsored by IRS and sanctioned by DOL gave plan fiduciaries confidence that they were acting prudently. Missing participant search services and their value vary widely today, based on industry representatives we interviewed. For example, representatives of one search firm told us they charged $1.25 for a search. However, an executive at a TPA firm told us another firm charged $35 for a Social Security Number-based search, which reliably connected with participants. PBGC estimates the cost of a commercial locator service to be $40 per search. Industry stakeholders told us that the steps currently required by existing guidance do not provide a straightforward way to send a letter about an unclaimed account to a missing plan participant. They described benefits that only IRS can provide through this service, such as the likelihood recipients will open a letter from IRS and the confidence fiduciaries have using an IRS- sponsored program. We discussed with IRS officials the commensurate fees charged in the private sector for missing participant searches and we discussed the variety of services and associated costs currently available. According to OMB Circular A-123, agencies and individual federal managers must take systematic and proactive measures to develop and implement appropriate, cost-effective management controls for results- oriented management. IRS has always charged a user fee for the letter forwarding program, and the fee has not changed since 1994. IRS officials told us resource constraints led them to revise the letter forwarding program. While IRS management controls will need to ensure that a program expansion is cost-effective, by reinstating the letter forwarding program for plan participants in a cost-effective manner, IRS can help support the retirement security of separated plan participants and plan sponsor efforts to meet their obligations under the IRC and ERISA. Information Provided to U.S. Workers on Unclaimed Retirement Accounts Is Often Outdated and Incorrect Certain information U.S. workers receive on unclaimed workplace retirement accounts based on data reported to IRS by plan sponsors is not reliable because plan sponsors are not updating the data over time as required. SSA maintains data on vested, unpaid retirement benefits left behind in workplace retirement plans by separated participants in its pension benefit record database. The information, including the name of the plan, the value of the benefit, and the contact information of the plan administrator, is reported by plan administrators to IRS, and IRS provides it to SSA. When an individual retires and claims Social Security benefits, SSA sends the individual a Notice of Potential Private Pension Benefit Information. The notice informs the recipient that they may have an unclaimed retirement account from a former employer and suggests that they may want to make an effort to determine whether or not the benefit actually does exist. SSA mails about 90,000 notices to new Social Security claimants each month. Separated participants can often find that no benefit exists, according to DOL and SSA documentation and stakeholders we interviewed. A TPA executive also told us separated participants are not always able to determine what happened to their accounts. IRS and SSA have a memorandum of understanding (MOU) in place establishing their agreements for collecting and managing these data. In the MOU, IRS and SSA agree to pursue improvements to the reporting process. The MOU states that, where appropriate and consistent with IRS directives, IRS will assess penalties under the IRC on plan sponsors who fail to file Form 8955-SSA according to instructions. The agencies have also agreed to contact and receive information from filers as necessary and appropriate to follow up regarding missing, incomplete, or incorrect information requested on the form. According to the Form 8955-SSA instructions, plan sponsors are required to report when benefits previously reported are paid, and therefore no longer due, to plan participants. Such updates allow the pension benefit record database at SSA, used to generate the Notice of Potential Private Pension Benefit Information, to reflect the fact that those benefits are no longer due. IRS officials said the data can be inaccurate because plan sponsors are not consistently reporting distributions, resulting in erroneous records of accounts accumulating in the database (see fig. 2). An executive at one TPA told us that plan sponsors generally remember to put participant names on Form 8955-SSA, but often fail to take the names off after benefits are paid. According to the TPA executive, if there are 1,000 names on the list of separated participants with vested benefits in the plan, 999 will have been paid by the time they receive the notice from SSA. Nonetheless, participants will generally inquire about a benefit when they receive the SSA notice because it is from the government, and they trust the notice and think the money is there, according to one TPA with whom we spoke. IRS officials told us that enforceable penalties can be imposed on plan sponsors for not including all required information on the form. The IRS website lists four possible actions related to incorrectly filing Form 8955- SSA that are subject to a penalty; however, a failure to report distributions is not on the list. IRS officials said if the agency were to add the failure to report distributions to the list the penalty would likely encourage some sponsors to update the data as required. IRS officials told us they do not currently know which plans are not reporting distributions. The Notice of Potential Private Pension Benefit Information leaves the responsibility for determining whether a benefit exists up to the participant and the agencies do not ask the participant for the results of their inquiries. SSA includes a note at the bottom of the notice encouraging new retirees to contact DOL with complaints, but the participant is not asked to follow up with IRS or SSA to identify plans associated with inaccurate data. Having this information would help IRS select plans to audit in order to update and improve the quality of data in SSA’s pension benefit record database. SSA could modify the notice participants receive to encourage them to inform IRS if they determine the information on the notice to be erroneous. DOL benefits advisers, who field calls from inquiring individuals after they receive a notice about a potential benefit that no longer exists, also have information on plans that may not be reporting distributions to separated participants on Form 8955-SSA as required. DOL officials told us they would need a formal MOU in place to facilitate such information sharing. Figure 3 illustrates these possible options for identifying plans not reporting distributions as required. Standards for internal control in the federal government state that agencies should communicate quality information externally so that external parties can help agencies achieve their objectives. Although IRS and SSA have agreed in the MOU to work together to promote efforts to improve internal controls, they are not collaborating to improve the likelihood that the Notice of Potential Private Pension Benefit Information will correspond to an actual benefit in the future. While IRS has authority over implementing and enforcing the Form 8955-SSA reporting requirements, IRS officials do not have access to SSA’s pension benefit record database to update records. IRS officials told us at one point they discussed with SSA a possible project that would allow plan sponsors to update all the records associated with their plan at once. SSA officials told us they could collaborate with IRS to update the data in the pension benefit record database. By working together, IRS and SSA can increase the likelihood that the Notice of Potential Private Pension Benefit Information corresponds to actual workplace retirement benefits in the future. Complex Tax Requirements and a Lack of Guidance Can Hinder U.S. Individuals’ Ability to Correctly Report Foreign Retirement Accounts Stakeholders Told GAO that U.S. Individuals Who Participate in Foreign Workplace Retirement Plans Face Multiple Challenges U.S. individuals who participate in foreign retirement plans can face a number of challenges with tax reporting requirements on their retirement savings. According to IRS officials and tax preparers with whom we spoke, these challenges are greater for U.S. individuals who live and work abroad full time than for corporate executives on temporary assignment in a foreign country. Individuals sent abroad for limited times by their employer often remain as participants in their employer’s U.S. workplace retirement plan and do not need to participate in a foreign workplace plan. According to IRS officials and tax professionals with whom we spoke, many of these executives may have tax filing assistance made available to them by their company, further reducing their reporting burden. Individuals who work in a foreign country may be forced to participate in a mandatory foreign retirement plan, depending on the country and the rules governing residency, according to officials with whom we spoke in our case study locations. In these instances, according to IRS officials, the individuals have no choice but to comply with U.S. tax reporting rules on their foreign retirement accounts. Those who live abroad long-term due to family or personal ties naturally accumulate foreign assets and savings, such as foreign retirement accounts. Tax preparers in all five case study locations we reviewed, as well as IRS officials, indicated that preparing a U.S. tax return for a participant in a foreign retirement plan is more complex than preparing a comparable U.S. tax return that does not include foreign assets. We were told that attempting to categorize a foreign retirement account for tax reporting under the IRC can be challenging because such accounts may be reported as one of several different designations that may or may not be eligible for tax-deferral in the United States. This contrasts with U.S. individuals participating in U.S. retirement plans that meet the criteria for tax-qualified status under the IRC, who generally receive a Form W-2 Wage and Tax Statement that automatically deducts retirement account contributions from gross wages. In addition, participating in a foreign retirement plan can initiate a complex set of U.S. reporting requirements on retirement assets, such as participants having to report contributions and earnings or having to file additional forms and schedules for their retirement account, which is typically not required of taxpayers with U.S.- based retirement plans. IRS officials told us that the onus is on U.S. individuals who participate in foreign retirement plans to comply with these complex reporting requirements. As a result, these participants often need to turn to expert tax preparers to prepare their U.S. tax return even if they ultimately do not have to pay taxes. Statutory changes on reporting foreign assets have further affected U.S. individuals who participate in foreign retirement plans. Stakeholders told us that reporting requirements under the Foreign Account Tax Compliance Act (FATCA) can increase the cost of tax preparation for U.S. individuals who participate in foreign retirement plans. For example, according to IRS guidance, these participants in foreign retirement plans must gather and examine monthly retirement account statements, convert the account balance to U.S. dollars, and determine if the total value of the account at the end of the year or anytime during the year caused the individual’s total asset value to exceed the reporting threshold. If the total assets meet the reporting threshold, the participant must report the value of their retirement account even if they are no longer contributing to the account. In contrast, participants in U.S. plans generally are not required to report the value of their U.S. workplace retirement accounts under FATCA or IRC section 6038D, according to IRS. We were also told of other consequences of FATCA for U.S. individuals abroad, such as a reduction in available financial services, as some banks refuse to do business with U.S. individuals because of FATCA’s reporting requirements. Lastly, once an individual decides to change jobs in a foreign country, transferring foreign retirement savings can be difficult. For example, in several of the case study locations we reviewed foreign officials and tax preparers told us that plans automatically transfer a retirement account to a different account within the plan or to a location outside the plan when an employee separates from their employer, which can have U.S. tax implications. Stakeholders said that existing U.S. tax law does not provide these participants with tax-deferral if they transfer their foreign retirement savings from one foreign workplace retirement plan to another—a benefit granted to U.S. participants in qualified U.S. retirement plans who make such transfers. This condition may act as a disincentive for U.S. individuals abroad to consolidate foreign retirement accounts and can cause challenges when individuals change jobs or are required by their retirement plan or employer to transfer their account. IRS Guidance Is Unclear Regarding How to Report Foreign Retirement Accounts While IRS has issued guidance providing information regarding foreign assets and pensions, IRS officials told us that the guidance is not specific on how foreign workplace retirement plans should be treated under the IRC, nor does it provide guidance for specific countries. One source of guidance is Publication 54, Tax Guide for U.S. Citizens and Resident Aliens Abroad, which discusses special tax rules for U.S. citizens and resident aliens who work abroad or who have income earned in foreign countries. Another source of guidance in the International Tax Gap Series describes how foreign pensions and annuity distributions are taxed. While both guidance sources provide taxpayers with some information on how to report foreign assets, neither describes in detail how taxpayers are to determine if their foreign workplace retirement plan is eligible for tax deferred status, or how to account for contributions, earnings, or distributions on their annual U.S. tax return, particularly whether and when contributions and earnings should be taxed as income. IRS also directs taxpayers to review tax treaties, if applicable, for provisions related to pensions, but IRS officials told us these treaties can vary from country to country and said that they can be difficult for non- experts to understand. For example, Treasury officials told GAO that the tax treaties for two of the five case study locations we selected have pension clauses and certain provisions that apply to U.S. residents of those countries saving for retirement that afford some tax protection. However, Treasury officials said that two of our other case study locations have treaties that do not provide tax protections for U.S. individuals on their foreign retirement accounts (one of our case study locations does not have an income tax treaty with the United States.) Tax preparers and IRS officials we spoke with indicated that it is difficult for U.S. individuals to know how to correctly apply tax treaty provisions to their foreign workplace retirement savings. In addition, these IRS officials and other retirement experts said a U.S. individual abroad without expertise in the IRC and tax treaties would have difficulty reporting their foreign retirement account correctly. Existing IRS guidance does not alleviate the confusion faced by U.S. individuals who participate in foreign retirement plans. Confusion regarding how to report foreign retirement accounts to IRS on a U.S. tax return or elsewhere is inconsistent with U.S. taxpayers’ rights, as described in the Taxpayer Bill of Rights, to pay no more than the correct amount of tax and to know what they need to do to comply with tax laws. (Government Service and Social Security), pensions and other similar remuneration beneficially derived by a resident of a Contracting State in consideration of past employment shall be taxable only in that State. (Government Service and Social Security), annuities derived and beneficially owned by a resident of a Contracting State shall be taxable only in that State. The term “annuities” as used in this paragraph means a stated sum paid periodically at stated times during a specified number of years or for life, under an obligation to make the payments in return for adequate and full consideration (other than services rendered).” Disagreement exists among the professional tax preparers with whom we spoke about the correct method for reporting foreign retirement accounts on a U.S. tax return. IRS officials told us that U.S. tax law generally does not recognize foreign retirement plans as tax-qualified and IRS does not recognize any retirement accounts outside the United States as having tax-qualified status. IRS officials we spoke to said that only plans meeting the specific requirements of 401(k) or other requirements describing retirement plan qualification may achieve tax-qualified status in the United States. As a result, according to IRS guidance, U.S. individuals participating in foreign workplace retirement plans generally cannot deduct contributions to their account from their income on their U.S. tax return. This is true even if the retirement account is considered a tax- deferred retirement account in the country where the individual works, and even if the account is similar in nature to those found in a U.S.-type retirement plan, such as a 401(k) plan. IRS officials told us that it should generally be unnecessary to file a foreign retirement account as a Passive Foreign Investment Company (PFIC) if the foreign retirement plan is covered by a tax treaty with the United States, but acknowledged that some tax advisors in foreign countries advise their U.S. clients to consider their interest in such plans as an investment in a PFIC. For example, in one of the case study locations we reviewed, a tax preparer said that he advises U.S. individuals who participate in such plans to report their foreign retirement account as a PFIC in their U.S. tax filing, and that contributions and earnings are subject to be taxed at the higher tax rate generally applicable to PFICs. Other tax preparers we spoke to in that location said that this is a matter of some discussion among tax preparers and that they reported retirement plans as an employees’ trust. The National Taxpayer Advocate told us that receiving incorrect tax advice from a foreign tax preparer may not be a sufficient mitigating circumstance to avoid penalties for reporting a foreign retirement account incorrectly on a tax return. While reasonable reliance on a tax professional with respect to the details of a return is generally a mitigating circumstance for errors on a return, according to the National Taxpayer Advocate, tax preparers in other countries are usually not considered qualified preparers by IRS. U.S. taxpayers who file an incorrect tax return can lose money by accruing penalties. IRS officials told us that individual taxpayers are responsible for understanding their filing requirements and for determining how to correctly file their tax returns, regardless of whether they live in a foreign country or the United States. In its mission to help taxpayers meet their tax responsibilities, IRS could issue guidance concerning how U.S. individuals are to correctly report their foreign retirement assets. The Taxpayer Bill of Rights states that as part of the right to a fair and just tax system, taxpayers have a right to expect that system to consider circumstances that affect their ability to provide timely information. IRS officials told us they had been considering issuing improved guidance in some areas pertaining to the taxation of foreign retirement accounts. However, without clearer specific guidance from IRS describing how to correctly report foreign retirement assets on a U.S. tax return, U.S. individuals who participate in foreign workplace retirement plans continue to run the risk of filing incorrect returns due to confusion over how to properly classify and report their accounts. Clarifying how U.S. individuals who participate in foreign workplace retirement plans should report their retirement assets on their annual U.S. tax return will help ensure these taxpayers can meet their tax reporting obligations. Complying with Reporting Requirements Can Be Costly for U.S. Individuals Who Participate in Foreign Workplace Retirement Plans Federal law requires U.S. individuals to report specified foreign financial assets, including any applicable retirement and pension accounts they own, if these assets, in the aggregate, are above the regulatory threshold. Similarly, the Report of Foreign Bank and Financial Accounts (FBAR) requires information with respect to foreign accounts above a certain amount. As a result, U.S. individuals who participate in foreign retirement plans may need to hire tax preparers to prepare returns in compliance with these U.S. laws, and, according to tax preparers with whom we spoke, the cost for having a complete tax return professionally prepared for an individual holding a foreign retirement account ranges from $1,800 to as high as $16,000. Determining how a foreign retirement account should be reported is time consuming even for experts. Tax preparers must prepare multiple items, including the tax return itself, and additional schedules and forms pertaining to the retirement account, according to the preparers with whom we spoke. For example, Form 3520 may be required if the account is being reported as a foreign trust. In addition to preparing tax forms, one tax preparer we spoke to said that preparers may have to spend time trying to obtain other documents necessary to prepare a U.S. tax return, for example, detailed retirement account statements. Since the implementation of IRC section 6038D, individuals have increased exposure to penalties, and failure to report a foreign retirement account when required may bring significant financial penalties, even if no taxes were due on the retirement account in question. For example, according to the IRS website, failure to report foreign financial assets on Form 8938 as required may result in a penalty of $10,000 and an additional penalty of up to $50,000 for continued failure to report after IRS sends the individual a notification of failure to report. As a result of this reporting requirement, U.S. individuals who participate in workplace retirement plans abroad may incur substantial costs to correctly file their returns and risk diminishing their retirement security if they fail to correctly report their foreign retirement assets. Even in cases where the individual owes no U.S. tax, tax preparation can costs thousands of dollars. Three tax preparers and representatives of one investment firm that provides pension advice with whom we spoke noted that even if a U.S. individual who participates in a foreign workplace retirement plan did not ultimately owe any taxes, they are required to report their foreign retirement assets under both FATCA and FBAR. Tax preparers in four of our case study locations as well as in the United States mentioned FATCA’s requirements as an added challenge when reporting foreign retirement accounts on U.S. tax returns. Additionally, one investment industry association representative we interviewed said that FATCA casts a wide net and that many “accidental Americans” and U.S. individuals abroad were challenged to comply with its requirements. Some of the tax preparers we spoke with said many individuals taking steps to come into tax compliance as a result of FATCA may happen to have U.S. citizenship but may never have lived or worked in the United States as adults. The National Taxpayer Advocate told us that the high cost of tax preparation amounted to an “advanced penalty” for U.S. individuals who live abroad. In a written testimony to Congress in 2015, she stated that FATCA has created unique challenges for U.S. taxpayers abroad and presented evidence in Volume 1 of the Taxpayer Advocate Service Fiscal Year 2016 Objectives Report to Congress that there was little evidence that foreign filers are any more likely to be non-compliant than taxpayers in the general taxpayer population. The National Taxpayer Advocate specifically identified concerns with FATCA as an area of focus in the Fiscal Year 2016 Objectives Report to Congress, and stated that taxpayers’ rights to a fair and just tax system, and to pay no more than the correct amount of tax, are being adversely affected by FATCA. IRS officials we spoke with indicated they are aware of the difficulties some taxpayers are experiencing with these reporting requirements, but said the agency is required to implement the law. They also said that retirement accounts are usually the primary asset for individuals abroad and that from an individual enforcement perspective, these reporting requirements help to ensure a “line of sight” year over year on participants’ foreign pension arrangements. IRS officials expressed concern that unless U.S. individuals are required to report foreign retirement accounts via Form 8938, they will seek to avoid proper reporting on their tax returns when distributions are made. IRS officials told us they have had extensive conversations about providing a possible exemption from reporting requirements under IRC section 6038D for certain U.S. individuals in foreign countries. IRS decided the ability to review a taxpayer’s foreign retirement data each year through filing a Form 8938 would allow regulators to evaluate whether contributions, earnings, and distributions were being identified and reported accurately. IRS officials stated that the agency’s goal is to build a database with Form 8938 information on individual taxpayers with foreign assets. IRS officials told us that, unlike individuals, foreign financial institutions in many countries are exempt from reporting retirement accounts under FATCA. IRS officials said this is because such foreign retirement accounts are typically at low risk for tax evasion and Treasury officials told us that the exemption for foreign financial institutions was provided to reduce burden on such institutions. This sentiment was echoed by foreign government officials and retirement experts abroad, who said a retirement account is generally at low-risk for tax evasion both because governments regulate retirement accounts and individuals attempting to evade taxes through a retirement account would have to wait many years before seeing any benefit. With respect to IRC section 6038D, according to IRS officials and Form 8938 instructions, if a fair market value is not readily available for a foreign workplace DB plan, it does not have to be included in the taxpayer’s calculation of the aggregate foreign assets used to determine whether the taxpayer meets the threshold to file Form 8938. If other foreign financial assets, in the aggregate, exceed the threshold, IRS officials said an individual must list their DB plan on Form 8938, but may list a zero balance if no distributions have been made. Given that IRS does not always require reporting of foreign retirement plans on Form 8938 if the plans cannot be readily valued, providing a broader exemption for other types of workplace plans or for other appropriate circumstances from the calculation of the foreign asset threshold could help ease the reporting burden on U.S. individuals. This would assist those individuals who hold most of their wealth in the form of foreign retirement savings in other types of workplace retirement plans, to avoid potentially high penalties that could diminish their retirement savings. IRS has not systematically analyzed data from Form 8938 on foreign retirement accounts owned by U.S. individuals. As a result, they may not have evidence showing the effect of these reporting requirements on U.S. individuals who participate in foreign workplace retirement plans, for instance, how many enforcement actions related to retirement accounts resulted from filing Form 8938. Without IRS systematically analyzing Form 8938 data on foreign retirement accounts owned by U.S. individuals, the agency will continue to lack an understanding of how these accounts change over time and if they are definitively low-risk for tax evasion. Understanding the effects of these reporting requirements can provide IRS with information to consider whether IRS could offer individuals some form of exemption from reporting on their foreign retirement accounts. Currently, there is no way for IRS to clearly distinguish different types of accounts being reported on Form 8938. To do so would require the Form 8938 to be revised in order to allow taxpayers to clearly specify that the account being reported is a foreign retirement account or pension. In addition, U.S. individuals participating in foreign workplace retirement plans, many of whom count their retirement savings as their primary financial asset, according to IRS officials, will continue to be caught up in IRS’ enforcement efforts aimed at catching tax evaders. These U.S. individuals may continue to face potentially high tax preparation fees to complete the filing of Form 8938 and may be liable for penalties for failure to report foreign retirement accounts that may pose little or no risk for tax evasion. Absent Specific Treaty Provisions, Current U.S. Tax Law Does Not Exempt from Taxation Transfers in Foreign Workplace Retirement Plans IRS officials told us that U.S. individuals who participate in foreign workplace retirement plans may not realize that a routine transfer of their foreign retirement assets within plans or from one plan to another should be reported as a taxable event, resulting in an incorrect filing and/or potential penalties. Changing jobs and transferring, or “rolling over” retirement savings to another qualified retirement plan is generally a tax- protected transaction for participants in U.S.-based retirement plans. However, IRS officials told us that a U.S. individual who participates in a foreign retirement plan may owe U.S. taxes for similar transfers within or between foreign workplace retirement plans. Retirement plans in some countries routinely initiate administrative transfers of a participant’s retirement savings between accounts within the plan, to the employee’s new plan, or to a designated institution outside the plan when the participant separates from their employer, according to officials in several of our case study locations. However, IRS officials told us the IRC does not recognize foreign retirement plans as tax-qualified plans, and because these plans are not able to meet the criteria for qualification, tax-deferred transfers or rollovers may not be possible unless a tax treaty provides otherwise. IRS generally considers routine administrative transfers of retirement assets that occur between or within foreign retirement plans to be distributions to the participant and therefore taxable income. According to IRS officials with whom we spoke, the transfer of retirement assets within or between plans implies that the participant has some access to and control over their retirement funds. Tax preparers and regulators in three of our case study locations told us that such transfers routinely take place (see appendix III). In these situations, IRS officials told us that deferring taxes on retirement contributions and earnings under IRC section 402(b) pertaining to foreign trusts would no longer be applicable because that section of the IRC does not cover transfers—only contributions and earnings within a given foreign trust. Instead, according to IRS officials, the transfer would generally constitute a “constructive receipt of funds” by the participant and would be reportable and taxable. As a result, a U.S. individual who participates in a foreign retirement plan could owe U.S. tax on the entire amount of their retirement savings when they separate from their employer and their account is transferred to another account within the plan or to a different workplace retirement plan (see fig. 4). Treasury officials said they have been aware of this issue for some years, having discussed it in multiple negotiations with other countries, and have taken steps to incorporate a solution in U.S. model income tax conventions dating as far back as 1996.Treasury officials told us that the 2016 U.S. Model Income Tax Convention includes a clause that would generally exempt from U.S. income tax such transfers if they qualify as tax-deferred transfers under the laws of the other country. According to Treasury officials, few of the treaties currently in force address this issue and many countries do not have tax treaties with the United States. 2016 U. S. Model Income Tax Convention Language Would Exempt Certain Transfers of Foreign Retirement Assets from Taxation The Department of the Treasury developed the U. S. Model Income Tax Convention to be the starting point for negotiating tax treaties with other countries. Language in an actual treaty results from that negotiation and therefore may not include this language. According to Department of the Treasury officials, few treaties currently contain this language. “2016 United States Model Income Tax Convention, Article 17, Paragraph 2(b) Where a citizen of the United States who is a resident of ______ is a member or beneficiary of, or participant in, a pension fund established in _____, the United States may not tax the income earned by the pension fund as income of the individual unless, and then only to the extent that, it is paid to, or for the benefit of, that individual from the pension fund (and not transferred to another pension fund established in _______ in a transfer that qualifies as a tax-deferred transfer under the laws of _______).” IRS officials told us that if no treaty exists between the United States and the country where the U.S. individual is participating in a foreign workplace retirement plan, or the treaty does not specify how to treat these transfers, there is generally no form of transfer that will receive U.S. income tax-deferral. In these situations, IRS officials said, there is no way that the plan can structure the transfer to prevent the U.S. individual who is transferring assets within or between foreign plans from receiving a distribution and being subject to tax liability. Even in cases where a tax treaty is in place, the treaty may not provide special treatment for the transfer of retirement assets. This would be the case in at least two of the five case study locations we examined, where despite a tax treaty in place, we were unable to identify any provisions that address these types of transfers. In these cases, according to IRS, the U.S. individual must fall back on the IRC, which does not provide tax-deferral on such transfers. As a result, a U.S. individual who participates in a foreign workplace plan would lose any tax-deferrals on the transfer. IRS officials and tax preparers told us that the transfer issue can cause tax consequences for holders of foreign retirement assets, but one tax preparer we spoke with noted that U.S. tax laws were not written with foreign retirement plans in mind. As a result, tax preparers said it can be difficult to determine how to report foreign workplace retirement assets under the IRC, making routine administrative transactions costly for U.S. individuals who participate in these plans. They said this is because some or all of the account balance may be subject to tax and retirement account asset growth would be lower due to the loss of tax-deferral. Each time retirement assets are transferred, the transfer may be viewed as a distribution, and new contributions and growth could be subject to tax and a loss of tax-deferral. IRS officials also told us that the potential taxation of transfers between foreign plans may cause some individuals to avoid consolidating foreign retirement accounts. Renegotiating a tax treaty can be time consuming and, according to Treasury officials, is unlikely to happen based on one issue, such as the transfer of retirement savings abroad. Treasury officials in the Office of Tax Policy said that the agency’s approach to address these transfers would be to evaluate the issue on a treaty-by-treaty basis. However, this approach may not provide relief because there is no guarantee the country negotiating a treaty with the United States will agree to include provisions for transferring retirement savings on a tax-deferred basis. In order to provide more immediate relief, these Treasury officials said Congress could pass legislation that would allow routine account transfers between two foreign workplace retirement plans in the same country to be free from U.S. tax if that country has a tax treaty with the United States. However, they cautioned that such efforts should be focused on foreign retirement plans that have already been examined by Treasury, for example, through the process of negotiating a tax treaty or as defined in FATCA IGAs, in order to avoid creating a tax evasion loophole. For example, foreign workplace retirement plans could be defined as those recognized by an existing tax treaty or other plans as deemed appropriate by Treasury’s Office of Tax Policy. According to Treasury officials, transfers within or between such plans in the same country could be protected from unnecessary taxation by, for example, modifying Section 402(b) or other provisions of the IRC. Officials said that without legislation, U.S. individuals who participate in foreign workplace retirement plans must follow current law, which does not provide tax- deferral for transfers within or between foreign plans, even those that may be eligible for tax-deferred contributions and earnings in the foreign jurisdiction. However, by changing the IRC, Congress can ensure that U.S. individuals who participate in foreign workplace retirement plans can consolidate their accounts in a tax-deferred manner without being taxed on the entire balance when their account is transferred. Conclusions Plan participants in the current workplace retirement plan environment can accumulate multiple retirement accounts and possibly lose track of them over their careers. The shift to DC plans and the mobility of the American workforce have led to an increase in the number of workplace retirement accounts, with many workers having multiple accounts over the course of their careers. Yet currently, with millions of small retirement accounts left behind by participants with previous employers, plan sponsors are experiencing challenges locating missing participants. DOL has agreed to evaluate the possibility of convening a taskforce to consider the establishment of a national pension registry, in part to address the difficulty of linking missing participants to their former accounts. However, until this effort brings results or another comprehensive solution to unclaimed accounts emerges, there are a variety of improvements federal agencies may make in the short term to help eliminate the inefficiencies in the current system that may reduce participants’ retirement savings. Since DOL audit findings show that ongoing plans have challenges staying in touch with missing participants, and DOL has provided guidance on missing participants for terminating DC plans, providing such guidance for ongoing DC plans will help ensure that separated participants will receive information about their benefits. In addition, IRS guidance on tax withholding does not address distributions of small unclaimed accounts sent to nonresponsive participants that are not always received by those participants. Some stakeholders mistakenly believe that IRS automatically credits all taxes withheld from such distributions toward taxes due. Following IRS guidance, plans generally withhold taxes on cash-outs from such accounts that the participant may not receive. By reviewing the issue of distributions made to participants who are unlikely to receive them, IRS has an opportunity to issue guidance clarifying the applicable tax withholding requirements in those situations. IRS also has the potential to offer a service that delivers letters that participants are likely to open, is trusted by plan fiduciaries, and can help connect missing participants with their benefits. IRS was forwarding fewer than 50 letters at a time for plan sponsors at no charge, but decided to stop forwarding letters about unclaimed accounts in 2012. IRS can consider again helping connect participants with unclaimed accounts using the letter forwarding program. Lastly, IRS and SSA can take steps to address situations in which sponsors fail to update data to reflect payment of retirement accounts, rendering the data unreliable. Under the existing agreement between IRS and SSA with respect to the Form 8955- SSA data, the agencies can take steps to ensure participants have a more reliable source of information on their benefits in the future. U.S. individuals who work abroad and participate in a foreign workplace retirement plan face challenges with reporting their accounts. Managing such accounts can be costly as individuals use expensive tax preparers for reporting their foreign retirement savings to IRS. These U.S. individuals are required to pay taxes on their worldwide income, but can become caught in a web of complex U.S. tax requirements governing how they report their foreign workplace retirement savings. By providing guidance on how to appropriately report foreign workplace retirement accounts, IRS can help U.S. individuals comply with these requirements and minimize their reporting burden. IRS can also initiate a systematic analysis of Form 8938 data on foreign retirement accounts owned by U.S. individuals. Such data would help IRS gain a better understanding of how these accounts change over time, and to determine if they pose a low-risk for tax evasion. The outcome of this analysis could allow IRS to consider offering these individuals an exemption from reporting requirements on their foreign retirement accounts, further easing the burden U.S. individuals face reporting their foreign retirement assets. Lastly, transferring accounts between foreign retirement plans can have negative tax consequences that threaten the ability of U.S. individuals abroad to save for retirement. Congress may wish to consider whether it can assist U.S. individuals who participate in foreign workplace retirement plans by permitting these individuals to transfer their retirement savings to a different account within the plan or to another foreign workplace retirement plan on a tax-deferred basis when they change jobs or separate from their foreign employer. Doing so would permit these U.S. individuals in foreign workplace retirement plans to receive the tax-deferred benefits available to other U.S. plan participants who reside in the United States and who participate in qualified retirement plans. Matter for Congressional Consideration We are making the following matter for congressional consideration. Congress should consider legislation modifying the Internal Revenue Code to allow routine account transfers within the same foreign workplace retirement plan or between two foreign workplace retirement plans in the same country to be free from U.S. tax in countries covered by an existing income tax treaty that provides for favorable U.S. tax treatment of foreign workplace retirement plan contributions. Recommendations We are making a total of seven recommendations, including one to DOL, five to IRS, and one to SSA. a. The Secretary of Labor should issue guidance on the obligations under ERISA of sponsors of ongoing plans to prevent, search for, and pay costs associated with locating missing participants. (Recommendation 1) b. The IRS Commissioner should review taxation issues relating to distributions involving incorrect participant addresses and uncashed benefit checks and clarify for the public the Internal Revenue Code’s requirements in these circumstances. (Recommendation 2) c. The IRS Commissioner should consider revising the letter forwarding program in a cost-effective manner to again provide information on behalf of plan sponsors on unclaimed retirement accounts to participants. (Recommendation 3) d. The IRS Commissioner should clarify how U.S. individuals are to report their foreign retirement accounts. The clarification could include addressing how these accounts should be designated and how the taxpayer should report contributions, earnings, and distributions made from the account. (Recommendation 4) e. The IRS Commissioner should systematically analyze data reported through Form 8938 filings on foreign retirement accounts owned by U.S. individuals with the goal of developing an evidence-based understanding of how these accounts change over time and what level of risk these accounts pose for tax evasion. To assist with this analysis, IRS should consider revising Form 8938 to more clearly distinguish between retirement accounts and other types of accounts or assets being reported by taxpayers under current reporting requirements. (Recommendation 5) f. The IRS Commissioner should take steps to improve the likelihood that the Notice of Potential Private Pension Benefit Information corresponds to actual retirement benefits in the future, for example, by working with the Social Security Administration as necessary. (Recommendation 6) g. The Social Security Administration Commissioner should take steps to improve the likelihood that the Notice of Potential Private Pension Benefit Information corresponds to actual retirement benefits in the future, for example, by working with IRS as necessary. (Recommendation 7) Agency Comments and Our Evaluation We provided a draft of this report to the Department of Labor, the Department of the Treasury, the Internal Revenue Service, the Social Security Administration, the Pension Benefit Guaranty Corporation, and the U.S. Department of State. DOL, Treasury and IRS, and PBGC provided technical comments, which we have incorporated where appropriate. DOL, IRS, SSA, and PBGC also provided formal comments, which are reproduced in appendices IV, V, VI, and VII, respectively. State did not have any comments. DOL agreed with our recommendation that additional guidance may be helpful to aid plan sponsors and plan fiduciaries of ongoing plans in meeting their existing fiduciary obligations to search for missing participants and to pay benefits. SSA agreed with our recommendation to take steps to improve the likelihood that the Notice of Potential Private Pension Benefit Information corresponds to actual retirement benefits in the future, for example, by working with IRS as necessary. In its written comments, IRS stated that it generally agreed with the report and its findings. IRS specifically cited that the report identifies several challenges for participants to manage their retirement savings, such as updating former employers with address changes to continue receiving information about retirement plan accounts with former employers and responding to former employers regarding retirement plan accounts. IRS also stated that U.S. individuals participating in foreign retirement plans often do not know how to correctly report foreign retirement accounts and associated income due to complex federal requirements and treaty provisions governing the taxation of foreign retirement accounts. This recognition by IRS of the complex federal requirements and treaty provisions governing the taxation of foreign retirement accounts is in line with GAO’s concerns about U.S. individuals with foreign workplace retirement accounts having trouble with routine account transfers within the same foreign workplace retirement plan or between two such plans in the same country. We have asked Congress to consider modifying the Internal Revenue Code to allow routine account transfers within the same foreign workplace retirement plan or between two foreign workplace retirement plans in the same country to be free from U.S. tax in countries covered by an existing income tax treaty that provides for favorable U.S. tax treatment of foreign workplace retirement plan contributions. Congress’ ability to modify the Internal Revenue Code in such a way can help U.S. individuals participating in foreign workplace plans to better save for retirement by allowing them to consolidate accounts in a tax- deferred manner without being taxed on the entire balance when their account is transferred. IRS agreed with two of our recommendations to improve the management of retirement savings. Specifically, IRS agreed to review taxation issues relating to distributions involving incorrect participant addresses and uncashed benefit checks and to clarify for the public the Internal Revenue Code’s requirements in these circumstances. We believe that IRS’ consideration of this recommendation and any subsequent actions the agency takes to clarify the issue will help to address questions about tax withholding from distributions in situations where the participant may be missing or where a distribution check remains uncashed after a period of time. IRS also agreed to work to improve the likelihood that the Notice of Potential Private Pension Benefit Information corresponds to actual retirement benefits in the future, and agreed to take steps to ensure that the data reported on Form 8955-SSA are accurate and to advise plan sponsors of any changes to reporting these data. We commend IRS for recognizing the importance of addressing this issue for taxpayers and for its willingness to take steps to ensure the accuracy of data reported by plans in the United States on vested benefits belonging to separated employees. Lastly, IRS agreed with our recommendation to clarify how U.S. individuals are to report their foreign retirement accounts, which could include how the taxpayer should report contributions, earnings, and distributions made from the account. We encourage IRS to take the necessary steps to dispel any confusion U.S. individuals may have over how to properly classify and report their foreign retirement accounts on a U.S. tax return—such clarification should help ensure that these taxpayers can meet their tax reporting obligations. IRS disagreed with two of our recommendations, citing the limited number of IRS staff and resources needed for the agency to implement these recommendations. First, IRS disagreed with our recommendation to consider revising the letter forwarding program in a cost-effective manner to again provide information on behalf of U.S. plan sponsors on unclaimed retirement accounts to participants. IRS commented that the IRS address of record for a participant would likely be of no greater value than addresses available through alternatives such as commercial locator services. However, our report does not cite the accuracy of IRS addresses, but rather other benefits that make a program revision worth considering, specifically the likelihood that individuals will open IRS correspondence, and the trust DOL places in the service as way for plan fiduciaries to meet their obligations. IRS also stated that the limited number of IRS staff and resources impact the feasibility of reinstating this program for plan participants. GAO continues to believe that expanding the letter forwarding program would be beneficial, and we encourage IRS to consider cost-effective ways to do so. IRS also disagreed with our recommendation to analyze data provided through Form 8938 filings on foreign retirement accounts owned by U.S. individuals with the goal of developing an evidence-based understanding of how these accounts change over time and what level of risk these accounts pose for tax evasion. Our recommendation further stated that IRS should consider revising Form 8938 to assist with this analysis. In its comments, IRS did not disagree with this recommendation on its merits; IRS only cited a lack of resources to implement the recommendation. Specifically, IRS noted that although the modification to the Form 8938 suggested in this recommendation may seem minor, systemically collecting and analyzing the data would require resources beyond those currently available to IRS. However, as we describe in the report, IRS indicated to us that they already collect foreign account filing data through the Form 8938 and that the current reporting requirements help the agency to “keep a line of sight” on U.S. individuals’ foreign pension arrangements. IRS told us that without such data being reported, U.S. individuals with foreign retirement accounts may seek to avoid proper reporting on their tax returns when distributions are made. However, without agreeing to take steps to analyze these data reported by taxpayers, the question remains why IRS continues to collect such information—which we show in the report to present a substantial reporting burden on taxpayers—if the agency has no plan to analyze the data in order to make an informed decision about the risk for tax evasion that such accounts present. It is also unclear to us how IRS would maintain a line of sight on foreign retirement accounts belonging to U.S. individuals without analyzing the data reported by taxpayers on such accounts. While we recognize that resource limits can impede an agency from taking on additional work and projects, we continue to believe that when staff and resources become available, IRS should modify the form and conduct a systematic analysis of these data—data that current law requires taxpayers to report—in order to assess the risk of tax evasion that foreign retirement accounts pose. Without such an analysis, IRS will have no basis to reach an evidence-based understanding of how these accounts change over time and what level of risk they pose for tax evasion. Further, as we have shown in the report, this reporting can be costly for U.S. individuals and could potentially lead to a decrease in their retirement savings. Without such an analysis by IRS, U.S. individuals who own foreign retirement accounts will continue to face these substantial reporting burdens without the knowledge that the data they are required to provide will be put to good use by the federal government. As agreed with your offices, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies to the Secretary of Labor, Secretary of the Treasury, Commissioner of Internal Revenue, Director of the Pension Benefit Guaranty Corporation, Acting Commissioner of the Social Security Administration, the Secretary of State, and other interested parties. In addition, the report will be available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-7215 or jeszeckc@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix VIII. Appendix I: IRS Instructions on Reviewing Tax Treaty Provisions Related to Foreign Retirement Accounts IRS advises U.S. taxpayers to review tax treaty provisions carefully to better understand how to report their foreign income, including the distribution of savings from foreign retirement accounts. IRS specifically advises taxpayers to read the residency article in a tax treaty to find any special rules pertaining to reporting and taxing foreign income, including distributions from foreign workplace retirement plans. When deciding whether a tax treaty applies to a taxpayer, the taxpayer should first identify their tax residency (Article 4 under most treaties). According to IRS, a taxpayer’s residency determines how treaty articles on pensions and annuities will be applied and taxpayers should use the domestic laws of each country to identify residency. If, after applying the domestic law of each country, the taxpayer determines they are a resident of both countries, the tiebreaker rules of the applicable treaty are applied to determine residency based on the country in which the taxpayer has closer personal and economic relations, the country of habitual abode for the taxpayer, or the country in which the taxpayer is a national, according to IRS. If none of the above tiebreaker rules apply, the treaty generally provides that residency will be decided by the competent authorities of each country upon request by the taxpayer. Taxpayers are also advised to read all the protocols of the treaty to see if the residency rules have been amended by a later protocol. As a general rule, according to IRS, the pension/annuity articles of most tax treaties allow the country of residence (as determined by the residency article) to tax the pension distribution or annuity under its domestic laws, unless the tax treaty provides an exception to that rule. According to IRS, some treaties, for example, provide that the country of residence may not tax amounts that would not have been taxable by the other country if the individual was a resident of that country. There also may be special rules for lump-sum distributions. If the taxpayer is a U.S. citizen, IRS guidance provides that they also may need to refer to the “saving clause” (typically found in Article 1) for special rules that allow the United States to tax income in some cases as if the treaty had not entered into force. Appendix II: Estimated Number of U.S. Citizens Living Abroad Researchers and federal officials have identified a range of estimates for U.S. citizens living outside the United States. (See fig. 5). The U.S. Department of State (State) estimates that as of April 2015, 8.7 million U.S. citizens live abroad. Table 2 shows estimates of the number of U.S. citizens living abroad by geographic area that State’s Bureau of Consular Affairs recently released to the Federal Voting Assistance Program. Appendix III: Efforts to Link Participants to Dormant or Unclaimed Retirement Accounts in Case Study Locations We gathered the information in this appendix for each case study location by reviewing relevant documentation, publicly available research and reports, and interviewing relevant stakeholders, including government officials, plan sponsors, and service providers. We did not conduct an independent legal analysis to verify the information provided about the laws or regulations in the locations we selected for this study. Instead, we relied on appropriate secondary sources and interviews with relevant officials to support our work. We provided this information to appropriate officials in each case study location for their confirmation. In the five case study locations we reviewed, participants, including U.S. individuals working in those locations, stay connected to their foreign workplace retirement savings through centralized institutions, direct contact with plans or government agencies, or through public pension registries. The low prevalence of unclaimed retirement accounts that we found in these locations is likely due, in part, to participants using these mechanisms to stay connected to their retirement savings. Unclaimed Retirement Accounts Are Maintained Within Centralized Institutions or a Participant’s Former Employer’s Plan In two of the five locations we reviewed, Australia and Switzerland, plans transfer dormant accounts belonging to separated employees to a centralized institution that is actively monitored by regulators. These accounts generally remain within these institutions until claims for benefits are made by the participant. For example, Swiss officials told us that in the event of a change of employment, the pension scheme (i.e., plan) of an insured person (i.e., participant) transfers the accumulated assets on behalf of that person to the pension scheme of the new employer. Vested benefits institutions are used to hold the assets when a person ceases to be subject to occupational benefits (workplace retirement) plans owing to termination of employment, e.g. in connection with a career break or being laid off. In these cases, the pension scheme mandatorily transfers the assets to a vested benefits institution. This procedure ensures that the accumulated assets remain blocked in the pension cycle until the insured person joins a new pension scheme or an insured event occurs (old age, disability, or death). Once the person recommences employment and thus becomes subject to mandatory occupational benefits plans again, the termination benefits must be transferred by the vested benefits institution to the new pension scheme. If the pension scheme member or insured person does not become re-employed, the vested benefits institution keeps the assets until an insured event occurs (retirement, disability, or death). Further, if a separated employee fails to inform their former plan that they have a new employer, the participant’s former plan automatically transfers the account after 6 months and within 2 years to the Substitute Occupational Benefit Institution. This institution is a non- profit entity that the Swiss federal government commissioned in 1985; it works closely with the Swiss federal government to maintain Swiss retirement assets for participants and is charged with certain governmental responsibilities. For retirement assets transferred to the Substitute Occupational Benefit Institution, account balances are not merely preserved until claimed or transferred, but grow according to returns on Switzerland’s central fund investments. Figure 6 describes how accounts of separated employees in Switzerland are transferred to designated locations when they become dormant or unclaimed. Swiss plans also transfer accounts belonging to separated employees to the employee’s new plan once they receive instruction from the employee. This transfer along with account transfers to the Substitute Occupational Benefit Institution or a vested benefits institution, such as a bank or insurance company, contributes to the low prevalence of lost retirement accounts in Switzerland because participants do not accumulate multiple retirement accounts with different plans when changing jobs throughout their career. In Australia, plans (also referred to as schemes or super funds) transfer unclaimed super accounts belonging to lost members (e.g., separated employees) to a centralized government institution, the Australian Taxation Office (ATO). These accounts generally remain within the ATO until claims for benefits are made by the member. While their money is being held by the ATO it earns interest at the consumer price index rate. In three of the five locations GAO reviewed—Canada, Hong Kong, and the UK—the participant’s former employer’s plan maintains dormant accounts until claimed or transferred to a new plan. For example, in Hong Kong, according to its retirement schemes (plan) regulator, the Mandatory Provident Fund Schemes Authority (MPFA), whenever employees, including U.S. individuals working in Hong Kong, change to a new employer, they need to open a new Mandatory Provident Fund (MPF) “contribution account” under the MPF scheme in which the new employer participates to accumulate MPF contributions in respect of the new employment. If an employee who has ceased employment with an employer does not take action to transfer the benefits accrued from the previous employment to the new “contribution account” with their new employer’s scheme (i.e., plan) or a “personal account” in an MPF scheme of the employee’s choice, their former employer’s scheme will automatically transfer their accumulated MPF benefits from the contribution account to a personal account within the original scheme for continuous investment. Government officials told us that MPF schemes keep the benefits of the scheme members (i.e., plan participants) within the scheme until the scheme member returns to make a claim or to issue instructions to transfer benefits in the account to another MPF scheme. The MPFA advises scheme members that failing to consolidate the MPF benefits accumulated from previous employments can result in accumulating multiple MPF accounts that can be difficult to manage—this can result in accounts becoming lost over time. To address this challenge, the MPFA conducts regular publicity programs and publishes pamphlets reminding scheme members that when they change employers they should consolidate the benefits under the previous employment to any existing personal accounts or to the new contribution account under the MPF scheme of their new employer. In two of the five locations we reviewed—Australia and Switzerland— plans are required to regularly report to regulators on unclaimed accounts, missing participants, and account transfers made for separated employees, including those made on behalf of U.S. individuals. For example, in Australia, plans are required to communicate information on unclaimed accounts to the ATO. Specifically, every 6 months plans are required to identify and report members who meet the definition of “lost” and unclaimed accounts considered “uncontactable or inactive” to the ATO. Further, plans are also required to transfer unclaimed accounts to the ATO when certain unclaimed super money criteria are met. In Switzerland, before the end of January each year, occupational benefits institutions and institutions that manage vested benefits accounts or policies are required to report to the 2nd Pillar Central Office all persons for whom assets were held in December of the previous year. Plans in two of the locations we reviewed provide separated participants information on account transfers that can help them stay connected to their retirement savings. For example, in Hong Kong, MPF schemes (plans) provide a transfer statement to members once the transfer of benefits to another MPF scheme is completed. The MPF scheme that receives the transfer must, as soon as practicable after receiving the transferred benefits, provide the member written notice confirming the transfer and stating the monetary value of those benefits. In Switzerland plans must regularly contact their participants and if unable to do so, must inform the 2nd Pillar Central Office, who will try to reestablish contact between the plan and their participants. Unclaimed Retirement Accounts Are Accessed Through Contact with Government Agencies or Through Public Pension Registries Participants in three of the five locations we reviewed can access information on their retirement accounts by contacting a government agency. According to government and retirement plan officials in Australia, participants can access their retirement account details by logging onto the myGov platform, which is a secure way to access government online services. Participants, including U.S. individuals, who have registered online via myGov and have their personal accounts linked to ATO online services can view their retirement accounts online and can claim their money at any time. For those that choose not to register for myGov, they can use the Departing Australia Superannuation Payment online service to claim their super funds once they have departed Australia and their visa has ceased to be in effect. In Hong Kong, MPFA officials told us that scheme members seeking information on their personal accounts or on unclaimed retirement benefits with any MPF scheme (plan) can approach MPFA to request a search of the Personal Accounts Register or Unclaimed Benefits Register, respectively. The MPFA’s website includes instructions for initiating these inquiries. In Switzerland, government officials told us that participants, including U.S. individuals, can directly contact the 2nd Pillar Central Office, which can locate all the institutions holding vested benefits on the participant’s behalf. In two of the five case study locations, Australia and the UK, participants can access information on their retirement accounts by using pension registries or other government supported services. For example, in the UK, the government provides all participants, including U.S. individuals with a UK retirement account, access to the Pension Tracing Service to help them locate their lost retirement accounts. The UK government has also established other organizations and services to help participants locate their lost retirement accounts. The Pensions Advisory Service is an independent organization that is funded by the UK government. Officials told us that the service was implemented because retirement accounts and pensions in the UK had become excessively complicated. The service sometimes receives questions from participants living abroad, such as in the United States, or from U.S. individuals living in the UK. Service officials told us that it is particularly challenging for these foreign participants to know how to repatriate their retirement benefits and to locate missing retirement accounts. Government officials told us that the UK government is committed to ensuring that members of the public can access good-quality, free-to-client, impartial financial guidance and debt advice which is currently provided by three different organizations. These officials said that a bill was introduced in June 2017 that would set up a new single financial guidance body to provide guidance and information on all matters relating to occupational and personal pensions. Officials said they expect that this single financial guidance body to go live no earlier than October 2018. The UK government is also currently developing a new pension online tool, the Pensions Dashboard. The dashboard is being developed as a joint project between the UK government and the country’s retirement industry; 17 of the UK’s largest pension firms developed a prototype demonstrating that the technology for the dashboard works. The goal of the dashboard is to allow participants to log into one portal to locate all of their pension data, including information on the value and the location of different retirement savings accumulated throughout their career. Currently, a UK ID verification system is available to UK residents to review their tax bills and other financial information online, and officials are considering permitting participants to use this system with proper credentials to access the dashboard. In time, UK government officials said that the dashboard may replace the UK’s Pension Tracing Service, but not for many years. Other officials added that they are uncertain whether the dashboard will include all plans. One concern is that many lost accounts may be with old defined benefit plans or small defined contribution plans that do not have online systems that can be integrated into the dashboard. As a result, some of the plans most likely to have lost participants may also be the least likely to participate in the dashboard. Appendix IV: Comments from the Department of Labor Appendix V: Comments from the Internal Revenue Service Appendix VI: Comments from the Social Security Administration Appendix VII: Comments from the Pension Benefit Guaranty Corporation Appendix VIII: GAO Contact and Staff Acknowledgments GAO Contact Charles Jeszeck, (202) 512-7215 or jeszeckc@gao.gov. Staff Acknowledgments In addition to the contact named above, Tamara Cross (Assistant Director), Ted Burik (Analyst-in-Charge), Ted Leslie, and Jessica Rider made key contributions to this report. Also contributing to this report were Susan Aschoff, James Bennett, Amy Bowser, Sherwin Chapman, Sarah Cornetto, Brian James, Kristy Kennedy, Jonathan McMurray, Sheila McCoy, Jennifer Lutzy McDonald, Dan Meyer, Mimi Nguyen, Amrita Sen, Deborah Signer, Andrew Stephens, Walter Vance, Kathleen Van Gelder, Adam Wendel, and Seyda Wentworth.

Why GAO Did This Study Saving for retirement can be difficult. However, when participants lose their workplace retirement accounts when they change employers or participate in a workplace retirement plan abroad they can encounter additional challenges in securing adequate retirement savings. GAO was asked to review steps federal agencies might take to assist participants with these challenges. This report examines key challenges U.S. participants face with: (1) unclaimed retirement accounts in the United States, and (2) complying with U.S. tax reporting requirements on their foreign retirement savings. GAO reviewed relevant federal laws and regulations, and reviewed selected tax treaties. GAO interviewed stakeholders in the United States and in Australia, Canada, Hong Kong, Switzerland, and the United Kingdom—chosen because these locations host relatively large populations of U.S. individuals and have well-developed workplace retirement systems. What GAO Found Plan participants in the United States face challenges after they change jobs, including not receiving communications from their plan sponsor and being vulnerable to unforeseen tax consequences that can result in a loss of retirement savings. GAO previously reported that when participants leave savings in a plan after separating from a job, the onus is on them to update former employers with their new address and to respond to their former employer's communications. GAO found that although an employer may incur costs searching for separated participants, there are no standard practices for the frequency or method of conducting searches. GAO reported that from 2004 through 2013, over 25 million participants in workplace plans separated from an employer and left at least one retirement account behind, despite efforts of sponsors and regulators to help participants manage their accounts. Department of Labor (DOL) officials told GAO that some sponsors do not search for participants when disclosures are returned as undeliverable. DOL has issued guidance on searching for missing participants for some plans that are terminating, but guidance does not exist on what actions DOL expects ongoing plan sponsors to take to keep track of separated participants. A key element of DOL's mission is to protect the benefits of workers and families. However, without guidance on how to search for separated participants who leave behind retirement accounts, sponsors may choose to do little more than remove unclaimed accounts from the plan when possible, and workers may never recover these savings. Stakeholders told GAO that U.S. individuals who participate in foreign workplace retirement plans face challenges reporting their retirement savings for tax purposes because of complex federal requirements governing the taxation of foreign retirement accounts and a lack of clear guidance on how to report these savings. For example, stakeholders told GAO it is not always clear to U.S individuals or their tax preparers how foreign workplace retirement plans should be reported to the Internal Revenue Service (IRS) and the process for determining this can be complex, time-consuming, and costly. In the absence of clear guidance on how to correctly report these savings, U.S. individuals who participate in these plans may continue to run the risk of filing incorrect returns. Further, U.S. individuals in foreign retirement plans also face problems transferring retirement savings when they switch jobs. In the United States, transfers of retirement savings from one qualified plan to another are exempt from U.S. tax. However, foreign plans are generally not tax-qualified under the Internal Revenue Code, according to IRS officials, and such transfers could have tax consequences for U.S. individuals participating in foreign retirement plans. Officials from the Department of the Treasury (Treasury) told GAO that a change to the U.S. tax code could improve the tax treatment of transfers between foreign retirement plans that Treasury has already examined. Without action to address this issue, U.S. individuals may not consolidate their foreign retirement accounts or may have to pay higher U.S. taxes on transfers than taxpayers participating in qualified plans in the United States, threatening the ability of U.S. individuals to save for retirement abroad. What GAO Recommends GAO recommends Congress consider addressing taxation issues affecting the transfer of retirement assets between plans within the same foreign country. GAO is making seven recommendations, including that DOL issue guidance to help ongoing plan sponsors search for separated participants, and that IRS issue guidance to clarify how U.S. individuals should report foreign retirement savings to the IRS. The agencies generally agreed with GAO's recommendations. IRS disagreed with two of GAO's recommendations.

gao_GAO-18-563T

gao_GAO-18-563T_0

Background TSA Processes for Allocating TSOs across Airports TSA allocates TSOs to airports using its Resource Allocation Plan, which is intended to provide each airport with the optimum number of TSOs needed to screen passengers for threats to aviation security, such as prohibited and other potentially dangerous items. To implement passenger screening and pursue efficient operations, in addition to relying on TSOs, TSA works with officials from airlines and airports, as well as officials from associations that represent airlines and airports. At airports, FSDs and their designees work with individual airport operators and airlines to, among other things, adjust TSA resources (i.e., TSOs and screening assets such as metal detectors) in response to increases in passenger throughput at each checkpoint, and monitor passenger wait times at checkpoints. At TSA headquarters, the Office of Security Operations (OSO) has primary responsibility for operation of the Resource Allocation Plan and allocation of TSOs across airports. To allocate staff to the nearly 440 TSA-regulated airports in the United States, OSO is to use a combination of computer-based modeling and line-item adjustments based on airport- specific information. First, the agency is to work with a contractor to evaluate the assumptions—such as rates of expedited screening—used by the computer-based staffing allocation model to determine the optimal number of TSOs at each airport based on airport size and configuration, flight schedules, and the time it takes to perform checkpoint and baggage screening tasks. Second, after the model has determined how many TSOs are required for each airport, headquarters-level staff are to make line item adjustments to account for factors such as differences in staff availability and training needs that affect each airport. In 2007, we reviewed the Resource Allocation Plan (referred to as the Staffing Allocation Model at that time) and recommended, among other things, that TSA establish a mechanism to ensure periodic assessment of the assumptions, such as passenger and checked baggage screening rates, underlying the plan. TSA agreed with the recommendation, and in December 2007 developed and implemented a plan to periodically assess the plan’s assumptions. TSA Processes for Collecting Wait Time and Throughput Data at Airports At each airport, TSA is to collect throughput data on the number of passengers screened under both expedited and standard screening and monitor passenger wait times at screening checkpoints. TSA airport officials are to submit passenger throughput and wait time data on a daily basis to OSO’s Performance Management Division at TSA headquarters, which compiles the data through the Performance Measurement Information System, TSA’s web-based data collection system. TSA required FSDs and their designees to collect actual wait times from 2002 through 2007 and beginning again in July 2014. From 2008 through June 2014, TSA required that FSDs collect data on wait time ranges, such as between 20 to 29 minutes or greater than 30 minutes. TSA Information Sharing Efforts with Stakeholders In February 2018, we reported that TSA headquarters officials have taken steps intended to improve information sharing with stakeholders—officials from airlines and airports, as well as officials from associations that represent airlines and airports—about staffing and related screening procedures at airports. For example, we reported that TSA holds daily conference calls with stakeholders at selected airports intended to ensure timely communication and to help identify and address challenges in airport operations such as increases in passenger wait times. Additionally, we reported that TSA conducted a series of presentations and meetings to discuss the Resource Allocation Plan, security enhancements at airports, and airport screening processes, among other things. TSA Uses Passenger Wait Time and Throughput Data to Monitor Airport Operations on a Daily Basis In February 2018, we reported that TSA collects passenger wait time and throughput data and uses those data to monitor daily operations at airports. TSA’s Operations Directive (directive), Reporting Customer Throughput and Wait Times, provides instructions for collecting and reporting wait time and passenger throughput data for TSA screening lanes. Regarding wait time data, according to the directive, FSDs or their designees at all Category X, I, and II airports must measure wait times every operational hour in all TSA expedited and standard screening lanes. The directive requires wait times to be measured in actual time, using a verifiable system such as wait time cards, closed circuit television monitoring, or another confirmable method. The directive indicates that wait times should be measured from the end of the line in which passengers are waiting to the walk through metal detector or advanced imaging technology units. According to TSA officials at that time, at the beginning of each hour, wait time cards are handed to passengers at the end of the checkpoint line and are collected when a passenger reaches the metal detector or imaging unit. Closed circuit television is monitored from a location other than the checkpoint, such as at the airport’s coordination center. According to TSA headquarters officials, TSA does not require FSDs or their designees to collect wait times from a statistical sample of passengers throughout the hour, but rather requires that one wait time is collected for every operational hour in all screening lanes. If more than one wait time is collected during the hour, the directive indicates that the maximum wait time should be reported. TSA officials at airports we visited for our February 2018 report stated that TSOs return completed wait time cards to supervisors, who then enter the information into a shared spreadsheet and eventually into the Performance Measurement Information System. Each hour’s reported wait time is then applied to all of a lane’s throughput for that given hour. FSDs or their designees at Category III and IV airports may estimate wait times initially, but the directive requires them to measure actual wait times when wait times are estimated at 10 minutes or greater. The directive also requires FSDs or their designees to collect passenger throughput data directly from the walkthrough metal detectors and advanced imaging technology units. According to TSA headquarters officials, the machines have sensors that collect the number of passengers who pass through each hour, and TSOs retrieve the data directly from the units. All airports regardless of category are required to enter their wait time and throughput data daily into the information system no later than 3:30 AM Eastern Time of the next calendar day so that the data can be included in the morning’s Daily Leadership Report (discussed in more detail below). To monitor operations for all airports, TSA compiles a daily report utilizing a variety of data points from the information system, including wait time and throughput data. The Office of Security Operations’ Performance Management Division disseminates the Daily Leadership Report to TSA officials, including regional directors and FSDs and their designees every morning detailing the previous day’s wait times and throughput figures, among other data points. The Performance Management Division includes a quality assurance addendum with each Daily Leadership Report, indicating missing or incorrect data, to include wait time and throughput data, and TSA has procedures in place intended to ensure officials at the airports correct the data in the Performance Measurement Information System within 2 weeks. In addition to the Daily Leadership Report, we reported that TSA utilizes wait time and throughput data to monitor airport operations at 28 airports in near real time. In May 2016, TSA established the Airport Operations Center partly in response to the long screening checkpoint lines in the spring of 2016 at certain airports. The center conducts near real time monitoring of the operations of 28 airports that, according to TSA headquarters officials, represent the majority of passenger throughput nationwide or are operationally significant. TSA requires the 28 airports monitored by the center to enter passenger wait time data and throughput data hourly (whereas the remaining airports are only required to submit data once daily, by 3:30 AM Eastern Time, as described above) so that officials can monitor the operations in near real time. In addition, TSA officials at airports are required to report to the center when an event occurs—such as equipment malfunctions, weather-related events, or unusually high passenger throughput—that affects airport screening operations and results in wait times that are greater than TSA’s standards of 30 minutes in standard screening lanes or greater than 15 minutes in expedited screening lanes. If an airport is undergoing a period of prolonged wait times, we found that officials at the Airport Operations Center reported coordinating with the Regional Director and the FSD to assist in deploying resources. For example, over the course of the summer of 2016, after certain airports experienced long wait times in the spring of 2016 as confirmed by our analysis, the center assisted in deploying additional passenger screening canines and TSOs to those airports that experienced longer wait times. The center disseminates a morning and evening situational report to TSA airport-level officials and airport stakeholders summarizing nationwide wait times, highlighting wait times at the top airports and any hot spots (unexpected passenger volume or other operational challenges) that may have occurred since the most recent report was issued. In addition to the near real-time monitoring of 28 airports, the center also monitors operations at all other airports and disseminates information to airports and stakeholders as needed. For our February 2018 report, to determine the extent to which TSA exceeded its wait time standards, we analyzed wait time data for the 28 airports monitored by the Airport Operations Center for the period of January 2015 through May 2017 for both standard and expedited screening. Our analysis showed that TSA met its wait time standard of less than 30 minutes in standard screening at the 28 airports 99.3 percent of the time for the period of January 2015 through May 2017. For expedited screening for the same time period at the same airports, we found that 100 percent of the time passengers were reported to have waited 19 minutes or less. Additionally, our analysis confirmed that the percentage of passengers in standard screening who waited over 30 minutes increased in 2016 during the months of March, April, and May as compared to 2015 at all 28 airports. Our analysis also confirmed that reported wait times increased in the spring of 2016 at selected airports, as mentioned in the news media. For example, in May 2016, approximately 22 percent of passengers at Chicago O’Hare International airport and 26 percent of passengers at Chicago Midway International airport waited over 30 minutes in standard screening as opposed to zero percent for both airports in May 2015, which accounted for the longest wait times in the spring of 2016. These two airports were part of the 28 airports for which we analyzed wait time data for the period of January 2015 through May 2017. TSA Airport Officials Reported Using a Variety of Tools to Respond to Increases in Passenger Wait Times and Throughput In February 2018, we reported that FSDs and their staff at the airports we visited identified a variety of tools that they utilize to respond to increases in passenger wait times and/or throughput. TSOs from the National Deployment Force —teams of additional TSOs—are available for deployment to airports to support screening operations during major events and seasonal increases in passengers. For example, TSA officials at one airport we visited received National Deployment Force officers during busy holiday seasons and officials at another airport received officers during the increase in wait times in the spring and summer of 2016. TSA officials at selected airports used passenger screening canines to expedite the screening process and support screening operations during increased passenger throughput and wait time periods. For example, TSA officials at one airport we visited emphasized the importance of passenger screening canines as a useful tool to minimize wait times and meet passenger screening demands at times when throughput is high. Officials at another airport we visited relied on these canines in busy terminals during peak periods. According to officials at two of the airports we visited, the use of passenger screening canines helped them to reduce wait times due to increased passenger volumes in the spring and summer of 2016. TSA officials at selected airports also utilize part-time TSOs and overtime hours to accommodate increases in passenger throughput and wait times. For example, according to officials at all eight of the airports we visited, they used overtime during peak travel times, such as holiday travel seasons, and officials usually planned the use of overtime in advance. Additionally, TSA officials at four of the airports we visited told us they used part-time TSOs to help manage peak throughput times throughout the day. According to TSA officials at two of the airports we visited, they moved TSOs between checkpoints to accommodate increases in passenger throughput at certain checkpoints and to expedite screening operations. For example, TSA officials at one airport we visited have a team of TSOs that terminal managers can request on short notice. Officials at the other airport estimated that they move TSOs between terminals about 40 times per day. Chairman Katko, Ranking Member Watson Coleman and Members of the Subcommittee, this concludes my prepared statement. I would be pleased to respond to any questions that you may have at this time. GAO Contacts and Staff Acknowledgments For questions about this statement, please contact William Russell at (202) 512-8777 or russellw@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this statement. Individuals making key contributions to this statement include Ellen Wolfe (Assistant Director), Joel Aldape, Brendan Kretzschmar, and Natalie Swabb. Key contributors for the previous report that this testimony is based on are listed in the product. This is a work of the U.S. government and is not subject to copyright protection in the United States. The published product may be reproduced and distributed in its entirety without further permission from GAO. However, because this work may contain copyrighted images or other material, permission from the copyright holder may be necessary if you wish to reproduce this material separately.

Why GAO Did This Study TSA employs about 43,000 Transportation Security Officers (TSO) who screen over 2 million passengers and their property each day at airports in the United States. TSA allocates TSOs to airports using both a computer-based staffing model and information from airports intended to provide each airport with the optimum number of TSOs. In the spring of 2016, long screening checkpoint lines at certain U.S. airports raised questions about TSA's process for allocating TSOs to airports. This testimony addresses (1) how TSA monitors wait times and throughput, and (2) tools TSA uses to respond to increases in passenger wait times. This testimony is based on a report GAO issued in February, 2018: GAO, Aviation Security: TSA Uses Current Assumptions and Airport-Specific Data for Its Staffing Process and Monitors Passenger Wait Times Using Daily Operations Data . GAO-18-236 , Washington, D.C.: February 1, 2018. For that report, among other things, GAO analyzed TSA documentation and passenger wait time and throughput data. What GAO Found In February 2018, GAO reported that the Transportation Security Administration (TSA) uses data to monitor passenger wait times and throughput, the number of passengers that are screened at each airport checkpoint, on a daily basis and responds to increases. For example, TSA's Airport Operations Center (AOC) monitors daily wait times and passenger throughput from 28 airports that TSA officials say represent the majority of passenger throughput nationwide or are operationally significant. Furthermore, TSA officials at airports are required to report to the AOC when an event occurs--such as equipment malfunctions--that affects airport screening operations and results in wait times that are greater than 30 minutes in standard screening lanes. For its February 2018 report, GAO analyzed wait time data for the AOC-monitored airports for the period of January 2015 through May 2017 and found that TSA's reported wait times met its standard of less than 30 minutes in standard screening 99 percent of the time. Within that time frame, two airports accounted for the longest wait times in the spring of 2016. TSA officials also identified several tools, such as passenger screening canines, that they reported using to respond to increases in passenger wait times at these airports. What GAO Recommends GAO is not making any recommendations.

gao_GAO-18-620

gao_GAO-18-620_0

Background Generally, the responsibility for reducing lead in drinking water and ensuring safe drinking water overall is shared by EPA, states, and local water systems. EPA is responsible for, among other things, national implementation of the Lead and Copper Rule, setting standards, overseeing states’ implementation of the rule, and conducting some enforcement activities. However, most states have primary responsibility for enforcing the requirements under SDWA as amended. Water systems are generally subject to requirements under SDWA as amended, such as the Lead and Copper Rule, and are responsible for managing and funding the activities and infrastructure needed to meet those requirements. Such infrastructure includes storage facilities and drinking water mains and may include other pipes such as service lines. There are 1 million miles of drinking water mains in the country, according to a 2017 American Society of Civil Engineers study. As figure 1 illustrates, service lines are the smaller pipes that connect the water mains to homes and buildings. According to EPA guidance, service lines also include any smaller pipes used for connecting a service line to the water mains (e.g., gooseneck pipes which are also known as pigtails). Service lines can generally be made of lead, steel, copper, or plastic. Service lines can be fully owned by the water system (publicly owned) or by the homeowner (privately owned), or ownership can be shared. In most communities, lead service lines are partially owned by the water system and partially owned by the homeowner. With shared ownership, the water system typically owns the service line from the water main to the curb stop, and the homeowner owns the service line from the curb stop into the home. In such cases, each party is responsible for maintaining the part of the service line that it owns. In some circumstances, if lead levels are higher than the Lead and Copper Rule allows and other measures do not alleviate the problem, the Lead and Copper Rule requires water systems to replace lead service lines under the systems’ control. The Lead and Copper Rule does not require homeowners to replace the portion of lead service lines they own, but if they choose to do so they are generally responsible for the associated costs. The Lead and Copper Rule allows for a partial replacement by the water system when an owner of a home or building is unable or unwilling to pay for replacement of the portion of the service line not owned by the water system. The Total Number of Lead Service Lines Is Unknown, and National, State, and Local Estimates Vary The total number of lead service lines is unknown and while national, state, and local estimates exist, approaches used to count lead service lines vary. The total number of lead service lines is unknown because, among other things, the Lead and Copper Rule does not require all water systems to collect such information. National, state, and local estimates exist, but the methods used to arrive at these estimates vary, making it challenging to compare estimates. The Lead and Copper Rule Does Not Generally Require Water Systems to Maintain Complete Information about Lead Service Lines or Report Such Information to EPA The total number of lead service lines is unknown, in part because the Lead and Copper Rule does not require all water systems to develop and maintain a complete inventory of lead service lines, and there are no national repositories of information about lead service lines. According to EPA headquarters officials we interviewed in 2017, the materials inventory required under the Lead and Copper Rule is not intended to be a census of lead service lines (and other lead pipes such as goosenecks/ pigtails). Instead, it is intended to provide sufficient information to develop a plan for periodically obtaining tap samples. For example, according to 2008 EPA guidance to water systems, if a system contains lead service lines, then, if possible, half of the sample sites should include those served by a lead service line. The Lead and Copper Rule requires water systems to conduct complete inventories only if the water system is required to begin replacing lead service lines. In these instances, water systems are required to expand the materials inventory to a complete inventory that identifies the total number of lead service lines for the purpose of tracking replacements over time. As we reported in 2017, based on the available data, the majority of the 68,000 water systems subject to the Lead and Copper Rule at the time of our review had not been required to replace lead service lines and therefore were not required to conduct complete inventories. Moreover, there are no national repositories for information about lead service lines. In September 2017, we recommended that, as a part of revisions to the Lead and Copper Rule, EPA require states to report data on lead pipes (including lead service lines) and incorporate these data in the agency’s Safe Drinking Water Information System. EPA agreed with the recommendation but has not implemented it. In May 2018, EPA noted that it was in the process of reviewing comments received through consultations with state and local officials and tribes. According to EPA officials, final revisions to the Lead and Copper Rule are expected by February 2020. We continue to believe that EPA should collect data about lead pipes (including lead service lines) from states. By doing so, EPA and congressional decision makers would have important information at the national level on what is known about lead infrastructure in the country, thereby facilitating oversight of the Lead and Copper Rule. National, State, and Local Estimates of Lead Service Lines Exist, and Those We Reviewed Had Significant Limitations; but the Methods Used to Arrive at These Estimates Vary The total number of lead service lines is unknown, and while some entities have developed estimates of lead service lines at the national, state, or local water system level, the estimates we reviewed have significant limitations to their reliability. Moreover, the approaches used to arrive at these estimates vary, making it challenging to compare estimates. Nationally, according to EPA’s October 2016 Lead and Copper Rule Revisions White Paper, there are an estimated 6.5 million to 10 million homes served by lead service lines. This range of estimates, based in part on data from a study for the 1991 Lead and Copper Rule, has significant limitations. In appendix I we explain why EPA’s estimate may not accurately reflect the total number of lead service lines, nationwide. An April 2016 American Water Works Association study estimated 6.1 million lead service lines nationwide. The authors of this study extrapolated the number based on survey responses from 978 water systems in 2011 and 2013. While this study is the most recent attempt to provide a national estimate, it has significant limitations. First, the sample was not statistically representative of all 68,000 water systems subject to the Lead and Copper Rule. Rather, the water systems that responded to the American Water Works Association’s survey are not a statistical sample. Second, according to the study’s authors, survey responses were based on water systems’ best guesses of the number of lead service lines in their systems. However, since water systems have not been required to maintain inventories of lead service lines, many of them do not know the exact number. For these reasons, we are not confident that the number accurately reflects the total number of lead service lines nationwide. An American Water Works Association official told us that the organization is not planning to update the study. EPA officials told us that they were not aware of a more recent study than the association’s 2016 study. In addition, EPA officials said in May 2018 that the results in the American Water Works Association study likely represent a lower-bound estimate for the number of lead service lines in the country because the sample was not generalizable, and had other data quality issues. EPA officials in one region we interviewed said that estimates of lead service lines can decrease or increase as a water system replaces lead service lines and as a water system does or does not count lead service lines on private property. The Lead and Copper Rule does not require states to collect statewide information about lead service lines, but at least two states collected data from water systems in their states and published reports with these data: A 2016 report by the Massachusetts Department of Environmental Protection’s Drinking Water Program reported 22,023 lead service lines and 15,809 lead goosenecks and pigtails statewide. The report counted goosenecks and pigtails separately from lead service lines. Officials from the Massachusetts Department of Environmental Protection told us that the state has about 2 million service lines total; therefore, about 1 percent of the total service lines are lead. A 2017 report by the Washington State Department of Health estimated 1,000-2,000 lead service lines statewide and 8,000 goosenecks statewide. According to Washington State officials, they continued to update their estimates in early 2018 with selected water utilities. Generally, the purpose of both studies, as stated in each report, was to identify areas in which water systems would need technical assistance in complying with the Lead and Copper Rule or state requirements. However, for the purposes of estimating the number of lead service lines, complete details were not available about the methodologies and some systems that did respond were only able to provide rough guesses rather than precise counts of lead service lines. EPA headquarters officials told us that Massachusetts and Washington were at the forefront of states’ efforts to gather information about lead service lines. EPA officials also told us that they were not aware of any other states with published reports estimating the number of lead service lines. However, at least two states have also collected information about lead service lines but have not published the information in official reports, at the time of our review. For example, in 2016, officials in Indiana and Maryland sent questionnaires to water systems in their states asking for information about the number of lead service lines. A representative of a water association told us that, generally, water systems were in the beginning stages of conducting complete inventories of lead service lines. However, some local water systems also have estimates. For example, EPA officials told us that water systems in the states of Ohio, Michigan, and Washington had estimates of lead service lines. In May 2018, a representative of the Greater Cincinnati Water Works water system estimated there were approximately 7 percent of publicly owned and approximately 18 percent privately owned lead service lines out of a total of 240,000 service lines in the area served by that water system. In March 2018, representatives of the Greater Cincinnati Water Works water system said that their estimates of lead service lines are best characterized as what is known at any given point in time. These representatives also told us that they collect this information on a continual basis from historical and on-going maintenance records, reports of lead service lines by customers, and the water system’s lead service line replacement program, among other sources. To conduct complete inventories and develop estimates, water systems have used varying approaches, which can hinder comparisons among states and water systems. The publicly available reports that existed as of May 2018 provide some insight into the various approaches water systems have used. For example, to identify lead service lines, water systems have used visual inspection or a combination of visual inspections, existing water system records, and discussions with homeowners. In addition, water systems have used various definitions of lead service lines. For example, water systems have counted: only active service lines delivering water to customers, or both active and inactive (no longer delivering water to customers) service lines; or only the publicly owned lead service lines, or both the publicly and privately owned portions of the lead service lines; or only lead service lines or the lead service lines and goosenecks/pigtails separately. Most States Reported Fulfilling EPA’s Request, but Potential Challenges Remain that EPA Information Sharing Could Help to Address While most states informed EPA that they intend to fulfill the agency’s request to work with water systems to publicize inventories of lead service lines, EPA has identified potential challenges to these efforts. Nonetheless, the agency has not followed up with all states since 2016 to share information about how to address these challenges. Most states that said they intended to fulfill EPA’s request to encourage water systems to publicize materials inventories reported in subsequent letters to or meetings with EPA that they did so; however, as of May 2018, most large waters systems had not made such information public. Most States Reported Fulfilling EPA’s Request to Encourage Water Systems to Publicize Materials Inventories, but Most Large Water Systems GAO Reviewed Did Not Do So Our analysis of states’ written responses to EPA’s 2016 request, and information obtained through interviews with EPA officials as of February 2018, found that most (43) of the 50 states indicated an intent to fulfill EPA’s request, 3 states said that they may consider it, and 4 states did not intend to fulfill EPA’s request. Of the approximately 43 states that responded that they would fulfill EPA’s request, almost all (39) reported in subsequent letters to or meetings with EPA that they had encouraged water systems to publicize their materials inventories or other information about lead service lines. In these letters and meetings, states also reported taking other actions to increase their knowledge about lead service lines such as requesting that water systems update the materials inventory required by the Lead and Copper Rule, creating online repositories of maps of lead service lines, posting reports on lead service lines, and issuing requirements for water systems to collect information on lead service lines. For example, in May 2016, the governor of Washington issued a directive requiring the state’s Department of Health to work with certain water systems to identify all lead service lines and lead components within 2 years. Figure 2 shows the number of states that reported fulfilling EPA’s request or taking other related actions. Because EPA asked states to prioritize large water systems (those servicing populations greater than 50,000), we reviewed the websites for the 100 largest water systems. As of January 2018, we found 12 of these water systems had publicized information on the inventory of lead service lines; the rest had not. The information on the websites for the 12 water systems varied. For example, the water system for Tulsa, Oklahoma posted a map that highlighted where lead service lines may be present. Water systems such as Cincinnati, Ohio, Boston, Massachusetts, and Washington, D.C., provided interactive maps that showed locations identified as having lead service lines. See figure 3 for examples of the interactive maps of lead service lines that some selected large water systems have provided to the public. Water systems that serve populations greater than 50,000 but were not among the 100 largest water systems at the time of our review may have also publicized information on the inventory of lead service lines. For example, the water systems for Akron, Ohio, Flint, Michigan, and Providence, Rhode Island each publicized an interactive or other type of map of lead service lines. EPA Identified Potential Challenges to Publicizing Materials Inventories but Has Not Followed Up with All States about How to Address Such Challenges Since 2016 EPA officials in the regional offices provided a range of reasons why water systems may be challenged in conducting inventories of lead service lines and making any information about lead service lines public, however, it has not followed up with all states about how to address such challenges since 2016. In September 2017, we reported that the six states that would not fulfill EPA’s 2016 request had highlighted challenges in finding historical documentation about lead pipes to create plans for collecting tap water samples or in dedicating staff resources to do so. In January and February 2018, some officials whom we interviewed in EPA’s 10 regional offices agreed that these would be challenges for states and water systems. The officials also mentioned additional potential challenges in conducting complete inventories of lead service lines or publicizing information about lead service lines. Table 1 describes the challenges mentioned by EPA officials in the 10 regional offices. Since the February 2016 letter, EPA followed up in July 2016 with a letter to the Association of State and Territorial Health Officials and Environmental Council of States, which represents all states. In that letter, EPA provided two examples of state practices that increase public transparency: some drinking water systems are providing online searchable databases that provide information on known locations of lead service lines, or are providing videos that show homeowners how to determine whether their home is served by a lead service line. The letter also said that EPA would continue to work with states to ensure that the identification of the locations of lead service lines remains a priority for drinking water systems. However, EPA has conducted limited follow-up since then, mainly, EPA headquarters and regional officials said, because they have focused their efforts on ensuring states appropriately comply with the Lead and Copper Rule. As previously noted in this report, posting materials inventories or other information about the location of lead service lines is not a requirement of the Lead and Copper Rule. In May 2018, EPA headquarters officials we interviewed said that they learned of some states’ and water systems’ efforts toward making information about lead service lines available to the public since 2016, through conferences and discussions with states. These headquarters officials told us that they have shared such efforts with those states who, in 2016, said they did not intend to fulfill EPA’s 2016 request. For example, EPA shared how states that were publicizing information about lead service lines were addressing privacy concerns with states that originally said they would not fulfill EPA’s request. However, as of January 2018, most of the 100 largest water systems had not made their materials inventories or additional maps or updated inventories public. According to EPA’s February 2016 letter, the agency’s objective in encouraging states to work with water systems to post, on a public website, the water system’s original materials inventory along with any additional updated map or inventories of lead service lines was to assure the public that lead risks were being addressed. Under federal standards for internal control, management should externally communicate the necessary quality information, so that external parties can help to achieve the entity’s objectives. By sharing information with all states about the approaches that some states and water systems are using to successfully identify and publicize information about lead service lines, including responses to potential challenges, EPA could encourage states to be more transparent to the public and support the agency’s objectives for safe drinking water. Conclusions Lead service lines present a significant risk of lead contamination in drinking water. Publicizing drinking water systems’ knowledge about lead service lines, and other lead infrastructure, would facilitate oversight of the Lead and Copper Rule. In September 2017, we recommended that, as a part of revisions to the Lead and Copper Rule expected by February 2020, EPA require states to report data on lead pipes (including lead service lines) and incorporate these data in the agency’s Safe Drinking Water Information System. EPA agreed with the recommendation, and we continue to believe that EPA should require data about lead pipes (including lead service lines) from states. Most states reported that they had encouraged their water systems to publicize information about lead service lines in response to EPA’s February 2016 requests. EPA headquarters officials told us that they had learned of some states’ and water systems’ efforts since 2016 and shared this information with the few states that said that they would not take action in response to EPA’s letter. This information did in fact help at least one state take action, according to information we received from EPA and the state. By sharing information with all states about the approaches that some states and water systems are using to successfully identify and publicize information about lead service lines, including responses to potential challenges, EPA could encourage states to be more transparent to the public and support the agency’s objectives for safe drinking water. Recommendation for Executive Action The Assistant Administrator for Water of EPA’s Office of Water should share information with all states about the approaches that some states and water systems are using to successfully identify and publicize information on lead service lines, including responses to potential challenges. (Recommendation 1) Agency Comments We provided a draft of this report to EPA for review and comment. In its comments, reproduced in appendix II, EPA agreed with our recommendation. The agency also highlighted a recently developed website that showcases efforts to identify and replace lead service lines and said that it will continue to ensure states and water systems are aware of this resource. We are sending copies of this report to the appropriate congressional committees, the Administrator of EPA, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff members have any questions about this report, please contact me at (202) 512-3841 or gomezj@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. Key contributors to this report are listed in appendix III. Appendix I: Objectives Scope and Methodology Our objectives were to examine (1) what is known about the number of existing lead service lines nationally, and among states and water systems; and (2) state responses to EPA’s February 2016 request to work with water systems to publicize inventories of lead service lines and any steps EPA has taken to follow up on these responses. To examine what is known about the number of existing lead service lines nationally, and among states and water systems, we relied on interviews and publicly available reports for which we could assess the reliability of the data. We reviewed the requirements under the Lead and Copper Rule for assessing the number of lead service lines. We interviewed officials from EPA’s Office of Water and the following water organizations concerning what these officials knew about the number of lead service lines nationally and among states and water systems: the American Water Works Association, Association of State Drinking Water Administrators, and Regional Community Assistance Partnership. We also interviewed an official with the Environmental Defense Fund regarding the available information about the number of lead service lines nationally and among states and water systems. We selected these organizations because they are all members of the Lead Service Line Replacement Collaborative, a consortium that provides information about voluntary lead service line replacement for states and water systems. On behalf of the Lead Service Line Replacement Collaborative, the organizations we spoke with are collecting examples of states’ and water systems’ experiences in conducting inventories of lead service lines, as the first step in replacing lead service lines. Using information from these interviews, we identified three published studies from the American Water Works Association, the state of Massachusetts, and the state of Washington. We interviewed the authors of the studies to determine the reliability, completeness, and accuracy of the data presented in the studies. For the 2016 American Water Works Association study, we determined that the data were of undetermined reliability because the responses of the water systems surveyed were not generalizable to all water systems and the study authors could not verify the accuracy of the information. Specifically, the sample in the 2016 American Water Works Association study was not based on a statistical sample, and therefore the sampling error was not calculated and information was not available to determine whether responding water systems were similar to nonresponding water systems. For example, the estimate is based on survey responses from 978 of the approximately 23,000 water systems that existed around the time of the surveys, and therefore may not represent all water systems nationwide. In addition, since many water systems do not have complete inventories of their lead service lines, the accuracy of data that water systems submitted in response to the survey is difficult to verify. For example, our interview with the study authors indicates that the information provided by water systems varied in quality, with some systems basing their responses on rough estimates. We based our determination about the data using the criteria of Total Survey Error, which is a framework for assessing the validity and reliability of survey estimates. It includes sampling error (the difference between the population and the sample), nonresponse error, measurement error (the difference between the true response and the response provided by the respondent) and coverage error (the discrepancy between the list of individuals that is used to select a sample and the target population). EPA’s 2016 Lead and Copper Rule Revisions White Paper also identified an estimate of lead service lines. According to EPA officials, this estimate used data from the 2016 American Water Works Association study and a 1988 American Water Works Association study cited in the regulatory impact analysis for the 1991 Lead and Copper Rule. The 1991 estimate also had significant limitations in measurement error and representation error as well as a lack of documentation about key aspects of the methodology. As such, we determined the estimate was not reliable for the purposes of establishing the total number of lead service lines in existence as of 1991. The two state-specific studies represent reasonable efforts to estimate the number of lead service lines in these states. However, they generally could not verify the accuracy of the information provided by these systems because, as we note elsewhere in this report, water systems may not know the number of lead service lines they have. Therefore, for the state-specific studies, we also determined that the data were also of undetermined reliability. Finally, while the Greater Cincinnati Water Works water system did not publish a report about lead service lines, we collected the information through an in-person interview and corroborated the information through a review of the water system’s geographic information system database. The Greater Cincinnati Water Works’ GIS database includes the location and material information for all of the water system’s distribution system. According to the Greater Cincinnati Water Works website, the water system continues to update its map as it obtains more information from its customers. Based on these steps we deemed the data provided by the water system to be sufficiently reliable for the purposes of describing the estimate reported by representatives of the Greater Cincinnati Water Works system. To examine states’ responses to EPA’s February 2016 request to work with water systems to publicize inventories of lead service lines and any steps EPA has taken to follow up on these responses, we relied both on the publicly available letters from each state to EPA and on interviews with EPA regional and headquarters officials. We did not interview state officials in all 50 states, but reviewed some state documents, where available. We used a standard set of open-ended questions to interview officials in EPA’s headquarters and in each of the 10 regional offices. To analyze states’ and EPA officials’ responses, we conducted two analyses. Specifically, we conducted two analyses to summarize updates in state responses to EPA’s February 2016 letter and EPA’s responses to challenges states and water systems may face in conducting and publicizing materials inventories. To confirm each analysis, one analyst independently summarized the information and another analyst verified the accuracy of the information. All initial disagreements were discussed and reconciled. All numbers in our analysis are considered approximate because interpretations of the states’ responses to EPA’s 2016 letter can differ, and states may have taken actions after our interviews with EPA regional officials, or may have taken actions that they did not report to EPA. Figure 4 shows the EPA regions and the states within those regions. We also reviewed EPA documents related to EPA’s request that states take certain actions following the events in Flint, Michigan. In addition, we reviewed federal regulations; EPA guidance to water systems on how to implement the Lead and Copper Rule; and other relevant documents such as an EPA white paper. Because EPA asked states to place an emphasis on working with large water systems to publicize their materials inventories or updated inventories or maps of lead service lines, we reviewed the websites of the 100 largest water systems by population. Our review was conducted in January to February 2018; and since then, additional water systems may have provided information to the public on lead service lines. We identified the largest water systems, based on population served, from data in EPA’s Safe Drinking Water Information System/Fed. EPA has stated on its website that the agency acknowledges challenges related to the data in the Safe Drinking Water Information System/Fed, specifically underreporting of some data by states. GAO has also reported on EPA’s challenges with the Safe Drinking Water Information System/Fed. Even with these challenges, the information on the populations served by water systems in the Safe Drinking Water Information System/Fed is generally reliable. We used a standard set of search terms on each website to ensure the consistency of our searches, as well as information from water organizations and EPA officials, where applicable. We counted a water system as having an inventory if the water system provided a map, interactive map, list of pipes or service lines, or numerical count of lead service lines available to the public. To ensure the completeness of this analysis, one analyst independently conducted the search of websites and another analyst verified the search. All initial disagreements were discussed and reconciled. We compared EPA’s actions to follow up on state responses with federal standards for internal control for information and communication. We conducted this performance audit from October 2017 to September 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Comments from the U.S. Environmental Protection Agency Appendix III: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Diane Raynes (Assistant Director); Tahra Nichols (Analyst in Charge); David Blanding, Jr.; Mark Braza; Lawrence Crockett, Jr.; Justin Fisher; Richard P. Johnson, and Jeanette Soares made key contributions to this report. In addition, Cynthia Norris and Dan Royer made important contributions. Related GAO Products Drinking Water: Additional Data and Statistical Analysis May Enhance EPA’s Oversight of the Lead and Copper Rule. GAO-17-424. Washington, D.C.: September 1, 2017. Water Infrastructure: Information on Selected Midsize and Large Cities with Declining Populations. GAO-16-785. Washington, D.C.: September 15, 2016. Water Infrastructure: EPA and USDA Are Helping Small Water Utilities with Asset Management; Opportunities Exist to Better Track Results. GAO-16-237. Washington, D.C.: January, 27, 2016. Drinking Water: Unreliable State Data Limit EPA’s Ability to Target Enforcement Priorities and Communicate Water Systems’ Performance. GAO-11-381. Washington, D.C.: June 17, 2011. Drinking Water: The District of Columbia and Communities Nationwide Face Serious Challenges in Their Efforts to Safeguard Water Supplies. GAO-08-687T. Washington, D.C.: April 15, 2008. Drinking Water: EPA Should Strengthen Ongoing Efforts to Ensure That Consumers Are Protected from Lead Contamination. GAO-06-148. Washington, D.C.: January 4, 2006. District of Columbia’s Drinking Water: Agencies Have Improved Coordination, but Key Challenges Remain in Protecting the Public from Elevated Lead Levels. GAO-05-344. Washington, D.C.: March 31, 2005. Drinking Water: Safeguarding the District of Columbia’s Supplies and Applying Lessons Learned to Other Systems. GAO-04-974T. Washington, D.C.: July 22, 2004.

Why GAO Did This Study The crisis in Flint, Michigan, brought increased attention to lead in drinking water infrastructure. Lead in drinking water primarily comes from corrosion of service lines connecting the water main to a house or building. In 1991, EPA issued the Lead and Copper Rule that required water systems to conduct a “materials inventory” of lead service lines. In light of the events in Flint, EPA sent a letter to all states in February 2016 encouraging them to work with water systems to publicly post the materials inventory, along with any additional updated maps or inventories of lead service linesactions the rule does not require. A House Committee report accompanying a bill for the Department of the Interior, Environment and Related Agencies Appropriations Act, 2017, includes a provision for GAO to review lead service lines. This report examines (1) what is known about the number of existing lead service lines among states and water systems and (2) states' responses to EPA's February 2016 request to work with water systems to publicize inventories of lead service lines and any steps EPA has taken to follow up on these responses. GAO reviewed existing studies of lead service lines, reviewed the websites of the 100 largest water systems, and interviewed EPA officials in headquarters and its 10 regional offices. What GAO Found The total number of lead service lines is unknown and while national, state, and local estimates exist, approaches used to count lead service lines vary. A 2016 American Water Works Association study estimated that nationally there were 6.1 million lead service lines, but the study has significant sampling limitations and, as a result, likely does not accurately reflect the total number of lead service lines nationwide. In addition, at least two statesMassachusetts and Washingtonpublished reports with estimates of lead service lines and reported 22,023 and 1,000-2,000 lead service lines as of 2016 and 2017, respectively. Certain water systems also have estimates, such as the approximately 7 percent of publicly owned lead service lines out of the area's total number of service lines cited by a representative for the system serving Cincinnati, Ohio and surrounding areas, as of May 2018. While most states informed the Environmental Protection Agency (EPA) that they intend to fulfill the agency's request to publicize inventories of lead service lines, EPA has identified potential challenges to these efforts. Of the approximately 43 states that responded that they would fulfill EPA's request, almost all (39) reported to EPA that, although they had encouraged water systems to publicize inventories, few systems had completed these actions. GAO found in January 2018 that, of the 100 largest water systems, 12 had publicized information on the inventory of lead service lines. According to EPA, among challenges in conducting inventories of lead service lines and publicizing information about lead service lines were concerns about posting on public websites information about lead service lines on private property; and a lack of records about the locations of lead service lines. EPA told GAO the agency was focused on state compliance with drinking water rules, and not following up with information on how states could address the challenges cited. By sharing information with all states about the approaches that some states and water systems are using to successfully identify and publicize information about lead service lines, including responses to potential challenges, EPA could encourage states to be more transparent to the public and support the agency's objectives for safe drinking water. What GAO Recommends GAO recommends that EPA share information about the successful approaches states and water systems use to identify and publicize locations of lead service lines with all states. EPA agreed with the recommendation.

gao_GAO-19-89

gao_GAO-19-89_0

Background Determining and Reporting on Core Capability Requirements DOD Instruction 4151.20, Depot Maintenance Core Capabilities Determination Process, requires the military services to apply a methodology to determine their core capability requirements—that is, to identify what core capabilities are required and what workload would be necessary to enable them to sustain these core capabilities at the depots. DOD’s instruction also requires the military services to determine the estimated cost of workloads to sustain the core capability requirement. The instruction describes a series of mathematical computations and adjustments that the military services are required to use to compute their core capability requirements, and to identify the projected workload needed to support these requirements. Specifically, the instruction requires that the military services identify the weapon systems required to execute the Chairman of the Joint Chiefs of Staff’s strategic and contingency plans, which, among other things, guide the use and employment of the military forces across all geographic regions and sustain military efforts over different durations of time. After the systems are identified, the military services compute annual depot maintenance capability requirements for peacetime, in direct labor hours, to represent the amount of time they will regularly take to execute required maintenance. A military service may adjust calculated direct labor hours to address redundant capability requirements that are so similar to one another that they share common base repair processes. DOD tracks core capability requirements using the following two metrics: direct labor hours, each of which represents 1 hour of effort directly allocated to a category of work; and work breakdown structure categories, which bundle types of work according to weapon systems and equipment. DOD uses work breakdown structure categories to organize data on its various core capability requirements and workloads, as well as to manage and report on its core capabilities. There are 10 first-level work breakdown structure categories, and these in turn are broken down into second-level subcategories, which are the major elements that make up the system or equipment in the first-level category. Figure 1 shows the 10 first-level categories of DOD’s work breakdown structure. For the full work breakdown structure, see appendix IV. Finally, the instruction requires the military services to provide a reason for all projected shortfalls, strategies to mitigate the effects of each projected shortfall, and actions taken by the services to rectify any projected workload or capability shortfall. A projected shortfall exists if a military service does not expect to have sufficient workload to sustain the required level of capability that has been identified. For example, an armed service may have identified 10,000 direct labor hours of core capability requirements for ground vehicles, but have only 4,000 hours of projected depot maintenance work for ground vehicles—resulting in a projected workload shortfall of 6,000 hours. DOD’s Biennial Core Reports and Our Prior Reviews In 2012 DOD submitted its first biennial core report to Congress, and we found that DOD did not provide sufficient explanations when reporting on the military services’ shortfalls in core capability requirements. In 2014 DOD submitted its second biennial core report to Congress, and we found that DOD did not have accurate and complete data in the report. In 2016 DOD submitted its third biennial core report to Congress, and we found (1) data errors; (2) inaccurate inter-service workload across the military services due to lack of coordination in reporting this information; (3) inconsistent calculations or display of workload shortfalls across the military services; and (4) inconsistent calculations of the estimated cost of planned workload across the military services. We made recommendations to address each issue. Further, we identified additional information that could increase the report’s transparency, and we suggested that Congress consider amending section 2464 to include additional elements to increase the transparency of future biennial core reports. Consistent with our recommendations, Congress amended section 2464 and added additional reporting requirements. We discuss DOD’s actions to address our specific recommendations to improve the completeness of its 2018 Biennial Core Report later in this report. DOD Addressed Eight of the Ten Reporting Elements and Plans to Address the Remaining Two in the 2020 Biennial Report In the 2018 Biennial Core Report, DOD and the military services addressed 8 of 10 required reporting elements, as shown in table 1 and discussed in more detail below. According to department officials, the department did not address two of the elements because changes to its guidance and processes for developing the 2018 report resulted in the 2016 and 2018 reports not being directly comparable. DOD officials stated that they plan to address these two elements in the 2020 Biennial Core Report. Military Services Identified Core Capability Requirements and Projected Workloads To address reporting elements 1 and 2, the military services presented their respective requirements and projected workloads in direct labor hours and associated costs, using the work breakdown structure. Table 2 shows DOD’s reported direct labor hours for the depots’ core requirements, as well as projected maintenance workloads and costs of workloads to sustain core requirements by military service. Military Services Identified Key Information by Work Breakdown Structure The military services presented core requirements and workloads, down to the second-level subcategories, to address reporting element 7. This structure represents all of the sub-specialties required to maintain core depot-level capabilities across the 10 categories of the work breakdown structure. For example, the aircraft category is broken down into 7 second-level subcategories: rotary, vertical/short take-off and landing, cargo/tanker, fighter/attack, bomber, unmanned systems, and aircraft engines. The Army, Navy, and Air Force also identified the items they placed into the “Other” category to address reporting element 9. The Marine Corps did not place any core requirements in the “Other” category in the 2018 Biennial Core Report and therefore was not required to address this reporting element. Specifically: The Army identified requirements associated with items such as air conditioners, food service hygiene equipment, chemical defense equipment, and water purification; The Navy identified requirements associated with specialty aircraft and aircraft components that are common across multiple platforms; and The Air Force identified requirements associated with specialty items such as surveillance aircraft, missile components, and communications/electronic equipment that do not fall under other distinct work breakdown structure subcategories. Military Services Identified Projected Shortfalls and Mitigation Plans The military services each identified projected shortfalls at the first- and second-levels of the work breakdown structure (elements 3 and 4), reasons for those shortfalls (element 3), and mitigation plans for the projected shortfalls (element 3). This includes—in some cases— leveraging excess core capabilities in one workload category to mitigate projected shortfalls in another category (elements 5 and 8). Specifically: Army: The Army reported a total projected shortfall of about 2.9 million direct labor hours, as shown in table 3. It identified projected shortfalls in 5 of the 10 first-level work breakdown structure categories, and in 13 of the 33 second-level categories. The Army identified a number of reasons for these projected shortfalls. Army officials stated that these reasons generally contributed to shortfalls across the various work breakdown categories. They also noted the challenge of calculating shortfalls based on comparing current workloads with predicted workloads that were based on potential future Army strategies. The Army identified the following specific reasons for shortfalls: DOD’s updated defense planning scenarios increased the Army’s equipment requirements. These additional requirements resulted in a greater total core depot requirement for the Army, which in turn contributed to projected shortfalls. The Army noted that DOD’s most recent Future Years Defense Program lacked sufficient depot maintenance funding (that is, money to pay for direct labor hours) to meet core capability requirements. The Army cited newly established software depot maintenance requirements as one of the reasons for its shortfall. Specifically, DOD updated requirements for reporting depot resources associated with upgrading and maintaining software in weapon systems. According to the Army’s 2018 core report submission, the Army previously determined this requirement based on the number of people assigned to the Army’s software sustainment activities. However, the Army revised its methodology for calculating its software sustainment workload to reflect actual workload, not just the number of people conducting the work. After identifying projected shortfalls, officials used that information to determine how best to close gaps and mitigate risks in future implementation. Specifically, the Army is currently working to move software-related direct labor hours from contractor to military sources, which will help the Army mitigate—that is, shrink—its projected shortfall by fiscal year 2020. The Army reported that it plans to mitigate many of its projected core shortfalls by using skill sets similar to those required for maintaining a core capability in repairing equipment for foreign militaries. Officials stated that the Army plans to hire and train maintenance personnel to conduct maintenance work associated with the foreign military sales program. This workload will also assist the Army in meeting its core capability requirements for Army systems, increasing the total projected workload, and decreasing estimated shortfalls. Additionally, the Army identified mitigations for specific shortfalls—for example, replacing old generators with a new system by fiscal year 2025 will mitigate its shortfall in support equipment. Navy: The Navy reported that it did not project an overall shortfall, nor did it project any shortfalls at the first- or second-level of the work breakdown structure, and therefore it did not provide mitigation plans. Navy and OSD officials noted that the Navy and the department differ regarding the definition of software sustainment. Specifically, a Navy official stated that the service views software sustainment as an engineering function, not a depot maintenance function. This official observed that while the Navy believes software sustainment to be critical to maintaining its weapon systems, it believes that managing software sustainment as depot maintenance is not the most effective approach for the Navy. As a result, the Navy did not report any software core capability requirement or projected workload for fiscal year 2019. OSD defined software maintenance and reporting requirements in its guidance requesting data from the military services for the biennial core report. In spite of differing perspectives between OSD and the Navy, OSD accepted the Navy’s core report submission, in which the Navy reported no core software maintenance capability requirements. Marine Corps: The Marine Corps reported that it did not project a total shortfall, but did project a shortfall of 82,971 direct labor hours in one second-level subcategory—that is, construction equipment—that falls in the ground vehicle first-level category. The Marine Corps identified a rationale and mitigation plan for its projected shortfall in construction equipment. The Marine Corps reported that general factors affecting maintenance workload and funding contributed to the shortfall, including: (1) After drawdowns from Iraq and Afghanistan, the Marine Corps repaired equipment to a desired level of combat effectiveness in line with current mission requirements and available resources. This led to fewer current maintenance needs and therefore reduced core maintenance workloads, creating projected shortfalls in some skill sets; and (2) The Marine Corps made changes to its force structure, which led to having more equipment in inventory, less equipment in use, and therefore less required maintenance. This created a shortfall in the skill set for construction equipment. To address this shortfall, the Marine Corps plans to use the excess workload in amphibious vehicles to mitigate the projected shortfall in construction equipment. Marine Corps officials stated that these second-level subcategories involve similar, tracked vehicles, which can be maintained using the same skill set. Air Force: The Air Force reported that it did not project a total shortfall, but did project shortfalls within the work breakdown structure, as shown in table 4. The Air Force identified projected overall shortfalls in 1 of the 10 first-level work breakdown structure categories, and in 7 of the 33 second-level work breakdown structure categories. The Air Force identified reasons and provided detailed explanations, as well as mitigation plans, for each projected shortfall. For example, it projected a shortfall in rotary workload according to Air Force officials because of staffing and supply issues with HH-60 Pave Hawk maintenance at Corpus Christi Army Depot. According to these officials, these maintenance issues have resulted in the Air Force’s using more contracted depot maintenance work on the HH-60 Pave Hawk in order to meet demand. As a result of the more extensive contracting of maintenance, planned workload at Corpus Christi Army Depot has been reduced, thereby creating a projected shortfall. The Air Force, Army, and Navy formed a team to address this projected shortfall. Air Force officials stated that contracts are being reduced and that they expect to resolve the maintenance issues before the 2020 Biennial Core Report. To address its projected shortfall in tactical missiles, the Air Force plans to identify Letterkenny Army Depot as the Technology Repair Center for this requirement, as the workloads are small in volume and the Letterkenny Army Depot can meet this requirement. In addition, the Air Force projected an overage of about 176,000 direct labor hours in strategic missiles. The Air Force believes that its projected workload in strategic missiles will allow it to maintain capability to repair tactical missiles—an area in which it projects a shortfall of about 42,000 direct labor hours. According to Air Force officials, the electronics on these two types of missiles are very similar and require the same skill set. DOD Did Not Address the Two Elements Concerning Progress in Implementing Mitigation Plans and Executing Reported Workloads DOD in the 2018 Biennial Core Report did not address progress made in implementing mitigation plans from the prior core report (element 6), nor did they address the degree to which projected workload reported in the prior core report was executed (element 10). According to Office of the Assistant Secretary of Defense for Logistics and Materiel Readiness (OASD L&MR) officials, they did not address these elements because the elements require DOD to compare information in the 2018 Biennial Core Report with information in the 2016 Biennial Core Report. Since DOD updated its guidance and processes for developing the 2018 Biennial Depot Core Report—in response to new statutory requirements and our prior recommendations—a meaningful comparison was not possible in the 2018 Biennial Core Report, according to OSD and military service officials. Additionally, DOD did not fully provide mitigation plans in its 2016 Biennial Core Report, as we reported in 2016. Therefore, DOD was unable to provide progress reports on 2016 mitigation plans. DOD officials told us that they plan to use the 2018 Biennial Core Report as a baseline for future biennial core reports, which will allow them to address elements 6 and 10. Specifically, they stated that they plan to provide progress reports on the mitigation plans they identified in the 2018 Biennial Core Report. Additionally, officials stated their intent to provide a comparison of the fiscal year 2019 projected workload reported in the 2018 Biennial Core Report with the actual workload for fiscal year 2019 contained in the 2020 Biennial Core Report. DOD’s 2018 Biennial Core Report Is Generally Complete DOD’s 2018 Biennial Core Report is generally complete in that it lacks any obvious errors and aligns with supporting information provided by the military services. Specifically, unlike previous biennial core reports, data submissions provided to DOD by the military services are identical to the data in the 2018 Biennial Core Report, and there are no transposition errors. Further, based on our review of the services’ submissions to OSD, data and other information provided by the military services were accurately and appropriately included in DOD’s 2018 Biennial Core Report. Finally, our analysis of the report and the military services’ submissions did not identify errors in the summation of the data. DOD’s focused efforts in 2017 and 2018 to develop better guidance and procedures assisted in improving the completeness of DOD’s 2018 Biennial Core Report—in part, according to DOD officials, due to our prior recommendations. Specifically, in 2017 the OASD L&MR began drafting new guidance to identify required depot maintenance core capabilities and the associated workloads needed to sustain those capabilities. This guidance was finalized and issued by the Office of the Under Secretary of Defense for Acquisition and Sustainment in May 2018. Officials from OASD L&MR and the military services told us that they used the methodology in this new guidance to complete the 2018 Biennial Core Report in late 2017 and early 2018. Officials told us that our prior recommendations, based on our reviews of the 2012, 2014, and 2016 biennial core reports, served to guide DOD’s update of its guidance and procedures. The changes made by Congress to section 2464 were also incorporated into DOD’s new guidance to ensure compliance with the 10 reporting elements, as we previously discussed. During the course of our review, we found that DOD had addressed all of the recommendations from our prior reports on the 2012, 2014, and 2016 Biennial Core Reports. First, in our review of the 2012 Biennial Core Report, we found that DOD did not include explanations for each identified projected shortfall. We recommended that DOD include in its biennial core report to Congress detailed explanations for why the military services did not have the workloads to meet core maintenance requirements for each projected shortfall identified in the report. Officials with OASD L&MR said that the May 2018 updated version of DOD Instruction 4151.20 was revised to require the submission of a detailed rationale for any and all shortfalls, and a plan to either correct or mitigate the effects of the shortfalls. The instruction states further that the detailed rationale and plan will identify the reason for the shortfall; contain a strategy to mitigate the effects of the shortfall (for example, specific transferrable workload, transfer of private- sector workload); and include actions to rectify any capability or workload shortfalls, including a description of planned capital investment, timing, and planned workarounds until the new capabilities or workloads are available. DOD’s 2018 Biennial Core Report as previously discussed provided rationales for shortfalls. Second, in our review of the 2014 Biennial Core Report, we found that some data were incomplete. We recommended that DOD review its processes and implement needed improvements to help ensure accuracy and completeness. In response to this and our other prior recommendations, DOD updated DOD Instruction 4151.20 to include additional steps and more controls that ensure more complete and accurate data submissions. According to OSD officials, changes to the guidance included deleting data fields unrelated to core requirements; streamlining and clarifying reporting instructions; ensuring that service submissions be reviewed and approved by general, flag, or senior executive service officials; determining the weapon systems or other platforms that are in the Chairman of the Joint Chiefs of Staff strategic and contingency plans; addressing inter-service workloads; having the worksheet automatically calculate shortfalls; and defining “software” and “software maintenance.” Most recently, in our review of the 2016 Biennial Core Report, we found (1) data errors; (2) inconsistent capture of inter-service workloads across the military services; (3) inconsistent calculations or transpositions of projected workload shortfalls across the military services; and (4) inconsistent calculations of the estimated cost of projected workloads across the military services. We recommended that DOD update its guidance—in particular DOD Instruction 4151.20—to require future biennial core reports to include instructions to the reporting agencies on how to (1) report additional depot workloads performed that have not been identified as core requirements; (2) accurately capture inter-service workloads; (3) calculate projected shortfalls; and (4) estimate the cost of projected workloads. DOD took steps to address each of these issues. Specifically, DOD did the following: Issued guidance stating that the total adjusted core capability requirements and the total projected public-sector depot maintenance workloads both reflect core workloads, as well as workloads that have not been identified as sustaining core. Developed and provided to each of the military services a worksheet on which to submit their projected inter-service workloads. OSD also held a meeting with all of the military services to resolve any discrepancies between their respective submissions. Created worksheets with formulas to automatically calculate the projected shortfalls at the subcategory level of the work breakdown structure for each service. Issued updated guidance to indicate that the estimated costs of the projected workloads to sustain the core capability requirements were to be included. According to OSD officials, these estimates are developed in accordance with financial management regulations and then applied to the estimated core sustaining workloads for each work breakdown structure, thereby providing a common baseline and process. In meetings with OSD and the military services, officials offered ideas for possible changes in future reports, such as including additional information on inter-service workloads to increase congressional visibility regarding coordination on depot maintenance across the military services. Additionally, OSD officials noted that they were considering the inclusion of additional information in future reports on how costs of projected workloads are calculated. Information on this is provided in DOD Instruction 4151.20, but not in its biennial core report. According to OSD officials, the department plans to consider these and other proposed changes from the military services and other stakeholders to its biennial core reporting process and supporting guidance. Given that DOD has made considerable progress by improving both the completeness of the 2018 Biennial Core Report and its guidance on the development of the report, we are not making recommendations at this time. Agency Comments We provided a draft of this report to DOD for comment. DOD provided technical comments, which we included as appropriate. We are sending copies of this report to appropriate congressional committees, the Secretary of Defense, and the Secretaries of the Military Departments. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact Diana Maurer at (202) 512-9627 or maurerd@gao.gov. Contact points for our Office of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix V. Appendix I: Complete Text of 10 U.S.C. § 2464(d) Appendix I: Complete Text of 10 U.S.C. § 2464(d) (d) Biennial core report. Not later than April 1 of each even-numbered year, the Secretary of Defense shall submit to Congress a report identifying, for each of the armed forces (except for the Coast Guard), for the fiscal year after the fiscal year during which the report is submitted, each of the following: 1. The core depot-level maintenance and repair capability requirements and sustaining workloads, organized by work breakdown structure, expressed in direct labor hours. 2. The corresponding workloads necessary to sustain core depot-level maintenance and repair capability requirements, expressed in direct labor hours and cost. 3. In any case where core depot-level maintenance and repair capability requirements exceed or are expected to exceed sustaining workloads, a detailed rationale for any and all shortfalls and a plan either to correct or mitigate the effects of the shortfalls. 4. Any workload shortfalls at any work breakdown structure category designated as a lower-level category pursuant to Department of Defense Instruction 4151.20, or any successor instruction. 5. A description of any workload executed at a category designated as a first-level category pursuant to such Instruction, or any successor instruction, that could be used to mitigate shortfalls in similar categories. 6. A description of any progress made on implementing mitigation plans developed pursuant to paragraph (3). 7. A description of core capability requirements and corresponding workloads at the first level category. 8. In the case of any shortfall that is identified, a description of the shortfall and an identification of the subcategory of the work breakdown structure in which the shortfall occurred. 9. In the case of any work breakdown structure category designated as a special interest item or other pursuant to such Instruction, or any successor instruction, an explanation for such designation. 10. Whether the core depot-level maintenance and repair capability requirements described in the report submitted under this subsection for the preceding fiscal year have been executed. Appendix II: Timeline of 10 U.S.C. § 2464 and Related GAO Reports In 1984 Congress passed legislation limiting the private contracting of certain core logistics functions. This law required the Department of Defense (DOD) to maintain a logistics capability to ensure a ready and controlled source of technical competence and resources. In 1988 Congress codified this law, as amended, at section 2464 of title 10 of the U.S. Code. While section 2464 has been amended multiple times since then, the requirement for DOD to maintain a core logistics capability that is government-owned and government-operated has persisted. In 2011 Congress added a requirement for DOD to provide a biennial core report. Most recently, in fiscal year 2018 Congress added additional elements that DOD is required to address in its biennial core reports. Among other things, changes to the statute are illustrated in figure 2 below. Appendix III: Scope and Methodology Section 2464(d) of Title 10 of the United States Code requires the Department of Defense (DOD), among other things, to submit to Congress a biennial report providing information on its core depot-level maintenance and repair capability requirements and workload. Specifically, section 2464(d) identifies 10 elements that DOD must address for each of the armed services (except for the Coast Guard) in its biennial report concerning depot-maintenance requirements and workload. Section 2464 also requires us to review DOD’s report for compliance with section 2464 and assess the completeness of the report. DOD submitted its most recent biennial core report to Congress on May 23, 2018. To determine the extent to which the DOD 2018 Biennial Core Report complies with section 2464(d), we analyzed the text of the report and obtained supporting information on DOD’s process to determine its core maintenance capability for fiscal year 2019. Two GAO analysts independently reviewed DOD’s report to determine the extent to which it addressed each element required by the statute. All initial disagreements between the two GAO analysts were discussed and resolved through consensus. For the military services, when the report explicitly included all parts of the required reporting element, we determined that DOD “addressed” the element. When the report did not explicitly include any part of the element, we determined that DOD “did not address” the element. If the report included some aspects of an element, but not all, then we determined that DOD “partially addressed” the element. We compared the types of information and data provided by each of the military services with the data that the Office of the Secretary of Defense (OSD) included in the 2018 Biennial Core Report, to assess consistency. We also discussed our preliminary analyses with OSD and military service officials to gain additional insight into their analysis and efforts to address the statutory requirements. To assess the report’s completeness, we obtained and analyzed the fiscal year 2019 data used in compiling DOD’s 2018 Biennial Core Report, including core capability requirements and projected sustaining workload expressed in direct labor hours and cost and other information, such as workload shortfall explanations. We compared the reporting agencies’ submissions with the reporting template in DOD Instruction 4151.20 in order to determine the extent to which the reporting agencies submitted the information required by DOD’s instruction, and we identified any inconsistencies or errors. In order to determine whether these data and information were complete, we performed a number of data check steps to identify transposition inconsistencies or errors, and we discussed our analyses with OSD and military service officials. These steps included (1) reviewing each military service’s submission to verify that it had consistently calculated and reported the direct labor hours identified as the total adjusted requirements and the workload needed to sustain depot maintenance core capability requirements; and (2) reconciling the information in the report against each military service’s submission, for accuracy. However, as in the past reviews of DOD’s biennial core reports, we did not assess the reliability of the underlying data provided by the military services for the 2018 DOD Biennial Core Report. The team also met with OSD and reporting agency officials responsible for overseeing the data collection and preparing the data submissions, to obtain clarification and understanding of the content of the submissions, as well as to discuss the department’s guidance and processes used to collect the data for the report. Lastly, we reviewed DOD’s actions to address our prior recommendations that were targeted at improving the completeness of DOD’s biennial report. We conducted this performance audit from May 2018 to November 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix IV: Category Levels from the Department of Defense’s (DOD) Depot Maintenance Core Capability Worksheet Appendix IV: Category Levels from the Department of Defense’s (DOD) Depot Maintenance Core Capability Worksheet Work Breakdown Structure Category 1. Aircraft 1.2 Vertical/Short Takeoff and Landing 2. Appendix V: GAO Contacts and Staff Acknowledgments GAO Contacts Diana Maurer, (202) 512-9627 or maurerd@gao.gov. Staff Acknowledgments In addition to the named contact above, John Bumgarner, Assistant Director; Thomas Gosling, Assistant Director; Pat Donahue, Amie Lesser, Shahrzad Nikoo, Bethann E. Ritter Snyder, Walter Vance, Cheryl Weissman, and Melissa Wohlgemuth contributed to this report.

Why GAO Did This Study DOD uses both military depots and contractors to maintain its complex weapon systems and equipment. Recognizing the depots' key role and the risk of overreliance on contractors, section 2464 of title 10 of the U.S. Code requires DOD to maintain a core logistics capability that is government-owned and operated, involving a combination of personnel, facilities, equipment, processes, and technology. Section 2464 requires DOD to provide a Biennial Core Report to Congress that addresses 10 reporting elements, including information on its core capability requirements and projected workload for the next fiscal year. Section 2464 includes a provision that GAO review DOD's Biennial Core Reports for compliance and completeness. In reviewing the 2018 Biennial Core Report, GAO assessed the extent to which DOD's report (1) addressed the 10 reporting elements required by section 2464(d), and (2) is complete. GAO reviewed and analyzed relevant legislation, DOD guidance, and the 2018 Biennial Core Report, and met with DOD and military service officials to discuss the processes used to develop the information in DOD's 2018 Biennial Core Report. What GAO Found In its 2018 Biennial Core Report, the Department of Defense (DOD) addressed 8 of 10 reporting elements. Specifically, DOD reported, by military service, its: depot maintenance workload required to sustain core maintenance capability requirements, based on contingency planning scenarios; projected fiscal year 2019 depot maintenance workloads; and projected fiscal year 2019 shortfalls (i.e., insufficient workload to sustain the required level of capability) and rationales and mitigations for those shortfalls. The Army reported a projected workload for fiscal year 2019 that would meet about 84 percent of its identified core capability—a shortfall of 2.9 million direct labor hours (see figure). The Army identified numerous reasons—such as newly established software depot maintenance requirements—for its shortfalls. Furthermore, the Army presented mitigation plans for its shortfalls, such as moving software-related work from contractor to military sources. The other services did not report overall shortfalls, but some services reported shortfalls associated with specific types of work. For example, the Air Force reported a shortage associated with the repair of tactical missiles. As a mitigation plan, the Air Force stated that it plans to use workload associated with repairing strategic missiles to maintain this capability, since the electronics on the two types of missiles are very similar and require the same maintenance skill set. DOD did not address two required reporting elements—progress in implementing mitigation plans from the 2016 biennial core report, and the degree to which projected workload reported in the 2016 biennial core report was executed. According to DOD officials, changes in its guidance and processes for developing the 2018 report resulted in the 2016 and 2018 reports not being directly comparable. However, DOD officials stated that they plan to address these two elements in the 2020 Biennial Core Report. DOD's 2018 Biennial Core Report is generally complete, in that it lacks obvious errors and aligns with supporting information provided by the services. DOD's concerted efforts to implement better guidance and procedures—in part, according to DOD officials, by implementing GAO's prior recommendations from 2012, 2014, and 2016—assisted in improving the completeness of the report.

gao_GAO-18-634

gao_GAO-18-634_0

Background Rural Hospitals and Areas In 2017, about 2,250 general acute care hospitals in the United States were located in areas that met FORHP’s definition of rural; these rural hospitals represented approximately 48 percent of hospitals nationwide and 16 percent of inpatient beds. These hospitals were spread across the 84 percent of the United States land area that FORHP classified as rural, and served the 18 percent of the United States population that lived in these areas. While there are significant differences across rural areas and populations, as a whole they differ from their urban counterparts in several ways. For example, rural areas have the following characteristics: Higher percentage of elderly residents. In 2014, 18 percent of the population was aged 65 or older in rural counties, compared with 14 percent in urban counties. Higher percentage of residents with limitations in activities caused by chronic conditions. In 2010-2011, 18 percent of adults in rural counties had limitations in activities caused by chronic health conditions, compared with 13 percent in large, central urban counties. Lower median household income. In 2014, the median household income in rural counties was approximately $44,000, compared to $58,000 in urban counties. Rural areas have also experienced several changes in recent years that have exacerbated these differences. For example, according to research by the United States Department of Agriculture, rural areas have experienced the following changes: Decreasing population. From 2010 through 2015, the population in rural areas declined, on average, by 0.07 percent per year, while the population in urban areas increased, on average, by 0.9 percent per year. Slow employment growth. From 2010 through 2015, rural employment grew at 0.8 percent per year, less than half that of urban areas (1.9 percent per year). Federal Response to Rural Hospital Closures in the 1980s Rural hospital closures are not a recent phenomenon. For example, we previously reported that between 1985 and 1988, 140 rural hospitals closed—approximately 5 percent of the rural hospitals in 1985. The large number of closures in the 1980s was preceded by a change in how Medicare paid hospitals. Specifically, in 1983, Medicare’s inpatient prospective payment system was created, whereby predetermined rates were set for each Medicare hospital discharge. The intent was to control Medicare costs by giving hospitals financial incentives to deliver services more efficiently and reduce unnecessary use of inpatient services by paying a hospital a predetermined amount. However, one consequence of the new payment system was that some small, rural hospitals experienced large Medicare losses and increased financial distress. Partially in response to the number of rural hospital closures, FORHP was established in 1987 to, among other things, advise the Secretary of HHS on the effects of current and proposed policies on the financial viability of small rural hospitals and on access to and quality of health care in rural areas; establish and maintain a clearinghouse for information on rural health coordinate rural health activities within HHS; and administer grants and other instruments to fund activities to improve health care in rural areas. HHS Administers Multiple Payment Policies and Programs That Support Rural Hospitals and Funds Research Centers to Monitor Closures and Access HHS Administers Payment Polices and Programs That Provide Financial Support to Rural Hospitals, Including Medicare Rural Hospital Payment Designations HHS officials identified several rural-specific HHS payment policies and programs as providing key financial support to rural hospitals, and in turn, rural residents’ access to hospital services. These key HHS payment policies and programs may be placed into three categories: (1) Medicare rural hospital payment designations; (2) rural grants, cooperative agreements, and contracts, and (3) new approaches in rural health care delivery and payment (see table 1). Medicare rural hospital payment designations. CMS administers five rural hospital payment designations, in which rural or isolated hospitals that meet specified eligibility criteria receive higher reimbursement for hospital services than they otherwise would have received under Medicare’s standard payment methodology. A rural hospital may qualify as a Critical Access Hospital, Sole Community Hospital, or Medicare Dependent Hospital—each of which has different eligibility criteria and payment methodologies. With the exception of Critical Access Hospitals, rural hospitals may also qualify as Low Volume Hospitals and Rural Referral Centers, in which eligible hospitals receive additional payments or exemptions. The largest of the five designations is the Critical Access Hospital program, which represented 56 percent of rural hospitals in 2017 and pays eligible small, rural hospitals based on their reported costs (instead of the standard rates under the inpatient prospective payment system). (See app. I, table 2, for a description of each of the five Medicare rural hospital payment designations.) CMS was unable to provide estimates of the additional Medicare payments rural hospitals received from each designation in 2017. According to CMS officials, CMS generally does not model the amount of additional Medicare payments resulting from rural hospital payment designations, except in years when there is a related payment policy change going through rulemaking. Rural grants, cooperative agreements, and contracts. FORHP administers multiple grant programs, cooperative agreements, and contracts that provide funding and technical assistance to rural hospitals. The largest of these is the Medicare Rural Hospital Flexibility grant program, in which FORHP provides funds to states to support Critical Access Hospitals to stabilize their finances, foster innovative models of care, and support other improvement activities. In 2017, 45 states received $25 million in Flex grants. FORHP officials noted that they can provide information to help states determine how to best target Flex grant funds, as there is not enough funding to financially assist all Critical Access Hospitals that are at risk of closing. (See app. I, table 3, for a description of the rural grants, and cooperative agreements and contracts identified by HHS officials.) New approaches in rural health care delivery and payment. CMS’s Center for Medicare & Medicaid Innovation (Innovation Center) tests new ways to deliver and pay for health care—including some focused on rural areas—with the goal of reducing spending and preserving or enhancing the quality of care for beneficiaries enrolled in Medicare, Medicaid, and the Children’s Health Insurance Program. As of June 2018, the largest of these rural models and demonstrations was Medicare’s Accountable Care Organization Investment Model. Groups of providers in rural and underserved areas participating in this model, potentially including small hospitals, agree to be held accountable for the cost and quality of care to their Medicare patients. The model tests providing pre-paid shared savings as an incentive for providers in rural and underserved areas to form Accountable Care Organizations and for these organizations to transition to arrangements with greater accountability for financial performance. For fiscal years 2012 through 2018, $96 million had been obligated to organizations participating in the model. Forty-five Accountable Care Organizations were participating in this model as of 2018. (See app. I, table 4, for a description of the new approaches in rural health care delivery and payment identified by HHS officials.) In addition to the HHS payment policies and programs specifically targeting rural areas, HHS officials also identified broader payment policies and programs that they stated can provide key support to rural hospitals and rural residents’ access to hospital services. These HHS payment policies and programs may be placed in four categories: Medicare and Medicaid base payments. These consist of the standard payments for hospitals services. Medicare and Medicaid uncompensated care payments. Both Medicare and Medicaid provide multiple types of additional payments to support hospitals that incur costs for services provided to uninsured and other low-income individuals for which the hospitals are not fully compensated. Medicare also provides bad debt payments to hospitals to reimburse them for a portion of Medicare’s beneficiaries’ unpaid deductibles and coinsurance, as long as the hospital makes a reasonable effort to collect the unpaid amounts. Other targeted HHS payment policies and programs. HHS administers other targeted payment policies and programs that support specific types of providers and areas, including, but not limited to, rural hospitals and areas. In particular, the Health Resources & Services Administration, an HHS agency, administers a drug discount program targeted at certain hospitals and other safety net providers. In addition, CMS administers bonus payments for certain physician services provided to Medicare beneficiaries in areas with a shortage of health professionals. State Innovation Models Initiative. The Center for Medicare & Medicaid Innovation’s State Innovation Models aim to achieve better quality of care, lower costs, and improve health for the population of the participating states or territory. Some states’ plans include testing new delivery and payment models specifically targeting rural areas. HHS Funds Research Centers That Monitor Rural Hospital Closures and Study Access HHS monitors rural hospitals’ financial viability and rural residents’ access to hospital services, primarily by funding rural health research centers that track rural hospital closures and study rural residents’ access to hospital services. To monitor rural hospitals’ financial viability, HHS funds and conducts several activities: Tracking rural hospital closures and monitoring profitability. The North Carolina rural health research center, a FORHP-funded rural health research center, tracks rural hospital closures and monitors rural hospitals’ profitability and other financial indicators. North Carolina’s researchers identify rural hospital closures through a multi- party agreement with FORHP, the American Hospital Association, and the National Rural Health Association, each of which alerts the research center once one learns about a closure. Research center staff then confirm the closure and ascertain whether the hospital converted to another facility type by searching the hospital website and calling a community leader, such as the mayor. The North Carolina rural health research center publishes a list of rural hospital closures since 2010 on its website. It also publishes reports on rural hospitals’ profitability, including the extent to which profitability varies by rural hospitals’ characteristics, and how rural hospitals’ profitability compares to the profitability of their urban counterparts. Monitoring Critical Access Hospitals’ financial indicators. The North Carolina rural health research center, through its role as part of the Flex Monitoring Team, develops and monitors various financial indicators for Critical Access Hospitals. Using the hospitals’ Medicare cost reports, the research center currently monitors 22 financial indicators under 6 domains—profitability, liquidity, capital structure, revenue, cost, and utilization. These financial indicator data are available to every Critical Access Hospital through an online tool that also helps those hospitals compare their financial performance to peer hospitals. The Flex Monitoring Team also publishes state-level summary data on Critical Access Hospitals’ finances that are available on its website. HHS also reviews and estimates the financial effect of policy changes on rural hospitals. In particular, FORHP officials review proposed and final rules for Medicare, Medicaid, and the Affordable Care Act’s health insurance exchanges to identify concerns from a rural health perspective. Drawing on the research it funds, FORHP officials may suggest policy modifications to CMS, such as exempting certain Medicare rural hospital designations from a proposed policy change. In addition to FORHP officials’ review, as required by statute, CMS conducts regulatory impact assessments that estimate the effect of policy changes on payments to hospitals, including small rural hospitals, and publishes key results as part of proposed and final rules. For example, as part of the fiscal year 2018 final rule on Medicare payment for hospital inpatient services, CMS estimated that the expiration of the Medicare Dependent Hospital designation would have decreased the payments to rural hospitals with that designation by 0.9 percent, or approximately $119 million. Subsequent to the final rule, the Medicare Dependent Hospital and Low Volume Hospital designations were both extended. To monitor rural residents’ access to hospital services, HHS relies on research conducted by the FORHP-funded research centers. Examples of recent research on rural residents’ access to hospital services conducted by FORHP-funded research centers include the following: Research on rural residents’ access to hospitals. In 2018 the North Carolina rural health research center published an analysis of populations in rural counties without access to an acute care hospital or other types of primary care facilities. North Carolina’s researchers estimated that about 4.4 million rural residents currently live in a county without an acute care hospital. Research on access to specific hospital services. The Minnesota rural health research center conducted a body of research on declining access to obstetric services in rural counties. These researchers found that between 2004 and 2014, the percent of rural counties without hospital obstetric services increased from 45 to 54 percent, through a combination of hospital and obstetric-unit closures. Research on options for ensuring rural residents’ access after a hospital closure. The Iowa rural health research center published a summary of currently available options for ensuring rural residents’ access to hospital services after a hospital closure, and additional policy options under consideration. The National Advisory Committee on Rural Health and Human Services, a 21-member citizens’ panel of nationally recognized rural health experts that advises HHS, also examined this topic, with a focus on alternative models to preserve rural residents’ access to emergency care in light of the recent surge in rural hospital closures. The committee noted that payments and grants to support rural hospitals were largely effective and stabilized rural hospital financial operations until 2013, when a new wave of rural hospital closures began. The report included recommendations regarding the design of alternative models, including that HHS seek public comments on the use of a combination of geographic distance and demographic or social determinants of health when setting eligibility criteria. To supplement the monitoring by FORHP-funded research centers, FORHP officials also track recent rural developments and reports from rural health stakeholders. FORHP officials said this monitoring adds a qualitative component to the quantitative research conducted by research centers. In particular, these activities often provide the first notice of a rural hospital closure or pending closure, and also help track changes to the status of former hospitals over time. HHS Uses the Results of Its Monitoring Activities to Inform Future Research and Grant Awards, and Disseminates This Information HHS uses the results of its monitoring activities on rural hospitals’ financial viability and rural residents’ access to inform related research, primarily conducted by HHS-funded research centers, and to determine future areas of research. For example, the North Carolina rural health research center has used the list of rural hospital closures it compiles and its monitoring of profitability to conduct research on predictors of rural hospitals’ financial distress. In addition, FORHP officials stated that, based on this monitoring, they have added topics to research centers’ agendas for subsequent years to gather more information on regulatory changes identified in its review of policy changes. Each year, specific research projects for the rural health research center are selected jointly by the center directors and FORHP. Topics are selected to have a timely impact on policy debates and decisions at both federal and state levels. Examples of added topics include North Carolina’s research on the financial importance of the Sole Community Hospital and Low Volume Hospital designations and Iowa’s research on the engagement of rural providers in Accountable Care Organizations. HHS has also used the results of its monitoring activities to update the types of services offered by certain grants and create new cooperative agreements for technical assistance. Specifically, for fiscal year 2016, FORHP officials updated the list of activities that Rural Health Network Development Planning grantees can spend funds on to include implementing innovative solutions to alleviate the loss of local services and enhance access to care in communities that have or are at risk of losing their local hospital. According to FORHP officials, the addition of this activity to the scope of the grant led to 11 of the 47 applicants from fiscal years 2016 and 2017 to come from rural communities with a recent rural hospital closure or perceived risk of closure. As another example, in response to increased funding, in 2018 FORHP announced a new cooperative agreement to provide targeted in-depth assistance to vulnerable rural hospitals within communities struggling to maintain health care services. The awardee of the Vulnerable Rural Hospitals Assistance Program must work with vulnerable hospitals and their communities on ways to ensure hospitals and communities can keep needed care locally, whether it is with a more limited set of services provided by the hospital, or by exploring other mechanisms for meeting community health care needs. FORHP disseminates the results of this research and successful rural health grants and other projects by funding cooperative agreements to maintain clearinghouses of information about rural health issues. These clearinghouses were originally designed to efficiently disseminate research findings from rural health research centers to the public and to help rural communities identify opportunities and information to provide better healthcare to their residents. According to one of these clearinghouses, since then, the focus has grown to developing evidence- based resources on rural health to share what works in rural communities, including toolkits and case studies. Recent Increases in Rural Hospital Closures Have Disproportionately Occurred in the South, With Multiple Factors Likely Contributing to These Closures From 2013 through 2017, More than Twice as Many Rural Hospitals Closed than in the Prior 5 Years Our analysis of data from the North Carolina rural health research center and CMS shows that, from 2013 through 2017, 64 of the approximately 2400 rural hospitals in the United States closed. These 64 rural hospital closures represented the following: More than twice the number of rural hospitals that closed during the prior 5-year period. From 2008 through 2012, 31 rural hospitals closed (see fig. 1). More than the share of urban hospitals that closed. The 64 rural hospital closures from 2013 through 2017—approximately 3 percent of all rural hospitals in 2013—exceeded the 49 urban hospital closures during the same time period—approximately 2 percent of all urban hospitals in 2013. More than the number of rural hospitals that opened. The 42 rural hospitals closed from 2014 through 2016 exceeded the 3 rural hospitals opened during the same time period. Approximately half of the rural hospitals that closed from 2013 through 2017—47 percent—ceased to provide any type of services. The remaining hospitals that closed during this period converted to other facility types, providing more limited or different services, such as urgent care, emergency care, outpatient care, or primary care. Rural Hospitals with Certain Characteristics— Including Those Located in the South—Accounted for a Disproportionate Share of Closures from 2013 through 2017 Our analysis of data from the North Carolina rural health research center and CMS shows that rural hospitals with certain characteristics—including those located in the South—accounted for a disproportionate share of the 64 closures that occurred from 2013 through 2017. Geography. Rural hospitals located in the South represented 38 percent of the rural hospitals in 2013, but accounted for 77 percent of the rural hospital closures from 2013 through 2017 (see fig. 2). Texas, one southern state, represented 7 percent of the rural hospitals in 2013, but accounted for 22 percent of the rural hospitals closures from 2013 through 2017. Medicare rural hospital payment designations. Medicare Dependent Hospitals – one of three Medicare rural hospital payment designations in which hospitals were eligible to receive a payment rate other than standard Medicare inpatient payment rate – were disproportionately represented among hospital closures. Specifically, Medicare Dependent Hospitals represented 9 percent of the rural hospitals in 2013, but accounted for 25 percent of the rural hospital closures from 2013 through 2017. Rural hospitals that did not receive one of these three Medicare rural hospital payment designations also represented a disproportionate share of the closures (see fig. 3). In addition, hospitals designated as Low Volume Hospitals had a disproportionate share of the rural hospital closures. Ownership. For-profit rural hospitals represented 11 percent of the rural hospitals in 2013, but accounted for 36 percent of the rural hospital closures from 2013 through 2017 (see fig. 4). According to literature we reviewed, hospitals with for-profit status had a higher probability of financial distress and were more likely to close. For example, a 2017 study found that for-profit hospitals were more than twice as likely to experience financial distress relative to government- owned and non-profit hospitals from 2000 through 2013. Bed size. Rural hospitals with between 26 and 49 inpatient beds represented 11 percent of the rural hospitals in 2013, but accounted for 23 percent of the rural hospital closures from 2013 through 2017. Critical Access Hospitals have 25 acute inpatient beds or less and make up a majority of the rural hospitals, but were less likely than other rural hospitals to close. FORHP officials identified the Critical Access Hospital payment designation – in which Medicare pays designated hospitals based on their costs – paired with the related Medicare Rural Hospital Flexibility grant program as the most effective HHS payment policy and program to support rural hospitals’ financial viability and rural residents’ access to hospital services. Fewer Patients Seeking Inpatient Care and Reductions in Medicare Payments Have Likely Contributed to Rural Hospital Closures According to literature we reviewed and stakeholders we interviewed, rural hospital closures were generally preceded and caused by financial distress. In particular, rural hospitals that closed typically had negative margins which made it difficult to cover their fixed costs. For example, one 2016 study found that rural hospitals that closed from 2010 through 2014 had a median operating margin of -7.41 percent in 2009. In contrast, rural hospitals that remained open during the same time period had a median operating margin of 2.00 percent in 2009. In addition, there is evidence that for-profit hospitals have been more sensitive to changes in profitability and more likely to experience financial distress, which could explain the disproportionate number of closures among rural hospitals with for-profit ownership type. The literature we reviewed and stakeholders we interviewed identified multiple factors that likely contributed to increased financial distress and closures among rural hospitals. One such factor was a decrease in patients seeking inpatient care at rural hospitals due to the following: Increased competition for the small volume of rural residents. Rural residents may choose to obtain services from other health care providers separate from the local rural hospital, for example from an increasing number of federally qualified health centers or newer hospital systems outside of the area. The competition for the small volume of rural residents between rural hospitals and other health care providers potentially increased due to the shift to paying for value instead of volume, and technology changes. This increased competition for a small volume of rural residents could explain disproportionate closures among hospitals receiving the Low Volume Hospital Medicare payment designation, hospitals that by definition have a low Medicare volume and that research has found have lower margins than other rural hospitals. In addition, representatives from the American Hospital Association told us that technological advances have allowed more services to be provided in outpatient settings. For example, changes in health care technology have expanded the provision of outpatient surgical procedures. Declining rural population. The years 2010 through 2016 marked the first recorded period of rural population decline. According to literature we reviewed and stakeholders we interviewed, the recent population decline in rural areas was likely associated with the recent decline in rural residents seeking inpatient services. Another factor highlighted by literature we reviewed and stakeholders we interviewed as contributing to rural hospitals’ increased financial distress was across-the-board Medicare payment reductions. Rural hospitals are sensitive to changes to Medicare payments because, on average, Medicare accounted for approximately 46 percent of their gross patient revenues in 2016. A 2016 study found that Medicare Dependent Hospitals’ operating margins decreased each year from 2012 through 2014, which could explain the disproportionate number of closures among the Medicare Dependent Hospital payment designation. The literature we reviewed and stakeholders we interviewed highlighted the recent Medicare payments cuts as contributing to rural hospital closures, which included the following: Reductions in nearly all Medicare reimbursements. Under sequestration – the cancellation of budgetary resources under presidential order implemented pursuant to the Balanced Budget and Emergency Deficit Control Act of 1985, as amended – each fiscal year since 2013, nearly all Medicare’s budget authority is subject to a reduction not exceeding 2 percent, which is implemented through reductions in payment amounts. According to stakeholders we interviewed, these payment reductions have contributed to negative margins for rural hospitals. Reductions in Medicare bad debt payments. Under the Middle Class Tax Relief and Job Creation Act of 2012, Medicare bad debt reimbursements for hospitals were reduced beginning in fiscal year 2013. According to stakeholders, Medicare bad debt cuts have been one of the most important factors contributing to the recent increase in rural hospital closures. The literature we reviewed and stakeholders we interviewed also identified factors that likely strengthened the financial viability of rural hospitals. Chief among these factors was the increased Medicaid eligibility and enrollment under the Patient Protection and Affordable Care Act. A 2018 study found that Medicaid expansion was associated with improved hospital financial performance and substantially lower likelihood of closure, especially in rural markets and counties with large numbers of uninsured adults before Medicaid expansion. Another 2017 study found that from 2008-2009 and 2014-2015 the drop in uninsured rates corresponded with states’ decisions to expand Medicaid on or before January 1, 2014. The increase in Medicaid coverage and decline in uninsured were both largest in the small towns and rural areas of those expansion states. Additionally, our analysis of data from the North Carolina rural health research center and CMS shows that from 2013 through 2017, rural hospitals in states that had expanded Medicaid as of April 2018 were less likely to close compared with rural hospitals in states that had not expanded Medicaid (see fig. 5). Agency Comments We provided a draft of this report to HHS for comment. The Department provided technical comments, which we incorporated as appropriate. As agreed with your offices, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies of this report to the appropriate congressional committees, the Secretary of Health and Human Services, the Administrator of Health Resources & Services Administration, the Administrator of CMS, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staffs have any questions about this report, please contact me at (202) 512-7114 or cosgrovej@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix II. Appendix I: Rural-Specific Payment Policies and Programs That Provide Key Support to Rural Hospitals Officials from the Department of Health and Human Services (HHS) identified several rural-specific HHS payment policies and programs as providing key support to rural hospitals, and in turn, rural residents’ access to hospital services. These key HHS payment policies and programs may be placed into three categories: Medicare rural hospital payment designations (table 2); Rural grants, cooperative agreements and contracts (table 3); and New approaches in rural health care delivery and payment (table 4). Appendix II: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Greg Giusto (Assistant Director), Alison Binkowski (Analyst-in-Charge), George Bogart, Zhi Boon, Leia Dickerson, Krister Friday, Mike Hoffman, Peter Mann-King, Beth Morrison, Vikki Porter, Merrile Sing, and Chris Woika made key contributions to this report.

Why GAO Did This Study Research has shown that hospital closures can affect rural residents' access to health care services and that certain rural residents—particularly those who are elderly and low income—may be especially affected by rural hospital closures. This report describes (1) how HHS supports and monitors rural hospitals' financial viability and rural residents' access to hospital services and (2) the number and characteristics of rural hospitals that have closed in recent years and what is known about the factors that have contributed to those closures. GAO reviewed documents and interviewed officials from HHS and HHS-funded research centers; analyzed data compiled by HHS and an HHS-funded research center, with a focus on 2013 through 2017—the most recent year with complete data; reviewed relevant literature; and interviewed experts and stakeholders. GAO identified hospitals as rural if they met the Federal Office of Rural Health Policy's definition of rural. GAO provided a draft of this report to HHS for comment. The Department provided technical comments, which GAO incorporated as appropriate. What GAO Found The Department of Health and Human Services (HHS) administers multiple payment policies and programs that provide financial support for rural hospitals and funds research centers to monitor closures and study access. Among the payment policies administered by HHS are special payment designations for rural hospitals in which rural hospitals that meet certain criteria receive higher reimbursements for hospital services than they otherwise would receive under Medicare's standard payment methodology. HHS-funded research centers monitor rural hospitals' profitability and other financial indicators, and study access to facilities and specific services. HHS uses the results of monitoring activities to inform future areas of research and disseminate information. GAO's analysis of data from HHS and an HHS-funded research center shows that 64 rural hospitals closed from 2013 through 2017. This represents approximately 3 percent of all the rural hospitals in 2013 and more than twice the number of closures of the prior 5-year period. GAO's analysis further shows that rural hospital closures disproportionately occurred in the South, among for-profit hospitals, and among hospitals that received the Medicare Dependent Hospital payment designation, one of the special Medicare payment designations for rural hospitals. According to literature GAO reviewed and stakeholders GAO interviewed, rural hospital closures were generally preceded and caused by financial distress. In particular, rural hospitals that closed typically had negative margins that made it difficult to cover their fixed costs. According to these sources, financial distress has been exacerbated in recent years by multiple factors, including the decrease in patients seeking inpatient care and across-the-board Medicare payment reductions. In contrast, according to the literature GAO reviewed and stakeholders GAO interviewed, rural hospitals located in states that increased Medicaid eligibility and enrollment experienced fewer closures.

gao_GAO-18-264

gao_GAO-18-264_0

Background The federal government’s increasing demand for IT has led to an increase in the number of federal data centers and a corresponding increase in operational costs. According to OMB, the federal government had 432 data centers in 1998, 2,094 in July 2010, and 9,995 in August 2016. Operating such a large number of centers has been, and continues to be, a significant cost to the federal government. For example, in 2007, the Environmental Protection Agency (EPA) estimated that the electricity costs to operate federal servers and data centers across the government were about $450 million annually. According to the Department of Energy (Energy), a typical data center has 100 to 200 times the energy use intensity of a commercial building. In 2009, OMB reported that server utilization rates as low as 5 percent across the federal government’s estimated 150,000 servers were a factor driving the need to establish a coordinated, government-wide effort to improve the efficiency, performance, and environmental footprint of federal data center activities. OMB and the Federal CIO Established FDCCI Concerned about the size of the federal data center inventory and the potential to improve the efficiency, performance, and the environmental footprint of federal data center activities, OMB’s Federal CIO established FDCCI in February 2010. This initiative’s four high-level goals were to reduce the overall energy and real estate footprint of government data centers; reduce the cost of data center hardware, software, and operations; increase the overall IT security posture of the government; and shift IT investments to more efficient computing platforms and technologies. In February 2010, OMB required all of the agencies participating in the FDCCI to submit a data center inventory and a consolidation plan. In October 2010, OMB also clarified the definition of a data center and noted that, for the purposes of FDCCI, a data center was to be defined as any room used for the purpose of processing or storing data that is larger than 500 square feet and meets stringent availability requirements. Under this definition, OMB reported that agencies had identified 2,094 data centers as of July 2010. However, in 2011, the Federal CIO expanded the definition to include a facility of any size and OMB published its revised definition in March 2012. Based on the revised definition, OMB estimated that there were a total of 3,133 federal data centers in December 2011. In addition, its goal was to consolidate approximately 40 percent, or 1,253 of these data centers, for a savings of approximately $3 billion by the end of 2015. Figure 1 provides an example of data center server racks at the Social Security Administration’s (SSA) National Support Center. The number of federal data centers reported by agencies has continued to grow since 2011. In March 2016, we reported that agencies had collectively identified a total of 10,584 data centers as of November 2015—an increase of about 7,500 data centers compared to OMB’s October 2011 estimate. According to the Federal CIO, the increase in the number of data centers was primarily due to the expanded definition of a data center and improved inventory reporting by the agencies. Further, OMB placed greater emphasis on data center optimization when it issued memorandum M-13-09 in March 2013. Specifically, OMB stated that, to more effectively measure the efficiency of an agency’s data center assets, agencies would also be measured by the extent to which their primary data centers are optimized for total cost of ownership by incorporating metrics for data center energy, facility, labor, and storage, among other things. Subsequently, in May 2014, OMB issued memorandum M-14-08, which established a set of data center optimization metrics to measure agency progress, along with target values for each metric. All agencies were expected to achieve the target values by the end of fiscal year 2015. IT Acquisition Reform Law Enhanced Data Center Consolidation and Optimization Efforts Recognizing the importance of reforming the government-wide management of IT, Congress enacted FITARA in December 2014. Among other things, the law required agencies to: Submit to OMB a comprehensive inventory of the data centers owned, operated, or maintained by or on behalf of the agency. Submit, by the end of fiscal year 2016, a multi-year strategy to achieve the consolidation and optimization of the agency’s data centers. The strategy was to include performance metrics that were consistent with the government-wide data center consolidation and optimization metrics. Report progress towards meeting government-wide data center consolidation and optimization metrics on a quarterly basis to OMB’s Administrator of the Office of Electronic Government. In addition, according to FITARA, the Office of Electronic Government at OMB was to: Establish metrics applicable to the consolidation and optimization of data centers (including server efficiency), ensure that agencies’ progress toward meeting government-wide data center consolidation and optimization metrics is made publicly available, review agencies’ inventories and strategies to determine whether they are comprehensive and complete, and monitor the implementation of each agency’s strategy. Develop and make publicly available not later than December 19, 2015, a goal broken down by year, for the amount of planned cost savings and optimization improvements to be achieved through FDCCI and, for each year thereafter until October 1, 2020, compare reported cost savings and optimization improvements against those goals. OMB Established DCOI In August 2016, OMB issued memorandum M-16-19, which established DCOI and included guidance on how to implement the data center consolidation and optimization provisions of FITARA. Among other things, the guidance required agencies to consolidate inefficient infrastructure, optimize existing facilities, improve their security posture, and achieve cost savings. For example, each agency was required to maintain a complete inventory of all data center facilities owned, operated, or maintained by or on its behalf, and measure progress toward defined optimization performance metrics on a quarterly basis as part of its data center inventory submission. OMB’s memorandum also directed each agency to develop a DCOI strategic plan that defined its data center strategy for fiscal years 2016 through 2018. Among other things, this strategy was to include a timeline for agency consolidation and optimization activities with an emphasis on cost savings and optimization performance benchmarks that the agency could achieve between fiscal years 2016 and 2018. For example, each agency was required to develop cost savings targets due to consolidation and optimization actions and report any realized cost savings. OMB required each agency to publicly post its DCOI strategic plan to its agency-owned digital strategy website by September 30, 2016, and to post subsequent strategic plan updates by April 14, 2017 and April 13, 2018. Further, the memorandum stated that OMB was to maintain a public dashboard (referred to as the IT Dashboard) to display government-wide and agency-specific progress in areas such as planned and achieved data center closures, consolidation-related cost savings, and data center optimization performance information. In this regard, OMB began including data center consolidation and optimization progress information on the IT Dashboard in August 2016. OMB’s memorandum also provided new guidance for the classification of a physical data center, expanding the definition of a data center. According to the revised definition, a room with at least one server that provides services (whether in a production, test, staging, development, or any other environment) should be considered a data center, while a room containing only print servers, routing equipment, switches, security devices (such as firewalls), or other telecommunication components, was not to be considered a data center. In light of this new definition, OMB directed each agency to perform a comprehensive review of its data centers and maintain a complete and updated data center inventory. Further, OMB directed each agency to categorize each of its data centers as either a tiered data center or a non- tiered data center. OMB’s memorandum defined a tiered data center as one that uses each of the following: a separate physical space for IT infrastructure; an uninterruptible power supply; a dedicated cooling system or zone; and a backup power generator for a prolonged power outage. According to the memorandum, all other data centers are to be considered non-tiered. Moreover, OMB guidance included a series of performance metrics in the areas of data center closures, cost savings, and optimization progress. Data center closures: Agencies are expected to close at least 25 percent of tiered data centers government-wide, excluding those approved as inter-agency shared services providers, by the end of fiscal year 2018. Further, agencies are to close at least 60 percent of non-tiered data centers government-wide by the end of fiscal year 2018. OMB’s guidance further notes that, in the long term, all agencies should continually strive to close all non-tiered data centers, noting that server rooms and closets pose security risks and management challenges and are an inefficient use of resources. Cost savings: Agencies are expected to reduce government-wide annual costs attributable to physical data centers by at least 25 percent, resulting in savings of at least $2.7 billion for fiscal years 2016 through 2018. Data center optimization: Agencies are expected to measure progress against a series of new data center performance metrics in the areas of server utilization, energy metering, power usage, facility utilization, and virtualization. Further, OMB’s guidance established target values for each metric that agencies are to achieve by the end of fiscal year 2018. OMB’s guidance further noted that agency progress against these performance metrics is to be measured by OMB on a quarterly basis, using agencies’ data center inventory submissions and OMB-defined closures, cost savings, and optimization targets. Agencies Have Taken Limited Action to Address GAO Recommendations from Prior Reports on Agencies’ Consolidation and Optimization Efforts Since the enactment of FITARA in December 2014, we have annually reviewed and verified the quality and completeness of each covered agency’s inventory and DCOI strategy. We have also published reports documenting the findings from each of these reviews. In addition, we have examined and reported on agencies’ efforts to optimize their data centers, as well as the challenges encountered and successes achieved. In a report that we issued in March 2016, we noted that agencies had reported significant data center closures—totaling more than 3,100 through fiscal year 2015—with the Departments of Agriculture (Agriculture), Defense (Defense), the Interior (Interior), and the Treasury (Treasury) accounting for 84 percent of the total. Although agencies fell short of OMB’s fiscal year 2015 consolidation goal, their plans identified about 2,100 additional centers planned for closure through fiscal year 2019. Agencies also reported significant consolidation cost savings and avoidances—totaling about $2.8 billion through fiscal year 2015, and expected to increase to over $8.0 billion in future years. The Departments of Commerce (Commerce), Defense, Homeland Security (DHS), Transportation (Transportation), and the Treasury accounted for 96 percent of the total planned savings. However, we pointed out that many agencies lacked complete cost savings goals for the next several years despite having closures planned. In addition, we reported that 22 agencies had made limited progress against OMB’s fiscal year 2015 data center optimization performance metrics, such as the utilization of data center facilities. Accordingly, we recommended that the agencies take actions to complete their cost savings targets and improve optimization progress. Of the 24 agencies to which we made recommendations, 14 agreed with our recommendations, 4 did not state whether they agreed or disagreed, and 6 stated that they had no comments. As of March 2018, 26 of the 32 recommendations from this report had yet to be fully addressed. In May 2017, we reported that the agencies continued to report significant data center closures—totaling more than 4,300 through August 2016— with Agriculture, Defense, Interior, and the Treasury accounting for 84 percent of the total. The agencies’ plans for 2016 had identified more than 1,200 additional centers planned for closure through fiscal year 2019. Agencies also reported significant consolidation and optimization cost savings and avoidances, which totaled about $2.3 billion through August 2016. However, reductions in the amount of achieved savings reported to OMB, particularly by the Treasury, resulted in a net decrease of more than $400 million in these savings, compared to amounts we previously reported in 2015. Further, our report noted that, as of December 2016, agencies’ total planned cost savings of about $656 million were more than $3.3 billion less, compared to the amounts that we reported in 2015, and more than $2 billion less than OMB’s fiscal year 2018 cost savings goal of $2.7 billion. This reduction in planned savings was the result of eight agencies reporting less in planned cost savings and avoidances in their DCOI strategic plans compared to the savings amounts previously reported to us in November 2015. The reduction also reflected the absence of cost savings information for one agency (Defense) that did not submit its strategic plan in time for our review. In addition, our May 2017 report identified weaknesses in agencies’ DCOI strategic plans. Of the 23 agencies that had submitted their strategic plans at the time of our review, 7 agencies—Agriculture, the Department of Education (Education), DHS, and the Department of Housing and Urban Development (HUD); the General Services Administration (GSA); the National Science Foundation (NSF); and the Office of Personnel Management (OPM)—had addressed all five required elements of a strategic plan, as identified by OMB (such as providing information related to data center closures and cost savings metrics). The remaining 16 agencies that submitted their plans either partially met or did not meet the requirements. We also pointed out that there were inconsistencies in the reporting of cost savings in the strategic plans of 11 agencies. Given these findings, we recommended that OMB improve its oversight of agencies’ DCOI strategic plans and their reporting of cost savings and avoidances. We also recommended that 16 agencies and Defense (which did not submit a plan in time for our review) complete the missing elements in their strategic plans, and that 11 agencies ensure the reporting of consistent cost savings and avoidance information to OMB. Of the 25 agencies (including OMB) to which we made recommendations, 12 agreed with our recommendations, 2 disagreed, and 11 did not state whether they agreed or disagreed. As of March 2018, 29 of the 30 recommendations had not been fully addressed. In a subsequent report that we issued in August 2017, we noted that 22 of the 24 agencies required to participate in the OMB DCOI collectively had reported limited progress against OMB’s fiscal year 2018 performance targets for the five optimization metrics. The 2 remaining agencies, Education and HUD, did not have agency-owned data centers and, therefore, did not have a basis to report on progress. Specifically, for each of the five targets, no more than 5 agencies reported that they had met or exceeded that specific target. This limited progress against OMB’s optimization targets was due, in part, to agencies not fully addressing our prior recommendations in this area. In addition, we noted in the report that most agencies had not yet implemented automated monitoring tools to measure server utilization, as required by the end of fiscal year 2018. Specifically, 4 agencies reported that they had fully implemented such tools, 18 reported that they had not yet done so, and 2 did not have a basis to report on progress because they did not have agency-owned data centers. We also noted that, although federal standards emphasize the need to establish plans to help ensure goals are met, none of the 18 agencies had fully documented plans for implementing automated monitoring tools. Accordingly, we recommended that OMB formally document a requirement for agencies to include plans, as part of existing OMB reporting mechanisms, to implement automated monitoring tools at their agency-owned data centers. We also recommended that the 18 agencies without fully documented plans take action, within existing OMB reporting mechanisms, to complete plans describing how they intend to achieve OMB’s requirement to implement automated monitoring tools at all agency-owned data centers by the end of fiscal year 2018. Of the 19 agencies (including OMB) to which we made recommendations, 10 agreed with our recommendations, 3 partially agreed, and 6 did not state whether they agreed or disagreed. As of March 2018, none of the 19 recommendations had been fully addressed. Agencies Report Mixed Results Relative to Achieving OMB’s Data Center Closure Targets As previously mentioned, in August 2016, OMB established a goal to close at least 25 percent of each agency’s tiered data centers and at least 60 percent of each agency’s non-tiered data centers by the end of fiscal year 2018. Related to doing so, agencies’ August 2016 inventories reported a total of 9,995 data centers and, in August 2017, 14 agencies reported an additional 2,067 facilities, for a total of 12,062 data centers. Based on this revised inventory, agencies will need to close 6,306 data centers (665 tiered and 5,641 non-tiered) to meet OMB’s goals by the end of fiscal year 2018. Toward this end, the 24 agencies participating in DCOI collectively have made progress on their data center closure efforts. Specifically, as of August 2017, the agencies reported that they had closed 5,805 tiered and non-tiered centers (48 percent). Figure 2 provides a summary of the total number of federal data centers and closures reported from 1998 to August 2017. Figure 3 provides a further breakdown of agencies’ data center inventories, as of August 2017, in terms of the total number of data centers that were closed, planned for closure, or not planned for closure. Nevertheless, while the agencies collectively had made progress toward OMB’s closure goals, the 24 agencies’ individual reported progress and plans showed mixed results when compared with OMB’s goal for each agency to close at least 25 percent of tiered data centers and at least 60 percent of non-tiered centers. Specifically, as of August 2017, 13 agencies reported that they had already met the goal of closing 25 percent of their tiered data centers, another 4 agencies reported that they plan to meet the goal by the end of fiscal year 2018, and 5 agencies reported that they do not currently plan to meet the goal. Further, as of August 2017, 7 agencies reported having already met the goal for closing 60 percent of their non-tiered centers, 6 agencies reported that they planned to meet the goal by the end of fiscal year 2018, and 10 agencies reported that they did not plan to meet the goal. Table 1 displays a breakdown of the number of reported tiered and non- tiered data centers and completed and planned closures by agency, as of August 2017. As shown in the table, the reported closures of Agriculture (2,233 data centers), Defense (834), and Treasury (1,713) together accounted for 4,780 (or 82 percent) of the 5,805 data center closures. However, no other agency accounted for more than 187 (or 3 percent) of those closures. In addition, the remaining 1,416 planned closures are to be carried out across 20 agencies. Further, among the agencies, 7 reported that they do not plan to meet one of their tiered or non-tiered closure goals, and 4 reported that they do not plan to meet either of the goals by the end of fiscal year 2018. Officials from these 11 agencies that do not plan to meet one or both of their goals provided various reasons for why they currently do not plan to do so. For example, officials in USAID’s Office of the Chief Information Officer stated that their agency had reported a number of server closets in overseas locations as non-tiered data centers to comply with OMB’s data center definition provided in its August 2016 guidance. The officials said that, as long as USAID maintains those locations, the agency needs the services provided in those server closets and will not be able to close them to meet OMB’s goal. However, the officials also said that the agency is exploring ways to replace the server closets using cloud services. Several agencies that viewed their goals as unattainable indicated that they were seeking revised closure goals. Specifically, officials from Interior’s Office of the Chief Information Officer stated that a number of the department’s non-tiered data centers were either mission critical or not cost effective to close. Thus, the officials said Interior was working with OMB to establish a revised closure goal. Similarly, Transportation’s Director for IT Compliance stated that the department was working with OMB to establish a revised closure goal. The department reported having 186 tiered data centers in Federal Aviation Administration control towers that it believes should be excluded from its count of data centers when OMB sets the department’s goal for closures. Further, our analysis determined that it may not always be realistic for an agency to meet OMB’s targets. For example, NSF, which does not have any tiered data centers, reported that it plans to close one (50 percent) of its two non-tiered data centers. However, the only way to meet OMB’s 60 percent threshold would be for the agency to close both of its non-tiered centers, which may not be an option, depending on the services provided by that one remaining center. Although OMB’s stated time frame for closing data centers currently remains as the end of fiscal year 2018, the recent extension of FITARA’s data center consolidation and optimization provisions through fiscal year 2020 provides agencies additional time to work toward meeting OMB’s closure targets. However, in some cases, these efforts may require significant restructuring of an agency’s business operations, or, as reported by several agencies, a revision of OMB’s goals in consideration of the agencies’ specific needs. Given that agencies had been working toward OMB’s DCOI goals for approximately one year as of August 2017, and because the extension of FITARA’s data center provisions pushes the sunset for these efforts out through fiscal year 2020, we are not making any related recommendations to those agencies that have not met the closure goals at this time. We plan to continue to monitor the agencies’ progress toward meeting the goals in our future work. Agencies’ Planned Savings Are Not Expected to Meet OMB’s Goal Since 2013, federal agencies have been required to report on data center cost savings. In this regard, OMB provided guidance regarding how agencies were to report cost savings and avoidances. Specifically, it required agencies to report both data center consolidation cost savings and avoidances, among other areas, as part of a quarterly data collection process known as the integrated data collection. FITARA also called for each agency to submit a multi-year strategy for achieving the consolidation and optimization of data centers that includes year-by-year calculations of investment and cost savings through fiscal year 2018, which has now been extended to 2020. In addition, in August 2016, OMB M-16-19 provided guidance on how agencies should implement the requirements of FITARA. Specifically, agencies were to develop a strategic plan that included information on historical cost savings and avoidances due to data center consolidation and optimization through fiscal year 2015. This guidance stated that agency strategic plans were also to include year-by-year calculations of target and actual agency-wide spending and cost savings on data centers from fiscal years 2016 through 2018. Further, the guidance established a DCOI government-wide cost savings goal of $2.7 billion for all federal agencies to achieve from fiscal years 2016 through 2018. This overall goal is then broken down into agency-specific targets on the IT Dashboard. As of August 2017, 20 agencies had reported through the integrated data collection that they had achieved $1.04 billion in cost savings for fiscal years 2016 and 2017, while 4 agencies reported that they had not achieved any savings. Further, the 20 agencies’ DCOI strategic plans identified an additional $0.58 billion, for a total of $1.62 billion in planned savings from fiscal years 2016 through 2018. Nevertheless, this total is about $1.12 billion below OMB’s goal of $2.7 billion for DCOI savings. Figure 4 provides a comparison of the 24 agencies’ total reported savings for fiscal years 2016 and 2017, and the planned savings through 2018, against OMB’s DCOI savings goal for fiscal years 2016 through 2018. Among the 24 participating DCOI agencies that reported achieving about $1.04 billion in savings, Commerce ($594.28 million), Defense ($141.36 million), and DHS ($106.51 million) were responsible for approximately $842 million (about 81 percent) of that total. No other agency reported saving more than $54.40 million. Table 2 provides specific data related to each agency’s planned and achieved savings for fiscal years 2016 and 2017, as of August 2017. In addition, the 24 agencies reported that they have planned an additional $0.58 billion in DCOI cost savings (for a total of $1.62 billion) through fiscal year 2018. However, as noted earlier, this total is approximately $1.12 billion below OMB’s $2.7 billion goal for the initiative. Table 3 provides a more detailed comparison between each agency’s planned savings, as reported in its DCOI strategic plan, and OMB’s agency- specific targets, as reported on the IT Dashboard. As shown in table 3, 6 agencies identified planned savings that are expected to meet or exceed their OMB targets, while 2 agencies that did not have an OMB target also identified planned savings. In contrast, 12 agencies reported that they are not currently planning to meet their targets, and 4 agencies did not have a savings target and are not planning any savings. These findings align with what we reported in March 2016, when we noted that 10 agencies had not established planned cost savings goals for fiscal years 2016 through 2018, even though they had closures planned during that time period. Accordingly, in that report, we recommended that these agencies complete their planned data center cost savings targets for fiscal years 2016 through 2018. Most of the agencies agreed with the recommendations or had no comments. Nonetheless, agencies continued to be challenged in identifying and reporting their cost savings. As of August 2017, 5 of the agencies had implemented our recommendations in this area. In the absence of consistent and full reporting of fiscal years 2016 through 2018 planned savings in agencies’ DCOI strategic plans, as required by FITARA and OMB, agencies’ total planned savings will likely continue to be understated. With less than a year for agencies to meet OMB’s current planned savings targets, we are re-emphasizing the need for agencies to implement our prior recommendations related to establishing and meeting their planned data center cost savings targets. Agencies Continue Reporting Limited Progress against OMB’s Data Center Optimization Metrics FITARA required OMB to establish metrics to measure the optimization of data centers, including server efficiency, and ensure that agencies’ progress toward meeting the metrics is made public. Pursuant to FITARA, OMB’s August 2016 memorandum established a set of five data center optimization metrics intended to measure agencies’ progress in the areas of server utilization and automated monitoring, energy metering, power usage effectiveness, facility utilization, and virtualization. According to OMB, while the server utilization and automated monitoring metric applies to agency-owned tiered and non-tiered data centers, the four remaining metrics apply only to agency-owned tiered centers. OMB’s memorandum also established a target value for each of the five metrics, which agencies are expected to achieve by the end of fiscal year 2018. OMB measures agencies’ progress against the optimization targets using the agencies’ quarterly data center inventory submission and publicly reports this progress information on its IT Dashboard. Table 4 provides a description of the five data center optimization metrics and target values. As of August 2017, 22 of the 24 DCOI agencies continued to report limited progress in meeting OMB’s fiscal year 2018 data center optimization targets identified on the IT Dashboard. As noted earlier, the remaining 2 agencies—Education and HUD—reported that they did not have any agency-owned data centers in their inventory and, therefore, did not have a basis to measure and report optimization progress. With regard to the data center optimization targets, the most progress was reported for the power usage effectiveness and virtualization metrics, with 5 and 6 agencies, respectively, reporting that they had met OMB’s targets. However, only 3 agencies or fewer reported meeting the energy metering, facility utilization, and server utilization and automated monitoring metrics. Figure 5 summarizes the 24 agencies’ progress in meeting each optimization target, as of August 2017. As of August 2017, SSA and EPA reported the most progress among the 22 agencies with a basis to report against OMB’s metrics targets— meeting 4 and 3 targets, respectively. Six agencies met either one or two targets, and 14 agencies reported meeting none of the targets. Further, of the 22 agencies, 9 were not able to report any progress against either the server utilization metric or power usage effectiveness metric, or both, because their data centers lacked the required monitoring tools to measure progress in these areas. OMB began requiring the implementation of these monitoring tools in August 2016; however, as of August 2017, these 9 agencies had not reported that they had implemented the tools at any data centers. The remaining 13 agencies reported that they had implemented the tools in at least one data center. Table 5 depicts the agencies and whether they met or did not meet each OMB target. Agencies’ limited progress against OMB’s optimization targets is due, in part, to not fully addressing our prior recommendations in this area. As discussed earlier, in March 2016, we reported on weaknesses in agencies’ data center optimization efforts, including that 22 agencies did not meet OMB’s fiscal year 2015 optimization targets. We noted that this was partially due to the agencies facing challenges in optimizing their data centers, including their decentralized organizational structures that made consolidation and optimization difficult, and competing priorities for resources. In addition, consolidating certain data centers was problematic because the volume or type of information involved required the data center to be close in proximity to the users. Accordingly, we recommended that the agencies take action to improve optimization progress, to include addressing any identified challenges. Most agencies agreed with our recommendations or had no comments. In response to our recommendation, 19 of the 22 agencies submitted documentation to us that described steps they intended to take to improve their data center optimization efforts. The planned steps included completion dates ranging between April 2016 and September 2019. Among the steps described by the agencies was developing internal scorecards to track and report on optimization progress, including progress at their component agencies, and launching more aggressive efforts to optimize data centers using virtualization and cloud computing solutions. However, as of February 2018, only 1 of the 22 agencies (Education) had fully addressed our recommendation. Agencies’ Planned Optimization Progress Will Not Meet OMB’s Fiscal Year 2018 Targets In addition to reporting current optimization progress on the IT Dashboard, OMB requires agencies’ DCOI strategic plans to include, among other things, planned performance levels for fiscal years 2017 and 2018 for each optimization metric. However, according to the 24 agencies’ DCOI strategic plan information as of August 2017, most are not planning to meet OMB’s optimization targets by the end of fiscal year 2018. More specifically, of the 24 agencies, only 4—Commerce, EPA, NSF, and USAID—reported plans to fully meet their applicable targets by the end of fiscal year 2018. Of the remaining agencies, 14 reported plans to meet some, but not all, of the targets; 4 reported that they do not plan to meet any targets; and—as already discussed—Education and HUD do not have a basis to report planned optimization milestones because they do not report having any agency-owned data centers. Figure 6 summarizes agencies’ progress, as of August 2017, in meeting OMB’s optimization targets and planned progress to be achieved by September 2018. The limited progress made by agencies in optimizing their data centers, combined with the lack of established plans to improve progress, makes it unclear whether agencies will be able to achieve OMB’s optimization targets by the end of fiscal year 2018. Considering that OMB is expecting at least $2.7 billion in cost savings from agencies’ optimization efforts, the ability of agencies to meet the optimization targets will be critical to meeting this savings goal. However, only four agencies are planning to meet all of their applicable targets. If the remaining agencies take steps to implement the prior recommendations we have made in this area, it should increase the likelihood that DCOI can achieve the expected benefits of optimization and the resulting cost savings. Agency Comments and Our Evaluation We requested comments on a draft of this report from OMB and the 24 other agencies that we reviewed. Of these, 5 agencies indicated that they agreed with our report and 20 acknowledged receiving the draft, but did not state whether they agreed or disagreed with the report. Agencies also offered various comments in support of the DCOI effort and actions taken to improve performance. Additionally, multiple agencies provided technical comments, which we have incorporated, as appropriate. The following five agencies agreed with our report: In emails received from Agriculture, Energy, VA, and NRC, the agencies agreed with the findings in the draft report. In written comments, SSA agreed with the draft report’s characterization of the agency’s DCOI efforts. SSA’s comments are reprinted in appendix II. In addition, the following 20 agencies did not state whether they agreed or disagreed with the report and offered other comments: In written comments, HUD and Transportation did not agree or disagree with the draft report. The departments’ comments are reprinted in appendices III and IV, respectively. In emails received from Commerce, Defense, Education, HHS, DHS, Justice, Labor, State, Treasury, EPA, NASA, NSF, OPM, SBA, and USAID, the agencies did not agree or disagree with the draft report. In written comments, Interior did not agree or disagree with the draft report’s findings. In addition, Interior provided a technical comment related to our calculation of the department’s achieved optimization savings for fiscal years 2016 and 2017. Specifically, the department provided updated savings figures for those years, which were based on Interior’s February 2018 quarterly submission to OMB. However, the effective date of the updated data (February 2018) is outside the scope of our review, which relied solely on data reported in August 2017, as detailed in appendix I. While we recognize the department’s efforts to ensure that its past reporting is updated and as accurate as possible, our report only presents data as reported by agencies in August 2017. Consequently, we believe that the achieved cost savings for Interior accurately reflect what the department reported at the time of our review. The department’s comments are reprinted in appendix V. In an e-mail received on March 27, 2018, GSA did not agree or disagree with the draft report’s findings, and provided comments questioning the methodology we used to determine the number of data center closures. Specifically, GSA stated that, by including facilities that are now closed in our reporting of the federal government’s inventory of 12,062 data centers, our report gives the impression that the government is currently operating 12,062 open data centers. GSA also asserted that we should include in our counts facilities identified as using cloud computing providers. In our report, we consistently state that the 12,062 data centers reflect the overall count of federal facilities identified since the launch of FDCCI in 2010. Further, we clearly identify the portion of the overall count of data centers that have since closed, or that are planned to close. By doing so, we provide a perspective of consolidation progress against the overall inventory over the past 8 years. Further, in regard to facilities using cloud providers, OMB’s August 2016 guidance specifically states that such facilities are not to be considered data centers. As such, we did not include them in our totals. Accordingly, we maintain that our methodology is reasonable and continue to believe that our report accurately reports the status of federal data center consolidation efforts. GSA also suggested that we confirm with OMB our statement that the FITARA Enhancement Act of 2017’s extension of the sunset provides agencies with an additional 2 years to accomplish the goals of DCOI. It is true that OMB has not issued guidance that extends the existing August 2018 DCOI deadlines. However, it is appropriate to note that the 2017 law extends the DCOI sunset date for the data center requirements that govern agencies. Finally, GSA expressed concern that the data on optimization metrics that we drew from the IT Dashboard were not representative of the data provided by agencies in their August 2017 quarterly submissions to OMB. Further, GSA noted that such a selection of mixed data sources may present an incomplete and inaccurate picture and recommended that we note the date on which we accessed the IT Dashboard data, note that those data are frequently updated, and specify the date of other sources of data that we used for our analysis. In this report, we present an analysis of agencies’ progress against OMB’s data center optimization metrics, using data that were taken from the IT Dashboard. These data are posted to the Dashboard after being automatically calculated from agencies’ quarterly submissions to OMB. For the purposes of this report, we pulled the data from the Dashboard a week after agencies’ August 2017 submissions were due to OMB. We then confirmed with the agencies that the data we collected were consistent with their August 2017 submissions to OMB and we labeled the data as being effective as of August 2017. We also analyzed other data from the agencies’ August 2017 quarterly submissions, such as the status of data center closures and associated cost savings, and showed the effective date of our analysis as being August 2017. Based on these actions, which are also discussed in the description of the scope and methodology of our audit work found in app. I, we believe that our presentation of the data on optimization metrics is accurate, appropriately labeled, and correctly reflects the status of agency efforts at a specific moment in time. In an e-mail received on April 18, 2018, OMB did not agree or disagree with our report, but offered several comments on our findings. Specifically, OMB noted that some agencies may be reporting their planned savings incorrectly in their DCOI strategic plans in that agencies may be reporting annual savings figures instead of the required cumulative figures. OMB further described plans to update the IT Dashboard in the near future to more accurately reflect planned DCOI savings and added that, as a result, the data in our report likely will not match the data OMB intends to publish. In conducting our analysis, on two occasions, we requested and received agencies’ validation of the results of our analysis of their planned cost savings. This process resulted in minor technical changes to some agencies’ data. We believe that our continued efforts to validate these data provide reasonable assurance as to the accuracy of the agency reported information that we analyzed. The implementation of OMB’s proposed changes to the IT Dashboard should provide yet another tool that can be used to improve how agencies report their cost savings—an important measure of DCOI’s success. Additionally, the comments noted that GAO and OMB use a different basis to calculate agencies’ data center closure targets, with OMB using a baseline inventory from the beginning of DCOI in 2016, which does not recognize changes in agencies’ inventories since that point. OMB asserted that, because GAO’s calculations account for changes in agencies’ inventories since the beginning of DCOI, OMB’s targets differ from GAO’s calculations. We recognize the difference in approach for calculating data center closures that we used, as compared to that used by OMB. As detailed in our report, when DCOI was launched in August 2016, agencies reported an overall inventory of 9,995 data centers. One year later, in August 2017, agencies reported more than 2,000 additional facilities, for a total of 12,062 centers. Because OMB’s closure targets required agencies to close a certain percentage of their data centers, basing agency goals on an outdated inventory (that did not include the additional facilities) does not give a true picture of progress towards consolidation of data centers. Conversely, our methodology (detailed in app. I) takes into account inventory growth and uses the same percentage-based closure goals defined in OMB’s guidance. Our methodology presents a more accurate status of progress based on the growth of the inventory over time. Consequently, we believe that our methodology allows us to present a reasonable status of agencies’ progress against OMB’s goals. We are sending copies of this report to interested congressional committees, the Director of OMB, the secretaries and heads of the departments and agencies addressed in this report, and other interested parties. In addition, the report will be available at no charge on GAO’s website at http://www.gao.gov. If you or your staffs have any questions about this report, please contact me at (202) 512-9286 or pownerd@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made major contributions to this report are listed in appendix VI. Appendix I: Objectives, Scope, and Methodology Our objectives for this engagement were to: (1) review agencies’ data center closures to date and plans for further closures, (2) evaluate agencies’ progress in achieving data center consolidation savings and describe plans for future savings, and (3) assess agencies’ progress against the Office of Management and Budget’s (OMB) data center optimization targets. To address the first objective, we obtained and analyzed August 2017 data center inventory documentation from the 24 departments and agencies (agencies) that participate in OMB’s Data Center Optimization Initiative (DCOI). To determine data center closures to date, we totaled agency reported closures from fiscal year 2010 through August 2017 and, to identify future closures, we totaled agency reported planned closures through fiscal year 2019. We also compared agencies’ completed and planned closures to OMB’s fiscal year 2018 consolidation goals, as documented in its August 2016 memorandum (M-16-19). To verify the quality, completeness, and reliability of the agencies’ data center inventories, we compared information on completed and planned data center closures to similar information reported on OMB’s Information Technology (IT) Dashboard—a public website that provides information on federal agencies’ major IT investments. We also checked for missing data and other errors, such as missing closure status information. Further, we obtained written responses from agency officials regarding actions taken to ensure the reliability of their inventory data, and discussed any discrepancies or potential errors identified to determine the causes or to request additional information. We determined that the data were sufficiently complete and reliable to report on agencies’ consolidation progress and planned closures. For the second objective, we obtained and analyzed cost savings and avoidance documentation from the 24 DCOI agencies. This documentation is required by OMB’s March 2013 and August 2016 memorandums and included the agencies’ quarterly reports of cost savings and avoidances posted to their digital services websites and their DCOI strategic plans. To determine cost savings achieved, we totaled agencies’ reported savings and avoidances from the start of fiscal years 2012 through August 2017, as found in the August 2017 quarterly reports posted to the agencies’ digital services websites. To identify future planned savings, we totaled the agencies’ projected savings and avoidances from fiscal years 2016 through 2018, as reported in their DCOI strategic plans. To assess the quality, completeness, and reliability of each agency’s data center consolidation cost savings information, we used the latest version of each agency’s update of the August 2017 quarterly cost savings report and DCOI strategic plan. We also reviewed the quarterly reports and DCOI strategic plans for missing data and other errors, such as missing cost-savings information. In addition, we compared agencies cost savings and avoidances with data from our most recent data center consolidation report. Further, we obtained written responses from agency officials regarding the steps taken to ensure the accuracy and reliability of their cost savings data. As a result, we determined that the data were sufficiently complete and reliable to report on agencies data center consolidation cost-savings information. For our third objective, we analyzed the August 2017 data center optimization progress information of the 24 DCOI agencies. This progress information was obtained from the IT Dashboard—an OMB public website that provides information on federal agencies’ major IT investments. We then compared the agencies’ optimization progress information against OMB’s fiscal year 2018 optimization targets, as documented in its August 2016 memorandum. Although OMB’s memorandum establishes a single optimization target value for the server utilization and automated monitoring metric, the IT Dashboard displays agencies’ progress for tiered and non-tiered data centers separately. To report consistently with OMB’s implementation memorandum, we combined the progress information for tiered and non-tiered data centers into a single assessment in this report. To assess the reliability of agencies’ optimization progress information on OMB’s IT Dashboard, we reviewed the information for errors or missing data, such as progress information that was not available for certain metrics. We also compared agencies’ optimization progress information across multiple reporting quarters to identify any inconsistencies in agencies’ reported progress. We discussed with staff from OMB’s Office of the Federal Chief Information Officer any discrepancies or potential errors identified to determine the causes. In addition, we interviewed OMB officials to obtain additional information regarding the steps taken to ensure the reliability of and validate the optimization data on the Dashboard. Moreover, we obtained written responses from agency officials regarding the steps taken to ensure the accuracy and reliability of the reported optimization progress. We discussed with agency officials any discrepancies or potential errors identified during our reviews to determine the causes or request additional information. We determined that the data were sufficiently reliable to report on agencies’ optimization progress. To assess the reliability of the DCOI strategic plans, we reviewed agencies’ documentation to identify any missing data or errors. We also compared the planned data center optimization milestones in agencies’ documentation against current optimization progress information obtained from the IT Dashboard. In addition, we discussed with agency officials any discrepancies or potential errors identified during our reviews of the DCOI strategic plans to determine the causes or request additional information. As a result of these efforts, we determined that the agencies’ strategic plan information was sufficiently reliable for reporting on plans to meet or not meet OMB’s fiscal year 2018 optimization targets. We conducted this performance audit from July 2017 to May 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Comments from the Social Security Administration Appendix III: Comments from the Department of Housing and Urban Development Appendix IV: Comments from the Department of Transportation Appendix V: Comments from the Department of the Interior Appendix VI: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, individuals making contributions to this report included Dave Hinchman (Assistant Director), Justin Booth (Analyst-in-Charge), Chris Businsky, Nancy Glover, Linda Kochersberger, and Jonathan Wall.

Why GAO Did This Study In December 2014, Congress enacted federal IT acquisition reform legislation that included provisions related to ongoing federal data center consolidation efforts. OMB's Federal Chief Information Officer launched DCOI to build on prior data center consolidation efforts; improve federal data centers' performance; and establish goals for inventory closures, cost savings and avoidances, and optimization performance. The 2014 legislation also included a provision for GAO to annually review agencies' data center inventories and strategies. Accordingly, GAO reviewed agencies' data center closures to date and plans for further closures; evaluated agencies' progress in achieving consolidation savings and described their plans for future savings; and assessed agencies' progress against OMB's data center optimization targets. To do so, GAO assessed the 24 DCOI agencies' data center inventories as of August 2017; reviewed their reported cost savings documentation; evaluated their data center optimization strategic plans; and assessed 22 agencies' progress against OMB's established optimization targets. Two agencies did not have a basis to report planned optimization milestones. OMB and the 24 DCOI agencies provided mixed responses to GAO's findings on the progress made towards initiative goals. GAO continues to believe that implementation of the recommendations made previously will help the agencies meet OMB's targets for cost savings and optimization of performance. What GAO Found The 24 agencies participating in the Office of Management and Budget's (OMB) Data Center Optimization Initiative (DCOI) reported mixed progress toward achieving OMB's goals for closing data centers by September 2018. Over half of the agencies reported that they had either already met, or planned to meet, all of their OMB-assigned goals by the deadline. This would result in the closure of 7,221 of the 12,062 centers that agencies reported in August 2017. However, 4 agencies reported that they do not have plans to meet all of their assigned goals and 2 agencies are working with OMB to establish revised targets. With regard to agencies' progress in achieving cost savings, 20 agencies reported, as of August 2017, that they had achieved $1.04 billion in cost savings for fiscal years 2016 and 2017. In addition, the agencies' DCOI strategic plans identify an additional $0.58 billion in planned savings—for a total of $1.62 billion for fiscal years 2016 through 2018. This total is approximately $1.12 billion less than OMB's DCOI savings goal of $2.7 billion (see figure). This shortfall is the result of 12 agencies reporting less in planned cost savings and avoidances in their DCOI strategic plans, as compared to the savings targets established for them by OMB. The 24 agencies reported limited progress against OMB's five data center optimization targets for server utilization and automated monitoring, energy metering, power usage effectiveness, facility utilization, and virtualization. As of August 2017, 1 agency had met four targets, 1 agency had met three targets, 6 agencies had met either one or two targets, and 14 agencies reported meeting none of the targets. Further, as of August 2017, most agencies were not planning to meet OMB's fiscal year 2018 optimization targets. Specifically, 4 agencies reported plans to meet all of their applicable targets by the end of fiscal year 2018; 14 reported plans to meet some of the targets; and 4 reported that they do not plan to meet any targets. In 2016 and 2017, GAO made a number of recommendations to OMB and the 24 DCOI agencies to help improve the reporting of data center-related cost savings and to achieve optimization targets. As of March 2018, 74 of these 81 recommendations had not been fully addressed.

gao_GAO-19-24

gao_GAO-19-24_0

Background Composition of Agent Orange Agent Orange is composed of two different chemical components—the n- butyl ester forms of 2,4-dichlorophenoxyacetic acid (hereinafter referred to as n-butyl 2,4-D) and 2,4,5-trichlorophenoxyacetic acid (hereinafter referred to as n-butyl 2,4,5-T)—that are manufactured separately and then combined to form the tactical herbicide. The U.S. EPA has determined that there was not adequate data either to support or to refute that the acid or ester forms of 2,4-D can cause cancer in humans. In 2015 the International Agency for Research on Cancer classified 2,4-D as possibly causing cancer to humans, since there was inadequate evidence in humans and limited evidence in experimental animals. According to an Institute of Medicine report, information on the toxic effects of 2,4,5-T alone is sparse. However, in the 2,4,5-T manufacturing process, the dioxin 2,3,7,8-tetrachlorodibenzo-p-dioxin (hereinafter referred to as 2,3,7,8-TCDD) is formed, particularly when the reaction temperature is excessive. The World Health Organization has determined that dioxins are highly toxic and can cause a variety of illnesses, including reproductive and developmental problems and damage to the immune system. The World Health Organization reports that 2,3,7,8-TCDD, a human carcinogen, is the most toxic dioxin-related compound. Moreover, according to the National Academies of Sciences, Engineering, and Medicine report, 2,3,7,8-TCDD has been shown by researchers to be very toxic in animals. Figure 1 depicts the proportion of the components of Agent Orange and the amount of 2,3,7,8-TCDD contamination that would be present in an average 55-gallon drum. Agent Orange Origins and Life Cycle The Crops Division of the U.S. Army Chemical Corps was established at Camp Detrick (now Fort Detrick), Maryland, in 1943 to conduct anti-crop research, development, and engineering. In 1944 the Crops Division was given the mission of developing chemical compounds to destroy or reduce the value of crops. These chemical compounds were intended to rapidly clear vegetation in military operations in order to eliminate concealed enemy positions, improve air and ground observations, and destroy or reduce the value of crops. Initial field trials at Camp Detrick were small-scale efforts involving test plots typically 6 by 18 feet in size, and the herbicides being tested were usually applied using a hand sprayer. Over the following three decades, DOD collaborated with the U.S. Department of Agriculture, universities, and private companies to conduct testing activities ranging from laboratory experiments to spray tests of larger-scale aerial dissemination of a variety of chemical compounds throughout the United States, U.S. territories, and abroad. The tactical herbicides used by the U.S. military in Vietnam were formulations based on tests of thousands of different chemical compositions at Camp Detrick in an effort to determine chemical agents and chemical compounds that would meet specific requirements. The U.S. military developed and tested six tactical “rainbow” herbicides that it used during the Vietnam War era—Pink, Purple, Green, Blue, White, and Orange. The chemical component n-butyl 2,4,5-T, which is known to have been contaminated with 2,3,7,8-TCDD, was present in four of these six tactical herbicides—specifically, Agents Pink, Purple, Green, and Orange. In late 1961, DOD began color-coding the herbicide formulations that it was testing in aerial spray trials in Vietnam and elsewhere in Southeast Asia. The tactical herbicides, which were used for a variety of different purposes, to include defoliation and crop destruction, were identified by colored bands placed around the drums, as shown in figure 2. Beginning in 1962, the U.S. Air Force received shipments of Agents Pink, Purple, and Green to supply the first spray missions for Operation Ranch Hand, the program for defoliation and crop destruction missions during the Vietnam War. Agent Purple was similar to the herbicide formulation that was later designated “Orange,” but it was more costly to purchase. Agents Blue and White were used in Vietnam extensively along with Agent Orange after 1964, but they were of a different chemical composition and did not contain any form of 2,4,5-T, the component that produced 2,3,7,8-TCDD as a by-product of the manufacturing process. Of the tactical herbicides, Agent Orange was used the most extensively in Vietnam. In 1964 DOD began to procure large quantities from U.S. manufacturers for military use in Vietnam. The first shipment of Agent Orange arrived in Saigon in February 1965 by merchant vessel. Together, nine manufacturers produced a total of approximately 13.9 million gallons of Agent Orange, and DOD is estimated to have used approximately 12.1 million gallons between 1965 and 1970 in operations in Vietnam, and much smaller quantities in Korea and Thailand. Evidence from animal and epidemiologic studies of adverse effects from Agent Orange exposure led the U.S. government to restrict the use of 2,4,5-T in April of 1970 and led DOD to temporarily suspend the use of Agent Orange. In 1972 the U.S. Air Force consolidated the approximately 1.36 million gallons of the herbicide that had remained unused in Vietnam and shipped them for storage on Johnston Island in the Pacific. DOD held its remaining stocks of Agent Orange—approximately 860,000 gallons—within the continental United States, at the Naval Construction Battalion Center Gulfport, Mississippi, until those stocks were also shipped toward Johnston Island in June 1977. All of these remaining stocks of Agent Orange were incinerated at sea aboard the M/T Vulcanus by September 1977. Comparison between Tactical and Commercial Herbicides In addition to the tactical herbicides used during the Vietnam War era, the U.S. military also used commercial herbicides to manage vegetation on its installations. The U.S. military managed tactical herbicides differently from commercial herbicides. According to DOD officials and archived military specifications, tactical herbicides were not authorized for use on lands owned by, or otherwise managed as military installations and were not to be diverted for domestic use. DOD developed military specifications for the tactical herbicides that provided detailed information on product requirements, quality assurance, packaging, and precautionary statements that prohibited domestic use. The tactical herbicides were centrally managed, first by the Army Chemical Corps and later by the U.S. Air Force Logistics Command. Agent Orange used in Vietnam was formulated for aerial spraying by aircraft and helicopter and applied at full strength without additional solvents at a rate of 3 gallons per acre. Agent Orange is soluble in diesel fuel and organic solvents, but it is insoluble in water, so equipment was cleaned using diesel fuel rather than water. Commercial herbicides, conversely, were widely available worldwide for use in vegetation management at military installations, to include controlling vegetation adjacent to flightlines or along perimeter fencing. Federal agencies developed federal specifications for these products to ensure that they met specific requirements, and these specifications were approved by the Commissioner, Federal Supply Service, in the General Services Administration for use by all federal agencies. According to DOD officials, during the Vietnam era there was no requirement for DOD to retain records concerning the use of commercial herbicides on military bases beyond 5 years. DOD officials also stated that DOD catalogued these herbicides available for use on military installations in the federal supply schedule under federal supply classification group 68, which contains chemicals and chemical products. In reviewing supply catalogues from that time period, DOD officials identified more than 35 different commercial herbicides that were listed in the federal supply system for use on DOD installations between 1960 and 1973. Some of these commercial herbicides contained 2,4-D; 2,4,5-T; or both, although they were not in the n-butyl form used in Agent Orange. These included at least 4 commercial herbicides that contained some form of 2,4,5-T, the component that contained the contaminant 2,3,7,8- TCDD. In addition, numerous commercial herbicides that were not in the federal supply system but were being widely used elsewhere for agriculture purposes contained the form of n-butyl 2,4,5-T found in Agent Orange and thus its associated contaminant, 2,3,7,8-TCDD. According to DOD officials, the commercial herbicides used on installations were mixed with diesel or water and sprayed by hand or truck. Tactical herbicides, however, were formulated for aerial spraying by fixed-wing aircraft or helicopter without being diluted. When the U.S. military was employing these tactical and commercial herbicides during the Vietnam War era, U.S. EPA had not yet been established, and the U.S. Department of Agriculture had oversight of commercial herbicides. The Federal Insecticide, Fungicide, and Rodenticide Act of 1947, then administered by the U.S. Department of Agriculture, governed the marketing and use of these commercial herbicides. Until amended in 1972, the Federal Insecticide, Fungicide, and Rodenticide Act review process was designed as a consumer protection measure that focused primarily on a product’s effectiveness, rather than on concerns about health or the environment. Agent Orange Legislative and Regulatory History The Agent Orange Act of 1991, as amended, requires a review of the available scientific evidence regarding the associations between certain diseases and exposure to tactical herbicides. The act specifically requires the VA to enter into an agreement with the National Academy of Sciences (the Academy), or with an alternative scientific organization, to review and evaluate the scientific evidence concerning the association between exposure to an herbicide agent and each disease suspected to be associated with such exposure. The Academy is required to submit periodic reports at least once every 2 years. The most recent report—the 2014 report—was issued in March 2016. The next report, which Academy officials told us would focus on inter-generational and trans-generational effects of exposure to herbicides, was at the time of our report scheduled to be issued in late 2018. In its biannual reports, the Academy identifies different levels of association between exposure to 2,3,7,8-TCDD or other chemical compounds in herbicides used in Vietnam and a wide range of health effects. These levels include the following: sufficient evidence of an association; limited or suggestive evidence of an association; inadequate or insufficient evidence to determine an association; and limited or suggestive evidence of no association. The Academy has identified that there is either sufficient evidence of an association with exposure to a tactical herbicide or limited or suggestive evidence of an association leading to certain diseases. For example, the Academy has identified both chloracne and non-Hodgkin’s lymphoma as having sufficient evidence of an association with exposure to a tactical herbicide, and both Parkinson’s disease and diabetes mellitus (type 2) as having limited or suggestive evidence of an association. Examples of diseases for which the Academy has found inadequate or insufficient evidence to determine an association include kidney disease and pancreatic cancer. In making determinations regarding the association between certain diseases and exposure to herbicide agents, the Secretary of Veterans Affairs is required to take into account the Academy’s reports. Once the Secretary finds that such an association existed, the Secretary is then required to prescribe regulations, providing that a presumption of service connection is warranted for that disease. The Agent Orange Act of 1991, as amended, also establishes a presumption of service connection, by reason of exposure to an herbicide agent, for diseases listed in the statute, to include Hodgkin’s disease and diabetes mellitus (type 2). This presumption applies to veterans who, during active military, naval, or air service, served in the Republic of Vietnam during the period beginning on January 9, 1962, and ending on May 7, 1975. Veterans who served in Vietnam and other specific locations and time frames and who have been diagnosed with those diseases are presumed to have incurred those diseases as a result of their service and are thus eligible for presumptive service connection for disability compensation. Figure 3 illustrates the diseases for which the Academy has found either sufficient, or limited or suggestive, evidence of an association. In addition, appendix II provides information on the 14 presumptive diseases that the VA currently identifies as being associated with exposure to Agent Orange or other tactical herbicides during military service for which veterans and their survivors may be able to receive disability compensation benefits. Veterans’ Benefits Under 38 U.S.C. § 1110, the United States will pay benefits to any veteran disabled for a disability resulting from personal injury suffered or disease contracted in line of duty, or for aggravation of a preexisting injury suffered or disease contracted in line of duty, in the active military, naval, or air service, during a period of war. The VA offers health registry exams, health care, disability compensation, and other benefits to eligible veterans who were exposed to herbicides during military service. According to the VA’s Claims Adjudication Procedures Manual, the claims evaluation process begins with the VA requesting any information missing from the veteran’s claim, such as the approximate dates and location(s) of service, claimed disability, and, for certain locations, the nature of the alleged exposure to herbicides. Generally, the veteran then has 30 days to submit the requested information. During the claims process, VA will check military records to confirm exposure to Agent Orange or other herbicides and qualifying military service. Certain diseases have already been presumed to be associated with herbicide exposure, and no further evidence of an association is needed. However, if the claimed disability is not a presumed condition, then VA will request that the veteran present scientific or medical evidence showing that the claimed condition is medically associated with herbicide exposure. If the veteran is not able to provide this information, the case is referred to DOD for verification of exposure to herbicides. Veterans’ claims can either be approved or denied based on the evidence submitted by the veteran, and, if needed, by DOD. The VA tracks its claims data for Agent Orange exposure according to whether the exposure occurred inside or outside of Vietnam, which includes the Korean demilitarized zone and certain locations in Thailand. According to VA officials, as of June 30, 2018, 557,653 living veterans and 199,451 deceased veterans have been granted benefits for diseases associated with Agent Orange exposure inside Vietnam, with 44,925 claims pending for veterans who served in Vietnam and believe they were exposed to Agent Orange. For diseases associated with Agent Orange exposure outside of Vietnam, VA had granted service connection decisions to more than 10,758 veterans and denied service connection decisions to more than 58,250 veterans, as of June 30, 2018. According to VA, there are an additional 23,400 claims pending for veterans who did not serve in Vietnam but believe they were exposed to Agent Orange. Environmental Cleanup In 1980 Congress passed the Comprehensive Environmental Response, Compensation, and Liability Act, which established the Superfund program—the federal government’s principal program to clean up hazardous waste sites. The U.S. EPA is responsible for administering the Superfund program, which places some of the most seriously contaminated sites on the National Priorities List, and has oversight for federal and non-federal sites on that list. Additionally, amendments to the act in 1986 require the Secretary of Defense to carry out the Defense Environmental Restoration Program, which was specific to DOD environmental cleanup activities at active installations, formerly used defense sites, and base realignment and closure locations in the United States. The cleanup process under the Environmental Response, Compensation, and Liability Act process generally includes the following phases and activities: preliminary assessment, site inspection, remedial investigation and feasibility study, remedial design and remedial action, and long-term monitoring. Through this process, DOD and U.S. EPA cleaned up some U.S. sites where Agent Orange was known to have been present after the sites were tested and confirmed to have been contaminated with 2,3,7,8- TCDD. For example, U.S. EPA identified a site in Jacksonville, Arkansas, where 2,4,5-T had been manufactured, that was contaminated with 2,3,7,8-TCDD. In addition, under the Defense Environmental Restoration Program, DOD cleaned up the Naval Construction Battalion Center Gulfport, Mississippi, where Agent Orange had been stored while awaiting shipment for use in Southeast Asia. The site had also been used to store Agent Orange drums that were awaiting shipment to Johnston Island for disposal. According to a DOD report, approximately 860,000 gallons of the herbicide were stored at the site. An Agency for Toxic Substances and Disease Registry report further states that spills that occurred during storage caused 2,3,7,8-TCDD contamination around several water areas. According to a 5-year review completed by DOD in 2017, capping of the contaminated soil at the site where herbicides were stored has been completed, and long-term monitoring of the soil and groundwater began in 2012 and continues today. DOD also cleaned up the Johnston Island site where Agent Orange was ultimately disposed of. Once drums of Agent Orange were stored at Johnston Island, environmental sea conditions caused them to corrode and leak. Initial cleanup activities assessed and monitored the area to track the chemical components remaining as a result of Agent Orange contamination. Site remediation and environmental monitoring continued throughout the 1970s until February 1989, when the Air Force, in accordance with the Defense Environmental Restoration Program, completed a final site cleanup at Johnston Island by destroying all remaining 2,3,7,8-TCDD-contaminated soil. Figure 4 shows drums of Agent Orange stored at Johnston Island. In addition, U.S. EPA listed on its National Priorities List two former Agent Orange manufacturing sites—the Kanawha River site in West Virginia previously owned by the Monsanto Company and a site in Newark, New Jersey, owned by the Diamond Alkali Company—due to high levels of contamination from various sources and threats to human health. In 2017, U.S. EPA entered into an agreement with the Monsanto Company on a cleanup plan to address 2,3,7,8-TCDD contamination at the Kanawha River Superfund Site in Putnam and Kanawha counties, West Virginia. The cleanup effort will focus on a 14-mile stretch within the Kanawha River. Cleanup work will include constructing a cap over more than 9 acres of contaminated river sediments. Similarly, the Diamond Alkali site in New Jersey contained 2,3,7,8-TCDD contamination at both the manufacturing site and the nearby Lower Passaic River. The site was found to contain high levels of 2,3,7,8-TCDD and was placed on the National Priorities List in 1984. As late as 2014, the site was still undergoing cleanup actions to prevent exposure to the contaminated soil and prevent further releases to the river. It is difficult to isolate the specific costs of cleaning up Agent Orange contamination under the Comprehensive Environmental Response, Compensation, and Liability Act, according to DOD and U.S. EPA officials. Moreover, cleanup plans address multiple contaminants, making it difficult to isolate the costs for cleaning up a specific contaminant, according to DOD and U.S. EPA officials. For example, the Diamond Alkali site had multiple contaminants from a number of companies that owned or operated facilities from which hazardous substances, including 2,3,7,8-TCDD and pesticides, were potentially discharged into the river and found in the soil and groundwater. Various cleanup actions were taken to address not only 2,3,7,8-TCDD contamination but the other contaminants as well. These actions included a groundwater collection and treatment system and capping to prevent exposure to contaminated soil (including contaminated soil that originated at the facility and soil that was brought to the facility from neighboring lots) and prevent further releases to the river. The Federal Government Has Some Information on the Procurement, Use, and Destruction of Agent Orange, and Available Documentation Indicates at Least One Vessel Carrying Agent Orange Transited through Guam to Vietnam, but Information Is Not Complete The federal government maintains information on Agent Orange, and available records indicate that DOD procured approximately 13.9 million gallons of the tactical herbicide, which was either used in U.S. military operations in Southeast Asia, used for testing, or destroyed. Our analysis of the available logbooks for 152 of the 158 shipments (approximately 96 percent) of Agent Orange to Southeast Asia that we identified indicates that the vessels carrying tactical herbicides generally stopped at foreign ports and sometimes at U.S. ports en route to Southeast Asia. Available primary source materials, such as shipment documentation, are incomplete because they were likely not maintained during and after the Vietnam era. However, based on the available information, we identified at least one ship carrying Agent Orange that stopped at Port Apra (now Apra Harbor) on Guam on its way to Vietnam, although we could not locate any evidence showing that any cargo was offloaded. Further, while DOD documents identify the use of commercial herbicides on Guam, they do not identify the use of tactical herbicides there. Available Records Indicate That All of the Agent Orange Procured Was Either Used in U.S. Military Operations, Used for Testing, Damaged, or Destroyed Available records that the federal government maintains indicate that DOD procured approximately 13.9 million gallons of Agent Orange between 1963 and 1968, of which it used an estimated 12.1 million gallons in Southeast Asia from 1965 to 1970; used a small amount for testing; and incinerated another 2.3 million gallons in 1977. Thus, the total quantity of Agent Orange that DOD procured was approximately equal to the total quantity that records indicate was tested in the United States and its territories, damaged during storage and shipment, and used during the Vietnam War, combined with the total quantity that records indicate was disposed of afterwards. Procurement and Use. Based on available records we reviewed, DOD procured approximately 13.9 million gallons of Agent Orange from nine chemical manufacturers between 1963 and 1968. In 1963 DOD used small amounts of Agent Orange for testing. DOD procurement officers then advised the Military Assistance Command, Vietnam, in late 1964 that they could fulfill the supply requirements for tactical herbicides with Agent Orange. Available records further indicate that of the approximately 13.9 million gallons of Agent Orange procured, DOD used an estimated 12.1 million gallons in operations in Vietnam from 1965 to 1970. In addition to the quantity used in Vietnam, Agent Orange usage also included quantities that were tested in the United States and its territories; used or tested in countries outside of Vietnam; lost during shipment and storage; or removed from the inventory and used to test different disposal options after its use was suspended. With the exception of the disposal testing amounts, no archival resources we could locate and obtain provided definitive usage figures. The last known shipment of Agent Orange to Vietnam was aboard the SS Frederick Lykes and arrived in May 1970. Restrictions on Use. In 1969 the National Environmental Health Service of the Department of Health, Education, and Welfare conducted testing of n-butyl 2,4,5-T—the component of Agent Orange whose manufacturing process produced 2,3,7,8-TCDD as a by-product—on mice, which raised concerns about health effects of the herbicide for women of child-bearing age. These concerns led to several U.S. government decisions that ended the use of tactical herbicides. Specifically, in 1969 DOD restricted the use of Agent Orange in Vietnam to keep it away from population centers. In April 1970 the federal government began restricting the use of 2,4,5-T in the United States. Exceptions were made for the control of weeds and brush on range, pasture, and forests, or on rights of way and other nonagricultural land. On April 15, 1970, DOD temporarily suspended the use of Agent Orange, including new procurement, acceptance of product on terminated contracts, transfer of stocks at Gulfport, and ocean shipping operations. Consolidation and Incineration of Remaining Stocks. After the U.S. government restricted the use of n-butyl 2,4,5-T—a component of Agent Orange—in 1970, DOD decided to consolidate the remaining 2.3 million gallons of Agent Orange stored in Vietnam and Gulfport, Mississippi, as well as any remaining amounts of n-butyl 2,4,5-T. According to an Office of Air Force History monograph, on January 16, 1971, DOD ordered the termination of all crop destruction missions by U.S. forces in Vietnam, and on September 27 of that year, the Chairman of the Joint Chiefs of Staff directed the Air Force to return all remaining stocks of Agent Orange to the United States and to dispose of them. Specifically, Agent Orange stocks in Vietnam were temporarily stored at U.S. Air Force bases at Da Nang, Phu Cat, and Bien Hoa until they were moved to Johnston Island in 1972. In 1972 the U.S. military moved approximately 1.36 million gallons of Agent Orange onto Johnston Island for storage. The cargo vessel SS Transpacific picked up this quantity at three Vietnamese ports from March 15 to April 1, traveled to Johnston Island, arrived on April 18, and completed offloading on April 28 before returning to the United States. This consolidated quantity of Agent Orange from Vietnam remained at Johnston Island until 1977. The Naval Construction Battalion Center Gulfport, Mississippi, was the final storage location in the continental United States for Agent Orange until the U.S. Air Force began the incineration of Agent Orange in 1977. There were approximately 860,000 gallons of Agent Orange at this location in 1977, which takes into account amounts lost in Hurricane Camille in 1969 or shipped away for testing, as described previously. The 1977 figure also takes into account 14,025 gallons transferred to the Naval Construction Battalion Center Gulfport, from Eglin Air Force Base, Florida, where the Air Force had tested formulations of Agent Orange for aerial spraying. In addition, available records show that quantities of the two components of Agent Orange were stored at the former Kelly Air Force Base, Texas, until 1972 before they were transferred to the U.S. Department of Agriculture for brush control projects. These reported amounts included 106,260 gallons of n-butyl 2,4-D and 38,940 gallons of n-butyl 2,4,5-T. These records also show that 173,910 gallons of Agent Blue were stored at the installation; see figure 5. DOD chartered the incinerator ship M/T Vulcanus and loaded the 860,000 gallons stored at Naval Construction Battalion Center Gulfport, Mississippi, beginning in May 1977. The vessel left Gulfport, Mississippi, in June 1977, and began incinerating the Agent Orange on board in July 1977 in a research burn to test the incineration process at sea near Johnston Island. In August 1977, the M/T Vulcanus loaded the remaining 1.36 million gallons stored at Johnston Island and conducted two more incineration operations just southwest of Johnston Island, as shown in figure 6. By September 3, 1977, all stocks of Agent Orange had been incinerated. Available Records Indicate That Vessels Transporting Agent Orange Stopped at Various Ports en Route to Southeast Asia, but Shipment Information Is Not Complete Our review of documentation for the shipment of almost 12.1 million gallons of the approximately 13.9 million gallons (approximately 87 percent) of Agent Orange procured by DOD found, based on available shipment documentation, that vessels transporting Agent Orange made stops at various ports on the way to Southeast Asia. However, shipment documentation is incomplete. Manufacturers of Agent Orange blended the two components of the herbicide—the n-butyl forms of 2,4-D and 2,4,5-T—and marked 55-gallon drums for shipment to Southeast Asia. Available records indicate that manufacturers produced Agent Orange according to military specifications and marked all drums for shipment directly to the receiving U.S. military unit in Vietnam. These specifications indicated the precise herbicide formulation of Agent Orange (n-butyl esters, 50 percent 2,4-D and 50 percent 2,4,5-T) and general instructions for marking the 55-gallon drums for shipment. For example, according to a historical monograph by the San Antonio Air Materiel Area, DOD specified that each drum was to be marked with a colored band or bands around the center as well as with transportation and contract data. Figure 7 shows an example of these drum markings. DOD then arranged for the transport of these drums, as well as drums of other tactical herbicides, by train from the manufacturers to several U.S. ports. DOD transportation officials accepted the product by signing a Material Inspection and Receiving Report that indicated the destination of the rail shipment and the final destination in Vietnam. DOD primarily chartered merchant marine vessels to ship the drums to Southeast Asia, but we identified one official Navy vessel, the USNS Lt. George W.G. Boyce, that carried Agent Orange to Southeast Asia. The first known shipment of Agent Orange left the port of New Orleans, Louisiana, on the SS Adabelle Lykes and arrived in Vietnam in February 1965. The last known shipment left the port of Gulfport, Mississippi, on the SS Frederick Lykes and arrived in Vietnam in May 1970. By that time, DOD had suspended all further shipments of Agent Orange. The photos in figure 8 provide examples of drums of Agent Orange being shipped by rail and tactical herbicides being loaded onto a cargo ship. The bulk of materiel used to support U.S. military forces in Vietnam, including tactical herbicides, was transported from the continental United States to Vietnam via ship. The vessels carrying the tactical herbicides generally stopped at foreign ports and sometimes at U.S. ports on the way to Southeast Asia. Our analyses of available shipment documentation indicate that at least 114 unique cargo vessels carried Agent Orange to Southeast Asia on at least 158 different voyages from 1965 through 1970. For each of these voyages, merchant vessel captains submitted logbooks to the U.S. port authorities at the end of each voyage. We were able to locate and obtain logbooks for 152 of the 158 shipments (approximately 96 percent) we identified. For 3 of the 6 voyages for which we were not able to locate logbooks, we obtained copies of the vessels’ shipping articles. We were not able to obtain shipping articles for the 3 foreign-flagged vessels because documents for such vessels were not turned in at U.S. ports. The Military Sea Transportation Service directly chartered merchant vessels to carry tactical herbicides during the Vietnam War. At least 28 vessels owned by the New Orleans, Louisiana-based Lykes Brothers Steamship Company transported Agent Orange between 1965 and 1970 from Gulf Coast ports to Southeast Asia. Lykes Brothers vessels were designed to handle cargo with cables that could place the cargo in a series of holds— numerous compartmented internal storage spaces. Tactical herbicides were stored vertically on pallets in these holds. The first large shipments of Agent Orange took place on the SS Adabelle Lykes, SS Elizabeth Lykes, and SS Mayo Lykes, traveling from the port of New Orleans, Louisiana, through the Panama Canal, and refueling in the Philippines before offloading a total of 1,782 55-gallon drums (approximately 97,000 gallons) in Saigon, Vietnam, in February and March of 1965. Our review of the logbooks and shipping articles for vessels carrying Agent Orange and other tactical herbicides showed that these vessels made stops at several U.S. and foreign ports, both in going to and in returning from Vietnam. For example, we identified vessels that stopped at several West Coast ports to load cargo before traveling to Vietnam, and others that made port calls to refuel in Hawaii. We also identified vessels that stopped at foreign ports such as Okinawa, Thailand, and Taiwan, as well as locations near the major U.S. Naval Supply Depots in Yokosuka, Japan, or Subic Bay, Philippines. These supply depots were major logistics hubs for U.S. military operations in East Asia, and they provided supplies to commercial ships that were chartered by DOD’s Military Sea Transportation Service through contracts with shipping companies. These companies would reserve cargo space for military cargo and include Saigon, Vietnam, as a destination, but the voyages were otherwise made for normal commercial activities. From those locations, the cargo vessels traveled to one or more ports in Vietnam. However, while the logbooks we reviewed identify when vessels left the various ports as they traveled to and from Vietnam, logbooks do not provide information on whether and how much cargo was loaded and unloaded at those ports of call, nor do they indicate whether tactical herbicides were offloaded at any ports before the vessels reached Vietnam. Available Shipment Documentation Indicates at Least One Vessel Carrying Agent Orange Went Through Guam en Route to Vietnam, but Archival Information Lacks Details or Is Not Complete Based on our review of available logbooks, we identified at least one vessel carrying Agent Orange that stopped at Guam en route to Vietnam and at least three vessels that stopped at Guam on the return from Vietnam. However, in our review of available shipment documentation, we found no evidence indicating that Agent Orange or any other tactical herbicides were offloaded from those vessels or used in the U.S. territories of Guam or the Northern Mariana Islands. Figure 9 indicates the timelines of the four vessels known to have carried Agent Orange that stopped at Guam either on their way to or returning from Vietnam, each of which is discussed in detail below. Available shipment documentation indicates that hundreds of vessels delivered supplies to the Naval Supply Depot, including supplies bound for Andersen Air Force Base, on Guam during the Vietnam War due to both installations’ strategic location in supporting the war effort. While the logbooks we were able to locate and review for vessels that transported Agent Orange to Southeast Asia between 1965 and 1970 do not show that these vessels typically stopped at Guam or the Northern Mariana Islands at any time during their voyages, we identified one ship carrying Agents Orange, Blue, and White that did stop at Guam on its way to Vietnam. Specifically, available records indicate that sometime around February 1, 1968, the SS Gulf Shipper stopped at Port Apra (now Apra Harbor) on Guam en route to Vietnam. Figure 10 shows a photo of the logbook from the SS Gulf Shipper indicating the ship’s ports of call en route to Vietnam. The logbooks do not provide details about whether cargo was moved on or off the vessels during these port calls, or whether tactical herbicides were offloaded at these ports before the vessels reached Vietnam. However, the SS Gulf Shipper’s logbook indicates that the stop at Guam could have been related at least in part to the repatriation of an injured crew member to the United States, and not to matters related to the loading or unloading of cargo. Further efforts to locate information on cargo movements for the SS Gulf Shipper, such as customs records, manifests, or bills of lading, were unsuccessful, because those records were not routinely retained. As such, we were not able to verify why the SS Gulf Shipper stopped at Guam, what its crew did while there, or whether any cargo was loaded or unloaded. We also identified at least three vessels that stopped on Guam on their return from Vietnam, based on our review of available logbooks. Specifically, around November 30, 1969, the SS Aimee Lykes stopped at Port Apra on Guam and offloaded an injured crew member into a small motorboat so that he could be hospitalized on Guam. In addition, around December 23, 1969, the SS Buckeye Atlantic stopped at Guam and offloaded two injured crew members. Lastly, around May 5, 1970, the SS Overseas Suzanne stopped at Guam and offloaded an injured crew member. Based on a review of the vessels’ logbooks, it is not clear whether the stops at Guam were for reasons other than offloading injured crew members—for example, reasons related to the loading or unloading of any cargo. Appendix III describes information that we were able to obtain regarding the quantities of herbicides known to have been shipped to Southeast Asia on the four vessels that we identified as having stopped at Guam (either on the way to or from Vietnam) between February 1968 and May 1970. As noted earlier, based on our review of available shipment documentation, we were able to identify approximately 87 percent of the shipments of Agent Orange to Southeast Asia, and to obtain logbooks for about 96 percent of the vessels known to have transported Agent Orange from U.S. ports to Vietnam. Because we were unable to obtain logbooks for every shipment of Agent Orange, we cannot conclude with certainty whether any ships other than the SS Gulf Shipper that were transporting the tactical herbicide to Vietnam, or the three ships returning to the United States from Vietnam—the SS Aimee Lykes, the SS Buckeye Atlantic, and the SS Overseas Suzanne—made port calls either at Guam or the Northern Mariana Islands. Additionally, we found and U.S Air Force officials agreed that it is unlikely that Agent Orange was shipped by air to or from Guam. The U.S. Air Force transported small quantities of tactical herbicides by air to Vietnam in 1961. However, we did not identify any documentation showing the transport of tactical herbicides by air to Vietnam after 1961. During our visit, officials at Andersen Air Force Base stated that it would have been possible to fly 55-gallon drums from Guam to supply operations in Vietnam, but that such an action would have been an inefficient method of transporting large quantities of herbicides. Agent Orange weighed approximately 600 pounds per drum, or about 11 pounds per gallon, a weight that, according to a 1966 memorandum from the Military Assistance Command, Vietnam, would have precluded large- scale transport of the herbicide by aircraft. DOD Documents Identify the Use of Commercial but Not Tactical Herbicides on Guam Available records show that DOD stored and used commercial herbicides on Guam, possibly including those containing n-butyl 2,4,5-T, during the 1960s and 1970s, but documents do not indicate the use of tactical herbicides on Guam. Commercial herbicides were available through the federal supply system for use on U.S. military installations worldwide. For example, the fuel supply for Andersen Air Force Base was delivered by ship to the port at Naval Base Guam and was then delivered to the Air Force base by a cross-island fuel pipeline—see figure 11. A detailed 1968 report by the Naval Supply Depot states that the Public Works Center sprayed herbicides semi-annually to control the vegetation along fuel pipelines between the depot and Andersen Air Force Base. Additionally, draft environmental assessments written in 1999 and 2009 by Naval Facilities Engineering Command, Pacific, indicate that commercial herbicides containing 2,4-D were present on Guam, and that commercial herbicides containing 2,4,5-T, which included the contaminant 2,3,7,8-TCDD, had been used for weed control along power lines and substations through 1980. Further, a 1969 master storage plan for the Naval Supply Depot includes sketches of storage facilities that specify the location of weed killers. Commercial herbicides approved for DOD procurement for use on installations were issued in 55-gallon drums and 5-gallon containers during the Vietnam War era, as were a range of other products, such as fuel oil and diesel. According to DOD officials, records for such purchases were not typically retained due to short record retention policies related to such routine supply transactions. During the course of our review, we received photographs and written statements from veterans alleging the presence of Agent Orange on Guam. However, based on our discussion sessions with veterans and civilians and our review of this documentation, we could not substantiate the presence or use of Agent Orange or other tactical herbicides on Guam. We asked veterans in our six discussion sessions about their potential for exposure to Agent Orange and where, if, and how they believe they were exposed. In their responses, some veterans in each of the six discussion sessions stated that they believe they were exposed to Agent Orange while deployed in Vietnam or other areas where a presumption of service for benefits has already been granted, while some veterans in three of the six discussion sessions stated that they believe they were exposed to Agent Orange while stationed on Guam. Specifically, some veterans in our discussion sessions described using herbicides or witnessing the spraying of herbicides at locations on Andersen Air Force Base and along the pipeline, as well as the burning of contaminated fuel as part of firefighting training on the installation. As we previously stated, according to DOD officials and archived military specifications, tactical herbicides were not authorized or available for use on lands owned by, or otherwise managed as military installations. However, commercial herbicides were widely available worldwide for use in vegetation management at military installations, to include controlling vegetation adjacent to flightlines or along perimeter fencing. Selected Comments by Veterans at Discussion Sessions Moderated by GAO Regarding Where They Believe They Were Exposed to Agent Orange or Its Components I feel like I was exposed on Guam. I was temporary duty there during the conflict and my duties were as a squadron controller that worked the schedules for the B-52 Bombers on Guam. I did venture into the loading area because I was with the aircrew on the Navy field at Andersen Air Force Base. I thought I was in contact with Agent Orange in Guam loading bombs in sites. We would move from one site to another and they would spray those areas before we got there. I never saw spraying but could smell it. One time I was near that and I broke out in boils and blisters on my face and arms. I was a fuel specialist I witnessed spraying going on at the barracks at Marbo Annex, 2 to 3 miles off the main Air Force base. It was sprayed all around the barracks. As my job, I worked at POL —where they stored all of the 55-gallon drums—fuels, pesticides, herbicides—in bulk storage. Those were constantly sprayed around—for maintenance and fire safety. Also, I would work on the flightline and at the pump houses—these were about 20 yards from the security fence. As I was working there, I witnessed spraying. DOD’s List of Herbicide Testing and Storage Locations Is Incomplete, and Veterans Have Expressed Confusion about How to Obtain Information on Potential Exposure DOD’s official compilation of herbicide testing and storage locations outside of Vietnam, which is posted on the VA’s website, is inaccurate and incomplete, and DOD does not have a process for managing the list. Further, while DOD and VA each have methods for communicating information to veterans and the public about Agent Orange, they do not have a formal process for communicating the most accurate available information to veterans about potential locations where they could have been exposed to Agent Orange or other tactical herbicides. DOD’s List of Locations Where Herbicides Were Tested and Stored Is Inaccurate and Incomplete DOD developed a list that identifies locations and dates where herbicides, including Agent Orange, are thought to have been tested and stored outside of Vietnam, which VA has made publicly available on its website, but this list is neither accurate nor complete. DOD’s list includes information on testing and storage locations, applicable dates, the herbicide or herbicide components tested, a description of the project, and DOD’s involvement. See appendix IV for the list that was posted on the VA website as of September 2018. When we began this review, DOD and VA officials were unable to identify the origin of the DOD list that is posted on the VA website, which does not have a date. A DOD official subsequently informed us that the list was initially created in 2003 by an individual in the Office of the Secretary of Defense in response to a congressional inquiry about the use of Vietnam-era herbicides at specific locations in the United States and overseas. DOD subsequently provided this list to VA, which in turn posted the information on its website. VA’s Claims Adjudication Procedures Manual related to Agent Orange directs VA officials to review the DOD list to determine whether herbicides were used as claimed as part of verifying potential herbicide exposure when a veteran alleges exposure at locations other than the Republic of Vietnam, the Korean demilitarized zone, or Thailand. However, in our review of several sources provided by DOD and VA officials, we identified multiple examples of inaccurate and incomplete information in DOD’s list, to include the following: Omission of specific testing and storage locations: We identified additional testing and storage locations in the United States and its territories that were not included on DOD’s list. For instance, we identified additional testing locations at Belle Glade, Florida, and Stuttgart, Arkansas, where researchers reported small-plot field tests of the components of Agent Orange on rice. In addition, we found examples of shipments of herbicides to Kelly Air Force Base, Texas, where Agent Orange components were stored following the cancellation of tactical herbicide contracts. None of these locations are included on DOD’s list. Lack of clarity in descriptive information: DOD’s list lacks clarity in descriptive information, making it difficult to identify which specific herbicides or components were tested and stored, as well as when and where. For example, the size and scope of some testing activities are unclear from the descriptions provided in DOD’s list, making it difficult to differentiate between small-scale and large-scale testing. Some testing events on DOD’s list are described in detail, including the amount of herbicide or components tested, while descriptions of other testing activities contain little information about what took place. Furthermore, we could not identify the chemical components of some of the agents on DOD’s list. We asked DOD and VA officials to identify those specific agents for us, and they were unable to do so. Specifically, neither DOD nor VA officials could identify the chemical composition of 26 different agents on the DOD list, making it difficult to determine whether these agents should be included on the list. Omission of additional time periods for identified locations: We identified additional testing events of Agent Orange or its components at locations that are on the DOD list but that cover additional time periods not reflected on the list. For instance, the DOD list identified testing that took place at Aberdeen Proving Grounds, Maryland, in July 1969. However, our review uncovered additional testing events that took place at Aberdeen Proving Grounds in 1963, 1965, and 1966. In addition to the lack of clarity and omissions that we identified, reports commissioned by DOD and VA since 2003 have also identified omissions in the list. For example, a report prepared for DOD in 2006 identified 40 different locations where Agent Orange was tested or stored outside of Vietnam. However, during our review, we found several examples of locations in the United States and its territories that were included in that 2006 report but are not included on the DOD list that is currently posted on the VA website. These include locations in Arkansas, California, New Jersey, New York, Maryland, Ohio, Oregon, Puerto Rico, Texas, and Utah. Similarly, a report prepared for VA in May 2013 described locations where Agent Orange exposure to Vietnam-era veterans has been alleged. This report summarized additional sites where veterans alleged Agent Orange was used, stored, or destroyed. It also included an assessment of the DOD information posted on the VA’s website—and indicated, notably, that information had not changed since the 2006 report to DOD. In the assessment, the report identified that the list contained many errors of dates, chemicals, locations, and the governmental agencies or institutions responsible for conducting the tests or military operations. The report suggested specific criteria for validating the presence of a tactical herbicide at a site, including evidence that a veteran actually came into contact with a tactical herbicide at that site. Even though they have received reports dating back more than a decade that identified issues with the accuracy and completeness of the list, neither DOD nor VA has taken steps to validate or correct the list, or to develop the criteria they would use to determine which locations and dates to include on the list. As previously stated, this list is posted on the VA’s Agent Orange website as a primary source for veterans seeking information on Agent Orange. Despite its inconsistencies, the list can be accessed from multiple places on the VA website, and we found that some veterans service organizations and other groups also post this incomplete and inaccurate list of testing and storage sites on their websites, as well as communicate this information to their members. Standards for Internal Control in the Federal Government state that agencies should use quality information to achieve their objectives. We found and DOD officials agreed that DOD’s list was not as accurate or complete as available records would allow because (1) there are not clearly identified responsibilities for validating the information on this list, (2) there is no process for updating the list as needed, and (3) criteria have not been developed and used to determine which locations and dates to include on the list. Until recently, neither DOD nor VA has taken responsibility for ensuring the accuracy and completeness of the list, which is being provided to veterans and the public on the VA website. Federal internal control standards state that management should establish an organizational structure, assign responsibility, and delegate authority to achieve the entity’s objectives. As noted earlier, DOD and VA officials were initially unable to identify the source or date of this list, and neither agency took action to respond to reports about the problems with it. During the course of our review, DOD took some initial steps to begin validating the accuracy and completeness of information on its list by reviewing primary source records for additional locations and events of herbicide testing and storage. However, thus far in its efforts, DOD has not identified responsibilities for completing the validation of the information included on the list, nor has it established a process for updating the list as any new information becomes available. Moreover, it remains unclear whether DOD’s review will cover all locations, including non-DOD sites, where testing and storage of Agent Orange or its components were thought to have occurred, or if it will focus only on U.S. military installations. Private companies, academic institutions, and other federal agencies were involved in the testing of herbicides at some of the non-DOD sites on the list, and, in some of those cases, Army personnel were involved in the testing at the non-DOD locations. For instance, testing was performed by DOD personnel at non- DOD locations in Georgia and Tennessee in the 1960s. Some non-DOD storage locations included various U.S. commercial ports, such as Mobile, Alabama, where Agent Orange was transferred by rail from the manufacturers to be stored until it was loaded onto vessels for shipment to Vietnam. According to a DOD official, DOD’s priority in its review of testing and storage locations is to focus on DOD installations. Although this official told us that the department expects to eventually identify non- DOD locations where the department was involved in herbicide testing and/ or storage through collaboration or funding, the official was not able to provide information on the time frames for conducting this review. Finally, DOD has not established a process for how this list will be updated once it has been validated and revised, when and if new information about Agent Orange testing and storage locations is identified. In our analysis of the DOD list, we were also unable to determine the criteria that DOD initially used to select which locations and time periods to include—particularly given that the testing varied in intensity and duration, and that the likelihood that personnel at a particular location could have been exposed to the herbicides or components was unclear. For example, some tests on the list included small laboratory experiments on a couple of plants using a very small amount of chemical agents, as in bench tests of various compounds at Forts Detrick and Ritchie, Maryland, in the 1950s, while other tests included gallons of Agent Orange or other chemical agent components that were used in field testing trials or to test aerial spraying, as in a defoliation effort in which 13 drums were sprayed by helicopter over an area covering 4 square miles. Similarly, the duration of testing events could have been over a total of 3 days, as with spray testing in Marathon, Florida, or over several months or even years, as with spray testing of several tactical herbicides at Eglin Air Force Base, Florida. Because of the variance in the size and duration of testing events; the specific areas where the testing events took place at the locations; and the number of personnel who actually came into contact with the chemical agents during the testing, the presence of a location on this list does not clearly indicate the likelihood or extent of potential exposure that individuals not involved would have had if they were simply present at the locations on the list at the times indicated. In May 2018, during the course of our review, a DOD official noted that DOD and VA formed a joint Herbicide Orange Working Group to address the issues with the DOD list and identify criteria for including information on this list. This group held its first meeting on May 31, 2018. As of July 2018, a DOD official noted that the group was working to identify appropriate steps to take, but that it was too soon to report specific actions that were being implemented, and that no documentation on the group’s efforts was available. Without assigned responsibility for ensuring an accurate and complete list of locations where Agent Orange or its components were tested and stored; a process for updating the list as needed; and clearly defined and transparent criteria for what to include on this list, DOD will not have reasonable assurance that it has identified the most complete information possible for VA to use when informing veterans and the public of the full extent of locations where Agent Orange exposure could potentially have occurred. As a result, veterans may not have complete information about the risk that they could have been exposed to Agent Orange during their military service, and VA may not have quality information when making important decisions on claims for veterans who may not be eligible for benefits. DOD and VA Have Communicated with Veterans and Others about Potential Exposure to Agent Orange, but Veterans Have Expressed Confusion Regarding How to Obtain Needed Information Both DOD and VA have communicated with veterans in response to inquiries about Agent Orange, but veterans have expressed confusion regarding how to obtain information to determine their potential exposure to Agent Orange. Further adding to this confusion are inconsistencies in the list of testing and storage locations, as discussed above. As the agency responsible for reviewing and validating veterans’ disability compensation claims for possible Agent Orange exposure, VA communicates with veterans largely through the agency’s website, which contains information on Agent Orange regarding related diseases, benefits, exposure locations, and resources. The VA also communicates through other means, including an annual newsletter and forums with veterans service organizations. DOD also receives inquiries from veterans about the potential that they could have been exposed to Agent Orange at DOD installations outside of Vietnam. In addition, DOD receives Freedom of Information Act inquiries and congressional requests for information on where Agent Orange was present. A DOD official stated that while they will respond to veterans’ inquiries, they typically direct veterans with Agent Orange inquiries to VA. In responding to these inquiries, both DOD and VA officials stated that they rely on the expertise of staff at the Armed Forces Pest Management Board to provide details to answer questions related to locations where exposure might have occurred. According to a DOD official, the board received 109 inquiries in 2017 alone. In addition, DOD’s Joint Services Records Research Center provides information to VA regional liaisons electronically in response to their questions about where and when specific units were stationed or on temporary duty. The center extracts operational records from various record repositories and, if the information is available, corroborates the descriptions of incidents described by veterans in their claims. According to DOD officials, unless an herbicide-related incident was documented in some sort of unit record, the center would not have information on where Agent Orange was present. Despite these various approaches for communicating information to veterans and the public, veterans we spoke with expressed confusion as to where to obtain information on their potential exposure to Agent Orange. Specifically, we asked veterans in our six discussion sessions about what they had heard from DOD, VA, or other federal agencies about the potential that they could have been exposed to Agent Orange or its components at locations where Agent Orange was manufactured, transported, stored, used, or destroyed. Veterans in each of the six sessions stated that, generally, the federal government has not reached out to them regarding Agent Orange, but that they instead have relied on their own research to learn more about their potential for having been exposed, adding to the confusion about where to obtain information on Agent Orange exposure. Other veterans, however, stated that they have received information from VA regarding potential exposure. DOD officials acknowledged that there is confusion among veterans about a variety of issues related to their potential for exposure to Agent Orange, including where to go for information. U.S. EPA and DOD officials stated that veterans are contacting multiple agencies to get information on herbicide exposure. Selected Comments by Veterans at Discussion Sessions Moderated by GAO Regarding What They Had Heard from the Federal Government about Negative Health Effects Associated with Exposure to Herbicides, Including Agent Orange or Its Components I’ve heard things from multiple sources—media, newspaper, television, people themselves. It has mainly been from my own research, not from a federal agency. Just based on the fact that I have heart disease and going through the VA process means I receive updates from VA on just about everything going on, including Agent Orange and all of the research they have done. I do know the Secretary is authorized by law from Congress late last year to add additional presumptive diseases associated with Agent Orange and how one would contract that. I had to do the research myself. It seems to be a secret with information coming out in spurts. When you have things happen to your body, they [the Department of Veterans Affairs] say it is not service connected. Sometimes when the government tries to explain something, they don’t give the whole thing and they give it piecemeal. It does not carry any essence of importance. I am not hearing anything from the federal government. Most of the information I get is from a USveterans.com website and I subscribe to a daily newsletter from the Vietnam Veterans of America and the Veterans of Foreign Wars. There is information on the VA website about conditions attributed to Agent Orange In that context, I went to the VA website and found that there are 21 states where Agent Orange was used, including on Hawaii in Kauai. It is because of this list that I became aware that people in Hawaii may have been exposed to Agent Orange. I learned that such exposure might increase the likelihood of having diabetes or cancer. I believe the list is still on the VA website and that there is also a list of units that were possibly exposed to Agent Orange. I have not been contacted by any government agency with regard to Agent Orange exposure or ill health. I first heard about Agent Orange and dioxin and cancer related issues/illnesses in late 1980s or early 1990s and later on after doing own research. Standards for Internal Control in the Federal Government state that management should internally and externally communicate the necessary quality information to achieve an entity’s objectives. The standard further states that management should evaluate the entity’s methods of communication so that the organization has the appropriate tools to communicate quality information throughout the entity on a timely basis. Additionally, DOD issued guidance in June 2017 establishing procedures for DOD components to implement when there is a scientifically plausible likelihood of a significant long-term health risk from a past environmental exposure to military personnel or civilians resulting from living or working on military installations. Even though the testing and storage of Agent Orange and its components occurred several decades ago, this instruction states that DOD components should provide targeted and effective health risk communication early and continuously, as new and credible information becomes available. However, DOD and VA officials stated that they have not developed a formal process for coordinating on how best to communicate information to veterans and the public regarding the presence of Agent Orange at locations outside of Vietnam. Officials stated that the DOD-VA Deployment Health Working Group—an existing forum for exchanging information—meets monthly to discuss health issues, including those related to Agent Orange. However, the working group is not focused on ensuring the availability and distribution of information on Agent Orange testing and storage locations. DOD’s and VA’s joint Herbicide Orange Working Group has the potential for being an effective forum for communicating this information; however, a DOD official noted that this is an ad hoc group, and as we discussed earlier, it has not yet determined the direction it will be taking for communicating with veterans regarding exposure to Agent Orange. By coordinating on how best to communicate this information, VA would be better positioned to provide veterans with information regarding their potential exposure to Agent Orange at locations where Agent Orange was known to have been present outside of Vietnam. Challenges Exist with Testing for Agent Orange Today Due to Degradation and Multiple Sources of Potential Contamination Testing to determine whether Agent Orange was present in a particular location is challenging because, for example, derivatives of Agent Orange—including the two components of Agent Orange (n-butyl 2,4-D and n-butyl 2,4,5-T) and the contaminant from the 2,4,5-T manufacturing process (2,3,7,8-TCDD)—degrade over time, and because derivatives of 2,4-D and 2,4,5-T can come from multiple sources. Regardless of these challenges, in response to a request by the Government of Guam, DOD developed a testing plan that was reviewed and accepted by U.S. EPA and Guam EPA to conduct a limited investigation into alleged Agent Orange use at three sites on Guam. Testing for Agent Orange Presents Challenges Due to Degradation and Multiple Sources of Potential Contamination Challenges Due to Degradation Testing to identify locations where Agent Orange may have been present is challenging because the components of Agent Orange degrade over time. It has been nearly 50 years since Agent Orange was last transported and used in support of military operations in Vietnam. According to scientific research, it is difficult to find traces of the two components of Agent Orange—n-butyl 2,4-D and n-butyl 2,4,5-T— because, under normal environmental conditions, the n-butyl forms break down rapidly into the acid forms. Scientific research indicates that the half-lives of the acid forms of the chemical components 2,4-D and 2,4,5-T in soil can range from several days to many months, depending on conditions. The World Health Organization has stated that the half-life of 2,4-D in soil is reported to range from 4 to 7 days in most soil types. According to the Centers for Disease Control and Prevention, the half-life of 2,4,5-T in soil varies with conditions, ranging from several weeks to many months. In addition, when Agent Orange is sprayed for defoliation, there are several things that can happen to it. For example, it can be washed out by rain, degrade in the presence of sunlight (photodegradation), or slowly turn into a vapor (volatize) from surfaces such as foliage. These factors reduce the chances of finding traces of Agent Orange components after 50 years. The amount of time it takes for the contaminant 2,3,7,8-TCDD to degrade is longer than that for the components of Agent Orange, although estimates vary. For example, according to the research cited by the Agency for Toxic Substances and Disease Registry, the half-life of 2,3,7,8-TCDD is approximately 9 to 15 years in surface soil and 25 to 100 years in subsurface soil. Further, 2,3,7,8-TCDD breaks down quickly when exposed to sunlight, providing one explanation for the shorter half- life in surface soil. Any 2,3,7,8-TCDD contamination from herbicide spraying—as opposed to being spilled onto the soil—would generally be expected to be found in surface soil, where it would be exposed to degradation due to sunlight. This reduces the likelihood of detecting this compound 50 years later. However, as discussed below, there are multiple sources of dioxins, including 2,3,7,8-TCDD, and the specific source of dioxin contamination is difficult to identify. Challenges Due to Multiple Sources of Potential Contamination Testing to identify locations where Agent Orange may have been present is challenging because there are multiple sources of 2,4-D and 2,4,5-T derivatives as well as multiple sources of the contaminant, 2,3,7,8-TCDD. Specifically, many commercial herbicides that were available at the time Agent Orange was used contained derivatives of 2,4-D; 2,4,5-T; or both. Additionally, 2,4-D derivatives are still used in commercial herbicides today. Therefore, even if testing were to show the presence of one of the two components of Agent Orange, it would be difficult to distinguish whether the chemicals were present from the use of commercial herbicides or the use of tactical herbicides. Further, because 2,4-D is still used in many commonly used herbicides sold today, the presence of this component could be due to a recent use of a commercial herbicide rather than a tactical herbicide used decades ago. Moreover, multiple sources of the contaminant 2,3,7,8-TCDD can be found in the environment today. DOD and U.S. EPA officials told us that if 2,3,7,8-TCDD is found in soil today, the source of the dioxin contamination could be a result of other sources besides Agent Orange. For example, according to the World Health Organization, dioxins— including 2,3,7,8-TCDD—are primarily released to the environment with the burning of materials such as wood and waste (see figure 12). Testing for the Components of Agent Orange on Guam Is Challenging In 2017 the Government of Guam coordinated with DOD to test for Agent Orange and other tactical herbicides at Andersen Air Force Base due to claims from veterans that they were exposed to Agent Orange while stationed on Guam during the 1960s and 1970s. In December 2017 DOD developed a draft testing plan in collaboration with U.S. EPA and Guam EPA to test for the acid form of the components 2,4-D and 2,4,5-T at three different sites on Andersen Air Force Base. The draft testing plan did not include testing for the presence of 2,3,7,8-TCDD. According to DOD and U.S. EPA officials, they are not testing for 2,3,7,8-TCDD because the test would not be able to conclusively link any positive results to the use of tactical herbicides, given that dioxins are also produced by, among other things, burning fossil fuels. These officials noted that, over time, large quantities of fuel have been burned at Andersen Air Force Base, and they stated their belief that if 2,3,7,8-TCDD were found, the likely source would be from combustion. The areas identified for testing included the fuel pipeline, a perimeter fenceline, and an area near some fuel storage tanks. See figure 13 for a photograph of the fenceline testing site near the fuel storage tanks on Andersen Air Force Base. Based on our initial review of the draft testing plan and a review of the scientific literature, we identified and discussed with DOD and U.S. EPA officials some challenges the two agencies would face in detecting the presence of Agent Orange on Guam due to two factors: (1) the short amount of time that it takes for 2,4-D and 2,4,5-T to degrade; and (2) the inability of testing to determine whether the presence of 2,4-D and 2,4,5-T is attributable to the use of Agent Orange or to some other source. Degradation of 2,4-D and 2,4,5-T: DOD officials and the jointly developed draft testing plan acknowledged that the planned testing would not be able to confirm the presence of Agent Orange, given that the components degrade over time. The draft testing plan indicates that the maximum half-lives of 2,4-D and 2,4,5-T are 14 days and 24 days, respectively, in soil and groundwater. Even given the possible variation in half-lives discussed above, it is likely that no detectable concentrations remain in soil today, given that the alleged period of use on Guam was in the 1960s and 1970s. Inability to distinguish whether the presence of 2,4-D and 2,4,5-T is attributable to the use of Agent Orange or some other source: Even if the results were to confirm the presence of either 2,4-D or 2,4,5-T in any form, it would be difficult to distinguish the source of the chemical, and whether its presence was attributable to the use of Agent Orange or some other source. For example, 2,4-D is still in use today, and 2,4,5-T was used in both tactical and commercial herbicides during the 1960s. In addition, if the components were found, the interpretation of those results could be complicated by, for example, natural variability in the potential half-lives and the possibility of more recent use of banned products. Further, the testing protocol will convert all forms of 2,4-D and 2,4,5-T, including the ester forms, to the acid forms, further complicating any attempt to identify the source of the compounds. We discussed with cognizant officials the challenges that we identified in the draft testing plan to determine how the information from the testing would be used to inform U.S. EPA, DOD, veterans, and the public about whether Agent Orange was present on Andersen Air Force Base. DOD officials subsequently stated that the questions raised by us and internally within DOD led them to reconsider the approach for testing for Agent Orange on Guam. For example, in December 2017, DOD officials told us that they would begin testing for Agent Orange and other tactical herbicides in March 2018. In late March 2018, a DOD official noted that the department had placed the testing on hold until they were certain that the methodology to be employed would meet scientific rigor and could be replicated in future testing efforts at other locations. In April 2018, DOD officials told us that the contract execution took longer than anticipated, and that soil sample testing would commence that month. In April 2018, DOD provided us with a copy of the final plan that was reviewed and approved by U.S. EPA and Guam EPA and was used to test for Agent Orange and other tactical herbicides at Andersen Air Force Base. When we reviewed the final testing plan and compared it with the draft previously provided, we found that some of the challenges we had initially identified in the draft testing plan, as described above, were still present. For example, based on our review of the final testing plan, with the proposed testing methodology, it would be difficult to determine if 2,4- D and 2,4,5-T came from Agent Orange or another source, and there were inconsistencies in the reported half-lives of the components of Agent Orange. At the same time, both DOD and U.S. EPA officials questioned the ability of any testing for 2,4-D or 2,4,5-T on Andersen Air Force Base to either confirm or deny the presence of Agent Orange on Guam. Specifically, the final testing plan states that more than 50 years have passed since the period of alleged use, and that a lack of detection provides no evidence that herbicides were not used historically. Moreover, U.S. EPA officials noted that the testing on Guam would not provide definitive proof of Agent Orange use on the island. Although DOD officials recognized these challenges and acknowledged the low probability of conclusively identifying the components of Agent Orange, they decided to move forward with testing to address veterans’ and the public’s concerns. In April 2018, samples were collected from the three areas at Andersen Air Force Base, according to DOD officials. Each sample was divided following procedures outlined in the final testing plan, resulting in two identical sample sets. A sample set was sent to two independent laboratories for analysis. According to officials from DOD and U.S. EPA, test results and associated quality control reports from both laboratories agreed on the results from two of the area samples, but did not agree on the third area sample. The jointly developed decision rules for the sampling and analysis plan required the results from both laboratories to agree in order to draw a conclusion on the presence or absence of Agent Orange. As a result, according to the officials, the DOD, U.S. EPA, and Guam EPA project team agreed in July 2018 to resample the one area where the two labs reported differing results. The project team is updating the sampling and analysis plan to address the various possible reasons for the differing laboratory results in order to provide a conclusive final testing result. DOD officials told us they do not anticipate completing the updates for the sampling and analysis plan, field sampling, analysis, and reporting until early 2019. As such, we were not able to comment on the results of the final testing in this report. Moreover, DOD officials said that, provided the final resampling results are negative, DOD does not have plans to conduct additional testing, because the testing was conducted in areas alleged to be the likeliest locations for the application of Agent Orange. However, an official from U.S. EPA said that the challenges associated with testing on Guam are not insurmountable and that the agency would like to continue this investigation. Given that (1) DOD, working with U.S. EPA and Guam EPA, made a decision to test for Agent Orange and other tactical herbicides; (2) DOD, U.S. EPA, and Guam EPA recognize the limitations associated with the testing; (3) the testing and analysis of results are still on-going; and (4) there is currently uncertainty regarding whether any additional testing will take place on Guam, we are not making any recommendations with respect to the testing plan or its execution. Conclusions DOD suspended the use of Agent Orange in Vietnam in 1970 and incinerated remaining stockpiles at sea in 1977, but concerns about the effects of exposure in U.S. locations have persisted. DOD developed a list that identifies locations and dates where herbicides, including Agent Orange, are thought to have been tested and stored outside of Vietnam, which VA has made publicly available on its website, but this list is neither accurate nor complete. Without assigning responsibilities for verifying the accuracy of the information included on the list; a process for ensuring that the list is updated, as new information is found; and clear and transparent criteria, indicating which locations should be included on the list, DOD and VA will not have assurance that they have the most complete information possible when informing veterans and the public of the full extent of locations where Agent Orange exposure could potentially have occurred. By relying on an inaccurate list, VA may not have quality information when making important decisions on claims for veterans who might or might not be eligible for benefits. Further, while DOD and VA both communicate with veterans in response to their Agent Orange inquiries, the two agencies do not have a formal process for coordinating on how best to communicate this information. Until DOD and VA develop a process for how best to coordinate to ensure that they are communicating information, veterans and the public may not have the information needed regarding their potential exposure to Agent Orange. Recommendations for Executive Action We are making six recommendations: four to the Secretary of Defense and two to the Secretary of Veterans Affairs. The Secretary of Defense should ensure that the Under Secretary of Defense for Acquisition and Sustainment assigns responsibility for ensuring that DOD’s list of locations where Agent Orange or its components were tested and stored is as complete and accurate as available records allow. (Recommendation 1) The Secretary of Defense should ensure that the Under Secretary of Defense for Acquisition and Sustainment develops a process for updating the revised list as new information becomes available. (Recommendation 2) The Secretary of Defense, in collaboration with the Secretary of Veterans Affairs, should develop clear and transparent criteria for what constitutes a location that should be included on the list of testing and storage locations. (Recommendation 3) The Secretary of Veterans Affairs, in collaboration with the Secretary of Defense, should develop clear and transparent criteria for what constitutes a location that should be included on the list of testing and storage locations. (Recommendation 4) The Secretary of Defense, in collaboration with the Secretary of Veterans Affairs, should develop a formal process for coordinating on how best to communicate information to veterans and the public regarding where Agent Orange was known to have been present outside of Vietnam. (Recommendation 5) The Secretary of Veterans Affairs, in collaboration with the Secretary of Defense, should develop a formal process for coordinating on how best to communicate information to veterans and the public regarding where Agent Orange was known to have been present outside of Vietnam. (Recommendation 6) Agency Comments and Our Evaluation We provided a draft of this report for review and comment to DOD, VA, U.S. EPA, the U.S. Department of Agriculture, and the U.S. Department of Health and Human Services. In its written comments, DOD concurred with each of our four recommendations directed to the Secretary of Defense and identified actions it plans to take to implement them. In its written comments, VA concurred with one recommendation directed to the Secretary of VA and described actions it would take to implement the recommendation. VA also non-concurred with one recommendation. In its written comments, the U.S. Department of Agriculture agreed with the report’s findings related to matters under the purview of agricultural research and programs, though we did not make any recommendations to the department. Comments from DOD, VA, and the U.S. Department of Agriculture are reprinted in their entirety in appendixes V through VII. We also received technical comments from DOD, VA, U.S. EPA, and the U.S. Department of Health and Human Services, which we incorporated as appropriate. Based on oral comments we received from DOD, we revised our recommendation regarding the development of clear and transparent criteria for what constitutes a location that should be included on the list of testing and storage locations to clarify that DOD and VA should collaborate on this effort. VA non-concurred with this recommendation, noting that DOD chairs the Herbicide Orange Working Group that will be responsible for developing the criteria (Recommendation 4). However, VA stated that as a member of the working group, it would work collaboratively with DOD as the lead. Doing so would meet the intent of our recommendation. In its overall written comments, VA stated that it was concerned that the report conflates the terms “commercial herbicides” with “tactical herbicides,” which the department noted were distinctive from one another. While VA stated that it does not dispute that some chemicals found in the VA regulation may be included in certain commercial herbicides, VA noted that exposure to tactical herbicides intended for military operations in Vietnam is required for VA to grant disability benefits on a presumptive basis. We recognize that the presumption for service- connection applies to exposure to tactical herbicides and nothing in our report states otherwise. VA also stated in its letter that the focus on commercial herbicides is not relevant for determining the list of locations where tactical herbicides were tested or stored. We agree and as we noted in this report, the U.S. military managed tactical herbicides used during the Vietnam War era differently from commercial herbicides in the federal supply system, which were widely available worldwide for use in vegetation management at military installations. To avoid conflating tactical and commercial herbicides, the report further notes that while some of these commercial herbicides contained 2,4-D; 2,4,5-T; or both, these commercial herbicides were not in the n-butyl form used in Agent Orange. However, commercial herbicides with 2,4,5-T likely contained some level of 2,3,7,8-TCDD. Moreover, we believe it is important to reiterate that numerous commercial herbicides that were being widely used elsewhere for agriculture purposes contained the form of 2,4,5-T found in Agent Orange and thus its associated dioxin contaminant, 2,3,7,8-TCDD. In its overall written comments, VA also recommended that GAO analyze its list to ensure that only locations where the presence of tactical herbicides has been confirmed are included on the list of locations. It is important to note that we do not maintain a list of herbicide testing and storage locations. As we noted in this report, DOD developed a list that identifies locations and dates where herbicides, including Agent Orange and its components, are thought to have been tested and stored outside of Vietnam, which VA has made publicly available on its website. We are sending copies of this report to the appropriate congressional addressees; the Secretaries of Defense, VA, Agriculture, and Health and Human Services; and the Administrator of U.S. EPA. In addition, the report will be available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact Brian Lepore at (202) 512-4523 or leporeb@gao.gov or J. Alfredo Gómez at (202) 512-3841 or gomezj@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix VIII. Appendix I: Objectives, Scope, and Methodology House Report 115–200 accompanying a bill for the National Defense Authorization Act for Fiscal Year 2018 included a provision that we review the government’s handling of Agent Orange on Guam. In response to both this provision and a separate request letter, this report examines (1) the extent to which the federal government has information about the procurement, distribution, use, and disposition of Agent Orange or its components at locations in the United States and its territories, including Guam; (2) the extent to which the Department of Defense (DOD) and the Department of Veterans Affairs (VA) have complete and accurate information about where Agent Orange and its components were tested and stored and communicated this information to veterans and the public; and (3) challenges associated with testing for Agent Orange. For objective one, we collected and reviewed available agency records and shipping documents on Agent Orange from the following sources: the U.S. National Archives and Records Administration; the U.S. National Agricultural Library; the U.S. Air Force Historical Research Agency at Maxwell Air Force Base, Alabama; staff historians at the Air Force Materiel Command at Wright- Patterson Air Force Base, Ohio, and Pacific Air Forces at Joint Base Pearl Harbor–Hickam, Hawaii; the Armed Forces Pest Management Board in Silver Spring, Maryland; the Defense Logistics Agency; the U.S. Army Corps of Engineers; and the Naval History and Heritage Command. The records we researched and collected include published and unpublished materials on the procurement, shipment, and disposition of Agent Orange, including U.S. military correspondence, logistics reports, and Navy and merchant vessel logbooks. We reviewed DOD documents related to Agent Orange contracts to determine the total quantity of Agent Orange that was produced by the nine manufacturers. To show how much Agent Orange was used in Vietnam, we used estimates from the National Academy of Sciences analysis of Operation Ranch Hand data. Details about the estimated quantity of Agent Orange that was destroyed in 1977 are available in public reports from DOD and the U.S. Environmental Protection Agency (U.S. EPA). We used a variety of archival sources to identify the shipping routes for Agent Orange, to include a database prepared for VA that lists records held in National Archives and Records Administration Record Group 341, which contains more than 200 boxes of unclassified records relating to tactical herbicides used in Vietnam. During our review of this record group, we identified and summarized the correspondence between and reports submitted by the U.S. military commands that managed the tactical herbicides, to identify details of tactical herbicide shipments and, to the extent that the data were available, to develop a consolidated list of shipments of Agent Orange, including vessel names, ports of embarkation and debarkation, time frames, and quantities. In some cases, individual source documents did not identify which specific tactical herbicides were being shipped. To the extent we were able, we used multiple sources to identify which shipments carried Agent Orange. For the purposes of this report, we refer to these records collectively as shipment documentation. Using this shipment documentation, we located and obtained from several regional facilities of the National Archives and Records Administration logbooks for the vessels that we had identified as having shipped Agent Orange—hereinafter referred to as logbooks—which accounted for approximately 83 percent of the shipments we found. Logbooks that were submitted to port authorities upon the vessels’ returns to the United States were consolidated at National Archives and Records Administration facilities including Fort Worth, Texas; Seattle, Washington; San Francisco and Riverside, California; New York, New York; Philadelphia, Pennsylvania; Boston, Massachusetts; Chicago, Illinois; and Atlanta, Georgia, as well as at Archives I in Washington, D.C., and Archives II in College Park, Maryland. These logbooks recorded basic details about each ship’s operation and route, which we analyzed to identify any shipments that stopped at locations in the United States or its territories before arriving in Vietnam. Because none of the logbooks we reviewed provided detail about the specific types of cargo that were loaded onto or unloaded from the vessels, we relied on available military correspondence and reports about those vessels to identify whether the ships carried Agent Orange. We attempted to locate the remaining 17 percent of the logbooks, or 27 shipments. Of those shipments, 3 were by foreign-flagged merchant vessels, which did not submit logbooks to U.S. ports. Working with officials from the U.S. Coast Guard, the agency that oversees the retention and archiving of logbooks, we coordinated with archivists at the Federal Records Centers to determine whether there were any unprocessed boxes of logbooks that had not yet been archived. When that effort did not turn up additional logbooks, we worked with archivists at Archives I to obtain copies of shipping articles—the articles of agreement between the captain of a ship and the seamen with respect to wages, length of time for which they are shipped, and related matters—for the remaining 24 shipments. While these documents focus on employment issues, the annotations include the locations where different personnel actions took place. We reviewed these documents to identify the locations and approximate dates of the ports of call during those voyages. We were able to obtain the shipping articles for the 24 remaining voyages, as well as for the one vessel that stopped in Guam on the way to Vietnam (SS Gulf Shipper) and the three that stopped in Guam on the way back (SS Aimee Lykes, SS Buckeye Atlantic, and SS Overseas Suzanne). Using the information on voyage ending dates and ports that we obtained from the shipping articles, we were able to work with the regional archives to obtain another 21 logbooks, bringing the total number of logbooks obtained to 152, or 96 percent of the shipments we identified. We relied on the shipping article information for the remaining three voyages (excluding the shipments on the three foreign-flagged vessels) to provide some information on the routes taken by those vessels. However, one limitation of relying on shipping articles for port information and dates is that locations are mentioned only if a personnel action—such as an injury, hospitalization, or desertion—took place. If no personnel action took place at a location on a vessel’s route, that port would not be listed in the shipping articles. To obtain specific information about the SS Gulf Shipper voyage that stopped in Guam en route to Vietnam, to include documentation on its cargo and whether or not cargo was loaded or unloaded at the ports on the way to Vietnam, we contacted officials at several agencies. In Guam, we contacted the Customs and Quarantine Service, the University of Guam’s Micronesian Research Center, and officials at Naval Base Guam for information on vessels that stopped in Guam during the Vietnam War era, and any cargo they carried. We also contacted archivists at the Federal Records Center in Seattle, Washington, where the SS Gulf Shipper logbook is archived, and the regional archives in Fort Worth, Texas, for additional information on the vessel itself and guidance on retaining and archiving cargo information. The National Archives had some information on the SS Gulf Shipper, such as sales documents and company correspondence records. However, the National Archives did not have records for the manifest or bills of lading, which may have documented any cargo offloaded from the ship. We contacted U.S. Customs and Border Protection for information on movements of vessels engaged in foreign trade in and out of ports, which is found in customs forms that are required to be archived after 30 years. We were unsuccessful in locating the customs forms for the SS Gulf Shipper’s voyage to Vietnam through Guam; however, an official noted that although these records provide manifest numbers and ports of sailing, the manifests themselves are not archived. An online search on the SS Gulf Shipper through the U.S. Maritime Administration’s website identified the transfer of vessel ownership over the years. We contacted the latest company that owned the vessel to see whether the company had retained any cargo manifests or other historical records as the ownership changed hands. However, we could not obtain this information because, according to a company official we contacted, the vessel’s records, along with other historical documents, were stored in an off-site storage facility in New Jersey, and were subsequently destroyed in a fire in 1996. We also looked at articles from Guam newspapers and news sources such as the Military Sea Transportation Service Vietnam Chronicles for any information about vessel comings and goings in Guam in early 1968 to see if they mentioned the SS Gulf Shipper or specific cargo being offloaded in Guam. None of these contacts or written sources provided information specific to any cargo that was being moved through Guam, or about this particular vessel. We also obtained original DOD reports and command histories that provided additional operational details about the procurement, distribution, use, and disposition of Agent Orange and its components. According to an Office of History, Air Force Logistics Command, monograph, the command directly responsible for managing Agent Orange was the Directorate of Aerospace Fuels at the San Antonio Air Materiel Area at the former Kelly Air Force Base, Texas, which was a sub-component of the U.S. Air Force Logistics Command during the Vietnam War. The unclassified San Antonio Air Materiel Area command histories for the years 1966 through 1973 include chapters with extensive documentation on “herbicide management.” We obtained copies of command histories from the Air Force Historical Research Agency at Maxwell Air Force Base, Alabama, and the Air Force Materiel Command at Wright-Patterson Air Force Base, Ohio. To obtain information regarding herbicide use on Guam, we obtained command histories for Naval Base Guam and an analysis and summary of the available documentation by the historian at Andersen Air Force Base. We also spoke with Navy and Air Force officials on Hawaii and Guam to identify any relevant records pertaining to such use. In addition, we met with and obtained information from officials representing the Office of the Governor of Guam and senior members and staff from the Guam Legislature. We also met with officials representing a veterans service organization. Finally, as discussed below, we spoke directly with veterans about their recollections of herbicide use on Guam, and any documentation they might have pertaining to such use. For objective two, we analyzed the archival search records provided by DOD to identify additional locations where Agent Orange or its components were tested and stored in the United States and its territories. We reviewed Army archives search reports of herbicide testing at Aberdeen Proving Grounds (including Edgewood Arsenal), Maryland; Dugway Proving Ground, Utah; Fort Chaffee, Arkansas; Fort Gordon, Georgia; Fort Meade, Fort Ritchie, and Fort Detrick, Maryland; and two Air Force studies related to herbicide equipment testing at Eglin Air Force Base, Florida, to determine whether there were additional sites and testing events that were not included on the DOD list found on the VA website. We also reviewed the proceedings of the First, Second, and Third Defoliation Conferences, technical and special reports, and published papers provided by the Armed Forces Pest Management Board to determine whether there were additional sites and testing events that were not included on the list. We compared the information about testing locations and dates on the DOD list found on the VA website with information found in a 2006 report on locations where Agent Orange was tested and stored. To determine the locations where Agent Orange or its components were tested and stored, we attempted to identify the chemical composition of all the agents on DOD’s list found on the VA website. We located information on the chemical composition of agents on the list in archives search reports for Forts Detrick, Meade, and Gordon; a glossary of pesticide chemicals from the Food and Drug Administration; journal articles; and the defoliation conference proceedings. We also interviewed DOD and VA officials about the chemical composition of agents on the list, the origins of the list, how the list is used, and the role of each agency in managing the list. We compared the results with information that DOD and VA provided publicly on testing and storage locations of tactical herbicides in the United States and its territories, and with DOD policies for conducting record research and responding to inquiries related to past environmental exposures. We also compared the accuracy and completeness of the list with Standards for Internal Control in the Federal Government, which state that management should internally and externally communicate the necessary quality information to achieve the entity’s objectives. We also reviewed the extent to which DOD and VA have communicated health information to DOD personnel and veterans. We compared the communication process that both DOD and VA use with DOD’s guidance on assessing long-term health risks, and with VA’s process for determining benefits based on veterans’ claims. We also compared DOD and VA actions with Standards for Internal Control in the Federal Government, which state that management should internally and externally communicate the necessary quality information to achieve the entity’s objectives. The standard further states that management should evaluate the entity’s methods of communication so that the organization has the appropriate tools to communicate quality information throughout the entity on a timely basis. We also reviewed documents from DOD and VA on communication with veterans, including the VA’s website on Agent Orange. Further, we interviewed cognizant agency officials from DOD and VA, including officials from the Armed Forces Pest Management Board and DOD’s Joint Services Records Research Center. For objectives one and two, to better understand veterans’ experiences with Agent Orange and other herbicides and the health effects of exposure to them, we conducted six small discussion sessions with a non-generalizable sample of veterans. Four of the discussion sessions were conducted in person in the following locations: two discussion sessions in Guam, and two discussion sessions in Hawaii. We conducted two additional discussion sessions that were moderated via telephone from Washington, D.C.: one of those had individuals participate both in person and by telephone, while the other was held solely by telephone. We selected Guam because of the provision in House Report 115–200 accompanying a bill for the National Defense Authorization Act for Fiscal Year 2018 for GAO to review the government’s handling of Agent Orange on Guam. We selected Hawaii because of its strategic location during the Vietnam War and because of the VA presence in the region. A total of 38 individuals attended the sessions, which ranged from 1 to 10 participants per session and lasted approximately 1 to 2 hours. These discussion sessions were consistently moderated by the same team member using a prepared script and documented by several other team members. To select candidates for participating in our discussion sessions, we worked with the Veterans Health Administration as well as veteran clinics and veteran centers at the selected locations to identify non-combat veterans who had served during the Vietnam era. In Guam, we also worked with the Guam Environmental Protection Agency to coordinate a discussion session. Attendees included Vietnam-era veterans who self- reported that they were in active service between 1961 and 1977 in Vietnam, the United States, and its territories, including Guam. As we became aware of other veterans who might be interested in these discussion sessions, including Vietnam combat veterans, we reached out to offer them the opportunity to participate in one of our discussion sessions. Our six discussion sessions included questions to individuals regarding what, if anything, they had heard from DOD, VA, or other federal agencies about links between exposure to herbicides and negative health effects, and whether attendees believed that they had been exposed to Agent Orange or its components at locations where Agent Orange was manufactured, transported, stored, used, or destroyed. We also asked individuals if they believed they had been exposed to Agent Orange in Guam, Vietnam, or another location, and if so, to describe the situation. At the start of the discussion sessions, the moderator told participants that their responses would be kept confidential and that we were not recording their statements. The moderator noted that we would be taking notes to make sure we accurately captured the conversations, but that we would not attribute statements directly to individuals. For those discussion sessions held in person in Guam and Hawaii, we also administered a brief, written questionnaire about individuals’ experiences during the Vietnam era (for example, duty locations, military occupation, rank), and what they had heard and experienced related to Agent Orange and other herbicides. Due to logistical obstacles, we were not able to administer the questionnaire to participants in sessions held via telephone. However, the information requested in the questionnaire was also covered in the discussion sessions themselves. Therefore, we did not analyze the information from the completed questionnaires. We also solicited from the veterans any documentation they might have that could support their allegations of the use of Agent Orange on Guam, but we did not receive documentation that corroborated the use of Agent Orange on Guam. In addition, we met with officers from the Vietnam Veterans of America to discuss how, if at all, veterans could have been exposed to Agent Orange beyond serving directly in Vietnam as part of Operation Ranch Hand, and how the organization disseminates information, especially on Agent Orange, to veterans. For objective three, we reviewed scientific literature and agency documents regarding the degradation and sources of the components of Agent Orange and an associated dioxin contaminant, 2,3,7,8-TCDD, as well as other sources of dioxins. This review included documents from the Agency for Toxic Substances and Disease Registry and reports and protocols from U.S. EPA, the World Health Organization, the Centers for Disease Control and Prevention, and the American Industrial Hygiene Association. We also reviewed the draft and final plans for testing for the presence of the acid forms of the components of Agent Orange—2,4-D and 2,4,5-T—on Guam. We compared the information outlined in the testing plan with scientific literature on the environmental fate of the components of Agent Orange and other Agent Orange testing methodologies. We interviewed officials from DOD, U.S. EPA, and Guam EPA about the testing plan for Guam and the science surrounding Agent Orange testing. We also conducted a site visit to Naval Base Guam and Andersen Air Force Base on Guam and interviewed DOD and Government of Guam officials involved in the planning for the testing for Agent Orange on Andersen Air Force Base. We visited the three selected sites where the initial testing took place and took photographs of those sites. We conducted this performance audit from May 2017 through November 2018, in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Comparison of the Department of Veterans Affairs (VA) List of Diseases Associated with Agent Orange against Those Identified by the National Academy of Sciences The VA recognizes 14 presumptive diseases associated with exposure to Agent Orange or other herbicides during military service for which veterans and their survivors may be eligible to receive disability compensation benefits. The list of diseases provided by the VA has generally incorporated the findings of reviews performed by the National Academy of Sciences (the Academy). The list includes 5 diseases that have been identified as having sufficient evidence of association and 9 that have been identified as having limited or suggestive evidence of association. In the Academy’s biannual reports, for a disease identified as having sufficient evidence of association, the evidence is sufficient to conclude that there is a positive association—that is, a positive association has been observed between herbicides and the outcome in studies for which chance, bias, and confounding could be ruled out with reasonable confidence. For a disease identified as having limited or suggestive evidence of association, the evidence is suggestive of an association between herbicides and the outcome but is limited, because chance, bias, and confounding could not be ruled out with confidence. Table 1 describes those 14 diseases and the extent of association identified by the Academy. The 2014 Academy biannual report, issued in 2016, listed four more diseases it categorized as having limited or suggestive evidence of association, as described in table 2. VA officials told us that these diseases are not included on the VA’s current list of presumptive diseases associated with exposure to Agent Orange or other herbicides because, as of October 25, 2018, the Secretary of Veterans Affairs had yet to make the determination based on the most recent biannual review (the 2014 report). According to the officials, the Secretary is also considering the inclusion of parkinsonism and Parkinson-like syndromes. Finally, according to the VA website, VA has recognized that certain birth defects among veterans’ children are associated with veterans’ qualifying service in Vietnam or Korea. For example, spina bifida (except spina bifida occulta) is associated with veterans’ exposure to Agent Orange or other herbicides during qualifying service in Vietnam or Korea. The affected child must have been conceived after the veteran entered Vietnam or the Korean demilitarized zone during the qualifying service period, and a child with spina bifida or covered birth defects who is a biological child of a veteran with qualifying service may be eligible for a monetary allowance, health care, and vocational training. The 2014 report moved spina bifida to the lower category of "inadequate or insufficient evidence to determine an association," as studies that have been released since the 1996 update do not support a link between the condition and exposure to herbicides. According to VA officials, VA does not currently plan to change its regulations based on this conclusion. Appendix III: Quantities of Herbicides Known to Have Been Shipped to Southeast Asia on Ships Identified as Having Stopped in Guam between February 1968 and May 1970 Based on available shipment documentation and logbooks, we identified one vessel—the SS Gulf Shipper—carrying Agents Orange, Blue, and White that stopped at Port Apra (now Apra Harbor) on Guam on its way to Southeast Asia. Additionally, we identified three vessels—the SS Aimee Lykes, the SS Buckeye Atlantic, and the SS Overseas Suzanne—that stopped in Guam on the return routes after having made various port calls in Southeast Asia. For each of these voyages, we obtained shipment documentation that outlined the quantities of herbicides that records indicate had been loaded onto the vessels while at port in the United States, and to the extent available, quantities of herbicides that were discharged in Southeast Asia. We also obtained logbooks that identified the routes the vessels took from U.S. ports to Vietnam and back, and identified any port calls en route. While we are unable to confirm the reliability of the information available in shipment documentation and logbooks, details on the quantities of herbicides that were documented to have been transported on these vessels during their routes are outlined below. SS Gulf Shipper: According to shipment documentation and the vessel’s logbook, the SS Gulf Shipper left the port of Mobile, Alabama, on January 9, 1968, and stopped at Port Apra (now Apra Harbor) on Guam and offloaded a mariner for repatriation to the United States on February 2, 1968. We are unable to state with certainty whether there were reasons why this vessel stopped in Guam beyond what was reported in available shipment documentation and the vessel’s logbook. The logbook further indicates that the SS Gulf Shipper then arrived in Saigon, Vietnam, approximately February 27, 1968, with subsequent stops in Cam Rahn Bay, Vietnam, approximately February 29, 1968, and Nha Trang, Vietnam, approximately March 2, 1968. According to available documentation, there is some discrepancy with regard to the amount of herbicides that records indicate were loaded onto the SS Gulf Shipper when it left the port of Mobile, Alabama. Specifically, shipment documentation indicates that 62,570 gallons of Agent Orange, 31,735 gallons of Agent White, and 4,620 gallons of Agent Blue—a total of 98,925 gallons of herbicides—were loaded onto the SS Gulf Shipper before it departed for Saigon, Vietnam. On the contrary, according to the available shipping documentation, the vessel’s manifest indicates that the vessel was carrying 86,270 gallons of herbicides, but does not break the total down by individual herbicide. The vessel’s manifest further indicates that the SS Gulf Shipper discharged 93,150 gallons of herbicide in Saigon, Vietnam, on March 1, 1968, which does not align with reported dates in the vessel’s logbook. However, we are unable to determine discharge quantities by specific herbicide—for example, the quantities of Agents Orange, Blue, or White discharged—because available documentation states that the breakdown of the herbicides would not be determined until arrival at the depot. Moreover, we are unable to account for the difference between the number of gallons of herbicides reported to have been loaded onto the vessel and the number of gallons reported to have been discharged in Saigon, Vietnam, or potentially any other location. SS Aimee Lykes: According to shipment documentation and the vessel’s logbook, the SS Aimee Lykes left the port of Beaumont, Texas, on October 4, 1969. The vessel arrived in Saigon, Vietnam, approximately November 9, 1969. The vessel made a subsequent stop at Da Nang, Vietnam, approximately November 23, 1969. Following its departure from Vietnam, the SS Aimee Lykes stopped in Apra Harbor on Guam approximately November 30, 1969, and offloaded an injured crew member. However, the logbook does not include Guam on its list of ports of call. Rather, there is a separate entry within the logbook that describes the vessel pulling into Apra Harbor and offloading the injured mariner into a small motorboat so that he could be hospitalized in Guam. Therefore, we cannot confirm whether the vessel docked at Port Apra during this voyage. According to available documentation, the SS Aimee Lykes left the port of Beaumont, Texas, with 880 gallons of Agent Orange on board—documentation does not indicate that there were any amounts of Agents White or Blue on this voyage. Based on the available documentation, we are unable to determine the quantity of Agent Orange that was discharged in Saigon, Vietnam, or potentially any other location. SS Buckeye Atlantic: According to shipment documentation and the vessel’s logbook, the SS Buckeye Atlantic left the port of New Orleans, Louisiana, on October 1, 1969. The vessel arrived in Saigon, Vietnam, approximately November 20, 1969. The vessel made a subsequent stop at Qui Nhon, Vietnam, approximately November 29, 1969. Following its departure from Vietnam, the SS Buckeye Atlantic stopped at various ports in Japan before stopping in Guam approximately December 23, 1969, and offloading two injured crew members, one who returned to duty and another who was repatriated to the United States. While on Guam, the SS Buckeye Atlantic also performed a fire and boat drill on December 26, 1969, before departing. According to available documentation, the SS Buckeye Atlantic left the port of New Orleans, Louisiana, with 17,105 gallons of Agent Orange on board. Based on the available documentation, we are unable to determine the quantity of Agent Orange that was discharged in Saigon, Vietnam, or potentially any other location. SS Overseas Suzanne: According to shipment documentation and the vessel’s logbook, the SS Overseas Suzanne left the port of New Orleans, Louisiana, on February 28, 1970. The vessel arrived in Saigon, Vietnam, approximately April 9, 1970. The vessel made a subsequent stop at Da Nang, Vietnam, approximately April 17, 1970, and at Cam Rahn Bay, Vietnam, approximately April 22, 1970. Following its departure from Vietnam, the SS Overseas Suzanne stopped in Taiwan and Japan before stopping in Guam approximately May 5, 1970, and offloading an injured crew member. The vessel then departed Guam on May 9, 1970. According to available documentation, the SS Overseas Suzanne left the port of New Orleans, Louisiana, with 80,795 gallons of Agent Orange and 48,537 gallons of Agent Blue on board. Based on the available documentation, we are unable to determine the quantity of Agent Orange that was discharged in Saigon, Vietnam, or potentially any other location. Appendix IV: The Department of Defense’s (DOD) List of Testing and Storage Locations Posted on the Department of Veterans Affairs (VA) Website Appendix V: Comments from the Department of Defense Appendix VI: Comments from the Department of Veterans Affairs Appendix VII: Comments from the U.S. Department of Agriculture Appendix VIII: GAO Contacts and Staff Acknowledgments GAO Contacts Staff Acknowledgments In addition to the contacts named above, Kristy Williams and Barbara Patterson (Assistant Directors), Karyn Angulo, Emil Friberg, Ashley Grant, Karen Howard, Kelly Husted, Richard Johnson, Amie Lesser, Keegan Maguigan, Jeff Mayhew, Dennis Mayo, Parke Nicholson, Shahrzad Nikoo, Josie Ostrander, Rebecca Parkhurst, Michael Silver, Anne Stevens, Rachel Stoiko, Roger Stoltz, and Cheryl Weissman made key contributions to this report. Related GAO Products Agent Orange: Limited Information Is Available on the Number of Civilians Exposed in Vietnam and Their Workers’ Compensation Claims. GAO-05-371. Washington, D.C.: Apr. 22, 2005. Agent Orange: Persisting Problems with Communication of Ranch Hand Study Data and Results. GAO/T-NSIAD-00-117. Washington, D.C.: Mar. 15, 2000. Agent Orange: Actions Needed to Improve Communications of Air Force Ranch Hand Study Data and Results. GAO/NSIAD-00-31. Washington, D.C.: Dec. 17, 1999. Agent Orange Studies: Poor Contracting Practices at Centers for Disease Control Increased Costs. GAO/GGD-90-122BR. Washington, D.C.: Sept. 28, 1990. Agent Orange: VA Needs To Further Improve Its Examination and Registry Program. GAO/HRD-86-7. Washington, D.C.: Jan. 14, 1986. VA’s Agent Orange Examination Program: Actions Needed To More Effectively Address Veterans’ Health Concerns. GAO/HRD-83-6. Washington, D.C.: Oct. 25, 1982.

Why GAO Did This Study The tactical herbicide Agent Orange was first produced in 1964, and some 12 million gallons were shipped from U.S. ports to Southeast Asia from 1965 to 1970. DOD suspended its use in 1970 and incinerated remaining stockpiles at sea in 1977. Congress has expressed long-standing interest in the effects of Agent Orange exposure. The House report accompanying a bill for the National Defense Authorization Act for Fiscal Year 2018 included a provision that GAO review the government's handling of Agent Orange on Guam. This report examines (1) information the federal government has about the procurement, distribution, use, and disposition of Agent Orange; (2) DOD and VA efforts to make information about where Agent Orange and its components were tested and stored available; and (3) challenges associated with Agent Orange testing. GAO reviewed agency policies, documents, and available archival records that GAO identified; interviewed DOD, VA, and other agency officials; and met with a non-generalizable sample of 38 veterans and a veterans service organization. What GAO Found Available shipment documentation indicates that nearly all of the Agent Orange procured was either used in U.S. military operations in Southeast Asia, used for testing, damaged, or destroyed. However, some records are incomplete, such as shipment documentation and logbooks that identify ports where vessels stopped on the way to Southeast Asia. GAO obtained and reviewed shipment documentation for over 12.1 million of the 13.9 million gallons of Agent Orange procured by the Department of Defense (DOD). GAO reviewed logbooks for 96 percent (152 of 158) of those shipments and identified that vessels stopped at various ports on the way to Southeast Asia, including at least one vessel carrying Agent Orange that stopped at Guam. While the logbooks GAO reviewed identify when vessels left various ports as they traveled to and from Vietnam, they do not show whether and how much cargo was loaded or unloaded at those ports. DOD's official list of herbicide testing and storage locations outside of Vietnam that is posted on the Department of Veterans Affairs' (VA) website is inaccurate and incomplete. For example, the list lacks clarity in descriptive information and omits both testing and storage locations and additional time periods covered by testing events. Also, the list has not been updated in over a decade, though DOD and VA have obtained reports on its shortcomings since 2006. Both DOD and VA communicate with veterans in response to inquiries about Agent Orange, but some veterans GAO met with expressed confusion regarding how to obtain information on potential exposure. DOD officials acknowledged this confusion and stated that veterans are contacting multiple agencies to obtain such information. However, DOD and VA have not established a formal process for coordinating on how best to communicate information to veterans and the public regarding the presence of Agent Orange outside of Vietnam. Without a reliable list with complete and accurate information and a formal process for DOD and VA to coordinate on communicating this information, veterans and the public do not have quality information about the full extent of locations where Agent Orange was present and where exposure could potentially have occurred. Challenges exist with testing for Agent Orange today due to degradation of the herbicide's two chemical components and a potential for sources of contamination other than the herbicide. According to scientific research, the half-life (average time for components to decrease by half of the original amount) of Agent Orange's two chemical components—n-butyl 2,4-D and n-butyl 2,4,5-T— in soil can range from several days to many months, depending on conditions. The suggested half-life of the dioxin 2,3,7,8-TCDD—a by-product of the 2,4,5-T manufacturing process—is much longer, but there are multiple sources of dioxins, including the burning of wood and waste. DOD and the U.S. and Guam Environmental Protection Agencies are testing for the acid form of the components of Agent Orange at Andersen Air Force Base on Guam. While acknowledging the low probability of conclusively identifying the components of Agent Orange on Guam, DOD has made a decision to move forward with testing to address veterans' and the public's concerns, and it expects to complete the updates for the sampling and analysis plan, field sampling, analysis, and reporting in early 2019. What GAO Recommends GAO is making six recommendations, including that DOD develop a process for updating its list of Agent Orange testing and storage locations, and that DOD and VA develop a process for coordinating the communication of information on where Agent Orange was known to have been present. DOD concurred with four recommendations. VA concurred with one recommendation and non-concurred with one recommendation.

gao_GAO-18-636T

gao_GAO-18-636T_0

Background VA’s mission is to promote the health, welfare, and dignity of all veterans in recognition of their service to the nation by ensuring that they receive medical care, benefits, social support, and lasting memorials. In carrying out this mission, the department operates one of the largest health care delivery systems in the United States, providing health care services to approximately 9 million veterans throughout the United States, Philippines, Virgin Islands, Puerto Rico, American Samoa, and Guam. In 2015, we designated VA health care as a high-risk area for the federal government, and we continue to be concerned about the department’s ability to ensure that its resources are being used cost-effectively and efficiently to improve veterans’ timely access to health care. In part, we identified limitations in the capacity of VA’s existing IT systems, including the outdated, inefficient nature of certain systems and a lack of system interoperability as contributors to the department’s challenges related to health care. Providing health care to veterans requires a complex set of clinical and administrative capabilities supported by IT. VA’s health information system—VistA—has been essential to the department’s ability to deliver health care to veterans. VistA contains an electronic health record for each patient that supports clinical settings throughout the department. For example, clinicians can use the system to enter and review patient information; order lab tests, medications, diets, radiology tests, and procedures; record a patient’s allergies or adverse reactions to medications; request and track consults; enter progress notes, diagnoses, and treatments for encounters; and enter discharge summaries. VistA was developed in house by clinicians and IT personnel in various VA medical facilities and has been in operation since the early 1980s. Over the last several decades, VistA has evolved into a technically complex system comprised of about 170 modules that support health care delivery at 152 VA Medical Centers and over 1,200 outpatient sites. In addition, customization of VistA, such as changes to the modules by the various medical facilities, has resulted in about 130 versions of the system—referred to as instances. According to VA, VistA modules are comprised of one or more software applications that support various health care functions, such as providing care coordination and mental health services. In addition to VistA, the department has other health information systems that must interface with VistA to send, exchange, or store related health (e.g., clinical and patient) data. Since 2001, VA has identified the need for enhancements and modifications to VistA and has pursued multiple efforts to modernize the system. Two major efforts have included the VistA Evolution program and, most recently, the planned acquisition of the same electronic health record system that the Department of Defense (DOD) is acquiring. In 2013, VA established VistA Evolution as a joint program between OI&T and VHA that was comprised of a collection of projects and efforts focused on improving the efficiency and quality of veterans’ health care. This program was to modernize the department’s health information systems, increase VA’s data exchange and interoperability capabilities with DOD and private sector health care partners, and reduce VA’s time to deploy new health information management capabilities. In June 2017, the former VA Secretary announced a significant shift in the department’s approach to modernizing VistA. Specifically, rather than continue to use VistA, the Secretary stated that the department planned to acquire the same Cerner electronic health record system that DOD has been acquiring. Accordingly, the department awarded a contract to Cerner in May 2018 for a maximum of $10 billion over 10 years. Cerner is to replace VistA with a commercial electronic health record system. This new system is to support a broad range of health care functions that include, for example, acute care, clinical decision support, dental care, and emergency medicine. When implemented, the new system will be expected to provide access to authoritative clinical data sources and become the authoritative source of clinical data to support improved health, patient safety, and quality of care provided by VA. As previously mentioned, this acquisition is being managed by VA’s EHRM program. According to program documentation, EHRM is also to deliver program management support and the infrastructure modernization required to install and operate the new system. According to EHRM program documentation, the department has estimated that an additional $5.8 billion in funding, above the contract amount, would be needed to fund project management support and infrastructure improvements over the 10-year period. This amount does not fully include government employee costs. Deployment of the new electronic health record system at the initial sites is planned for within 18 months of October 1, 2018, with a phased implementation of the remaining sites over the next decade. Each VA medical facility is expected to continue using VistA until the new system has been deployed at that location. VA Has Reported Obligating about $3.0 Billion to VistA and Related Activities from Fiscal Years 2015 through 2017 According to VA, the department’s costs for VistA and related activities are approximated by funding obligations of about $1.1 billion, $899 million, and $946 million in fiscal years 2015, 2016 and 2017, respectively, for a total of about $3.0 billion over 3 years to support the system. Specifically, VHA and OI&T reported obligations to cover the costs for the VistA Evolution program, including costs for development, operation and maintenance, and payroll for government employees over the 3 fiscal years. Further, in their efforts to fully determine the costs associated with VistA, VA officials also reported obligations for activities that supported VistA, but were not included in the VistA Evolution program. These other obligations were for investments in interoperability initiatives, such as increasing data standardization and data sharing between VA, DOD, and other government and non-government entities, and the Virtual Lifetime Electronic Record Health. These obligations also include other VistA- related technology investments, such as networks and infrastructure sustainment, continuation of legacy systems, and overall patient safety, security, and system reliability. Table 1 provides a summary of the total VistA and related obligations that VA identified for fiscal years 2015 through 2017. VA Is Working to Define VistA’s Scope and Identify Components to Be Replaced by the Cerner System Understanding the scope of VA’s current health information system is essential to effectively planning for the new system. However, according to VA officials, there is no single information source that fully defines the scope of VistA. Instead, existing definitions of the system, including the components that comprise it, are identified by multiple sources. These sources include the VA Systems Inventory, VistA Document Library, and VA Monograph. Each of these sources describes VistA from a different perspective. For example, the VA Monograph provides an overview of VistA and non-VistA applications used by VHA. The monograph also describes modules and their associated business functions, but does not document all customization at local facilities. The VA Systems Inventory is a database that identifies current IT systems at VA, including systems and interfaces that are related to VistA. The VA Document Library is an online resource for accessing documentation on VA’s nationally released software applications, including VistA. In the absence of a complete definition of VistA, EHRM program officials have taken a number of steps to define the system’s scope and identify the components that the Cerner system will replace. These steps have included conducting two analyses, performing preliminary site assessments, and planning for Cerner to perform a detailed assessment of each site where the new system will be deployed. Specifically, EHRM program subject-matter experts undertook an analysis that identified 143 VistA modules and 35 software applications as representing the scope of the system. They then compared the functionality provided by the VistA modules to the Cerner system’s capabilities to identify the VistA components that are expected to be replaced by the Cerner system. The analysis identified 131 (92 percent) of the 143 VistA modules and 32 (91 percent) of the 35 applications that are expected to be replaced by the Cerner system. For example, the analysis determined that the Care Management and Mental Health modules would be replaced by the new system. EHRM program officials also undertook a subsequent, broader analysis to identify, among other things, the scope of VistA, as well as the department’s other health IT systems that could also be replaced by the Cerner system. These other systems include, for example, dentistry and oncology applications. As part of this analysis, the department combined data from the VA Systems Inventory, the VistA Document Library, the VA Monograph, and other sources to identify the health information technology environment at a typical VA medical center. The resulting analysis of VA’s health IT environment identified a total of 330 applications that support health care delivery at a medical center, of which 119 applications (approximately 36 percent) have been identified as having similar functionality as a capability of the Cerner system. Further, 128 of the 330 applications are identified as VistA applications. Of the 128 applications designated as VistA, 58 (approximately 45 percent) have been identified as having similar functionality as a capability of the Cerner system, including pharmacy, laboratory, and scheduling capabilities. In addition to the analyses discussed above, VA has taken steps to understand differences in VistA at individual facilities. Specifically, according to EHRM officials, representatives from VA and Cerner have visited 17 VA medical facilities to conduct preliminary site assessments. The intent of these assessments is to obtain a broad perspective of the current state of the systems, applications, integration points, reporting, and workflows being utilized at individual facilities. These site visits identified VistA customization that may be site specific. The identification of such site specific customization is intended to help Cerner plan for implementation of its system at each location. According to EHRM program officials, full site assessments that are planned at each location in preparation for implementation of the Cerner system are expected to identify the full extent of VistA customization. VA’s Preparations for Transitioning from VistA to the Cerner System Are Ongoing Since the former VA Secretary announced in June 2017 that the department would acquire the same electronic health record system as DOD, VA has taken steps to position the department for the transition to the new system. These actions, which are ongoing, have included standardizing VistA, assessing the department’s approach to increasing interoperability, establishing governance for the new program and the framework for joint governance with DOD, and preparing initial program plans. Standardizing VistA VA’s goal is for all instances of VistA being used in its medical facilities to be standardized where practical. Such standardization is intended to better position the department to switch to the Cerner system. To increase standardization, the VistA Evolution program has been focused over the last 5 years on standardizing a core set of VistA modules related to interoperability which, according to the department, accounts for about 60 percent of VistA. In addition, the program has focused on identifying software that is common to each VistA instance. VA refers to this collection of standard software as the gold instance. As part of its effort to standardize VistA, VA has implemented a process to compare the system at each site with the gold instance. Sites that are identified as having variations from the gold instance must apply for a waiver to gain approval for continuing to operate a non-standard VistA instance. OI&T and VHA assess the waivers, which may be approved if a site needs non-standard functionality that is deemed critical to that site. Alternatively, waivers are not approved if the assessment determines that a site’s needs can be met by reverting to the gold instance of VistA. Assessing the Approach to Increasing Interoperability VA has identified increased interoperability as a key expected outcome of its decision to switch from VistA to the Cerner system. To ensure that the contract with Cerner will improve interoperability with community care providers (i.e., non-VA and third party providers), the former VA Secretary announced in December 2017 that the department had taken a “strategic pause” on the electronic health record acquisition process. During the pause, an independent study was undertaken to assess the approach to interoperability with the new acquisition. The assessment made recommendations to improve imported data, address data rights and patient safety risks, and improve data access for patients. VA agreed with all of the resulting recommendations and, according to EHRM program officials, included provisions in the contract with Cerner to address the recommendations. Establishing a Program Office and Governance Our prior work has identified strong agency leadership support and governance as factors that can increase the likelihood of a program’s success. Such leadership and governance can come from the establishment of an effective program management organization and a related governance structure. VA has taken steps to establish a program management office and drafted a structure for technology, functional, and joint governance of the electronic health record implementation. Specifically, in January 2018, the former VA Secretary established the EHRM Program Executive Office (PEO) that reports directly to the VA Deputy Secretary. According to EHRM program officials, this office supported the contract negotiations with the Cerner Corporation and is expected to continue to manage the program going forward. Program officials stated that the office is beginning the process of hiring full-time employees. In addition, to support the program office, the department has awarded a contract for project management support and has also reassigned a number of VA staff to the PEO. Further, VA has drafted a memorandum that describes the role of governance bodies within VA, as well as governance intended to facilitate coordination between DOD and VA. For example, according to the draft memorandum, within VA, the EHRM Steering Committee is expected to provide strategic direction for the efforts while monitoring progresses toward goals and advising the Secretary on the progress and performance of the EHRM efforts. This committee is to include the Deputy Secretary, the Undersecretary for Health, and the Chief Information Officer, among others, and is to meet quarterly or as necessary to make its reports to the Secretary. Additionally, according to EHRM program documentation, VA is in the process of establishing a Functional Governance Board, a Technical Governance Board, and a Governance Integration Board comprised of program officials intended to provide guidance; coordinate with DOD, as appropriate; and inform the Steering Committee. Further, a joint governance structure between VA and DOD has been proposed that would be expected to leverage existing joint governance facilitated by the DOD/VA Interagency Program Office. Nevertheless, while the department’s plans for governance of the EHRM program provide a framework for high-level oversight for program decisions moving forward, EHRM officials have noted that the governance bodies will not be finalized until October 2018. Preparing Initial Program Plans Program planning is an activity for ensuring effective management of key aspects of an IT program. These key aspects include identification of the program’s scope, responsible organizations, costs, and schedules. VA has prepared initial program plans, including a preliminary timeline for deploying the new electronic health record system to its medical facilities. The department also has a proposed 90-day schedule that depicts key program activities currently underway now that the contract has been awarded. For example, the department’s preliminary plans include an 8- year deployment schedule beginning with planned implementation at initial sites within 18 months of October 1, 2018. According to the executive director for the EHRM program, the department also intends to complete a full suite of planning and acquisition management documents to guide the program. These documents include, for example, a life cycle cost estimate, a data migration plan, a change management plan, and an integrated master schedule to establish key milestones over the life of the project. EHRM PEO officials have stated that the department intends to complete the development of its initial plans for the program within 30 to 90 days of awarding the contract (between mid-June and mid-August 2018), and intends to update those plans as the program matures. The plans are to be reviewed during the milestone reviews identified in the department’s formal project management framework. Critical Factors Underlying Successful Major Acquisitions Our prior work has determined that successfully overcoming major IT acquisition challenges can best be achieved when critical success factors are applied. Specifically, we reported in 2011 on common factors critical to the success of IT acquisitions, based on seven agencies having each identified the acquisition that best achieved the agency’s respective cost, schedule, scope, and performance goals. These factors remain relevant today and can serve as a model of best practices that VA could apply to enhance the likelihood that the acquisition of a new electronic health record system will be successfully achieved. Among the agencies’ seven IT investments, agency officials identified nine factors as having been critical to the success of three or more of the seven investments. These nine critical success factors are consistent with leading industry practices for IT acquisition. The factors are: Active engagement of senior officials with stakeholders. Qualified and experienced program staff. Support of senior department and agency executives. Involvement of end users and stakeholders in the development of requirements. Participation of end users in testing system functionality prior to formal end user acceptance testing. Consistency and stability of government and contractor staff. Prioritization of requirements by program staff. Regular communication maintained between program officials and the prime contractor. Sufficient funding. Officials for all seven selected investments cited active engagement with program stakeholders—individuals or groups (including, in some cases, end users) with an interest in the success of the acquisition—as a critical factor to the success of those investments. Agency officials stated that stakeholders, among other things, reviewed contractor proposals during the procurement process, regularly attended program management office sponsored meetings, were working members of integrated project teams, and were notified of problems and concerns as soon as possible. In addition, officials from two investments noted that actively engaging with stakeholders created transparency and trust, and increased the support from the stakeholders. Additionally, officials for six of the seven selected investments indicated that the knowledge and skills of the program staff were critical to the success of the program. This included knowledge of acquisitions and procurement processes, monitoring of contracts, large-scale organizational transformation, Agile software development concepts, and areas of program management such as earned value management and technical monitoring. Finally, officials for five of the seven selected investments identified having the end users test and validate the system components prior to formal end user acceptance testing for deployment as critical to the success of their program. Similar to this factor, leading guidance recommends testing selected products and product components throughout the program life cycle. Testing of functionality by end users prior to acceptance demonstrates, earlier rather than later in the program life cycle, that the functionality will fulfill its intended use. If problems are found during this testing, programs are typically positioned to make changes that would be less costly and disruptive than ones made later in the life cycle. Use of the critical success factors described above can serve as a model of best practices for VA. Application of these acquisition best practices presents opportunities for the department to increase the likelihood that its planned acquisition of a new electronic health record system will meet its cost, schedule, scope, and performance goals. In conclusion, VA continued to obligate billions of dollars for its VistA system. Recently, the department has undertaken important analyses to better understand the scope of the system and identify capabilities that can be provided by the Cerner electronic health record system it is acquiring. VA has additional key activities underway, such as establishing program governance and EHRM program planning. Based on these preliminary observations and as the department continues its activities to transition from VistA to the Cerner electronic health record system, critical success factors can serve as a model of best practices that VA could apply to enhance the likelihood that the acquisition of the new system will be successfully achieved. While it is early in VA’s acquisition of the Cerner system, it will be important for the department to leverage all available opportunities to ensure that its transition to a new system is carried out in the most effective manner possible. Our experience has shown that challenges can successfully be overcome through using a disciplined approach to IT acquisition management. Chairman Roe, Ranking Member Walz, and Members of the Committee, this concludes my prepared statement. I would be pleased to respond to any questions that you may have. GAO Contact and Staff Acknowledgments If you or your staffs have any questions about this testimony, please contact David A. Powner at (202) 512-9286 or pownerd@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this testimony statement. GAO staff who made key contributions to this statement are Mark Bird (Assistant Director), Jennifer Stavros-Turner (Analyst in Charge), John Bailey, Rebecca Eyler, Jacqueline Mai, Scott Pettis, and Charles Youman. This is a work of the U.S. government and is not subject to copyright protection in the United States. The published product may be reproduced and distributed in its entirety without further permission from GAO. However, because this work may contain copyrighted images or other material, permission from the copyright holder may be necessary if you wish to reproduce this material separately.

Why GAO Did This Study VA provides health care services to almost 9 million veterans and their families and relies on its health information system—VistA—to do so. However, the system is more than 30 years old, is costly to maintain, and does not support interoperability with DOD and private health care providers. Since 2001, VA has pursued multiple efforts to modernize the system. In June 2017, VA announced plans to acquire the same system—the Cerner system—that DOD is implementing. GAO was asked to summarize preliminary observations from its ongoing review of VistA and the department's efforts to acquire a new system to replace VistA. Specifically, the statement summarizes preliminary observations regarding (1) costs incurred for the system and related activities during the last 3 fiscal years; (2) key components that comprise VistA and are to be replaced; and (3) actions VA has taken to prepare for its transition to the Cerner system. The statement also discusses common factors critical to the success of IT acquisitions that GAO has previously identified. GAO reviewed its prior reports on the VistA modernization and on critical success factors of major IT acquisitions. GAO also reviewed records of obligations for VistA for fiscal years 2015, 2016, and 2017; analyzed VA documentation that describes the scope of VistA, and reviewed program documentation. What GAO Found According to the Department of Veterans Affairs (VA), the Veterans Health Information Systems and Technology Architecture (VistA) and related costs, as approximated by funding obligations, were approximately $1.1 billion, $899 million, and $946 million in fiscal years 2015, 2016 and 2017, respectively. These obligations total about $3.0 billion over 3 years to support the system. As identified by the department, the obligations were to cover the costs for three programs (VistA Evolution, Interoperability, and Virtual Lifetime Electronic Record Health) and other supporting investments for activities such as networks and infrastructure sustainment. The following table provides a summary of the total VistA and VistA-related obligations. GAO's preliminary results indicate that VA is working to define VistA and identify system components to be replaced by the new system. However, according to VA officials, there is no single information source that fully defines the scope of VistA. This situation is partly due to differences in VistA at various facilities. In the absence of a complete definition of VistA, program officials have taken a number of steps to define the system's scope and identify the components that the new system will replace. These steps have included conducting analyses, performing preliminary site (medical facility) assessments, and planning for a detailed assessment of each site where the new system will be deployed. Since VA announced in June 2017 that the department would acquire the same electronic health record system as the Department of Defense (DOD), GAO's preliminary results indicate that VA has begun taking actions to prepare for the transition from VistA. These actions have included standardizing VistA, clarifying the department's approach to interoperability, establishing governance for the new program and the framework for joint governance with DOD, and preparing initial program plans. VA is early in its effort to transition from VistA to the Cerner system and the department's actions are ongoing. In 2011, GAO reported on nine common factors critical to the success of major IT acquisitions. Such factors include ensuring active engagement of senior officials with stakeholders and having qualified, experienced program staff. These critical success factors can serve as a model of best practices that VA could apply to enhance the likelihood that the acquisition of a new electronic health record system will be successfully achieved.

gao_GAO-18-429

gao_GAO-18-429_0

Background Public Housing and Housing Choice Voucher Programs HUD administers its public housing and HCV programs—which serve eligible low- and very-low income households, the elderly, and persons with disabilities—through local PHAs. PHAs are typically municipal, county, or state agencies created under state law to develop and manage public housing units for low-income families. PHAs that participate in the programs contract with HUD to provide housing in exchange for federal grants and subsidies. In total, there were 3,825 PHAs as of December 2017, according to HUD data. PHAs may administer both public housing and HCV programs. HUD’s public housing program provides low-rent housing units to over 1 million eligible households. According to HUD, the majority of PHAs (approximately 3,300 of the 3,825 PHAs) across the country take part in its public housing program. These PHAs own and manage public housing properties, which can include high-rise and low-rise buildings, scattered single family properties, or be part of mixed-income housing developments. Some PHAs manage public housing programs with less than 100 units and others manage programs with more than 30,000 units. For the public housing program, PHAs handle admissions, calculate rents, and enforce leases, among other duties. Under the terms of their contracts with HUD, PHAs agree to administer their properties according to federal statute and HUD regulations, and in exchange they receive funding from HUD. These statutes and regulations provide PHAs with more discretion in developing certain policies, such as parts of the admissions process, and less discretion in developing other policies, such as the income determination process. PHAs are required to develop plans that describe their policies through a process that allows for—and responds to—community feedback. Approximately 2,200 PHAs across the country are responsible for managing the day-to-day operations of the HCV program, including determining the eligibility of households, approving applications, and distributing vouchers. The HCV program subsidizes housing costs for approximately 2.2 million households in the private rental market as of March 2018, according to HUD officials. In the HCV program, participants are able to find their own housing within the PHA’s jurisdiction, including most single-family homes, townhouses, and apartments. If the household moves out of the unit, it can move with continued assistance to another private rental unit. PHAs are required to state their admissions policies within their administrative plans and make these plans publically available. HUD’s Office of Public and Indian Housing (PIH) is responsible for implementing HUD’s public housing and HCV programs, among others. Forty-five PIH field offices across the country are charged with overseeing PHAs’ compliance with HUD rules. Within PIH, the Office of the Deputy Assistant Secretary for Public Housing and Voucher Programs develops national policy, allocates funding, and provides program direction for public housing and HCV programs. The Office of the Deputy Assistant Secretary for Field Operations oversees the field offices. Figure 1 shows the organizational chart for selected HUD divisions with responsibilities related to public housing and HCV programs. Fugitive Felon Initiative HUD OIG operates independently within HUD and reports to the Office of the Secretary. The OIG conducts audits, evaluations, and investigations to detect and prevent fraud, waste, and abuse; and promotes effective and efficient government operations. The HUD OIG Office of Investigations conducts work through a headquarters office and seven regional offices. The Office of Investigations initiates investigations about possible violations of laws or regulations in the administration of HUD programs and activities, or misconduct on the part of HUD employees or recipients of HUD funds. The HUD OIG Office of Investigations began the Fugitive Felon Initiative in fiscal year 2003, in response to a request from USMS and one of our prior recommendations. The Initiative began as a data-sharing effort between HUD OIG and USMS to identify fugitives that may be living in HUD-assisted housing. According to HUD OIG officials, the data-sharing responsibilities were transferred from USMS to the FBI in 2004. This initiative has been governed by three memoranda of understanding (MOU). Specifically, a 2002 MOU between HUD OIG and USMS facilitated sharing USMS a 2004 MOU between HUD OIG and the FBI established a process to share a larger set of warrant data from federal, state, and local law enforcement agencies; and a 2012 MOU between HUD OIG and the FBI clarified the purposes of the Fugitive Felon Initiative and the roles of HUD OIG and the FBI. For the purposes of this report, we refer to the Fugitive Felon Initiative as the data-sharing effort between HUD OIG and the FBI to locate and apprehend fugitives. The term “Fugitive Felon Initiative,” however, is HUD OIG’s name for the program. The FBI participates in the data-sharing efforts with HUD OIG through the FBI’s Fugitive Identification Notice Delivery project. This project leverages FBI data-sharing with a small number of federal agencies, including HUD, to identify the possible location of fugitives. The Fugitive Felon Initiative is a law enforcement initiative, and it operates separately from PHA processes for conducting criminal history screenings to determine eligibility for housing assistance. Consistent with the 2002 and 2004 MOUs, the 2012 MOU states that the primary purpose of the Fugitive Felon Initiative is to apprehend fugitives and the secondary purpose is for HUD OIG to investigate, identify, and refer for prosecution the fraudulent receipt of HUD benefits. The Fugitive Felon Initiative includes HUD programs administered by PHAs—including public housing and HCV—as well as additional HUD programs not administered by PHAs. Through the Fugitive Felon Initiative, the HUD OIG leverages FBI and HUD data to identify potential investigative leads into the possible location of fugitives. The FBI shares with the HUD OIG nationwide data on felony and misdemeanor warrants from the FBI’s Wanted Persons File. The Wanted Persons File is included in the FBI’s National Crime Information Center (NCIC) database. The HUD OIG also accesses data from HUD’s Public and Indian Housing Information Center (PIC) system and Tenant Rental Assistance Certification System (TRACS). These systems maintain data on tenants who receive housing assistance. The HUD OIG then cross-references the FBI and HUD data to identify potential investigative leads based on possible matches between these data sources. As shown in figure 2, after the FBI receives a list of potential investigative leads from the HUD OIG, the FBI is to verify that warrants associated with the leads remain active because some warrants may have been resolved during the period of time the HUD OIG cross-referenced the FBI and HUD data. For example, a warrant may no longer be active if the individual associated with the warrant was already arrested or if the case involving the warrant was dismissed. For warrants that remain active, the FBI disseminates these investigative leads by sending “lead letters” to the federal, state, or local law enforcement agencies that entered the warrant into NCIC. These lead letters provide information, such as a possible address for an individual with the outstanding warrant. HUD OIG also disseminates potential investigative leads to its regional offices. HUD OIG regions may assist law enforcement in apprehending a fugitive or make referrals to PHAs to take administrative action against a tenant. This referral informs the PHA that one of its tenants may be a fugitive or has been apprehended. To ensure law enforcement agencies have sufficient time to apprehend wanted persons, the 2012 MOU states that HUD OIG regions must wait 60 days after law enforcement agencies have received the investigative leads before making referrals to PHAs. PHAs then have discretion about whether to take administrative action against the tenant to terminate assistance. Federal Statutes and HUD Regulations Include Requirements for PHAs’ Criminal History Policies but Some Also Provide Discretion Federal Statutes and HUD Regulations Require PHAs to Conduct Criminal Background Checks and Mandate Denial of Housing Assistance for Certain Offenses PHAs must follow federal statutes and HUD regulations (federal requirements) in determining eligibility for public housing and HCV assistance for persons with criminal history records. These requirements include the following: Conducting criminal background checks for program applicants. PHAs are required to conduct criminal background checks on all applicants to the public housing and HCV programs. PHAs must conduct these checks in the state where the housing is located and also check for criminal history records in other states where the applicant and members of the applicant’s household are known to have resided. When recertifying tenants, PHAs are not required to conduct criminal background checks. According to HUD officials, there are barriers to conducting background checks when recertifying tenants such as limited staff resources and cost constraints. Obtaining sufficient evidence of criminal activity. In November 2015, HUD’s Office of Public and Indian Housing issued a notice on the use of arrest records and other issues related to denying and terminating housing assistance for individuals who have engaged in criminal activity. The notice stated that the fact that an individual was arrested is not sufficient evidence that the individual engaged in criminal activity and informed PHAs that arrest records could not be used as the basis for denying admissions, terminating assistance, or evicting tenants. In a Frequently Asked Questions document pertaining to the notice, HUD advised PHAs to review their plans and revise their policies, as needed, to comply with the Notice. PHAs may use other forms of evidence such as conviction records, police records, or witness statements to determine whether the individual engaged in disqualifying criminal activity. The notice also reminded PHAs that their policies and procedures for screening applicants and eviction or termination of assistance must comply with the Fair Housing Act and the Civil Rights Act, and that inconsistent application of standards or decisions based on partial or inaccurate information (such as arrest record information) may result in liability under these laws. Establishing a process that allows applicants and tenants to dispute adverse information. PHAs must provide applicants and tenants with notification and the opportunity to dispute the accuracy and relevance of a criminal record before denying admission or terminating assistance on the basis of such a record. Denying or terminating assistance for certain types of criminal-related offenses. HUD regulations mandate that PHAs deny admission to the public housing and HCV programs for six types of offenses, two of which require lifetime bans on admissions. Specifically, PHAs must permanently ban admissions for individuals convicted of producing methamphetamine on the premises of federally assisted housing and individuals subject to a lifetime registration requirement under a state sex offender program. For the other four mandatory denials—which are related to illegal drug use, drug-related crime, and alcohol abuse—PHAs have some discretion to determine whether the offense applies to an applicant or household member or to consider mitigating circumstances. While six offenses require denial of admissions, only one of these offenses—the offense related to methamphetamine production—also mandates termination of assistance, as shown in table 1. In addition, federal statute and HUD regulations require that PHAs include certain offenses that are grounds for denial or termination in their policies, but give PHAs discretion on when and how to act on them. For example, PHAs can, but are not required to, terminate assistance for “fugitive felons.” Table 1 provides a summary of criminal history-related restrictions for the public housing and HCV programs. PHAs Generally Have Discretion in Establishing Their Criminal History Policies PHAs generally have discretion in establishing their specific criminal history policies, apart from the specific federal requirements discussed above. Below are examples of how policies vary among the 10 PHAs we reviewed. PHAs Have Discretion by Design In the 1990s, Congress enacted legislation to deregulate federal housing assistance programs, which gave public housing agencies broader discretion in establishing their own policies for tenant selection, income and rent, and administrative operations for the public housing and Housing Choice Voucher programs. This included discretion on policies for screening applicants, denying admissions, and terminating assistance. Denials and terminations. PHAs may choose to deny or terminate assistance for additional offenses that are not specifically listed in federal requirements. All of the PHAs we reviewed had established policies to deny admissions or terminate tenancy for additional offenses. For example, in addition to the mandatory denials, one PHA had a written policy to deny admission to public housing to applicants or household members convicted of arson or child molestation and persons who committed homicide, armed robbery, trafficking, or domestic violence in the past 3 years. Another PHA would deny housing assistance if an applicant, tenant, household member, or guest had ever committed homicide, kidnapping, rape or sexual assault, indecency with a child, or arson. According to selected PHA’s written policies, other offenses for which PHAs may deny admission or terminate tenancy include selling, producing, or manufacturing illegal substances; violent behavior; property destruction; and fraud, bribery, or other crimes in connection with a federally-assisted housing program. Lookback periods. PHAs can establish periods of time before the admission decision during which an applicant must not have engaged in certain types of criminal activity, such as drug-related or violent crimes, known as lookback periods. Based on our interviews with selected PHAs, lookback periods generally ranged from 2 years to 7 years but were sometimes longer for offenses such as homicide or assault. For example, one PHA had a policy to deny housing assistance to individuals who have committed manslaughter, robbery, illegal possession of a firearm or deadly weapon, assault, or physical violence to persons or property within a 5-year period. Another PHA had a 5-year lookback period for felony convictions for burglary; a 10-year lookback period for felony convictions for assault, kidnapping, abduction, forcible sex, or arson; and a 20-year lookback period for convictions for first degree murder, according to its screening criteria for the public housing and HCV programs. Some PHAs began the lookback period on the date of the conviction, and others on the date the offense occurred. Representatives of three PHAs we interviewed said that they had revised their policies in the past 5 years to reduce their lookback periods. For example, from 2013 through 2016, one PHA reduced its lookback period for all offenses first from 10 years to 7 years, and then finally to 3 years. The officials said their neighborhood had a high incarceration rate and they wanted to give second chances to ex-offenders. Officials from another PHA said that in 2016, they changed their lookback period from 10 years to 5 years at the suggestion of their new deputy director. Use of arrest records. According to HUD’s 2015 guidance, PHAs cannot rely on arrest records to determine eligibility for housing assistance. However, they may still review arrest records and may make an adverse housing decision based on the conduct underlying an arrest if the conduct indicates that the individual is not suitable for tenancy and the PHA has sufficient evidence (in addition to the arrest record) that the individual engaged in the conduct. Officials from 9 of the 10 PHAs said that they did not rely on arrest records to determine eligibility for assistance. Officials from the remaining PHA told us they have used arrest records as the basis for denying assistance for certain offenses and believed they complied with HUD’s notice on the use of arrest records by providing the applicant or tenant the right to appeal the denial or termination. Of the 9 PHAs that did not rely on arrest records for determining eligibility for assistance, officials at 5 PHAs indicated that they obtained and reviewed information on arrest records, but that they did not take action to deny assistance or terminate tenancy based on an arrest record. Officials at 1 PHA stated that they only took action based on conviction records and officials at another PHA stated they do not use arrest records at all in making eligibility determinations. For cases where an applicant has charges pending, officials at 2 PHAs said that they may wait for the case to be closed prior to making an eligibility determination. Consideration of mitigating circumstances and other factors. PHAs sometimes consider mitigating circumstances for applicants or tenants who may otherwise be denied housing assistance. Officials from PHAs we interviewed took different approaches to allowing mitigating circumstances and other factors. For example, officials from one PHA said that it always considered mitigating circumstances and requested such information as part of the application process. Officials at another PHA said that after a denial letter is sent, applicants can provide evidence of mitigating circumstances during the appeals process. Another PHA’s officials said that in making eligibility decisions, they considered the severity of the crime and whether the individual completed rehabilitation. As allowed by federal requirements, some PHAs included in their policies factors to consider when determining whether or not to deny or terminate housing assistance. For example, one PHA’s policy stated that in making such determinations it considers several factors such as the seriousness of the case and the effects that denying assistance may have on other household members or the community. Officials from another PHA said that they allowed public housing residents to preserve their tenancy on the condition that the offending household member is permanently excluded from the public housing unit. Selected PHAs’ Coordination with Local Law Enforcement Officials we interviewed at 10 public housing agencies (PHA) said they coordinated with local law enforcement as part of their efforts to address criminal activity in public housing. Two PHAs have their own police departments. Three PHAs said that local police officers patrol their public housing properties and inform the PHA if there are any issues related to criminal activity. Officials at one of the larger PHAs we interviewed said that staff check arrest reports every night to see if any crimes were committed by their tenants. Officials at another PHA said that they had off-duty police officers regularly patrol their public housing properties and had security cameras on their properties that are monitored by local police. Timing and Frequency of Background Checks. In addition to federal requirements to conduct criminal background checks at time of application, PHAs may also choose to conduct such checks as part of an annual recertification process for persons already receiving rental assistance. Officials from 3 of the 10 PHAs we interviewed said that they conducted background checks on tenants during the recertification process. In addition, officials from 3 other PHAs said that they may conduct background checks if issues arise during a person’s tenancy or at any time. Officials from the other 4 PHAs we interviewed did not provide additional details on conducting tenant background checks. Methods Used to Obtain Criminal History Information. Federal statute and HUD regulations authorize PHAs to obtain criminal history information from law enforcement agencies. HUD has also recognized that PHAs may obtain this information through other means. HUD officials at one regional office estimated that most of the PHAs under their purview use private companies to obtain criminal history information. Of the 10 PHAs we interviewed, 6 said that they hired private screening companies to provide the PHA with a criminal history report for an applicant or tenant. Criminal History Records and Data Quality Challenges The completeness and accuracy of criminal history information is a known and persistent challenge for state and federal agencies and private companies that compile and sell this information to entities such as employers and public housing agencies. In its 2015 notice on the use of arrest records, the Department of Housing and Urban Development (HUD) affirmed its commitment to the goal of ensuring that individuals are not denied access to HUD-subsidized housing on the basis of inaccurate, incomplete, or otherwise unreliable evidence of criminal conduct. In addition, the Federal Interagency Reentry Council, of which HUD is a participating agency, reported that it plans to take steps to address widespread inaccuracies in criminal records, and that it would work with consumer reporting agencies to develop best practices for improving the accuracy of criminal records. Officials from one of the selected PHAs we interviewed said that the housing authority’s police department conducted the criminal background check and determined whether to approve or deny the applicant based on the results, consulting with the PHA if needed. Officials at two PHAs said a local law enforcement agency or state agency did the initial criminal background check to determine if the applicant has a criminal record, and if so, a private screening company may obtain the individual’s detailed criminal record. Another PHA said that their staff used state databases to conduct criminal background checks. HUD Has Not Yet Updated Its Guidebooks for PHAs with Newer Criminal History Policy Guidance, and Its Compliance Reviews Do Not Address Some Requirements HUD Has Not Yet Updated Its Guidebooks to Reflect New Criminal History Policy Guidance As of mid-May 2018, HUD officials stated that they were in the process of updating HUD’s HCV Program Guidebook and Public Housing Occupancy Guidebook (guidebooks), including updating sections of these guidebooks with new criminal history policies. However, the documentation HUD provided on these updates did not specifically address criminal history policies. The guidebooks serve as key reference documents and are designed to advise PHAs on the administration of the HCV and public housing programs, but have not been revised since 2001 and 2003, respectively. From 2011 through 2016, HUD issued notices and other documents that urged PHAs to move away from policies that deny admissions or tenancy to anyone who has engaged in criminal activity, and instead to seek policies that strike a balance between resident safety and the reentry needs of formerly incarcerated individuals and others with criminal history records. In 2011, the Secretary of HUD issued a letter to PHAs encouraging them to allow ex-offenders to rejoin their families in the public housing or HCV programs when appropriate. The letter reminded PHAs that they have broad discretion to set admission and termination policies for the public housing and HCV programs, aside from the federal requirements. The letter also reminded PHAs that they have discretion to consider other factors such as evidence of rehabilitation or participation in social service programs when screening applicants for suitability. HUD issued a notice in 2012 recommending that PHAs terminate the tenancy of persons living in federally assisted housing who were erroneously admitted while subject to a state lifetime sex offender registration requirement or who commit sex offenses while living in federally assisted housing. HUD recommended that PHAs ask at the time of annual recertification whether any member of the household is subject to a state lifetime sex offender registration program in any state. If the PHA finds that a member of the household engages in criminal activity, including sex offenses, while living in HUD-assisted housing, the PHA should pursue termination of tenancy, according to the notice. As previously discussed, HUD issued a notice on criminal history policies and the use of arrest records in 2015, stating that the fact that an individual was arrested is not sufficient evidence that the individual engaged in criminal activity. In addition, the notice stated that HUD does not require the adoption of “one strike” policies (for example, policies that deny admissions or tenancy to anyone who has engaged in criminal activity), and that in most cases PHAs have discretion to determine whether to deny admission or terminate assistance to applicants or households with criminal history records. In 2016, the HUD Office of General Counsel issued a document indicating that policies that exclude individuals based on arrests do not satisfy the Fair Housing Act’s burden of proof. The document further stated that housing providers should consider factors such as the type of crime and the length of time since conviction when making housing decisions based on criminal history records. As of mid-May 2018, HCV and public housing guidebooks were outdated because they did not reflect the letters and notices cited above. HUD has not updated the guidebooks in more than 15 years because they do not frequently update these documents. For example, according to HUD’s website, the 2003 Public Housing Occupancy Guidebook is the first update in over 20 years. We reported previously that HUD had struggled to maintain up-to-date and complete policies and procedures across its management functions. In March 2018, HUD officials told us they had begun the process of updating their HCV and public housing guidebooks, noting that PHAs have requested such an update. HUD officials said the eligibility chapters of the updated guidebooks will reflect the notices that HUD has provided to PHAs in recent years on criminal history policies. However, we requested documentation on HUD’s planned updates and the information we received did not clearly indicate that the new criminal history guidance would be incorporated into the guidebooks. Federal internal control standards state that management should communicate the necessary quality information to achieve the entity’s objectives. This can include ensuring appropriate means of communicating with external parties. Effective communications can take many forms, including guidance. By updating its HCV and public housing guidebooks to reflect newer criminal history guidance, HUD can ensure that these guidebooks serve as consolidated and up-to-date references for PHAs that accurately communicate HUD’s current guidance on criminal history policies. HUD’s Reviews for Some High-Risk PHAs Do Not Comprehensively Address Criminal History Policies HUD reviews the criminal history policies for the small number of PHAs it designates as high risk or very-high risk, but these reviews do not address all related federal requirements or their implementation. Using its National Risk Assessment, HUD designates each PHA on a quarterly basis as low, moderate, high, or very-high risk. The assessment uses quantitative and qualitative data sources to identify, mitigate, prevent, and anticipate potential risk in five categories: financial, physical, governance, management risks, and risks to the HCV program. This assessment does not include specific metrics related to PHAs’ criminal history policies, according to HUD officials. HUD uses the results to direct field staff resources towards higher-risk PHAs, such as providing these PHAs with technical assistance or conducting compliance reviews. HUD Field Staff May Have Cause to Review PHAs’ Criminal History Policies for Various Reasons Outside of the Department of Housing and Urban Development’s (HUD) Compliance Monitoring Checklist, HUD field staff may have cause to review a PHA’s criminal history policies for other reasons. Those reasons include complaints from applicants who were denied assistance for criminal history reasons or low occupancy rates, which could indicate that people do not want to live in particular public housing complexes for safety reasons, or that a PHA’s screening policies may be too stringent. HUD may also review a PHA’s criminal history policies through the annual plan submission process. According to HUD officials, about one-third of PHAs are required to submit annual plans, which describe PHAs’ policies governing resident or tenant eligibility, and selection and admission, among other policies. Although HUD does not routinely monitor PHAs’ compliance with federal requirements on criminal history policies, it does evaluate some aspects of compliance for those high-risk and very-high-risk PHAs that receive a compliance review. To conduct these compliance reviews, field staff use HUD’s Compliance Monitoring Checklist (checklist). The checklist, which was first piloted in 2016 among six PHAs, contains six questions field staff must cover that directly relate to PHAs’ criminal history policies. These include questions on the PHA’s policies for denying applicants for drug-related criminal activity and checking states’ sex offender registry lists. In 2017, HUD expanded the use of the checklist to 74 high-risk and very-high-risk PHAs. For 2018 reviews, HUD officials stated that each of HUD’s 45 field offices will be required to use the checklist for at least one high-risk PHA in their portfolio, meaning the checklist will be applied to at least 45 PHAs out of 626 PHAs designated as high risk and very-high risk (out of a total of 3,825 PHAs as of December 2017), according to HUD officials. HUD field offices can choose to use the checklist at more than one PHA, according to HUD officials. Prior to this checklist, HUD officials said HUD field staff collected information on PHAs’ criminal history policies through HUD’s Rental Integrity Monitoring reviews by which HUD field office staff collect and analyze PHA income and rent information, identify income and rent errors, and assess PHA policies and procedures in both the public housing and HCV programs. However, HUD no longer required these reviews after 2006, though field staff may still conduct them, according to HUD officials. As shown in table 2, the checklist generally directs field staff to obtain a copy of a PHA’s written policies related to criminal history. For two of the six questions, field staff are also directed to review supporting materials and interview PHA staff, but for the other four questions, no additional information must be obtained. According to HUD officials, field staff who conduct the reviews are experienced and know to obtain additional information even if it is not listed in the checklist guidance. Officials stated that the checklist was not intended to be a step-by-step guide. HUD’s checklist does not include items to assess PHAs’ compliance with additional aspects of PHAs’ criminal history policies. As shown in table 2, the checklist includes specific items related to federal requirements on drug-related criminal activity, sex offenders, and convictions for methamphetamine production for which PHAs are required to deny admissions for public housing and HCV programs. The checklist, however, does not cover the requirement related to the abuse of alcohol. In addition, HUD’s checklist also does not address the requirement that PHAs may not use arrest records as the basis for denying or terminating assistance. Officials from 8 of the 10 PHAs we interviewed stated that they were already implementing policies or changed their policies to follow HUD’s notice on arrest records. However, we found that 1 of the 2 remaining PHAs we interviewed had not yet updated its written policies, though officials at this PHA said they did not base any decisions on arrest records in practice. The other PHA’s policies state that a record of arrest(s) will not be used as the basis for the denial or proof that the applicant engaged in disqualifying criminal activity, but officials from this PHA said that they did use arrest records as the basis for denying assistance to persons. Specifically, PHA officials stated that they based assistance decisions on records of arrest for drug-related or violent activity if the arrest had not been dismissed, had not reached disposition, and occurred within the last 5 years. Officials from this PHA said that they comply with HUD’s 2015 notice by providing the applicant the right to appeal a denial or termination (officials said that appeals by applicants are rare). HUD’s checklist instructions direct Office of Public and Indian Housing (PIH) field staff to note regulatory violations that they observe when conducting compliance reviews using the checklist. However, officials in HUD headquarters stated that they could not provide information on any regulatory violations related to PHAs’ criminal history policies specifically because they have aggregate results from the 2017 checklist reviews, which do not specify the type of compliance issues identified by field staff. As a result, violations related to criminal history policies would be included under the general categories of PHA’s Admissions and Continued Occupancy Policies (for public housing) or Administrative Plans (for HCV). In addition, none of the HUD staff we interviewed from July through December 2017 from five of HUD’s field offices discussed any instances of noncompliance specifically related to PHAs’ criminal history policies. Field staff we interviewed identified a range of potential actions they might take if they found that a PHA’s criminal history policies did not meet HUD’s requirements. These actions could include providing technical assistance to the PHA, requiring the PHA to make corrective actions within a specified time frame, or requiring the PHA to rescreen applicants. HUD is required by law to assess the performance of PHAs in all major areas of management operations, including implementing effective screening and eviction policies and other anticrime strategies. In addition, federal internal control standards indicate that management should design control activities to achieve objectives and respond to risks. However, HUD’s checklist does not address PHAs’ criminal history policies in a comprehensive manner. For example, it generally does not require field staff to go beyond reviewing written policies and obtaining additional information on how the policies are being implemented. In addition, field staff are not required to address some federal requirements, such as PHAs’ use of arrest records. According to agency officials, HUD issued the arrest record notice in response to information indicating that PHAs were basing denial decisions on whether an individual had been arrested, which is not sufficient evidence of criminal activity. In our interviews of 10 selected PHAs, as discussed above, officials from one PHA described practices that were not in line with its written policy on the use of arrest records. Specifically, the officials stated that they make housing assistance decisions based on arrest records though their policies state they will not. Another PHA had not updated its written policy to reflect its practice of not basing decisions on arrest records. HUD officials stated that, due to resource issues, they developed the checklist to address high-risk areas, but that they planned to review the checklist again after the guidebooks are updated. By reviewing the checklist to determine what additional criminal history policy requirements should be included and revising the checklist instructions to direct staff to obtain information on PHAs’ implementation of criminal history policy requirements, HUD could improve its ability to identify areas of noncompliance. Noncompliance, according to HUD’s public housing guidebook, could lead to admission of ineligible families or unlawful discrimination. The Fugitive Felon Initiative Has Led to Apprehensions, but Its Implementation and Program Oversight Have Been Inconsistent The HUD OIG Identified and Shared Potential Leads on Locations of Fugitives with the FBI Through the Fugitive Felon Initiative, the HUD OIG and the FBI have shared information that has produced thousands of potential investigative leads on the location of fugitives who may live in HUD-assisted housing. From May through June 2017, the HUD OIG identified approximately 18,000 potential investigative leads using FBI warrant data from September 2016, according to HUD OIG officials and FBI data. The HUD OIG identified these leads by cross-referencing the approximately 2.4 million felony and misdemeanor warrants in the FBI’s Wanted Persons File with the approximately 10.6 million records in HUD’s PIC and TRACS data systems. Cross-referencing involves identifying corresponding records within the FBI and HUD data that show the same or similar names, the same date of birth, and the same sex. A HUD OIG official stated that this process is designed to be overly inclusive to minimize the risk of missing a potential investigative lead. In addition, because the Fugitive Felon Initiative uses data from HUD tenant files, fugitives who live in HUD-assisted housing but are not listed on the rental agreement would not be identified through this process, according to HUD OIG officials. As part of its activities under the Fugitive Felon Initiative, after cross- referencing the FBI and HUD data, the HUD OIG distributed potential investigative leads to HUD OIG regional offices and the FBI. According to HUD OIG officials, the list of potential investigative leads they sent to HUD OIG regional offices only included extraditable warrants for felony offenses. The FBI did not verify these potential investigative leads to determine if the warrants remained active. The list of potential investigative leads the HUD OIG sent to its regional offices differed from the list of leads the FBI distributed to law enforcement agencies. Specifically, the investigative leads the FBI distributed to law enforcement agencies contained only extraditable warrants for both felony and misdemeanor offenses that the FBI verified remained active, according to FBI officials. According to our analysis of HUD OIG data, many of the potential investigative leads the HUD OIG sent to its regional offices involved nonviolent offenses, though a small percentage included crimes such as assault or homicide. Specifically, from May through June 2017, the HUD OIG sent 4,814 potential investigative leads (about 27 percent of the approximately 18,000 potential investigative leads) to its regional offices. As shown in table 3, about one-third of these leads were for failure to appear in court or probation violations—the two most frequently occurring offenses. According to FBI officials, once they electronically receive the list of potential investigative leads from the HUD OIG, their system automatically removes potential leads when either (1) the warrant associated with the lead is no longer active or (2) the warrant associated with the lead is not extraditable. A warrant would no longer be active if an arrest or other warrant resolution occurred between the time the FBI sent the Wanted Persons File to the HUD OIG and the time the HUD OIG returned the list of potential investigative leads to the FBI. An investigative lead would not be extraditable if the fugitive’s address fell outside of the geographic extradition area. According to HUD OIG officials, the HUD OIG sent the FBI approximately 18,000 potential investigative leads in 2017. FBI data show that the warrants associated with 9,415 of these leads remained active once the FBI received the leads. Of the potential leads with active warrants, FBI data show that 4,957 of the warrants were extraditable and active. According to FBI officials, they sent lead letters— which notify law enforcement agencies of the possible location of a fugitive who may be receiving HUD assistance—for the leads associated with the extraditable warrants that remained active to the relevant law enforcement agency. Lead letters include information from HUD tenant data and the associated warrant, such as name, date of birth, Social Security number, warrant number, date of the lead, and a possible address for the individual. The FBI’s Investigative Lead Letters Facilitated Apprehensions The FBI’s investigative lead letters have led to over 1,200 fugitive apprehensions from fiscal years 2013 through 2017 as a result of the Fugitive Felon Initiative. FBI data show that the FBI sent lead letters to law enforcement agencies for active, extraditable warrants each time the FBI received a list of potential investigative leads from the HUD OIG from fiscal years 2013 through 2017. From fiscal years 2013 through 2017, the FBI sent approximately 45,100 lead letters to law enforcement agencies for extraditable warrants that remained active (out of approximately 66,000 total potential investigative leads FBI data show it received from the HUD OIG during this time period, which included extraditable and nonextraditable active warrants). Law enforcement agencies provide information to the FBI on the disposition of most warrants associated with lead letters. According to FBI officials, when the FBI provides a lead letter to law enforcement agencies, it includes an optional questionnaire on the disposition of the warrant. Law enforcement agencies return the questionnaire about 75 percent of the time, according to FBI data. Data from these questionnaires show that law enforcement agencies reported 1,260 fugitive apprehensions that were facilitated by information from the Fugitive Felon Initiative from fiscal years 2013 through 2017. Lead letters do not always result in apprehensions. For example, law enforcement agencies may have resolved the outstanding warrant through a separate investigation, been unable to locate the subject of the warrant, or decided to not extradite a subject located in another state, according to FBI data. There may also be additional apprehensions that occurred without the FBI’s knowledge if the law enforcement agency apprehended an individual but did not return the disposition questionnaire to the FBI. From fiscal years 2013 through 2016, law enforcement agencies reported numbers of apprehensions resulting from the Fugitive Felon Initiative ranging from 254 to 339 each year (see table 4). However, in fiscal year 2017, law enforcement agencies reported a substantial decrease in apprehensions to 77. FBI officials stated that this decrease was a direct result of the decrease in the frequency and speed with which the HUD OIG cross-referenced HUD and FBI data and provided potential leads to the FBI. Specifically, the HUD OIG did not cross-reference data for over a year during fiscal years 2016 and 2017, and the HUD OIG did not return the results to the FBI for 10 months after receiving warrant data from the FBI in September 2016, which resulted in many warrants no longer being active, according to FBI officials. HUD OIG officials stated that the lag in returning potential investigative leads to the FBI in July 2017 was due to staff turnover. HUD OIG officials stated they are developing a process so that staff turnover will not prevent the HUD OIG from cross-referencing the data in the future. Of the 77 apprehensions in fiscal year 2017 based on the HUD OIG’s potential investigative leads, our analyses showed that many were for nonviolent offenses. Specifically, about 57 percent were for failure to appear in court or probation violations. The next most frequent offenses included larceny, fraud, dangerous drugs, harassing communication, parole violation, and contempt of court. These offenses made up approximately 25 percent of all apprehensions. The HUD OIG Has Revised Its SOP to Address Inconsistent Regional Office Participation in the Fugitive Felon Initiative Participation in the Fugitive Felon Initiative among the HUD OIG’s regional offices was inconsistent and declined from fiscal years 2012 through 2016. In April 2018, the HUD OIG revised its Standard Operating Procedure (SOP) for the Fugitive Felon Initiative to define regional office responsibilities, improve consistency among regional offices’ participation, and leverage the FBI’s efforts. Inconsistent Participation The inconsistent participation of HUD OIG regional offices in the Fugitive Felon Initiative resulted from changes in HUD OIG investigative priorities, inconsistent data-sharing from HUD OIG headquarters, and resource constraints: Change in agency priorities. According to HUD OIG officials, beginning in 2012, the HUD Inspector General prioritized investigations that would have significant financial effects, such as fraud committed by PHA employees. Subsequently, four of the seven HUD OIG regional offices did not participate in the Fugitive Felon Initiative from 2012 through 2016, according to regional officials, while the other three regional offices participated by following-up on at least some of the potential investigative leads. In addition, most HUD OIG regional offices stopped participating in USMS fugitive task forces after 2012. Specifically, officials in six of the seven regional offices said that before 2012, they coordinated with or participated in USMS task forces to investigate potential leads they received from HUD OIG headquarters. An agent in one regional office who participated as a member on the USMS Regional Fugitive Task Force said that he gathered additional information on fugitives from law enforcement and assisted in the apprehension of fugitives. According to officials in that region, they stopped participating in the task force in 2012. Similarly, officials in four other regional offices that coordinated with or participated in USMS fugitive task forces either did not receive the data from HUD OIG headquarters after 2012 or stated that they discontinued their formal involvement in the USMS task forces around 2012. Officials in the sixth regional office stated that they continue to interact with the USMS fugitive task force. Officials in the seventh regional office reported not working with USMS on fugitive apprehensions. Officials we interviewed from three USMS fugitive task forces confirmed their prior interaction with three HUD OIG regions. According to HUD OIG and USMS officials, the three HUD OIG regional offices stopped working with the USMS fugitive task forces in 2005, 2012, and 2015, respectively. For example, officials from one task force stated that a HUD OIG agent was detailed to the task force until 2015 and provided them with related HUD information to locate potential fugitives. Inconsistent data-sharing. HUD OIG headquarters did not consistently share potential investigative leads with all regional offices after 2012, which affected their participation in the Fugitive Felon Initiative. HUD OIG headquarters did not track when it shared potential investigative leads with its regional offices, but our interviews indicate that regional offices did not consistently receive leads from HUD OIG headquarters. Officials from three regional offices stated that they continued to receive data on the potential leads from headquarters from 2012 through 2016, one received data on the potential leads from 2012 through 2014, one received the data upon request from 2012 through 2015, and two did not receive the data after 2012. Of the three regional offices that received the potential leads from 2012 through 2016, officials from two offices stated that they conducted further investigations or coordinated with law enforcement to pursue apprehensions of fugitives on at least some of the potential leads. Resource constraints. Resource constraints limited HUD OIG regional office participation in the Fugitive Felon Initiative, according to officials from six of the seven regional offices. Officials from two of these regions stated that their staff levels have been reduced in recent years, limiting the resources available to address the hundreds of potential investigative leads from HUD OIG headquarters. They noted that following up on each lead was time-consuming, requiring agents to reenter warrant information into NCIC, identify the law enforcement agency point of contact, and call the agency to provide the potential location of the wanted person. Officials from four regions that continued to receive the potential investigative leads after 2012 stated that they investigated a subset of leads, such as leads for violent offenses. Officials from another region that continued to receive the leads after 2012 stated they did not follow up on any of the leads they received due to work constraints. Revised Standard Operating Procedure In April 2018, the HUD OIG revised its SOP and added guidance for regional office participation in the Fugitive Felon Initiative. The prior version of the SOP (issued in 2016) did not specifically define regional activities. The 2018 SOP states that regional offices will be responsible for verifying that the warrant associated with the potential investigative lead is still active and coordinating with the law enforcement agency that originally entered the warrant into NCIC. In addition, regional offices will generally be required to conduct additional research by querying criminal databases, referring leads to PHAs for administrative action, and recording their efforts in the HUD OIG case management system. The 2018 SOP states that based on resource and staffing levels, HUD OIG regions may limit their participation in the Fugitive Felon Initiative to only “priority” leads. According to HUD OIG headquarters officials, regional offices are to follow up on priority leads by undertaking activities listed in the 2018 SOP such as coordinating with law enforcement agencies and referring leads to PHAs for administrative action. The Prioritized Fugitive Felon List is defined as leads associated with warrants for violent felonies, sexual assault, and narcotics distribution, as well as other offenses that may affect the health and safety of housing residents, children, national security, or law enforcement. The 2018 SOP also details a new process in which HUD OIG headquarters will provide regional offices with (1) the priority list of leads and (2) the nonpriority list of leads, which includes all leads associated with extraditable felony warrants not included in the priority list. The 2018 SOP also states that the HUD OIG will cross-reference FBI and HUD data twice each year and return the list of potential investigative leads to the FBI before sending it to HUD OIG regional offices. As a new step under the 2018 SOP, the FBI will verify whether each warrant on the list is active before sending the list back to the HUD OIG, which according to HUD OIG officials, is intended to reduce the number of leads with inactive warrants provided to regional offices. Because the HUD OIG only recently issued the new SOP, it is too early to assess its effectiveness in enhancing regional office participation in the Fugitive Felon Initiative. The HUD OIG Has Reporting Requirements for the Fugitive Felon Initiative, but Does Not Plan to Collect or Assess Data on Some Key Aspects of Regional Implementation The 2018 SOP includes some added requirements for HUD OIG headquarters to track and report some statistics related to its regional offices’ activities, but the HUD OIG does not plan to collect or assess data on some activities listed in the 2018 SOP that HUD OIG officials stated regional offices are required to undertake for the Prioritized Fugitive Felon List. Under the 2018 SOP, HUD OIG headquarters will be responsible for tracking and reporting statistics on the number of referrals, evictions, PHA actions, and positive matches. However, the 2018 SOP does not require the HUD OIG to track the extent to which its regional offices undertake all the activities that HUD OIG officials stated regions are required to undertake, such as contacting and coordinating with relevant law enforcement agencies for the leads on the Prioritized Fugitive Felon List. The HUD OIG’s 2018 SOP states that the development and use of the SOP is integral to a successful quality control system and that it provides pertinent information needed to perform a required task properly by facilitating consistency. Federal internal control standards state that management should establish activities to monitor the internal control system and evaluate results. HUD OIG headquarters officials stated that they do not plan to collect or assess information on the extent to which regional offices are implementing the new SOP because collecting such information would be resource intensive. However, we believe the HUD OIG could obtain more comprehensive information on its regional offices’ activities using current resources. For example, the 2018 SOP states that HUD OIG regions are to create a subject profile in the case management system on all confirmed hits. This indicates that the regions will track their efforts to implement the new SOP. As a result, HUD OIG headquarters could collect and assess this information on the extent to which regions are implementing the new SOP through periodic data calls to its regional offices. Collecting and assessing more comprehensive information would better enable the HUD OIG to (1) determine the extent to which HUD OIG regions are undertaking activities listed in the 2018 SOP, including activities agency officials stated regions are required to undertake for the leads on the “Prioritized Fugitive Felon List” and (2) identify any areas for improvement. Such assessments of regional office efforts would also inform HUD OIG headquarters of whether the new SOP is being implemented as intended and consistently, which is particularly important given the regions’ inconsistent participation in the initiative in the past. The HUD OIG and the FBI Have Not Consistently Shared Results of the Initiative The HUD OIG and the FBI have not consistently shared information on the results of the Fugitive Felon Initiative or agreed on the type of information that would be the most useful to share. The 2012 MOU for the initiative states that (1) the FBI is to provide apprehension and other fugitive felon statistics to the HUD OIG monthly and (2) the HUD OIG is to provide apprehension information and estimated program savings to the FBI every 30 days. FBI Apprehension Data. Prior to 2012, the FBI shared aggregate data on apprehensions that resulted from its lead letter process with the HUD OIG, but stopped doing this at the request of the HUD OIG, according to FBI officials. During our review and at the request of the HUD OIG, the FBI resumed sharing information on apprehensions with the HUD OIG in November 2017. However, rather than providing aggregate apprehension statistics, the FBI provided individual disposition letters to the HUD OIG on a weekly basis. While the disposition letters contain information on apprehensions, HUD OIG officials stated that aggregate statistics would better assist them in judging the effectiveness of the initiative. HUD OIG Apprehension Data. HUD OIG headquarters has not tracked the numbers of apprehensions of wanted persons under the initiative and therefore has not shared this information with the FBI. HUD OIG officials stated that it is not feasible for them to collect and share this information with the FBI every month. The HUD OIG’s April 2018 SOP also states that the HUD OIG will no longer share information on apprehensions with the FBI. As of April 2018, FBI officials said that they were not aware of any changes to the HUD OIG’s responsibilities for sharing apprehension information under the 2012 MOU. HUD OIG Program Savings Data. FBI data show that HUD OIG headquarters has not shared program savings data with the FBI since 2012. HUD OIG headquarters officials stated that they do not currently have a method for estimating program savings under the Fugitive Felon Initiative although they have calculated program savings in the past. FBI data show that the HUD OIG provided an estimate of program savings in 2012. The HUD OIG and the FBI have not agreed on whether sharing information on program savings would be useful in implementing the initiative. The 2012 MOU also does not specify for what purpose the HUD OIG should share information on program savings with the FBI. FBI officials stated, however, that if they received data on apprehensions and program savings in the future, they would use this information to report to FBI management to show the ongoing results from the initiative as well as benefits for law enforcement. In our prior work, we found that collaborating agencies should develop mechanisms to monitor, evaluate, and report results. Reporting on these activities can help the agencies identify areas for improvement such as policy and operational effectiveness. In the 2012 MOU, the HUD OIG and the FBI documented the information they would share on results; however, they have not consistently shared this information, according to HUD OIG and FBI officials. In addition, in its 2018 SOP, the HUD OIG stated that it would no longer collect or share data on apprehensions or program savings with the FBI, but this change is not reflected in the current MOU. By agreeing on what information on results would be useful to share, and consistently sharing this information, the HUD OIG and the FBI could enhance their ability to identify areas for improvement and evaluate the effectiveness of the initiative. The HUD OIG and the FBI Have Not Updated Their 2012 MOU to Reflect Program Changes In addition to not consistently sharing information on results, we found several other areas where the 2012 MOU between the HUD OIG and the FBI does not align with current processes for implementing the Fugitive Felon Initiative. The MOU also does not reflect changes made by HUD OIG’s April 2018 SOP, and the HUD OIG generally had not discussed these changes with the FBI. Prosecution for Fraud. According to HUD OIG officials, the HUD OIG generally does not pursue tenant fraud cases as part of the Fugitive Felon Initiative, although the MOU lists this as one of the purposes of the initiative. Specifically, the MOU states that in addition to apprehending fugitive felons, the secondary purpose of the initiative is to investigate, identify, and refer for prosecution individuals who fraudulently receive HUD benefits. However, according to HUD OIG headquarters and officials from one regional office, the HUD OIG generally does not pursue federal tenant fraud cases because these cases typically do not meet the dollar threshold for federal prosecution. New Data-Sharing Process. The HUD OIG’s 2018 SOP includes a new procedure in which the FBI will return verified investigative leads to the HUD OIG, but the MOU does not include this new responsibility for the FBI. As discussed earlier, the FBI will now be responsible for verifying whether each warrant on the list of potential investigative leads is active and then sending a list of investigative leads with active warrants to the HUD OIG for distribution to its regional offices. According to FBI officials, they have discussed this added step with the HUD OIG and are currently developing the capability to implement it. HUD OIG Referrals to PHAs. The MOU states that HUD OIG regional offices should not refer cases to PHAs for administrative action for 60 days after the FBI sends the lead letter to law enforcement. However, the MOU does not specify how HUD OIG regional offices will be notified about when the 60-day period begins. In addition, officials from HUD OIG regional offices had differing understandings of when this 60-day period begins, and officials from one region stated that they had only recently become aware that there was a 60-day waiting period. Further, the current MOU does not reflect new language in the HUD OIG’s 2018 SOP that allows HUD OIG regional offices to make referrals to PHAs if the subject of the warrant is on the Prioritized Fugitive Felon List and is apprehended before the 60-day period expires. HUD OIG Interaction with Law Enforcement. The HUD OIG and the FBI have not updated the 2012 MOU to reflect that, under the 2018 SOP, HUD OIG regional offices are now generally tasked with proactively contacting and coordinating with law enforcement. Further, according to HUD OIG officials, regional offices are required to proactively contact and coordinate with law enforcement for persons on the Prioritized Fugitive Felon list. However, the MOU only states that the HUD OIG will encourage law enforcement agencies to contact the HUD OIG’s regional Special Agents in Charge for assistance with fugitive apprehension activities. Our prior work has found that agencies that articulate their agreements in formal documents can strengthen their commitment to working collaboratively, and written agreements are most effective when they are regularly updated. The HUD OIG and the FBI articulated their agreement for the Fugitive Felon Initiative in the 2012 MOU, but the MOU has not been updated to reflect either of the agencies’ current implementation of the initiative or the HUD OIG’s updated April 2018 SOP, according to HUD OIG and FBI officials. As discussed previously, the HUD OIG’s April 2018 SOP includes program changes that affect the activities listed in the 2012 MOU, but according to HUD OIG officials, they have only raised some tentative changes with the FBI. According to FBI officials, as of April 2018 HUD OIG officials mentioned that they are interested in updating the MOU, but the HUD OIG has not discussed any specific changes with the FBI and has not made a formal request to update the MOU. HUD OIG officials stated that they are waiting to process the findings of this GAO report before finalizing program changes with the FBI. Jointly agreeing to any changes in HUD OIG and FBI responsibilities under the Fugitive Felon Initiative and updating the MOU to reflect these changes could improve collaboration between the HUD OIG and the FBI and improve implementation of the initiative. Conclusions Criminal history policies for federally assisted housing and the Fugitive Felon Initiative help ensure the safety of residents receiving rental assistance. In the past decade, HUD issued notices and other documents urging PHAs to strike a balance between resident safety and the reentry needs of individuals with criminal history records. By completing its planned updates of program guidebooks to reflect this guidance, HUD could help ensure that PHA staff know and follow HUD’s current guidance on criminal history policies. In addition, HUD could improve its ability to identify and address potential noncompliance by determining what additional criminal history requirements to include in its compliance reviews and obtaining additional information on how PHAs are implementing their policies as part of these reviews. Through the Fugitive Felon Initiative, the HUD OIG and the FBI undertook efforts that led to over 1,200 apprehensions of wanted persons in the past 5 years. During the course of our review, the HUD OIG updated its procedures for the initiative in an effort to better define regional office responsibilities and improve the consistency of their participation, as well as to leverage the FBI’s efforts. However, collecting and assessing more comprehensive information on the extent to which regional offices are implementing these new procedures would better enable the HUD OIG to determine the extent to which its regional offices are fulfilling their responsibilities and identify areas for improvement. In addition, by consistently sharing useful information on the results of the initiative, the HUD OIG and the FBI would have better information with which to evaluate the overall effectiveness of the initiative. Finally, the HUD OIG and the FBI could improve their collaboration by agreeing to changes in HUD OIG and FBI responsibilities under the initiative and updating the MOU to reflect these changes. Recommendations for Executive Action We are making a total of seven recommendations: two to HUD’s Office of Public and Indian Housing, three to HUD’s Office of the Inspector General Office of Investigation, and two to the FBI. Specifically: The HUD Assistant Secretary for the Office of Public and Indian Housing should complete its updates of the HCV Program Guidebook and Public Housing Occupancy Guidebook to reflect current guidance on criminal history policies for its public housing and HCV programs. (Recommendation 1) The HUD Assistant Secretary for the Office of Public and Indian Housing should review HUD’s Compliance Monitoring Checklist to determine if questions should be added to address additional federal criminal history requirements and revise checklist instructions to direct HUD staff to obtain information on PHAs’ implementation of these requirements during compliance reviews. (Recommendation 2) The HUD Assistant Inspector General for the Office of Investigation should collect and assess more comprehensive information on regional efforts to implement the activities listed in the 2018 SOP. (Recommendation 3) The HUD Assistant Inspector General for the Office of Investigation should, in collaboration with the FBI, determine what information on fugitive apprehensions and any estimated program savings that occur as the result of the Fugitive Felon Initiative would be most useful and consistently share such information with the FBI. (Recommendation 4) The HUD Assistant Inspector General for the Office of Investigation should, in collaboration with the FBI, update the Fugitive Felon Initiative MOU to reflect the agencies’ current activities and responsibilities. (Recommendation 5) The Director of the FBI should, in collaboration with the HUD OIG, determine what information on fugitive apprehensions that occur as the result of the Fugitive Felon Initiative would be most useful and consistently share such information with the HUD OIG. (Recommendation 6) The Director of the FBI should, in collaboration with the HUD OIG, update the Fugitive Felon Initiative MOU to reflect the agencies’ current activities and responsibilities. (Recommendation 7) Agency Comments and Our Evaluation We provided a draft of this report to HUD, the HUD OIG, and DOJ (including the FBI and USMS) for review and comment. HUD provided comments in an email and the HUD OIG provided comments, the latter of which are reproduced in appendix II. The FBI provided technical comments, which we incorporated as appropriate. USMS informed us that they did not have any comments. In an email received from a HUD PIH audit liaison on July 16, 2018, HUD stated that they agreed with our recommendation to reflect current guidance on criminal history policies in HUD’s updated public housing and HCV program guidebooks (Recommendation 1) and expect to publish the relevant updated chapters in December 2018. In response to our recommendation to review questions in HUD’s Compliance Monitoring Checklist and include instructions for obtaining information on the implementation of the requirements (Recommendation 2), agency officials stated that they reviewed the current checklist questions and determined that no additional questions or revisions are needed at this time. However, the officials did not provide supporting documentation on how they determined that the existing questions were sufficient. They also did not address the part of our recommendation related to HUD revising its checklist instructions to direct staff to obtain information on PHAs’ implementation of criminal history policy requirements. We believe these actions are needed to fully address our recommendation. In its written comments, the HUD OIG disagreed with our recommendation that it collect and assess more comprehensive information on regional office efforts to implement activities listed in the 2018 SOP (Recommendation 3). The HUD OIG stated that it is not feasible to capture information on regional offices’ activities without diverting resources from its primary mission, and that it would be burdensome to create additional mechanisms to monitor participation. We disagree. According to the 2018 SOP, the HUD OIG will be responsible for collecting and reporting statistics for some regional office activities, such as the number of referrals. As discussed in this report, we believe the HUD OIG could obtain more comprehensive information on additional required regional activities using existing resources, such as through periodic data calls to regions. Such assessments of regional office activities are particularly important given that regional offices had not consistently participated in the Fugitive Felon Initiative in the past. Accordingly, we believe our recommendation is still warranted. The HUD OIG also disagreed with our recommendation to determine what information on results of the Fugitive Felon Initiative would be the most useful to share in collaboration with the FBI (Recommendation 4). The HUD OIG stated that its ability to determine apprehensions and program savings is limited. However, the current MOU between the HUD OIG and the FBI states that the HUD OIG is to share this information with the FBI. In addition, in May 2018, HUD OIG officials stated that HUD OIG plans to track statistics on apprehensions that occur with HUD OIG involvement and eventually share these statistics with the FBI. The intent of our recommendation is for the HUD OIG and the FBI to collaborate to determine what information on results should be shared and then share such information consistently. We believe our recommendation provides sufficient flexibility for the HUD OIG and the FBI to determine what information on results would be feasible to collect, and maintain that such collaboration could better position the HUD OIG and the FBI to enhance their ability to identify any areas for improvement and evaluate the effectiveness of the initiative. The HUD OIG agreed with our recommendation to update the Fugitive Felon Initiative MOU to reflect the agencies’ current activities and responsibilities (Recommendation 5). In an email received on July 9, 2018, an FBI management and program analyst stated that the FBI agreed with our recommendation to determine what information on apprehensions resulting from the Fugitive Felon Initiative would be most useful to share and consistently share this information with the HUD OIG (Recommendation 6). The FBI also agreed with our recommendation to update the Fugitive Felon Initiative MOU to reflect the agencies’ current activities and responsibilities (Recommendation 7). As agreed with your office, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies to the Secretary of the Department of Housing and Urban Development, the Inspector General of the Department of Housing and Urban Development, the Attorney General of the United States, and other interested parties. In addition, the report will be available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions concerning this report, please contact Daniel Garcia-Diaz at (202) 512-8678 or garciadiazd@gao.gov, or Gretta Goodwin at (202) 512-8777 or goodwing@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. Key contributors to this report are listed in appendix III. Appendix I: Objectives, Scope, and Methodology This report examines (1) the statutory and regulatory requirements for public housing agencies’ (PHA) criminal history policies for public housing and Housing Choice Voucher (HCV) programs; (2) the extent to which the Department of Housing and Urban Development (HUD) provides guidance and monitors PHA implementation of criminal history policy requirements for public housing and HCV programs; and (3) the implementation of the Fugitive Felon Initiative by the HUD Office of Inspector General (OIG), in coordination with the Federal Bureau of Investigation (FBI). To describe the statutory and regulatory requirements (federal requirements) for PHAs’ criminal history policies, we reviewed federal statutes and HUD regulations for the public housing and HCV programs on providing housing assistance to persons with criminal history records and arrest warrants, including fugitive felons. We focused on the public housing and HCV programs because PHAs screen applicants and determine eligibility for these programs, whereas property owners are primarily responsible for these functions for other HUD rental assistance programs. In addition, the HCV program is the U.S. government’s largest rental assistance program. In addition, we interviewed officials from HUD headquarters as well as officials in five HUD field offices and 10 PHAs in four selected metropolitan areas: Chicago, Dallas/Ft. Worth, New York City, and Philadelphia. We used a cluster sampling technique to select the four metropolitan areas. In selecting these areas, we considered geographic location and proximity of HUD field offices and HUD OIG regional offices to each other, whether there were options to visit a variety of differently sized PHAs that had different characteristics (such as ones that managed both the public housing and HCV programs or had a law enforcement department), and participation of HUD OIG regional offices in the Fugitive Felon Initiative. In each of the four selected metropolitan areas, we selected two to three PHAs to visit, for a total nongeneralizable sample of 10 PHAs (see table 5). In selecting PHAs, we considered PHA size (as measured by the number of public housing and HCV units), whether the PHA implemented both public housing and HCV programs, distance in miles between a PHA and the HUD and HUD OIG metro area offices, and whether a PHA was in an urban or nonurban location and had a law enforcement department. We selected five HUD field offices by determining which field office oversees each of the selected PHAs. We interviewed officials from the 10 selected PHAs and reviewed their criminal history policies to better understand the federal requirements and how PHAs implemented them for the public housing and HCV programs. We did not conduct a compliance audit of the selected PHAs. We also interviewed officials from three housing associations (selected based on their expertise with the public housing and HCV programs) about federal requirements and PHAs’ implementation of the requirements. In addition, we interviewed a nonprofit organization that wrote a report on HUD’s criminal records policies and two private companies that conducted criminal background screening for PHAs to better understand criminal screening processes. To determine the extent to which HUD provides guidance and monitors PHA implementation of criminal history policy requirements, we reviewed HUD letters and notices for the public housing and HCV programs. We also reviewed HUD’s 2001 HCV Program Guidebook and 2003 Public Housing Occupancy Guidebook. We interviewed officials from the 10 selected PHAs for their perspectives on HUD’s guidance. We also reviewed HUD’s monitoring procedures for PHAs. Specifically, we reviewed documentation related to HUD’s National Risk Assessment as well as HUD’s Compliance Monitoring Checklist for reviewing PHA compliance with federal requirements, including requirements on providing housing assistance to persons with criminal history records. We interviewed officials from HUD headquarters and our sample of five HUD field offices about the agency’s efforts to monitor and oversee PHAs’ implementation of criminal history policy requirements (same selected regional offices discussed above). We assessed HUD’s guidance and compliance procedures in relation to federal requirements for criminal history policies in relation to federal statutes, HUD regulations concerning criminal history policies, and internal control standards. To determine the extent to which the HUD OIG, in coordination with law enforcement agencies, implements and monitors the Fugitive Felon Initiative, we reviewed memorandum of understanding (MOU) agreements between the HUD OIG and the FBI and between the HUD OIG and the U.S. Marshals Service (USMS) on their efforts to share and analyze data on HUD tenants and wanted persons and coordinate any apprehension efforts. We reviewed HUD OIG’s Standard Operating Procedure for the Fugitive Felon Initiative and interviewed officials from the FBI, HUD OIG, and USMS headquarters to obtain information on the processes these agencies follow as part of the initiative. We also interviewed officials from all seven HUD OIG Office of Investigation regional offices and relevant USMS Fugitive Task Forces in our four selected metropolitan areas to obtain information on their involvement with and perspectives on the Fugitive Felon Initiative. We worked with USMS Headquarters to identify the relevant fugitive task force that would have jurisdiction over the geographic area covered by a HUD OIG regional office. We assessed the HUD OIG’s and the FBI’s activities in relation to their current MOU, OIG’s Standard Operating Procedure for the Fugitive Felon Initiative, and federal internal control standards. We collected and analyzed data on the HUD OIG’s Fugitive Felon Initiative. Specifically, we analyzed the results of the HUD OIG’s 2017 efforts to cross-reference HUD tenant data and the FBI’s Wanted Persons File (from September 2016) to identify potential investigative leads into the possible location of fugitive felons. We summarized the types of offenses related to these potential investigative leads by grouping similar offenses together and identified the top 10 most frequently occurring offenses. Table 6 lists the subcategories of assault, burglary, fraud, forgery, larceny, and robbery. There were no subcategories associated with the other 4 offenses in the top 10 most frequently occurring (failure to appear, probation violation, parole violation, and dangerous drugs). To assess the reliability of the HUD OIG data, we interviewed knowledgeable agency officials, conducted electronic testing for missing data and obvious errors, observed the HUD OIG’s process for cross- referencing HUD tenant data and the FBI’s Wanted Persons File, and reviewed system documentation for the data systems the HUD OIG uses to cross-reference the data. We determined these data to be reliable for our purposes of describing the number of potential investigative leads produced by the initiative, the types of offenses associated with the potential investigative leads, and the HUD rental assistance programs in which identified fugitive felons participated. We also reviewed FBI data on the results of law enforcement agencies (as reported to the FBI from fiscal years 2013 through 2017) in apprehending fugitive felons based on potential investigative leads produced by the initiative. To assess the reliability of the FBI data, we interviewed knowledgeable agency officials and reviewed documentation for the data system the FBI uses to store and retrieve these data. We determined these data to be reliable for our purposes of describing the number of apprehensions that result from the potential investigative leads identified as part of the Fugitive Felon Initiative. We conducted this performance audit from January 2017 to August 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix III: GAO Contact and Staff Acknowledgments GAO Contacts Staff Acknowledgments In addition to the contacts named above, Allison Abrams (Assistant Director), Eric Erdman (Assistant Director), Charlene J. Lindsay (Analyst in Charge), Charlene Calhoon, Mara McMillen, David Ballard, Rudy Chatlos, Willie (Billy) Commons III, Marc Molino, Tovah Rom, and Tyler Spunaugle made key contributions to this report.

Why GAO Did This Study HUD has encouraged PHAs to balance resident safety with the housing needs of persons with criminal records when administering its rental assistance programs. PHAs are responsible for screening program applicants. The HUD OIG and the FBI implement the Fugitive Felon Initiative to identify and apprehend wanted persons receiving rental assistance. GAO was asked to review HUD's criminal history policies and the Fugitive Felon Initiative. This report examines (1) federal requirements for PHAs' criminal history policies, (2) HUD guidance and monitoring of these requirements, and (3) implementation of the Fugitive Felon Initiative. GAO reviewed federal statutes and regulations and interviewed officials from HUD, the HUD OIG, and the FBI; analyzed Fugitive Felon Initiative data from 2013 through 2017; and interviewed staff at a nongeneralizable sample of 10 PHAs (selected based on size and other factors). What GAO Found Federal requirements for public housing agencies. Federal statutes and Department of Housing and Urban Development (HUD) regulations require public housing agencies (PHA) to conduct criminal history checks on individuals applying for rental assistance under HUD's public housing and Housing Choice Voucher programs and deny assistance for six types of offenses. Mandatory denials include convictions for producing methamphetamine on the premises of federally-assisted housing and lifetime sex offender registrants. Otherwise, PHAs generally have discretion in establishing their criminal history policies and may deny assistance for other offenses or factor in mitigating circumstances. HUD monitoring of public housing agencies. From 2011 through 2016, HUD issued new guidance to PHAs on criminal history policies, but these changes are not reflected in HUD's program guidebooks for PHAs. These guidebooks serve as key reference tools, but have not been updated in over 15 years. Updating them would help HUD more accurately communicate its criminal history policies. While HUD officials said their current efforts to update the guidebooks will reflect recent criminal history policy notices, documentation provided by the agency on these updates did not specifically address criminal history guidance. In addition, HUD's compliance reviews of high-risk PHAs do not address some criminal history policy requirements, such as the prohibition on using arrest records as the basis for determining eligibility. Further, these reviews are largely limited to examining PHAs' written policies and do not cover how PHAs implement those policies. More comprehensive compliance reviews would improve HUD's ability to identify areas of noncompliance with criminal history policy requirements. Fugitive Felon Initiative. From fiscal years 2013 through 2017, the HUD Office of Inspector General (OIG) and the Federal Bureau of Investigation (FBI) shared data through the Fugitive Felon Initiative, which led to the apprehension of more than 1,200 wanted persons who may have lived in HUD-assisted housing. However, GAO found that the HUD OIG had not defined its regional office responsibilities under the initiative and that four of the seven HUD OIG regions did not participate from 2012 through 2016. The HUD OIG revised its procedures for the initiative in April 2018 to include regional office responsibilities, such as coordinating with law enforcement agencies. According to HUD OIG officials, regional offices are now required to coordinate with law enforcement agencies on a priority list of investigative leads, which include warrants for violent felonies, sexual assault, and narcotics distribution. However, the HUD OIG does not plan to assess regional office implementation of several requirements. Collecting and assessing more comprehensive information on regional office activities would help the HUD OIG determine the extent to which regions are undertaking required activities. In addition, the HUD OIG and the FBI have not consistently shared information on the initiative's results—such as apprehension statistics and program savings—which could help evaluate the effectiveness of the initiative. Further, the HUD OIG's and the FBI's current activities to implement the initiative differ in some areas from the agreed-upon responsibilities listed in their 2012 memorandum of understanding. Updating the memorandum to reflect current responsibilities under the initiative could help improve collaboration between the agencies and improve implementation. What GAO Recommends GAO is making seven recommendations, including that HUD update PHA guidebooks and improve monitoring procedures; that the HUD OIG assess more comprehensive information on the implementation of the Fugitive Felon Initiative; and that the HUD OIG and the FBI consistently share information on the initiative's results and update their memorandum of understanding to reflect current responsibilities. HUD and the FBI generally agreed. The HUD OIG did not agree with two of our recommendations. GAO maintains the recommendations, as discussed in the report.

gao_GAO-18-67

gao_GAO-18-67_0

Background Electricity Operation and Delivery in the United States and Canada The electricity operation and delivery system—collectively referred to as the grid—in the United States and Canada includes four functions: generation, transmission, distribution, and system operations (see fig. 1). Electricity is generated at power plants by burning fossil fuels; through nuclear fission; or by harnessing renewable sources such as wind, solar, geothermal energy, or hydropower. Once wholesale electricity is generated, it enters the bulk power system—a network of high-voltage, high-capacity transmission systems—where it is transformed to a higher voltage and flows through transmission lines, generally over long distances, to areas where it is transformed to a lower voltage and sent through the local distribution system for use by various customers. Throughout this process, system operations are managed by a system operator, such as a local utility. Below is additional information on the functions of the electric grid, including equipment that may be vulnerable to GMD and HEMP. Electricity generation. Power plants generate electricity by converting energy from other forms—such as coal, natural gas, or wind—into electricity. While they produce electricity once operating they are vulnerable when power outages occur because initially starting a power plant after an outage typically requires an external source of electricity to operate the control systems—electronics that are integral to their operations. Some power plants have the capability to restore operations by employing a “black start,” which is the process of restoring a plant to operation without relying on off-site sources of electricity, usually through using dedicated diesel generators to provide the electricity needed. However, not all plants have this capability and in the event of a power outage could therefore be vulnerable to lengthy system disruptions. Electricity transmission. Power plants are generally geographically distant from the areas where electricity is consumed. To move electricity from where it is produced to where it is used, electricity is sent over transmission lines; together, these lines form a network, or grid. To transport energy over long distances with reduced power losses, suppliers increase voltages—the “force” that makes electricity flow through a conductor—and utilize high-voltage transmission lines, operating from 230 up to 765 kilovolts (kV) in North America. According to the Quadrennial Energy Review, as of January 2017, there were approximately 240,000 miles of high-voltage transmission circuit lines in the contiguous United States. During a solar storm, high-voltage transmission lines can act as “antennae” that allow GMD to enter the electric system. Transformers. Transformers are critical to the efficient and effective delivery of electricity to customers and, under certain circumstances, can be vulnerable to the effects of GMD and HEMP. Transformers facilitate the efficient transfer of electricity over long distances through the transmission system by converting electricity to different voltages along the delivery system—either up or down, depending on the design and function of the transformer (see sidebar). Figure 2 depicts a large power transformer. paired with equipment—for example, a protective relay—that is designed to take them out of service temporarily when the effects of an electromagnetic event reach the grid. If transformers were temporarily taken out of service for preventative purposes, it could lead to an interruption of electricity service to consumers. However, if transformers—especially those more vulnerable due to age, condition, or design—are not taken out of service during an electromagnetic event they are at risk of being permanently damaged when additional electrical current flows into them, causing excessive localized heating and damage to internal components. (See fig. 3 for an example of transformer windings that were damaged from localized heating associated with a GMD event.) Transformers that become permanently damaged during an electromagnetic event can also contribute to interruptions in service. According to DOE, replacing a damaged transformer can be challenging because they are custom-designed and interchangeability and availability of spares is limited. If a usable spare transformer is not immediately available, obtaining a replacement transformer is often a long and costly process, usually involving long delivery lead times due to their size and weight, limited inventory, a complex procurement and manufacturing process, and other factors. According to DOE, in 2014 the average lead time to obtain a large power transformer was between 5 and 16 months, but could take more than 20 months in the event of supply disruptions or delays in procuring raw materials or key parts; larger, more sophisticated models are generally manufactured abroad. According to a transformer manufacturer, depending on the function of the transformer, the voltage rating, and the model, in 2017 the approximate price of a large power transformer, weighing from 170 to 410 tons, ranged from approximately $2 to $7.5 million in the United States. Distribution system. The final stage in the electric power system is the distribution system, which carries electricity out of the transmission system to industrial, commercial, residential, and other consumers. The distribution system includes equipment that can be damaged during electromagnetic events, but the extent of the risk is limited because distribution lines are generally too short and of too low voltage to pose a risk to distribution equipment. System operations: Operation of the electricity system is managed by entities such as a local utility, which this report collectively refers to as system operators. Because electric energy is not typically stored in large quantities, system operators must constantly balance the generation and consumption of electricity to maintain reliability. To do this, system operators utilize a system of sensors and controls to monitor power consumption and generation from a centralized location. Operators use computerized systems to send signals to power plants and other grid components to adjust their output to match changes in consumption. Electromagnetic events can interrupt or damage some of the equipment system operators use, which can cause a disturbance in control systems (for example, such events can cause relays to unintentionally operate, which can disable system protection equipment). Because the electric power system increasingly operates at or near reliability limits during peak demand periods, a relatively modest disturbance to the system can potentially pose a risk to system reliability. In the United States, the electrical infrastructure is primarily operated by private industry, which owns approximately 85 percent of the nation’s critical electrical infrastructure. In contrast, Canada’s electrical infrastructure is primarily organized along provincial lines with large, government-owned, integrated public utilities playing a leading role in the generation, transmission, and distribution of electricity. Electromagnetic Events – GMD and HEMP Based on our review of relevant studies and interviews with cognizant government and industry officials, there are differing opinions on the potential impact electromagnetic events could have on the electric grid and the risk of long-term, widespread damage. However, they generally agree that more study on the effects of electromagnetic events is needed. The following section describes (1) the nature and potential impact of GMD, U.S. efforts to monitor it, and the frequency of past occurrences; and (2) the nature and potential impact of HEMP events. GMD – Description, Potential Impact, Monitoring, and Historical Occurrences Naturally occurring solar weather events can create electromagnetic impacts—or GMD—of sufficient intensity that can adversely affect the electric power system. Solar weather events include, for example, large coronal mass ejections (CME), which are energetic eruptions in the sun’s atmosphere that can cause the release of a large mass of charged particles from the sun into space. When a large CME travels from the Sun to the Earth it can interact with and create disturbances in the Earth’s geomagnetic field, referred to as a geomagnetic storm; the resulting impact on Earth is commonly referred to as a geomagnetic disturbance, or GMD. Figure 4 illustrates how solar weather can create a GMD. Strong GMDs can create large geomagnetically induced current (GIC) on the grid. The degree to which GMD and accompanying GIC affect the electric power system depends on several factors, including the magnitude of the GMD, design and geomagnetic latitude of the power system, and geology of the local area, among other things. According to NERC, the most likely consequence of a strong GMD and the accompanying GIC is the loss of voltage stability, although GMD can also damage components of the system, including high-voltage transformers. In the United States, the National Oceanic and Atmospheric Administration’s (NOAA) National Weather Service manages the Space Weather Prediction Center (SWPC), which is responsible for monitoring and providing services on space weather, including geomagnetic storms. SWPC uses a variety of ground and space-based sensors, as well as imaging systems, to monitor conditions on the Sun and to observe and forecast geomagnetic activity around the world. SWPC uses this information to issue Watches, Warnings, and Alerts for geomagnetic storms through e-mail and website postings to those who are impacted by space weather, such as owners and operators of the electric grid, spacecraft operations, users of radio signals, and others. In addition, SWPC provides immediate telephone notification and confirmation of imminent and ongoing geomagnetic storms to all NERC reliability coordinators through a NERC hotline. To communicate the magnitude of geomagnetic storms (disturbances in Earth’s magnetic field) and to determine whether geomagnetic alerts and warnings should be issued, SWPC relies on a real-time estimate of the Planetary K-index (Kp-index), which ranges from Kp = 0, or quiet, to Kp = 9, or extreme storm intensity. (See appendix II for more information on SWPC’s notification system as well as their estimates of overall impact of geomagnetic storms to the electric power system, by storm level.) Figure 5 shows the range of planetary geomagnetic activity, by solar cycle and Kp level, from 1933 through 2017. As shown in this figure, recent activity—between 2007 and 2017, approximately equivalent to the average length of a solar cycle—exhibited the fewest occurrences of GMD events (minor, moderate, strong, severe, and extreme) of any solar cycle in nearly a century. -index) data show the maximum fluctuations in the magnetic field observed from a network of selected magnetometers—instruments that measure relative change of a magnetic field at a particular location—relative to a quiet day. HEMP – Description and Potential Impact According to the 2008 EMP Commission, a nuclear EMP is the burst of electromagnetic radiation that results from the detonation of a nuclear device, which can disrupt or destroy electronic equipment. The threat primarily focused on by the 2004 and 2008 EMP Commissions, and specifically addressed in this report, is the high-altitude EMP (HEMP). A HEMP event is caused by the detonation of a nuclear device above the atmosphere, from about 40 to 400 kilometers (approximately 25 to 250 miles) above the Earth’s surface. A HEMP attack does not cause direct physical impacts at the Earth’s surface, such as injury or damage directly from heat, blast, or ionizing radiation, but instead creates an intense electromagnetic pulse. The components of HEMP—commonly identified as E1, E2, and E3—can disrupt or damage critical electrical infrastructure, such as computers, electronics, and transformers. EMP can also be produced using nonnuclear weapons, but these generally have a short range and are not a focus of this report. Electricity System Regulation and Oversight Responsibility for regulating electricity is divided between states and the federal government. Most electricity consumers are served by retail markets that are regulated by the states, generally through state public utility commissions or equivalent organizations. As the primary regulator of retail markets, state commissions approve many aspects of utility operations, such as the siting and construction of new power plants, as well as approving the prices consumers pay and how those prices are set. Prior to being sold to these retail consumers, such as households and businesses, electricity is bought, sold, and traded in wholesale electricity markets by companies that own power plants, as well as utilities and other companies. Wholesale interstate electricity markets are regulated by FERC. Historically, FERC-approved wholesale electricity rates based on utilities’ costs of production plus a rate-of-return that it determined to be reasonable. Beginning in the late 1990s, FERC took a series of significant steps to restructure the wholesale electricity markets to increase the role of competition—market forces of supply and demand— in setting wholesale electricity prices, a process referred to as electricity restructuring. Subsequently, FERC has provided authority for many entities—for example, independent owners of power plants—to sell electricity in wholesale markets at prices determined by supply and demand. These entities can now compete with existing utilities and one another to sell electricity in wholesale markets, but have no assurance that their costs will be recovered. While electricity restructuring has introduced a measure of market-based pricing to the generation of electricity, transmission (and distribution, regulated by states) are still subject to regulation on a cost-recovery basis. FERC has jurisdiction over transmission rates on the federal level, and state regulators have jurisdiction over the charges that utilities incorporate in customers’ rates in order to recover their transmission costs. As part of the restructuring process, FERC also encouraged the voluntary creation of new entities called Regional Transmission Organizations (RTO) and Independent System Operators (ISO) to manage regional networks of electric transmission lines as grid operators—functions that, in these areas, had traditionally been carried out by local utilities. These RTOs, in many cases, established and manage wholesale electricity markets for electricity buyers and sellers to participate in. As grid operators, RTOs are also responsible for managing transmission in their regions, which includes establishing and implementing rules and pricing related to transmission, among other things. As we reported in 2003, 24 states also introduced retail competition to electricity markets they regulate and allow former utilities and new companies to compete to serve customers; since then, the states where retail competition is occurring have changed. In states with retail competition, in general, electricity rates for generators other than the original utility are not structured to guarantee recovery of generation-related costs. In addition to its role in regulating aspects of the electricity market, FERC is also responsible for reviewing and approving standards to ensure the reliability of the bulk power system. FERC designated NERC to develop and enforce these reliability standards, subject to FERC review and approval. These standards outline general requirements for planning and operating the bulk power system to ensure reliability. (See appendix III for information on NERC reliability standards requiring electricity suppliers to address the potential impact of GMD on the reliable operation of the U.S. electric grid.) NERC and its Regional Entities, along with FERC, can all independently enforce reliability standards. Within the boundary of each regional entity, there are one or more NERC-certified reliability coordinators. Reliability coordinators are charged with the task of continuously assessing the reliability of the transmission system. The coordinator has the authority to direct stakeholders—transmission operators, generators, and others involved with the electric grid’s operations—to take action to preserve the reliability and integrity of the bulk power system. U. S. and Canadian Electricity Suppliers We Contacted Have Identified Information about GMD Effects, but Have Less Information about HEMP Effects Electricity Suppliers Have General Information about GMD Effects and Some Selected Suppliers Have Taken Steps to Evaluate Their Networks U.S. and Canadian government and industry organizations have studied and publicly reported on potential GMD effects on the electric grid. These studies have covered the general threats to the nation’s electric grid from GMD but do not cover the unique aspects of individual suppliers’ generation and transmission networks that could potentially make them more or less vulnerable to GMD events. In addition, these studies typically identified areas in which more research is needed regarding the GMD threat and potential mitigation measures that would inform suppliers’ own assessments of the potential impact of GMD events on their unique networks. These select studies we identified included those performed by NERC, DOE, EPRI, and other private industry groups and generally examined the areas of vulnerability for the grid with respect to GMD events, potential impact on the grid from these events, possible mitigation measures, and areas needing further research. While noting the need for further research, some of these studies vary with regard to their assessment of the likelihood of long-term, widespread damage to the grid from these events. The following is a summary of some of these selected studies performed since 2010 and grouped by the entities responsible for performing them: NERC. In June 2010, NERC issued a report, based on a joint effort with DOE, which included a plan to form a task force of government and industry efforts to examine GMD. This resulted in the formation of the NERC GMD Task Force consisting of government and industry officials to examine the GMD threat to the nation’s power grid. The task force’s work in evaluating the potential impact of GMD events resulted in NERC’s subsequent February 2012 report which outlined its plans for working with industry on new reliability standards for protecting the grid against GMD events. This report concluded, among other things, that the failure of a large number of transformers during a severe GMD event was unlikely, although certain older transformers, along with generator step-up transformers, could be particularly susceptible. As a result of this work, and as directed by FERC, NERC developed the EOP-010-1 and TPL-007-1 GMD reliability standards. Also, as a result of this work, NERC issued a GMD Planning Guide for electricity suppliers, which assists the suppliers in carrying out studies of their individual vulnerabilities to a GMD event. DOE/National Labs. Since 2010, DOE has been engaged in a number of efforts regarding GMD. For example, in 2011, DOE enlisted the Pacific Northwest National Laboratory (PNNL) to study the potential effect of GMD on long, high-voltage transmission lines and associated mitigation measures that could potentially be employed. Also, in April 2014, DOE reported on the results of its study of the vulnerabilities of large power transformers to GMD and other threats and the challenges facing the replacement of these transformers in the wake of such events. EPRI. In conducting research for its private industry membership, the Electric Power Research Institute engaged in a number of studies from October 2010 to March 2014. These efforts began with an effort to examine potential impacts from GMD through an assessment of various risk factors. EPRI’s later efforts involved the development of approaches for modeling the impacts from GMD on the grid to allow suppliers to better protect their networks from these events. Other private industry. Private entities conducted studies in January 2010 and November 2011 for Oak Ridge National Laboratory and DHS, respectively, that examined prior GMD events and assessed the potential future impact of these events on the grid along with areas of vulnerability and potential mitigation measures. Other private studies included those examining which regions of North America are most vulnerable to GMD events in addition to the potential impact on the insurance industry and society in general from these events. See appendix IV for additional details on these and other select studies performed by government and industry regarding protection of the grid from GMD events. These past research efforts have generally identified the degree to which the electric power system is affected by a GMD event. The level of impact from these events can depend on various factors including, among other things, magnitude of the event, design and geomagnetic latitude of the power system, and geology of the local area. Further, these studies identified that GMD can have a broad range of impacts when it is introduced to a power system, ranging from minor events such as radio interference and control malfunctions to wide-scale disruptions. NERC has identified two predominant risks to the system: (1) potential voltage instability in the transmission system caused by insufficient reactive power support and (2) possible damage to system components. The first risk and, according to NERC, the most likely consequence of a strong GMD event and accompanying GIC, is the insufficient reactive power support, which can lead to voltage instability and power system collapse. Reactive power support is necessary to stabilize the transfer of electricity through the electric power system, from generation to consumption. With regard to the second risk, several components of the electric system are susceptible to damage from GMD and GIC, but government and industry officials agree that the vulnerable components with the greatest potential consequence in the event of loss are transformers, which typically exist at substations throughout suppliers’ transmission networks. High-voltage transmission lines act as “antennae,” allowing GIC to enter the electric power system, disrupting normal operations and, in some cases, damaging equipment. Transformers, in turn, run the risk of overheating during a GMD event. According to NERC, restoration times for these two risk scenarios are significantly different. Restoration time from voltage collapse—i.e., system blackout—would be a matter of hours to days, while the replacement of transformers, as previously discussed, could take months or potentially years. Therefore, the failure of large numbers of transformers, while less likely, would have considerable impacts on portions of the electric power system. While general information on the potential impact of GMD events on the electric grid is available from the aforementioned government and industry reports, individual suppliers must assess the potential impact on their own, unique networks. For example, of the 13 selected suppliers we spoke with, 11 reported performing analyses to evaluate the potential impact of GMD on their specific generation systems or transmission networks. The 11 suppliers that had performed these analyses did so in advance of all suppliers being required to analyze the vulnerabilities of their networks as part of their compliance with NERC’s second-stage GMD reliability standard TPL-007-1. The nature of the analyses the 11 suppliers engaged in required the use of modeling software to determine the specific vulnerabilities of their networks which further allowed them to design their own mitigation measures, if warranted, to address any vulnerabilities identified and prevent equipment damage or power outages. Suppliers we contacted noted that potential GMD mitigation measures included installation of specific equipment to assist with network stability and voltage regulation. As noted previously, past research efforts have indicated that GMD events can have a variety of impacts ranging from minor malfunctions to wide-scale disruptions. For example, 3 of the 11 suppliers we contacted that had performed an analysis of their networks’ potential vulnerabilities had also reported prior impacts on their networks from a GMD event. Of these three suppliers, two (including Hydro-Quebec) reported major power interruption or equipment damage from the event. The remaining supplier reported a brief power outage on one transmission line during the same 1989 GMD event that caused a major power outage for Hydro- Quebec; however, this power outage did not result in any significant loss of electricity to the supplier’s customers. Outside of the 1989 event, this same supplier reported minor power fluctuations and voltage drops from smaller GMD events. Most suppliers we contacted that had assessed their networks’ vulnerabilities to GMD expressed confidence in their ability to avoid major damage or power interruptions from future GMD events. Specifically, 6 of the 11 selected suppliers that had performed analyses of their networks’ vulnerabilities to GMD reported that, going forward, they expected that any effects from a future GMD event on their networks would likely be minimal (i.e., no significant damage or power interruption). Six suppliers also thought that procedures and technology currently in place afforded better protection from these events than in the past. For example, one northern U.S. supplier we contacted had, after acquiring new GMD analysis software, studied its system and concluded that it could easily withstand the GMD “benchmark event” established by NERC in its TPL- 007-1 reliability standard and that its current technology and procedures were adequate to deal with the threat. Also, another supplier studied its system and is using the results to inform future decisions on transformer purchases to obtain technology that is more resistant to GMD. Government and Industry Are Taking Steps to Better Understand Effects of HEMP on the Grid According to U.S. government and industry officials we spoke with, completed research and available information on the vulnerability of the grid to HEMP, along with its potential effects, is less extensive and lags behind industry understanding of GMD. These officials noted that the understanding of HEMP and how it can affect the electricity system is general in nature and not specific to the commercial electric grid. Specifically, the Department of Defense has developed information regarding the potential effects of HEMP on military assets and facilities. According to DOE, the most detailed HEMP testing has been performed on military communication and weapons systems, not on the commercial electric grid. In a number of studies since 2010, both government and private industry have examined the HEMP threat to the grid while also noting the need for further research to fully understand the specific threats to components of the grid that would allow suppliers to protect against these events. While noting the need for further research, some of these studies vary with regard to their assessment of the likelihood of long-term, widespread damage to the grid from HEMP. See appendix IV for additional details on government and industry studies on the threat to the grid from EMP events including HEMP. The government and private industry studies generally note the threat to the grid presented by the E1, E2, and E3 pulse components of HEMP as follows: E1. The E1 pulse is capable of destroying microelectronics (such as computers), communication and control systems, and other electronic equipment that can disrupt the grid and other critical sectors. According to DOE, E1 can also generate very large and damaging voltage surges in power lines. Figure 6 depicts the potential impact from an E1 pulse, and shows the higher the altitude the greater the potential radius of the impact from an E1 pulse. E2. The E2 pulse, similar to lightning, has an ability to impair or destroy control features that are not protected from lightning. However, the grid typically has protections in place to address the lightning threat to major components. E3. The E3 pulse is similar to GMD and also creates similar disruptive currents in transmission lines which can cause grid instability and heating that damages transformers. Few electricity suppliers we contacted reported taking steps to examine how HEMP could impact their systems. Specifically, 3 of 11 selected suppliers who responded to our inquiry on this topic reported performing a study of the potential impact of HEMP events on their network infrastructure. Two of these three suppliers reported studying the potential impact of HEMP events on their network in conjunction with these suppliers’ design of hardened control centers expected to be resistant to HEMP and other hazards. One of the two suppliers that designed control centers resistant to HEMP did so due to a concern over being able to maintain power to certain critical customers for which the loss of power would have national security implications. The other supplier that had designed an HEMP-resistant control center did so as part of an “all hazards” approach to protecting its transmission infrastructure. The third supplier that had studied the potential impact of HEMP on its system did so as part of a combined study, required by its state legislature, on the threats posed by both GMD and HEMP. Specifically, this supplier examined the potential impact of the HEMP E3 pulse on its system which is similar to GMD, and, therefore, is expected to involve similar mitigation measures. The supplier stated that the lack of available modeling and analysis tools prevents them from fully understanding the potential impact of all components of HEMP— particularly the E1 and E2 pulses. Four of the 8 suppliers we contacted who stated they had not studied the potential impact of HEMP on their networks indicated a desire to see EPRI complete its ongoing EMP research before engaging in studies of their own networks. Further, all eight suppliers who stated they had not performed any studies of the potential impact of HEMP on their networks also noted a lack of key information on the nature and risk of the HEMP threat that would allow them to complete studies of their networks and develop corresponding mitigation measures. Six of the suppliers cited the classified nature of much of the available information maintained by the federal government on the EMP threat—particularly HEMP—as a contributing factor to the industry’s lack of needed information on the threat. In addition, according to NERC officials, while they have developed reliability standards directing suppliers to study the vulnerabilities of their networks to GMD and establish procedures for dealing with those events, it has yet to produce similar standards for EMP or HEMP due to the lack of information available to industry on the EMP threat and how it may impact the grid. According to DOE, more research is needed to fully investigate and evaluate how an electric utility could protect itself from, or mitigate the effects of, HEMP on its systems. DOE also noted that government and industry have ongoing research efforts to better understand these potential effects and develop possible mitigation measures. For example, DOE has three ongoing research efforts related to HEMP. First, DOE is collaborating with DHS to advance the understanding of HEMP effects on the grid through research at the Los Alamos National Laboratory. Second, DOE has funded efforts underway at the Idaho National Laboratory focused on developing potential HEMP strategies, protections, and mitigations for the electric grid—including hardening of infrastructure, blocking of currents, developing a strategy for stocking and prepositioning of spare parts, as well as developing operational and emergency planning tools. Finally, DOE has enlisted the Oak Ridge National Laboratory in analyzing the vulnerability of the grid to a HEMP event, along with the potential damage from such an event, and how it would impact on the reliability and delivery of electric power. The analysis will examine resilience options such as hardening some facilities, stockpiling some parts, and contingency planning. In addition to these research projects, DOE officials told us both Los Alamos National Laboratory and Lawrence Livermore National Laboratory are working to produce unclassified information on the characteristics of the electromagnetic signals associated with HEMP that could be shared with electricity suppliers to better inform their planning efforts. In addition to its ongoing research efforts, DOE and industry have taken steps to enhance understanding of HEMP issues. In particular, DOE and industry issued the Joint Electromagnetic Pulse Resilience Strategy (Joint Strategy) in July 2016 to study HEMP and improve the sharing of information on HEMP that would be useful to industry. According to DOE, central to development of the Joint Strategy was an effort to enhance shared government-industry understanding of the current status of risks from, and preparedness for, HEMP events. DOE added that this effort is important because what is currently known about HEMP effects to the grid has been developed from computer models designed for other purposes (e.g., understanding Department of Defense system effects), or is classified and thus difficult to share with industry. Specifically, the Joint Strategy includes five strategic goals to guide DOE and industry in minimizing HEMP impacts and improving resilience of the grid to these events. These strategic goals are (1) improving and sharing understanding of HEMP: threat, effects, and impacts, (2) identifying priority infrastructure, (3) testing and promoting mitigation and protection approaches, (4) enhancing response and recovery capabilities relating to a HEMP attack, and (5) sharing best practices across government and industry both nationally and internationally. Following development of the Joint Strategy, both DOE and EPRI (working with DOE, on industry’s behalf) committed to developing separate, but coordinated, action plans that would implement the five strategic goals for studying HEMP and providing needed information to industry. DOE’s Electromagnetic Pulse Resilience Action Plan (DOE Action Plan), issued in January 2017, delineates the steps that DOE will take to address HEMP risks and emphasizes the federal government’s ability to clarify and communicate HEMP threats and impacts, reduce HEMP vulnerabilities, and facilitate the energy sector’s response and recovery after HEMP events. DOE stated that its Action Plan also considers the over 90 recommendations made in the 2008 EMP Commission report and at least partially addresses 10 of the 15 recommendations directly related to the electric power system made by the EMP Commission in their report. See appendix V for additional detail on the DOE Action Plan, including its relationship to the EMP Commission’s work. As noted in the Joint Strategy, EPRI’s industry action plan—initiated in April 2016—is a complement to the DOE Action Plan and includes research to be performed to (1) detail the potential impacts of HEMP on the bulk power system, (2) examine potential industry actions to mitigate HEMP risks, and (3) inform industry investment decisions regarding those mitigation options. According to DOE and EPRI, the research that is outlined in the industry action plan is ongoing and scheduled for completion over a 3-year period with the first two reports being issued in September 2016 and February 2017. EPRI officials added that this research is intended to provide the electric industry with what it needs— specifically, an unclassified, science-based approach to HEMP with regard to (1) threat characterization, (2) testing results, (3) modeling and simulation, and (4) recommended strategies for mitigating the impacts of HEMP including prudent and practical hardening and recovery options. To meet these goals, EPRI, together with participating suppliers, have undertaken this 3-year long research effort and expect to complete this work in 2019. This research effort is comprised of the following tasks: HEMP threat characterization. For the first part of this task, EPRI is identifying the state of knowledge of unclassified HEMP research for all three components of the HEMP environment (i.e., the E1, E2 and E3 pulse components of HEMP). This portion of EPRI’s research was achieved by the issuance of the aforementioned September 2016 report. The remaining two components of this task are ongoing and include (1) identifying characteristics of the electromagnetic signals associated with HEMP that can be used to assess the potential impacts on bulk power system components, and (2) investigating the physics of HEMP’s transmission to, and impact on, power system infrastructure. Electric infrastructure EMP vulnerability. This task involves identifying the vulnerability of transmission systems and support assets (e.g., protection and controls systems, communications, transformers, etc.) exposed to the HEMP threat by performing laboratory tests. EPRI will test various infrastructure components at two EMP test labs by subjecting them to E1 pulses. According to EPRI, initial results for this task are possible by the end of 2017. Electric infrastructure impacts. For this task, EPRI is assessing the potential impacts of a HEMP attack on the bulk power system by combining the system modeling-related efforts in the first task above with the equipment testing results of the second task above. Under this task, EPRI is also developing assessment techniques, models, and tools for assessing the impacts of a HEMP attack. The aforementioned February 2017 report assessing the potential effects of the E3 pulse component of HEMP on U.S. bulk-power transformers represents a portion of the work under this task. In this report, EPRI found that a small number of geographically-dispersed transformers (14 out of the tens of thousands included in EPRI’s analysis) were potentially at risk for thermal damage from the E3 pulse. EPRI produced a companion report assessing the potential impacts of the E3 pulse on the stability of the bulk-power system (i.e., the potential for voltage collapse) in December 2017 to be followed by the results of the first E1 pulse assessment at a later date. Mitigation, hardening, and recovery. Under this task, EPRI is assessing various mitigation and hardening approaches that can be employed to reduce the impacts of HEMP on bulk-power system reliability—including examinations of potential unintended consequences of these approaches and cost effectiveness. As an initial step, EPRI is developing interim guidance on hardening substations based on military and international standards that is scheduled to be completed by the third quarter of 2017. Risk-based decision support. For this task, EPRI is developing methodologies and tools to support risk-informed decisions regarding the implementation of HEMP hardening and mitigation measures. Trial implementation. Once hardening measures have been identified, EPRI’s supporting member utilities will have the opportunity to evaluate implementation of these measures on aspects of their networks. This task will develop a collection of leading industry practices with regards to HEMP mitigation and hardening. EPRI is to communicate the effectiveness of these measures including lessons learned. Project member and stakeholder communication. Under this task, EPRI will communicate the results of its research project to its supporting members and stakeholders in order to share new learning in a timely manner. U.S. and Canadian Electricity Suppliers Have Made Improvements to Provide Protection against GMD and HEMP Events, Including Some That NERC Requires and Monitors Selected Suppliers Have Made Technology Improvements Primarily to Enhance Overall Network Reliability which Can Also Help Protect Against GMD and HEMP Events Overall, 10 of the 13 selected suppliers we contacted reported making technological improvements to provide a range of system reliability benefits, some of which can also provide collateral benefits for protecting against GMD and HEMP events. These 10 suppliers purchased and maintained their own transmission-related equipment, while the remaining three suppliers were reliability coordinators who did not purchase or own their equipment. Various examples of these technological improvements for improved system reliability—that had the added benefit of protecting against GMD or HEMP events—were reported by the suppliers we contacted and include the following: Replacement of older transformers for various reasons, including susceptibility to GMD. Overall, 7 of the 13 suppliers we contacted noted that transformer replacement occurs for a variety of reasons, including increased efficiency. However, seven of the ten suppliers that purchased their own equipment added that, when they acquire new transformers, they generally selected models that have the added benefit of being more resilient to the effects of GIC during a GMD event. These seven suppliers reported that their specifications for the acquisition of new transformers specifically included qualities to make them more resilient to GIC. The suppliers also told us they are adhering to these specifications whenever they replace an older, less resilient, transformer as part of ongoing system upgrades. One supplier reported that they have undertaken a broad review of the transformers used in their system and taken steps to systematically reduce the number of unique units as part of a broader effort to make their system more consistent. They told us they have worked, to the extent possible, to standardize their transformer designs since implementing a new transformer purchasing program in 2008 which included upgrades such as more stringent specifications for protection against GMD. This supplier told us these efforts would also make it easier and less costly to maintain spares and to replace individual transformers that could be damaged from GMD or HEMP events. Participation in spare transformer programs to facilitate timely recovery of suppliers’ networks after transformer failures, including those caused by GMD and HEMP events. Of the 10 selected suppliers we contacted who purchased their own equipment, 6 reported having participated in at least one spare transformer program. For example, five of these suppliers participated in the Edison Electric Institute’s (EEI) Spare Transformer Equipment Program (STEP) which was intended as a coordinated approach to developing a shared inventory of spare transformers and streamlining the process of sharing transformers with affected companies. This program requires participating utilities to maintain a specific number of transformers up to 500 kV to be made available to other utilities in case of a critical substation failure. According to program documentation, any investor-owned, government-owned, or rural electric company in the U.S. or Canada may participate in the EEI STEP. The sixth supplier did not participate in an outside spare transformer program such as EEI’s, but, instead, maintained its own, in-house program. Investment in series capacitors to enhance network efficiency. Eight of the 10 selected suppliers we contacted, who purchased their own equipment, stated that they had added series capacitors to their networks. Seven of these eight suppliers told us they had acquired series capacitors to enhance the efficiency of their networks and help with network stability and voltage regulation. These suppliers stated that these devices offer the added benefit of mitigating the impacts of GMD and HEMP events because series capacitors block GIC, therefore preventing GIC from affecting certain parts of the transmission system. For example, one Canadian supplier, whose customers were almost totally dependent on electricity for heat during the winter, reported installing these technologies to improve overall network reliability but recognized the benefits of the technology for helping alleviate the threat of GMD events—which, according to DOE, is particularly acute at its far northern latitude. Installation of digital relays with enhanced functionality. Four of the 10 suppliers we contacted who acquired their own equipment had replaced, or were in the process of replacing, older electro- mechanical protective relays used in their grid control systems with newer digital relays. Unlike electro-mechanical relays—which can fail to operate properly under certain conditions resulting from a GMD event—digital relays can be programmed to properly respond to these conditions. FERC officials confirmed that digital relays may offer some degree of protection during GMD events, but cautioned that they are likely more susceptible than the older electro-mechanical relays to the E1 pulse of HEMP events. Construction of hardened control centers to protect against a variety of threats, including HEMP. Two of the 10 suppliers we contacted that purchased their own equipment had built, or were planning to build, control centers specifically designed to be resilient to the effects of EMP and other threats. For example, one electricity supplier’s customers included critical national security agencies and others in the Washington, D.C. area—resulting in the supplier’s desire to protect against the HEMP threat. The second supplier was in the process of designing its own hardened control center to guard against both EMP and other threats posed by extreme weather events occasionally occurring in its area of the country. In addition to technological improvements to provide a range of system reliability benefits, some suppliers are considering investments in technology specifically focused on blocking harmful GIC produced during GMD events. This GMD mitigation technology is referred to as a “GIC blocking device” and is still being tested. Since this technology is for the sole purpose of blocking GIC produced during GMD events, its cost may be directly attributed to GMD mitigation. One of our 13 selected electricity suppliers had installed such a prototype device on its high- voltage transmission system as part of an ongoing field trial to assess its performance and overall system impact in order to determine the effectiveness of the device under different operating conditions. Four selected suppliers expressed concern that GIC blocking devices can have unintended consequences on the stability or reliability of their transmission networks which could limit their overall benefits. Two of these suppliers stated that, before considering the installation of these blocking devices, they would perform analysis to determine their effectiveness in suppressing GIC at the system level and the impact on the functioning of their transmission system. Suppliers Have Developed Operating Procedures for the Initial GMD Reliability Standard and Recognize They May Need to Take Further Steps for the Next Standard NERC’s initial reliability standard EOP-010-1 requires certain suppliers to have GMD operating procedures to mitigate the potential effects of GMD events on the reliable operation of the transmission networks for which they are responsible. As of May 2017, the 13 suppliers we contacted told us they were all subject to the requirements of the EOP-010-1 standard and had GMD operating procedures in place to comply with the standard. Moreover, three of the 13 suppliers functioned as reliability coordinators and told us that all of the suppliers they oversaw in their territory also had operating procedures in place in accordance with EOP- 010-1. Officials with the reliability coordinators stated they reviewed their suppliers’ operating procedures to ensure they did not conflict with the procedures of other electricity suppliers in the coordinators’ geographic areas of responsibility. In addition, NERC’s Compliance Registry indicates that 188 electricity suppliers in the United States and Canada are potentially subject to the EOP-010-1 standard. NERC officials stated that, based on audit reports reviewed from its Regional Entities that included EOP-010-1, suppliers with transformers fitting the criteria specified in EOP-010-1 have developed the operating procedures required by the standard. NERC officials also stated that the EOP-010-1 standard requires electricity suppliers’ operating plans and procedures to mitigate the effects of GMD events on the reliable operation of the grid—as well as for the reliability coordinators to coordinate these plans and procedures within their area of responsibility. NERC officials stated that, as part of their compliance review for the standard, the NERC regions will assess the reasonableness of these plans and procedures. According to NERC, the standard provides the suppliers the flexibility to develop the procedures they think they need for their respective networks. NERC officials added that the quality of the measures put in place to address vulnerabilities to GMD would be further addressed under NERC’s second-stage GMD standard, TPL-007-1. NERC’s initial GMD-related reliability standard, EOP-010-1, went into effect in April 2015. NERC’s next reliability standard, TPL-007-1, includes requirements that will be phased in over a 5-year period from July 2017 to January 2022. The TPL-007-1 standard lists a total of seven requirements of which all but one are directed at planning coordinators and transmission planners whose planning area includes certain high- voltage transformers. In general, these requirements detail further steps suppliers must take to periodically model their networks and assess the vulnerable points of their networks to GMD. Depending on the vulnerabilities suppliers identify in conducting future assessments in accordance with TPL-007-1, suppliers will be required to develop corrective action plans, starting in January 2022, to ensure their generation or transmission networks meet certain performance requirements during a GMD event (e.g., no cascading blackouts). According to NERC, corrective actions in each plan may include (1) operational procedures, (2) enhanced training, (3) installation of devices (e.g., GIC blocking devices), (4) modification of devices (e.g., modifying equipment for greater GIC resilience), (5) removing vulnerable devices (e.g., old transformers), and (6) spare transformer programs. See appendix III for additional detail on TPL-007-1’s 7 requirements for certain electricity suppliers along with implementation dates for each. NERC Has a Process to Verify Compliance with Reliability Standards, Including Those Related to GMD NERC has an established process to verify electricity suppliers’ compliance with reliability standards, including EOP-010-1 and TPL-007- 1. Annually, NERC identifies and prioritizes risks based on the potential impact to reliability across its eight North American regions and the likelihood that such an impact might be realized. This process results in an annual compilation of risk elements for the coming year that are reflected in NERC’s implementation plan for compliance monitoring of reliability standards throughout its eight regions. In this implementation plan, NERC obtains input from the regions on risks inherent in their geographic areas of responsibility, and NERC links these areas of risk with specific reliability standards. For example, since becoming effective in 2015, NERC officials stated that the EOP-010-1 standard has been an annual area of focus in the implementation plan under the “extreme physical events” risk area. NERC’s overarching implementation plan provides a template for the regions to follow in developing their own regional implementation plans. NERC’s eight Regional Entities build on NERC’s guidance on risks facing all regions by assessing risks to the reliable operation of the bulk power system in their specific geographic areas of responsibility and identifying the reliability standards associated with those local areas of risk that they will focus on in their compliance monitoring efforts for the upcoming year. Further, according to NERC officials, each NERC Regional Entity performs individual risk assessments for each of the electricity suppliers in their areas of responsibility which further inform their approach to compliance monitoring for each of these suppliers—including which tools to use when assessing compliance. According to NERC, these individual risk assessments, along with the overarching and region- specific risks, inform the regions compliance monitoring oversight plan for each supplier. At the end of this planning process, NERC approves each region’s implementation and audit plans and submits the audit plans to FERC. As of August 2017, NERC’s regions had conducted 63 compliance audits of suppliers that included the EOP-010-1 reliability standard out of the total of 188 electricity suppliers potentially subject to the standard in the United States and Canada. According to NERC officials, the EOP-010-1 reliability standard went into effect in April 2015, and, as noted previously, NERC Regional Entities conduct compliance audits of individual suppliers—including those that must comply with EOP-010-1—at least once every 3 years. Therefore, due to this reason and the fact that these audits are just one of several options for NERC to consider in compliance monitoring, not every supplier subject to EOP-010-1 has been the subject of a compliance audit that included that standard in its scope as of the date of this report. NERC regions conducted these compliance audits on both reliability coordinators and transmission operators registered in the U.S. that were subject to EOP-010-1. As of September 2017, NERC had reported a total of two instances of non-compliance with the EOP-010-1 standard since its inception in April 2015. Electricity suppliers self-reported these two instances of non-compliance to NERC, and they were not the result of a compliance audit. NERC concluded that these incidents posed minimal risk to the reliability of the bulk power system. The two suppliers engaged in mitigation activities (e.g., training of personnel and modification of procedures) to address their non- compliance with the standard, which was verified by NERC’s Regional Entities. NERC concluded that no further action was needed in these two cases. Selected Suppliers Reported that Costs for Protecting against GMD and HEMP Events Have Been Relatively Small to Date, and Most U.S. Suppliers Are Expected to Be Able to Recover Costs through Charges to Electricity Customers Selected Suppliers Told Us Costs for Protecting against GMD and HEMP Events Have Been Relatively Small to Date, but Costs May Increase as Suppliers Comply with Future NERC Requirements Selected electricity suppliers told us the costs they have incurred to date for protecting against GMD and HEMP events have been small relative to their overall system costs. One supplier said that the costs they have incurred are generally associated with projects that provide broader system reliability or other benefits not specific to GMD or HEMP events. Based on interviews with selected suppliers, there are several types of projects that protect against GMD and HEMP events at different levels of costs: Projects providing collateral GMD or HEMP protection at no specific, incremental cost. As noted previously in this report, selected suppliers have installed several types of equipment for the purposes of transmission efficiency or benefits of general stability, and this equipment also provides collateral protection against GMD or HEMP events. This equipment has included series compensation systems installed on transmission lines, replacement of older electro- mechanical protective relays used in the suppliers’ grid control systems with newer digital relays, and acquisition of spare transformers or participation in shared spare transformer programs which improves their ability to quickly restore transmission systems from any cause, including GMD or HEMP events. Total project costs may vary widely depending on the amount and type of equipment suppliers choose to install, but according to suppliers we interviewed and information from transformer manufacturers, costs for this equipment can range from thousands of dollars per digital relay to tens of millions of dollars for a series compensation system. Projects providing supplemental GMD or HEMP protection at minimal added cost. As also noted previously in this report, some suppliers we interviewed said they have added specifications for improved protection against GMD or HEMP events as part of larger equipment procurement or construction projects and that this improved protection typically came at a relatively small increase in total project price. For example, several suppliers told us that transformers and other transmission equipment used to control voltage levels can be made more resistant to GIC by using certain designs or materials, and one supplier said this would increase equipment costs by 2 to 3 percent or less. In addition, the two suppliers we interviewed who have designed new control centers that are to be hardened against a range of hazards—including extreme weather (earthquakes, tornadoes, hurricanes, lightning), physical attacks, and HEMP events—told us that adding HEMP protection to the design of new control centers has increased total project costs from about 5 to approximately 20 percent. Projects built primarily for GMD or HEMP protection. As also noted previously in this report, one supplier has installed a prototype GIC-blocking device, designed specifically to protect against GMD events, as part of a pilot effort to test its operational impacts. The costs of deploying these devices are expected to be better understood after the pilot effort is completed, but based on its initial results, the supplier expects that the total cost for a well-designed GIC-blocking device would be at least $500,000, excluding installation and other costs and one device could be required to protect each transformer. Suppliers we interviewed told us they have also developed plans or procedures to mitigate for GMD. According to suppliers, in general these plans emphasize reducing the (1) level of power provided by individual power plants and (2) amount of power flowing over power lines to levels below their operating limits. For example, the plan for one coordinator—a grid operator—requires that they immediately take action to reduce the transfer of power down to GMD Operating Plan-designated limits; if these limits are approached or exceeded, selected power plants are directed to reduce the levels of power provided and, if necessary, the grid operator modifies the levels of power flowing through the system until designated transfer limits are reached. According to suppliers, lowering these power levels can reduce the temperatures of key equipment such as transformers and provide for greater flexibility to operate the system during an event. In some cases, such plans can require increased use of power plants that are more costly to operate, potentially increasing overall system costs. The costs of emergency operating procedures implemented in response to electromagnetic events are likely to vary considerably on a case-by-case basis, depending on such factors as the level of demand and the generation resources available during the event. In terms of customer costs, U.S. suppliers we interviewed said that the costs they have incurred for GMD or HEMP protection thus far would represent a negligible increase in rates paid by customers. For example, one supplier we interviewed serves about 4.5 million retail customers, and officials from that supplier estimated the cost of hardening a planned control center against HEMP to be at least $10 million. If this cost is fully passed on to customers and paid for in a single year, we calculated that it would amount to a total of about $2 for the average customer’s electric bill for that year. In the future, suppliers could face increased costs for protecting against GMD, depending on the corrective actions needed to address vulnerabilities, which suppliers are to identify in accordance with reliability standard TPL-007-1. The standard does not require suppliers to complete vulnerability assessments and develop corrective action plans until 2022, and suppliers told us it is too early to know what types of corrective actions may be required. However, the costs associated with some types of potential actions could be high. In particular, examples of potential corrective actions provided in the standard, such as installing new equipment or modifying existing equipment for improved GIC resilience, could be costly according to some suppliers we interviewed. For example, high-voltage transformers can cost tens of millions of dollars each. If suppliers identify multiple transformers that are vulnerable to thermal impacts from GIC flows, replacing or modifying them would be costly. Similarly, a supplier may need to install GIC-blocking devices throughout their network to effectively protect against a GMD event because the devices re-direct GIC flows elsewhere in the network. Therefore, a blocking device strategy could be costly if suppliers determine that large numbers of their transformers are vulnerable. Based on our prior review of federal efforts to enhance electric grid resiliency and federal emergency management programs, and interviews with agency and industry representatives, there are no sources of direct federal funding specifically to reimburse suppliers for costs they incur for protecting against GMD or HEMP events. DHS officials told us there are two DHS grant programs that could be used to indirectly support suppliers’ efforts to prepare for GMD or HEMP events. However, DHS directly awards these grants to state, local, or tribal governments, and DHS officials told us that it is rare for these grant funds to be passed through to private companies and they have no record of instances in which electricity suppliers received funding for grid preparedness efforts. Regulated U.S. Suppliers’ Costs for Protecting against GMD are Generally Recoverable, but Cost Recovery is Less Certain for HEMP Events Federal and State Regulators Have Made Specific Assurances about Recovering GMD-Related Costs At the federal level, in FERC’s September 22, 2016, order approving NERC’s TPL-007-1 reliability standard, FERC stated that cost recovery for prudent costs associated with or incurred to comply with the standard would be available to suppliers for whom FERC approves rates. Two suppliers we interviewed said that because FERC requires suppliers to comply with the standard and has provided specific assurance that prudent costs will be recoverable, they do not expect challenges recovering such costs. According to FERC officials, FERC determines whether suppliers’ investments are prudent on a case-specific basis, in part by considering whether the supplier acted reasonably given industry norms. FERC officials also stated that for most transmission rates, it does not conduct in-depth reviews of the reasonableness and prudence of each cost item unless a stakeholder such as a ratepayer advocacy group, large customer, or state public utility commission challenges the suppliers’ rate filing with FERC. FERC officials told us they were not aware of any cases in which stakeholders challenged GMD-related costs. Some suppliers we interviewed said that the revisions to TPL-007-1 that FERC required in Order 830—particularly, revisions to the benchmark GMD event suppliers must use in their vulnerability assessments—could result in added costs for suppliers. For instance, one supplier expressed concern that they could have to begin work to assess vulnerabilities and protect against the first version of the benchmark event, and that the revised standard would require them to re-do such work using a new benchmark event, at additional cost. In response to such concerns, FERC stated that it could not yet determine what impacts the revisions might have on the actions suppliers would have to take to comply, because NERC had not yet developed or proposed the revisions. However, FERC re-affirmed that cost recovery for prudent costs associated with or incurred to comply with reliability standard TPL-007-1, and future revisions to the standard, will be available to regulated suppliers. Representatives from the state regulators we interviewed said they allow recovery of prudent generation or distribution costs for regulated utilities for improvements needed to meet federally-required reliability standards, such as NERC’s GMD reliability standards. In addition, some of the selected suppliers told us that they use federally-required reliability standards to justify necessary investments when filing a rate case with state regulators. As with FERC, state regulators we interviewed said they determine the prudence and reasonableness of costs on a case-specific basis. Suppliers’ Ability to Recover Future HEMP-Related Costs is Uncertain Due to Limited Understanding of HEMP Risks and Mitigation Efforts To the extent suppliers and regulators determine that HEMP events pose a risk to bulk power system reliability, FERC may allow recovery of prudent costs for protecting against EMP events. However, according to FERC officials, determining prudence for costs associated with new, emerging areas such as HEMP mitigation could be challenging because regulators and suppliers have limited understanding of HEMP risks. In 2004, FERC publicly assured suppliers that it will allow for recovery of prudent costs necessary for ensuring the reliability of the bulk power system. Specifically, FERC issued a policy statement assuring public utilities that FERC will approve applications to recover prudently-incurred costs necessary to ensure bulk power system reliability, including prudent expenditures for compliance with good utility practices—practices engaged in or approved by a significant portion of the electricity industry or that could be expected to accomplish the desired result at a reasonable cost. Two suppliers we interviewed said that they expect FERC would allow them to recover transmission costs they deemed necessary for protecting against HEMP events. FERC officials told us that they are not aware of any cases to date where suppliers have sought recovery of transmission costs associated with HEMP protection through FERC-approved rates, so they do not know what challenges they might encounter in determining whether these costs are prudent. Also, unlike GMD events, suppliers and electricity industry stakeholders told us there are not yet tools for assessing suppliers’ vulnerability to HEMP events, standards for protecting against these events, or tools for assessing the effectiveness of protective remedies. Suppliers and state regulators we met with said more information is needed to understand HEMP risks and mitigation efforts in order to determine to what extent costs would be recoverable. Electricity industry stakeholders and suppliers told us that they are sensitive to the fact that their costs are typically borne by customers, and more complete knowledge of HEMP risks would allow them to invest responsibly in HEMP protection from both a reliability and cost perspective. Similarly, one state regulator we interviewed has not yet received any rate filings from suppliers that include costs associated with HEMP protection. However, one supplier said that their state regulators prioritize reliability, and they expect the regulators would allow recovery of costs for HEMP protection if suppliers determined such protection was needed. As with FERC, state regulators said that when rate filings involve new technologies or practices, there is more uncertainty regarding costs and benefits and it can be more difficult for regulators to determine prudency. For example, one state regulator told us that DHS is doing work to understand risks associated with HEMP events, and what protections such events may necessitate. The regulator said they would like to see the results of this work before suppliers invest in mitigation equipment, so there can be more certainty that the costs will be considered prudent. Independent Generators May Face Challenges Recovering Costs for Protecting Against GMD and EMP Events Independent generators—generators that sell power in wholesale electricity markets and are not part of an integrated utility—do not have a mechanism assuring cost recovery for reliability improvements, including such as GMD and HEMP protection. FERC officials stated that these generators sell electricity at prices determined by supply and demand in markets that FERC has determined are sufficiently competitive or that have adequate procedures in place to mitigate the effect of companies to manipulate prices, such as could be the case for a company with a large market share. As such, according to electricity industry and FERC officials, independent generators do not have the assurances of cost recovery that traditionally-regulated suppliers do. If they invest in protecting their facilities from the potential effects of GMD and HEMP, the prices independent generators obtain for selling electricity so as to be competitive in the wholesale markets may be too low to allow them to fully recover their costs. According to data from DOE’s Energy Information Administration, independent generators represented nearly 47 percent of electric generation facilities and generated about 39 percent of utility- scale electricity in the U.S. in 2015. FERC officials said they recognize that independent generators could face challenges recovering costs for step-up transformers—generator equipment which, if it is vulnerable to GMD, may need to be replaced or modified in accordance with NERC standard TPL-007-1. Independent generators must balance the need to recover costs associated with these transformers with the need to offer prices for their electricity that are competitive in wholesale markets. According to suppliers, until studies are completed to identify how companies will comply with TPL-007-1 it is unclear the extent of the risk to step-up transformers owned by independent generators and the extent of the challenges of paying for steps to mitigate those risks. Agency Comments We provided a draft of this report to DOE, DHS, NOAA, NRC, FERC, and NERC for their review and comment. DOE, DHS, NRC, FERC, and NERC provided technical comments, which we incorporated as appropriate. We are sending copies of this report to the Secretaries of Commerce, Energy, and Homeland Security, the Chairmen of the Federal Energy Regulatory Commission and the Nuclear Regulatory Commission, and the Chief Executive Officer of the North American Electric Reliability Corporation. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact Chris Currie at (404) 679-1875 or curriec@gao.gov or Frank Rusco at (202) 512-3841 or ruscof@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix VI. Appendix I: Scope and Methodology In conducting our work, we interviewed representatives from 13 of the 181 U.S. and Canadian electricity suppliers—entities that own or operate generation or transmission infrastructure—subject to the North American Electric Reliability Corporation’s (NERC) 2014 geomagnetic disturbance (GMD) reliability standard and which conduct planning and generation, transmission, and distribution operations. We selected these 13 electricity suppliers based on input from the U.S. Department of Energy (DOE), NERC, industry associations, and research institutions as to which suppliers had taken steps to prepare for and mitigate impacts from electromagnetic events. We also considered, among other things, the following supplier preparedness and mitigation actions and characteristics: (1) efforts or plans to install mitigation equipment or technology; (2) efforts or plans to develop specific mitigation processes, procedures, or other operational actions; (3) infrastructure, such as length and voltage of transmission lines; (4) high-voltage equipment, including transformers over 230 kilovolts (kV); (5) geomagnetic latitude; and (6) experience with GMD-related service disruptions. We included 3 Canadian electricity suppliers among the 13 suppliers we interviewed due to their (1) experiences with past geomagnetic disturbance (GMD) events, (2) research on the impacts of GMD, and (3) actions taken to prepare for and mitigate GMD events. We conducted site visits to 6 of the 13 suppliers to better understand their experiences with past GMD events and identify actions they have taken to prepare for and mitigate GMD and High-Altitude Electromagnetic Pulse (HEMP) events, among other things. During these visits we met with organization officials; observed operations and facilities, such as control centers hardened to mitigate effects from HEMP events; and viewed equipment potentially vulnerable to GMD, such as high-voltage transformers. While we cannot generalize the information we learned from these selected suppliers to all U.S. and Canadian suppliers, they provided insight on what electricity suppliers may know regarding the potential impacts of electromagnetic events on the electric grid, as well as steps suppliers may be taking to prepare for and mitigate such impacts. The selected U.S. suppliers also identified opportunities available to them for recovering costs for protecting against electromagnetic events. Based on input from DOE, NERC, supplier, and industry officials, and because of these organizations’ specialized knowledge and experience with the electricity industry, we also interviewed representatives from six industry organizations—five industry associations and one industry research organization—two transformer manufacturers, one software modelling company specializing in simulations of high-voltage power system operations, and one designer of a prototype geomagnetically induced current (GIC)-blocking device. To determine the extent to which U.S. and Canadian electricity suppliers have identified information about the effects of GMD and HEMP on the electric grid, we reviewed selected U.S. and Canadian government studies issued—or commissioned by—DHS, DOE, U.S. National Laboratories, Natural Resources Canada, the Federal Energy Regulatory Commission (FERC), and NERC since 2010 regarding, among other things, the vulnerability of transmission and generation infrastructure and equipment to GMD and HEMP events, possible measures to mitigate the effects of GMD and HEMP, and areas requiring further research. We also reviewed relevant studies published since 2010 from the Electric Power Research Institute (EPRI) and private contractors referred to us by government, supplier, and industry representatives. We identified these studies based on feedback from all entities listed above and through references in reports and other documentation. While we did not compile a comprehensive list of all studies of the effects of GMD and HEMP on the U.S. and Canadian electric grid, industry experts indicated that we had identified relevant studies published on this subject since 2010. We also interviewed knowledgeable officials from these U.S. and Canada government agencies, national laboratories, and industry organizations to clarify our understanding of the issues addressed in these studies. We assessed the methodologies used in the relevant reports and determined them to be sufficiently rigorous to provide information about the potential effects of GMD and HEMP events on the electric grid. To better understand the effects of solar weather on the electric grid, how GMD is measured, and mechanisms in place for notifying electricity suppliers of potentially dangerous solar storms, we interviewed representatives from the National Oceanic and Atmospheric Administration’s (NOAA) National Weather Service, the U.S. Geological Survey (USGS), and the National Aeronautics and Space Administration (NASA). We also reviewed relevant documentation on processes and procedures. -index) is a near real-time estimate of the official Planetary K-index maintained by the GFZ German Research Centre for Geosciences. events occurring from 1933 through 2016. We also interviewed Department of Homeland Security (DHS) officials regarding the Department’s efforts to address requirements in the National Defense Authorization Act for Fiscal Year 2017. To obtain perspectives on efforts individual electricity suppliers have taken to better understand the effects of GMD and HEMP, we interviewed officials from 13 U.S. and Canadian suppliers regarding the extent to which they had evaluated the impact of electromagnetic events on their specific generation systems or transmission networks and what they had learned from these evaluations. With respect to ongoing efforts to research the effects of HEMP, we reviewed DOE and EPRI’s Joint Electromagnetic Pulse Resilience Strategy and the U.S. Department of Energy Electromagnetic Pulse Resilience Action Plan and interviewed relevant DOE and EPRI officials regarding these plans. Further, we interviewed officials from various national laboratories regarding their ongoing efforts to fully investigate and evaluate how an electric utility could protect itself from, or mitigate the effects of, HEMP on its systems. We also interviewed officials from the Nuclear Regulatory Commission (NRC) regarding efforts to assess the ability of a nuclear power plant to achieve safe shut down following a GMD or EMP event and the extent to which plants are required to implement strategies or guidelines in the event of a prolonged loss of offsite power, similar to what could be caused by a GMD or EMP event. Finally, we reviewed the 2008 Commission to Assess the Threat to the United States from Electromagnetic Pulse Attack (EMP Commission) report with recommendations on preparing for and recovering from a possible EMP attack. In October 2017, we also requested an interview with a representative from the EMP Commission but did not receive a response to our requests. To identify steps selected U.S. and Canadian electricity suppliers have taken to protect against GMD and HEMP events and understand how NERC has monitored these efforts, we reviewed FERC orders and NERC reliability standards that require certain suppliers to take steps to assess and prepare for GMD impacts. We interviewed FERC and NERC officials to discuss these standards and reviewed public comments submitted by stakeholders during the FERC rulemaking process. We also interviewed officials from 13 U.S. and Canadian electricity suppliers to identify steps they had taken to comply with NERC reliability standards as well as any additional actions to prepare for and mitigate potential GMD and HEMP effects, such as replacement of older equipment or investment in spare transformer programs. Additionally, we reviewed relevant federal guidance on preparing for GMD and HEMP events, such as DHS’s Electromagnetic Pulse protection guidelines and NERC’s Geomagnetic Disturbance Planning Guide. To identify the extent to which NERC has monitored electricity suppliers’ steps to comply with NERC reliability standard EOP-010-1, we reviewed NERC monitoring processes, including procedures for developing an annual, nationwide implementation plan for conducting monitoring activities. NERC officials provided the number of compliance audits conducted between April 2015—when NERC, through Regional Entities to which it has delegated enforcement authority, first began reviewing suppliers for compliance with EOP-010-1—and August 2017 that included the EOP-010-1 reliability standard. We contrasted the number of compliance audits with the total number of suppliers potentially subject to NERC’s GMD reliability standard EOP-010-1. We assessed the reliability of the data on the total number of suppliers subject to EOP-010-1 by interviewing agency officials regarding data sources, system controls, and any quality assurance steps performed by officials before the data were provided; we found the data to be sufficiently reliable to provide the number of suppliers subject to EOP-010-1 since it went into effect. We also discussed with cognizant NERC officials the organization’s processes for collecting and reporting comprehensive data on the status of their overall compliance monitoring efforts. To identify what opportunities exist for U.S. electricity suppliers to recover costs for protecting against GMD and HEMP events, we reviewed FERC regulations and orders related to cost recovery, such as suppliers’ costs for spare transmission equipment services. We also interviewed FERC officials and representatives of selected state regulators whose jurisdictions include suppliers we interviewed, regarding procedures available to electricity suppliers to recover costs for actions taken to prepare for and mitigate GMD and HEMP effects. We asked these officials to discuss previous, current, and potential future regulatory actions—orders or rate cases they have overseen—involving recovery of costs for actions taken to protect against GMD and HEMP events. Further, we interviewed cognizant DHS and DOE officials to identify the extent to which financial incentives—such as preparedness grants—are available to U.S. electricity suppliers to offset the costs of preparation and mitigation efforts. As part of our review of actions taken by ten selected U.S. electricity suppliers to prepare for and mitigate the impact of electromagnetic events, we interviewed officials regarding the extent to which they had recovered costs expended on preparedness and mitigation efforts and what, if any, options they were considering to recover such costs in the future. While the information provided by these selected electricity suppliers is not generalizable to the U.S. industry, it illustrates examples of actions selected suppliers have taken to recover costs for GMD and HEMP mitigation and preparedness efforts. In addition, we interviewed representatives from various trade associations to identify challenges suppliers face in recovering costs for mitigation and preparedness efforts. We conducted this performance audit from May 2016 to February 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: National Oceanic and Atmospheric Administration (NOAA) Notifications for Geomagnetic Disturbances In the United States, the National Oceanic and Atmospheric Administration’s (NOAA) National Weather Service manages the Space Weather Prediction Center (SWPC), which is responsible for monitoring and providing services on space weather, including geomagnetic storms. SWPC uses a variety of ground and space-based sensors, as well as imaging systems, to view and estimate geomagnetic activity around the world, and to issue Watches, Warnings, and Alerts for geomagnetic storms through e-mail and website postings to those who are impacted by space weather. Additionally, in the event of imminent geomagnetic storms, SWPC issues immediate voice notification and confirmation to all North American Electric Reliability Corporation (NERC) reliability coordinators through a special hotline. is a “range index”—a measure of variation that saturates at K= 9—Dst is an unbounded measure of solar storm effects on the Earth’s magnetic field. In this report, we use the K- index, the history of which spans three solar cycles more than the Dst-index. = 9 (extreme)—to determine whether geomagnetic alerts and warnings should be issued. SWPC’s primary notifications include: Watches: Watch forecasts for impending geomagnetic storms— coronal mass ejections (CME)—are issued when the highest predicted Kp-index for a day is between K = 5 or higher and are posted approximately 1 to 2 days before a storm reaches Earth. According to SWPC, Watch forecasts are less reliable in predicting storm intensity and timing than other types of forecasts, but are considered useful for longer-range notification. Watch forecasts are based primarily on space and ground-based solar observations as well as modelling predictions. Warnings: Warnings of geomagnetic storms are issued when estimated K-indices of K = 4 or higher are expected; they are generally issued 20 to 40 minutes in advance and are based on real- time observations of the solar wind conditions affecting earth. SWPC considers Warning notices as more reliable than Watch forecasts in terms of measuring storm intensity and timing. -index to determine the G-scale level (G1 through G5), in which “K= 5” corresponds to “G1” and “K= 9” corresponds to “G5.” A K of 0 to 4 is below storm levels and is labeled as “G0.” For purposes of consistency, we use the K- index in this report. Alerts: Alerts are near real-time indications that a specific storm threshold—K = 4 or above—is reached; they are based on SWPC’s minute-by-minute estimate of GMD activity. Alerts are derived from ground-based magnetometer observations from eight locations around the world. According to SWPC, Watches, Warnings, and Alerts are to be issued as activity occurs and therefore can be issued very frequently during high- activity intervals and not at all during quiet periods. SWPC issues these notifications when storm levels reach a specific estimated K level. Table 1 shows the estimated K-indices that trigger each SWPC notification product as well as the estimated impacts to the electrical power system. Appendix III: North American Electric Reliability Corporation (NERC) Geomagnetic Disturbance (GMD) Reliability Standards In May 2013, the Federal Energy Regulatory Commission (FERC) directed NERC to develop reliability standards requiring electricity suppliers to address the potential impact of GMD on the reliable operation of the U.S. bulk power system. In June 2014, FERC approved standard EOP-010-1, submitted by NERC, requiring that certain suppliers prepare for the effects of GMD events by developing contingency operating plans, procedures, and processes. FERC approved a second standard—TPL- 007-1—in September 2016, also submitted by NERC, requiring certain suppliers to assess the vulnerability of their transmission systems to GMD events; suppliers that do not meet certain performance requirements must develop a plan to achieve the performance requirements. Table 2 summarizes the specific requirements in NERC’s stage 1—EOP-010-1— and stage 2—TPL-007-1—standards, the electricity industry entities responsible for them, and their effective dates for the requirements. Appendix IV: Select Government and Industry Studies on Electromagnetic Events An electromagnetic event can result from a naturally occurring, large- scale geomagnetic disturbance (GMD), caused by severe solar weather, or from human-made sources, such as the high-altitude detonation of a nuclear device to create a high-altitude electromagnetic pulse (HEMP). Table 3 provides details on a select number of geomagnetic-related studies performed since 2010 with respect to their objectives, findings, and recommendations. These studies include details on (1) areas of vulnerability for the grid with respect to GMD events, (2) potential impact on the grid from these events, (3) possible mitigation measures, and (4) areas needing further research. For example, as shown in the table, the North American Electric Reliability Corporation’s (NERC) and the Department of Energy’s (DOE) June 2010 report included a plan to form a task force of government and industry efforts to examine GMD. This resulted in the formation of the NERC GMD Task Force, consisting of government, industry, and academic experts, to examine the GMD threat to the nation’s power grid. The task force’s work in evaluating the potential impact of GMD events resulted in NERC’s subsequent February 2012 report (also shown in table 3) which outlines its plans for working with industry on new reliability standards for GMD events, among other things. As a result of this work, and as directed by the Federal Energy Regulatory Commission (FERC), NERC developed the EOP-010-1 and TPL-007-1 GMD reliability standards. Also as a result of this work, NERC issued a GMD Planning Guide for electricity suppliers, which assists the suppliers in carrying out studies to assess the effects of GMD on their individual networks. Table 4 provides details on a select number of unclassified HEMP-related studies performed since 2010 with respect to their objectives, findings, and recommendations. These studies include details on (1) areas of vulnerability for the grid with respect to HEMP events, (2) potential impact on the grid from these events with respect to all three HEMP pulses (E1, E2, and E3), (3) possible mitigation measures, and (4) areas needing further research. Appendix V: Details on the U.S. Department of Energy’s (DOE) Electromagnetic Pulse (EMP) Resilience Action Plan DOE’s EMP Action Plan (DOE Action Plan), issued January 2017, describes 19 actions to be taken by September 30, 2021, to enhance the resilience of the electric power grid to high-altitude electromagnetic pulse (HEMP) effects. DOE stated that its Action Plan considers the over 90 recommendations made in the 2008 Commission to Assess the Threat of the United States from Electromagnetic Pulse (EMP) Attack (EMP Commission) report and at least partially addresses 10 of the 15 recommendations directly related to the electric power system made by the EMP Commission in their report. See table 5 for these 10 EMP Commission recommendations from 2008 and corresponding components of DOE’s 2017 Action Plan. As of November 2017, based on our review of implementation dates for specific actions in DOE’s plan, the agency had yet to complete 15 of the 19 actions detailed in the Action Plan but had initiated efforts under the plan to identify gaps in HEMP knowledge and coordinate government and industry information sharing with the electricity sector and other critical industry sectors. Future work DOE expects to address under the plan will include (1) evaluating existing models used to estimate EMP impacts to the grid, (2) the adequacy of backup power generation in the wake of an EMP event, (3) establishing a national capability for conducting EMP testing of existing grid components, (4) identifying and evaluating mitigation and protection measures for various grid components, and (5) assessing the feasibility of testing different hardening techniques for substations for EMP scenarios. The DOE Action Plan includes deliverables and due dates for the 19 action items detailed in the plan which, according to DOE, are subject to the availability of necessary funding. See table 6 for details on these deliverables, and associated dates, for each action item. Appendix VI: GAO Contacts and Staff Acknowledgments GAO Contacts Staff Acknowledgments In addition to the individuals named above, Jon Ludwigson (Acting Director), Ben Atwater (Assistant Director), and Barbara Guffy (Analyst-in- Charge) managed this assignment. Frederick K. Childers, Jonathan Felbinger, Daniel Friess, Alexandra D. Gebhard, Michael Harmond, Eric Hauswirth, Richard Hung, Miles Ingram, and Heidi Nielson made key contributions to this report.

Why GAO Did This Study A severe GMD or HEMP event could potentially have significant impacts— including power outages—on the nation's electric grid, which could affect other sectors that depend on electricity, such as communications. In response, NERC created two regulatory standards requiring certain U.S. and Canadian suppliers to assess their vulnerability to GMD and take appropriate steps in response. GAO was asked to review electricity industry actions to prepare for and mitigate electromagnetic risks. This report examines, among other things, (1) to what extent U.S. and Canadian electricity suppliers have identified information about GMD and HEMP effects on the grid, (2) what steps selected U.S. and Canadian suppliers have taken to protect against GMD and HEMP, and (3) what opportunities exist for U.S. suppliers to recover costs for protecting against GMD and HEMP. GAO examined government and industry studies and interviewed federal and industry officials about potential GMD and HEMP effects on grid infrastructure; reviewed regulatory standards, monitoring processes, and NERC compliance audit data from April 2015 through August 2017; reviewed federal regulations and interviewed state regulators on cost recovery issues; and interviewed officials from a nongeneralizable sample of 13 U.S. and Canada electricity suppliers, selected based on factors such as GMD experience and preparation for GMD and HEMP events. GAO provided a draft of this report to five federal agencies and NERC. Technical comments provided were incorporated as appropriate. What GAO Found U.S. and Canadian electricity suppliers—electricity generation and transmission owners and operators—have identified information on the potential effects of a severe geomagnetic disturbance (GMD), resulting from a solar storm, but have identified less information about the potential effects of a high-altitude electromagnetic pulse (HEMP), resulting from the detonation of a nuclear device, on the electric grid. There is general agreement that more research is needed on both GMD and HEMP. Government and industry have publicly reported on the potential impacts of GMD on the grid. For example, one study identified two main risks: (1) potential voltage instability, causing power system collapse and blackouts; and (2) possible damage to key system components. However, these studies do not address the unique aspects of individual suppliers' networks. Recognizing this, 11 of the 13 selected suppliers GAO contacted said they had assessed their network vulnerability; of these 11, 6 expected GMD effects to be relatively small. In contrast, Department of Energy (DOE) and industry officials told GAO that information on HEMP effects is limited in that suppliers lack key information to fully understand HEMP effects on their networks. Historically, study of HEMP effects focused on impacts to military equipment rather than the commercial electric grid. Recently, DOE and industry began research to better understand HEMP effects. Of the 11 suppliers who responded to GAO about their HEMP efforts, 3 reported having studied the impact of HEMP on their networks and 2 of the 11 had integrated, or planned to integrate, HEMP-resistant features into new control centers. Of the 13 selected suppliers GAO contacted, 10 reported making technological and operational improvements to enhance overall network reliability that also provided some protection against GMD and HEMP risks. For example, suppliers reported making technological improvements such as replacement of some older transformers and unprotected control centers. As of May 2017, all 13 suppliers stated they had complied with a GMD regulatory standard issued by the North American Electric Reliability Corporation (NERC)—the federally designated regulatory authority responsible for developing and enforcing reliability standards–-to develop operating procedures to mitigate GMD effects. A second regulatory standard—which is to be implemented in phases through 2022—will generally require suppliers to further assess their vulnerability to GMD. Selected U.S. suppliers told GAO that costs they have incurred to protect against GMD and HEMP have been relatively small so far and they expect to recover those costs through customer rates. Suppliers could face future increased costs depending on corrective actions needed to comply with the second GMD regulatory standard. Federal and state regulators indicated that regulated U.S. suppliers' costs for protecting against GMD are generally recoverable through customer rates, but recovery is less certain for protection against HEMP because less is known about HEMP risks. Further, some suppliers could face challenges to cost recovery. Specifically, independent owners of power plants—those that sell power in wholesale electricity markets and are not part of an integrated utility—must recover reliability improvement costs through their sales of electricity and are not assured of cost recovery; federal regulators told GAO they are aware this could be a challenge for these independent owners.

gao_GAO-18-101

gao_GAO-18-101_0

Background MILCON Appropriation and Obligation Process DOD’s MILCON appropriations are used to fund the acquisition, construction, installation, and equipping of temporary or permanent public works, military installations, facilities, and real property needed to support U.S. military forces in the United States and overseas. As with other DOD activities, no funds may be appropriated in any fiscal year or obligated or expended for MILCON activities unless such funds have been specifically authorized by law. Each year, the National Defense Authorization Act authorizes amounts to be appropriated in each of the 18 programmatic MILCON appropriations accounts. Individual or conference committee reports accompanying each fiscal year’s National Defense Authorization Act provide specific congressional direction on authorized funding levels designated for specific construction projects supported by the various MILCON accounts. Similarly, conference committee reports or explanatory statements accompanying each fiscal year’s appropriations acts establish appropriated funding levels for MILCON projects. The process through which the active component requests funding for construction projects is supported by DOD’s Form 1391 Military Construction Project Data (Form 1391). The Form 1391 is to be used to support each project proposed for inclusion in the MILCON appropriations request submitted concurrently with all other DOD appropriations requests annually. The forms are to be used for both new projects as well as urgent unforeseen projects. The Form 1391 describes the scope, total project costs, and estimates of specific project elements. Costs associated with other project elements such as contingency and supervision, inspection, and design are also to be captured and included in the total requested amount. Finally, the Form 1391 is to include a description of the proposed construction and a requirements statement indicating what requirement the project provides. Project budget estimates are initially developed at the installation level and are provided to the next responsible level for review, validation, refinement, prioritization, and approval. Administrative support is to be provided when requested across the departments, but ultimately the installation is the originator and the primary responsible entity in developing the completed Form 1391. MILCON appropriations are generally available for obligation for 5 fiscal years, at which time the appropriation expires. For 5 years after they expire, appropriations are available for limited purposes, such as liquidating obligations made during the period of availability or adjusting contract costs. After these 5 years, any remaining unexpended amounts, whether obligated or unobligated, are canceled and returned to the U.S. Treasury. Once funds are returned to the U.S. Treasury, they are no longer available for any purposes. DOD obligates its appropriations throughout the period in which the appropriation is available. An “unobligated balance” is the difference between the total appropriation amount and total obligations made against the appropriated amounts. An “unexpended balance” is the total of obligated but unliquidated and unobligated amounts. According to DOD officials, available but unobligated amounts no longer needed may be either rescinded by Congress or reprogrammed to other MILCON projects that the active component identifies as needing additional funding. Reprogrammed amounts may be used to fund other projects where there are shortfalls; for projects authorized by Congress but not specifically funded through the appropriations process; for emergency projects, such as for facilities destroyed by fires. DOD’s flexibility to reprogram without congressional approval is limited by the amount to be reprogrammed to a particular project. DOD’s Financial Management Regulation requires prior congressional approval for a reprogramming that would result in an increase exceeding 25 percent of a project’s authorized base amount or $2 million, whichever is less. Prior approval is not required when established costs or project-related thresholds are not reached. According to DOD officials, reprogrammings requiring congressional approval are called “above-threshold reprogrammings” and those that do not are called “below-threshold reprogrammings.” DOD Construction Agents DOD designates construction agents for the military departments and defense agencies with primary responsibility for developing and refining MILCON proposals and cost estimates, and to manage the design and construction of projects. Typically, the Army Corps of Engineers is the construction agent for Army MILCON-funded projects and the Naval Facilities Engineering Command is the construction agent for Navy and Marine Corps MILCON-funded projects. Either of those DOD entities can be the construction agent for the defense agencies and activities, such as for the Missile Defense Agency or Defense Education Activity, with the approval of the military department having jurisdiction of the real property facility. However, both the Army and the Navy may use each other’s construction agent if it is in the interest of efficiency and cost- effectiveness or when otherwise considered appropriate. The Air Force may use either the Army Corps of Engineers or Naval Facilities Engineering Command for its projects. Additionally, the Air Force Civil Engineer Center, although not a designated construction agent, reviews and approves requirements for Air Force MILCON cost estimates, and in some cases may design and construct Air Force projects where both the Air Force and the commander of the assigned construction agent agree that it is the most efficient, expeditious, and cost-effective means to complete the project. DOD Guidance for MILCON Within DOD there are two levels of military construction guidance: the Unified Facilities Criteria and component-level guidance. The Unified Facilities Criteria are overarching, DOD-wide technical manuals and standards used for planning, design, construction, restoration, and maintenance of DOD facility projects. The Unified Facilities Criteria was designed to standardize and streamline the process for developing, maintaining, and disseminating criteria in support of MILCON. The Unified Facilities Criteria contains guidance describing methods, procedures, and formats for the preparation of construction cost estimates and construction contract modification estimates, among other types of guidance. The Unified Facilities Criteria is to be used to the greatest extent possible by all the DOD regardless of funding source. In addition to the Unified Facilities Criteria, the military departments and agencies have also developed their own internal guidance on MILCON, providing further direction on conducting activities such as cost analysis and determining facility requirements. Our Cost Assessment Model We developed the GAO Cost Estimating and Assessment Guide: Best Practices for Developing and Managing Capital Program Costs (Cost Guide) to assist federal agencies in developing reliable cost estimates and also as a tool for evaluating existing cost estimating procedures. To develop the Cost Guide, our cost experts assessed measures applied by cost estimating organizations throughout the federal government and industry and considered best practices for the development of reliable cost estimates. While the Cost Guide has a focus on developing cost estimates in the context of government acquisition programs, it outlines best practices that are generally applicable to cost estimation in a variety of circumstances. These best practices can be used to assess (1) the specific project cost estimates an agency develops to determine whether they meet the four characteristics—comprehensive, well-documented, accurate, and credible—for being reliable and (2) an agency’s cost estimating guidance and procedures to see how well they incorporate all the steps needed for producing a high-quality cost estimate. Figure 1 shows the four characteristics and associated best practices for each that define a reliable cost estimate and table 1 shows the 12 steps identified in the Cost Guide that, if followed correctly, should result in high-quality cost estimates that management can use for making informed decisions. The Active Component Obligated and Expended Most of Its Military Construction Appropriations Received during Fiscal Years 2005 through 2016 During fiscal years 2005 through 2016, Congress appropriated about $66 billion in MILCON funds to the active component and, as of September 30, 2016, the active component had obligated all but about $5.1 billion and expended all but about $11 billion of those funds. Of the $5.1 billion that remains unobligated, about $4.6 billion was unexpired and available for new obligations (i.e., from fiscal year 2013 through 2016 appropriations). Table 2 shows the active component’s combined MILCON appropriations, obligations, and unexpended funds from fiscal year 2005 through fiscal year 2016. In general, during the early first few years of a MILCON appropriation available for 5 years, it is often likely that most of the funds will remain unobligated. For example, as shown in table 2 above, of the nearly $3.9 billion appropriated for military construction for the active component from the fiscal year 2016-2020 appropriation, only about $1.1 billion had been obligated as of September 30, 2016. This is not surprising given the time that it takes to award, obligate and disburse funds for projects. Ultimately, though, as an appropriation nears its expiration date, all or nearly all of the amounts have generally been obligated. In fact, as shown in table 2, for each MILCON appropriation received by the active component prior to fiscal year 2013 (fiscal years 2005 through 2012), less than 2 percent of each year’s appropriation was unexpended as of September 30, 2016. In appendix II, we provide additional analysis of the active component’s unexpended and unobligated balances, by appropriation year and by military department. Although ultimately, the active component obligates and expends most of its MILCON appropriations, the active component can experience a wide range of differences between initial cost estimates and final costs during the execution of individual MILCON projects, resulting in savings or shortfalls depending on the project. For example, we found that from fiscal year 2010 through fiscal year 2016, the active component achieved about $4.2 billion in MILCON project savings as a result, for example, of canceled projects, projects with lower than expected contractor bids, or the use of less expensive building materials. In appendix III, we provide additional analysis of the active component’s estimated initial costs and the contract award amounts that were funded by MILCON appropriations for fiscal year 2010 through fiscal year 2016. The Active Component Reprogrammed Hundreds of Millions of Dollars in Military Construction Appropriations in Fiscal Years 2010- 2016 The active component reprogrammed about $1.6 billion in MILCON appropriations to fund shortfalls caused by emergency projects, projects that were authorized but did not receive specific appropriations, and projects needing additional funding in fiscal years 2010 through 2016. Of this amount, the Army reprogrammed about $789 million of about $14 billion in appropriated MILCON funds; the Navy, about $535 million of about $14 billion in appropriated MILCON funds; and the Air Force, about $295 million of about $7 billion in appropriated MILCON funds. Table 3 shows the number and amounts of above-threshold reprogrammings by the active component for fiscal years 2010 through 2016. As seen in table 3, for any given year there are typically hundreds of millions of dollars reprogrammed. There are generally multiple active or canceled projects that result in cost savings, which may be used to fund authorized but not specifically funded projects. Below are three examples where the active component funded MILCON projects with amounts reprogrammed from other projects: Repair Shop at Andersen Air Force Base, Guam: This is an Air Force project to construct a pacific air resiliency low observable/corrosion control/composite repair shop in Guam. It is an authorized project that did not receive specific funding during the appropriation process but was fully funded by reprogrammed cost savings from active construction projects. Congress authorized $34.4 million for the repair shop in fiscal year 2015; however, no funds were specifically appropriated for the project. According to Air Force officials, since this was their top unfunded military construction priority, they used $34.4 million in savings achieved from other projects to construct the repair shop. Table 4 lists the three projects whose MILCON funds were reprogrammed for the repair shop at Andersen Air Force Base in Guam. Training Facility at the Naval Air Station at Mayport, Florida: This is a Navy project to construct a littoral combat ship training facility in Florida. It is a specifically funded project requiring additional funds that received reprogrammed amounts from a canceled project. In fiscal year 2014, the initial cost as listed on the Form 1391 was estimated to be $20.5 million, but project costs increased by 41 percent to an estimated $28.9 million, according to a fiscal year 2016 reprogramming request to Congress. As detailed in the reprogramming request, the Navy attributed the increased cost to underestimated mission simulator and communication line requirements. To fund the increased costs, the Navy used $8.3 million in savings from a canceled project to complete the facility. Table 5 lists the canceled project that resulted in funds being reprogrammed for the training facility at Mayport. Barracks at Presidio of Monterey, California: This is an Army project to construct a trainee barracks in California. It is a specifically funded project in need of additional funds that received reprogrammed amounts from active and canceled construction projects. In fiscal year 2011, the initial cost for the project as listed on the Form 1391 was estimated to be $63 million, but project costs increased by 51 percent to $95 million, according to a fiscal year 2015 reprogramming request to Congress. As detailed in the reprogramming request, the Army attributed the increased costs to a 3-year delay in construction and the need to move the project to a small, steep-terrain site. The reprogramming request further noted that the delay in construction was due to the discovery at the proposed construction site of a seismic fault and a plant that is an endangered species. To fund the increased costs, the Army sought to reprogram funds from the savings achieved from the active and canceled projects. Table 6 lists the projects that generated the reprogrammed funds used for the barracks at Presidio. DOD’s Cost Estimates for Selected Construction Projects Were Not Reliable and DOD’s Guidance Does Not Fully Incorporate the Steps Needed for Developing Reliable Estimates Our analyses of the cost estimates for three selected projects shows that the cost estimates were not reliable, and DOD’s cost estimating guidance does not fully incorporate all the steps needed for producing reliable estimates. We examined the cost estimates of three high-value military construction projects and noted that the initial cost estimates increased for all three projects, with cost estimates for two of the projects increasing by over 30 percent and the other, by about 7 percent. Specifically: Strategic Command Operations Building, Offutt Air Force Base, Nebraska. The project to construct a nuclear, space, and network command and control operations building for the command at Offutt Air Force Base, Nebraska, increased from an initial cost estimate in fiscal year 2012 of $564 million to $601 million in fiscal year 2014 (or a 7-percent increase). According to a fiscal year 2014 reprogramming request to Congress, the Air Force attributed the increased cost to the fact that the project team did not appreciate the full scope, complexity, and risk of such an information technology- intensive project. These cost issues are similar to challenges we have reported on for other information technology-intensive MILCON projects. The Air Force is the project owner and the Army Corps of Engineers is the construction agent for this project. Command Headquarters and Cyberspace Operations Building, Fort Meade, Maryland. The project to construct a command headquarters and cyberspace operations building with sensitive compartmented information facility in Fort Meade, Maryland, increased from an initial cost estimate in fiscal year 2013 of $84 million to $110 million in fiscal year 2015 (or a 31-percent increase). As detailed in the fiscal year 2015 reprogramming request, the Navy attributed the increased cost to higher than expected construction costs due to increased demand on the labor workforce in the Washington, D.C./Baltimore area and underestimated electrical power requirements. The Navy is the project owner and the Army Corps of Engineers is the construction agent for this project. Elementary School Camp Foster, Japan. The project to replace an elementary school at Camp Foster, Japan increased from an initial cost estimate in fiscal year 2012 of $79 million to $107 million in fiscal year 2014 (or a 35-percent increase). As detailed on the fiscal year 2014 reprogramming request, the Department of Defense Education Activity attributed the increased cost to the volatile construction climate in Japan caused by natural disasters; Japanese government policies, economic stimulus, and reform; and the planned developments for the 2020 Tokyo Olympic Games. Although this project is not owned by any of the military departments, it is being managed by the Army Corps of Engineers in its role as a DOD construction agent through which it plays an important role in the development of the construction cost estimate. The Department of Defense Education Activity is the project owner and the Army Corps of Engineers is the construction agent. To determine the reliability of the cost estimates for these three selected projects, we assessed the cost estimates against the best practices for developing a reliable estimate in our Cost Guide. As previously discussed, the Cost Guide defines the four characteristics— comprehensive, well documented, accurate, and credible—of a reliable cost estimate and the associated best practices related to each characteristic. In conducting these assessments, we examined both the Form 1391 estimate (i.e., the estimate used to develop the budget) and the independent government estimate i.e., (the estimate used to award the contract) for each project. Our analysis of the cost estimates for the three selected projects shows that the cost estimators did not follow all the best practices listed for each of the four characteristics. As a result, none of the characteristics were fully or substantially met. To be reliable, a cost estimate must substantially or fully meet each of the four characteristics. As the Cost Guide states, if any of the characteristics are not met, minimally met, or partially met, then the cost estimate does not fully reflect the characteristics of a high-quality estimate and cannot be considered reliable. Table 7 provides the results of our assessment of the cost estimates for each of the three selected projects. The Cost Guide also identifies 12 steps that, when incorporated into an agency’s cost estimating procedures and guidance, are more likely to result in reliable and valid cost estimates. However, our analysis of DOD’s department-wide cost estimating guidance—the Unified Facilities Criteria—found that the criteria did not include all of these 12 steps. The Unified Facilities Criteria incorporates some of the 12 steps to some degree, but not others, and as a result DOD is at a greater risk of developing estimates that are not reliable. Table 8 provides our assessment of the extent to which DOD’s Unified Facilities Criteria incorporates the 12 steps needed to develop a high-quality, reliable cost estimate. Each of the military departments is required to follow the Unified Facilities Criteria to the greatest extent possible when designing and constructing facilities. However, as shown by the table above, there are shortcomings in these criteria when compared with our Cost Guide. Despite these shortcomings, the military departments have gone beyond the Unified Facilities Criteria and developed their own guidance that more closely aligns with our Cost Guide. For example, for both the “determining the estimating structure” and “obtain the data” steps, we found that all three military departments had developed their own guidance that more closely aligned with the 12 steps than the Unified Criteria did. In addition, some military departments are also making improvements to their cost estimating processes, but these improvements have not been fully implemented yet. For example, the Air Force Civil Engineer Center is implementing a cost estimate improvement plan to include the training of nearly 700 airmen and has conducted a study that directly ties the 12 steps in the Cost Guide to the associated tasks to be completed by the Air Force cost estimator to meet each individual step. However, the actions contained in the cost improvement plan have not been fully implemented and still remain in the concept phase. Similarly, although the Army Corps of Engineers is investigating expanding the use the of cost and schedule risk analysis—which could align with the best practices in the Cost Guide—that the Army currently conducts for selected civil work construction projects to its high-cost military construction projects, the Army has not formally required the use of these tools. In appendix IV, we describe the guidance the military departments have developed beyond the Unified Facilities Criteria. The Cost Guide is designed to establish a consistent methodology that is based on best practices and that can be used across the federal government for developing, managing, and evaluating capital program cost estimates. Air Force and Army Corps of Engineers officials noted that there may be instances in which following all the 12 steps of the Cost Guide for every MILCON project would not be appropriate to the risk level of the project. For example, it may not be realistic or to the military departments’ benefit for the military departments to conduct a sensitivity and uncertainty analysis or develop an independent cost estimate for all the construction projects they initiate every year, especially for low-cost projects. We agree that it may not be suitable to fully apply all 12 of the cost estimating steps in the Cost Guide to all MILCON projects. However, incorporating the 12 steps into the Unified Facilities Criteria would establish consistency across DOD in the cost estimating process by ensuring that, for each MILCON project, each step in the Cost Guide would at least be considered. Furthermore, DOD could choose to establish thresholds—based on, for example, the dollar values of the projects—to guide the services in implementing the 12 steps for the most valuable projects. Skipping or not considering any step of the 12-step cost estimating process, especially for high-value projects such as those in our case studies, increases the risk that cost estimates may use improper assumptions, lack appropriate definition, or be otherwise unreliable. Without improving the Unified Facilities Criteria with respect to cost estimating processes, DOD and the services will not be positioned well to provide reliable cost estimates to DOD and congressional decision- makers. Conclusions Each year DOD receives billions of dollars in MILCON appropriations to use for projects in the United States and overseas. The quality of project cost estimates are of great importance since those estimates are the basis for DOD’s requests for appropriations. While DOD’s policy is that MILCON cost estimates be prepared as accurately as possible in order to reflect the full cost of constructing DOD facilities, DOD’s Unified Facilities Criteria—the department’s primary construction criteria for developing cost estimates—does not fully incorporate all of the steps needed for producing reliable cost estimates. Until DOD incorporates the 12 steps of high-quality, reliable cost estimating into this department-wide construction criteria, DOD and congressional decision-makers may not have reliable estimates to inform their decisions regarding appropriations and the oversight of projects. Recommendation for Executive Action We are making one recommendation to DOD: The Secretary of Defense should ensure that the Assistant Secretary of Defense for Energy, Installations, and the Environment work with DOD’s construction agents, military departments, and other offices to improve DOD’s MILCON cost estimating guidance (i.e., DOD’s Unified Facilities Criteria) by fully incorporating all the steps needed for developing high- quality reliable cost estimates. (Recommendation 1) Agency Comments and Our Evaluation We provided a draft of this report to DOD. In written comments, which are reprinted in their entirety in appendix VI, DOD partially concurred with our recommendation. DOD also provided technical comments that have been incorporated into the report as appropriate. DOD partially concurred with our recommendation to improve its cost estimating guidance by fully incorporating all 12 steps needed for developing high-quality, reliable estimates. DOD stated that it did not believe that it is suitable to fully apply all 12 steps to any construction project due to characteristics of the military construction program that DOD believes differ from those of major system or weapon acquisition programs. However, DOD also stated that it concurred with the intent and general applicability of the twelve steps to military construction and that DOD cost estimating guidance lacks specificity in several of these areas. DOD acknowledged that expanding its cost guidance to more fully incorporate these steps would benefit the military construction program, and that it is planning to address this by revising its cost guidance during Fiscal Year 2019. In our report, we recognize that it may not be appropriate to fully apply all 12 steps to each construction project. For example, it may not be realistic or to the military departments’ benefit to conduct a sensitivity and uncertainty analysis or develop an independent cost estimate for all the construction projects they initiate every year, especially for low-cost projects. Accordingly, we did not recommend that DOD fully apply all 12 steps to each construction project, but rather that it fully incorporate the 12 steps into the Unified Facilities Criteria so that, at least, each step is considered for each project. DOD could then choose to establish thresholds—based on, for example, the dollar values of the projects—to determine for which the 12 steps should be fully applied or other circumstances in which some steps might not be applicable. We believe DOD’s planned revisions will meet the general intent of our recommendation. We are sending copies of this report to the appropriate congressional committees; the Secretary of Defense, the Secretaries of the Army, Navy, and Air Force. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff has any questions about this report, please contact me at (202) 512-4523 or leporeb@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix VI. Appendix I: Scope and Methodology To examine the active component’s military construction (MILCON) obligations and expended balances, we reviewed MILCON appropriations found in appropriations acts, including accompanying explanatory statements and conference committee reports from fiscal year 2005 through 2016. Further, we analyzed the obligation and disbursement data of the active component‘s MILCON accounts, appropriation status reports, bid savings reports, as well as annual reports from the U.S. Department of the Treasury. We also collected and compared project data from each of the active component on projects that had been initiated and completed during fiscal year 2010 through fiscal year 2016. Specifically, we compared the initial estimate as shown on the Form 1391—the form DOD uses to submit requirements and justifications in support of its funding requests to Congress—with the contract award amount and analyzed any differences between the two. To examine the amount of MILCON reprogramming during fiscal years 2010 and 2016 by the active component, we reviewed DOD’s requests to Congress to reprogram MILCON funds from one project to another. We calculated the total number of times such requests were made and the dollar amounts for fiscal year 2010 through fiscal year 2016. We selected this time frame because the reprogramming requests were readily available from DOD. In addition, we judgmentally selected three projects from this same time frame and reviewed accompanying Forms 1391 and the reprogramming requests associated with the projects to illustrate instances in which savings from one MILCON project funded another project. We collected and analyzed data for fiscal years 2005 through 2016 on the active component MILCON appropriations, obligations, and disbursements and we collected reprogramming data for fiscal years 2010 through 2016. We assessed the reliability of the data by interviewing knowledgeable officials about the data and the steps that they had taken to verify the data’s accuracy. We determined that the data were sufficiently reliable for our objectives. To determine the extent to which DOD’s MILCON cost estimates are reliable and DOD’s guidance for producing estimates fully incorporates all of the steps needed for developing reliable estimates, we compared the process for developing three selected projects with the characteristics and best practices for developing a reliable estimate identified in GAO’s Cost Estimating and Assessment Guide: Best Practices for Developing and Managing Capital Program Costs (the Cost Guide). This guide is a compilation of cost estimating best practices drawn from across industry and federal government. We selected our projects from the universe of projects that we reasonably expected could have begun execution (i.e., projects initiated during fiscal years 2012-2014); projects that were underway, but not substantially completed (i.e., between 10- and 75- percent complete); and projects that constituted a significant financial investment (i.e., projects with appropriations of $75 million or greater). Ultimately, of 690 total projects we identified DOD-wide, 13 met these criteria and, from this sample, we selected the 3 projects included in this report: (1) the construction of a replacement elementary school at Camp Foster, Japan; (2) the construction of a Strategic Command operations building at Offutt Air Force Base, Nebraska; and (3) the construction of a Marine Corps command headquarters and cyberspace operations building in Fort Meade, Maryland. In conducting the assessments for these three selected projects, we examined the processes used to develop both the Form 1391 estimate (i.e the form DOD uses to submit project-level requirements and justifications in support of its MILCON funding requests to Congress ) and the independent government estimate (i.e., the estimate used to award the contract) to determine whether the project cost estimates had the characteristics of a high-quality and reliable cost estimate, as defined in the Cost Guide. These projects are not intended to be a projectable sample, but to illustrate how cost estimates are assessed against best practices. Although the Camp Foster project is not owned by any of the active component, the construction and planning of the project is being led by the Army Corps of Engineers in its capacity as a DOD construction agent and, as such, we decided to include it in our review. Additionally, we reviewed DOD’s Unified Facilities Criteria and the active component’s respective guidance related to MILCON cost estimating and compared them with the steps needed for developing reliable estimates identified in the Cost Guide. We also interviewed military project cost estimators and active component construction agents to discuss the requirements and guidance they follow in preparing, documenting, and reviewing project cost estimates. Table 9 details the documents we reviewed for our cost estimating assessments. We conducted this performance audit from January 2016 to March 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Active Component’s Unobligated, Unexpended Balances, and Execution of Military Construction Appropriations, Fiscal Years 2005 through 2016 In this appendix we provide the supporting details on the active component’s unobligated and unexpended balances of military construction (MILCON) appropriations for fiscal years 2005 through 2016. We include details on unobligated and unexpended balances by appropriation year and include individual tables for each military department of the active component. Overall, the active component had high obligation and expenditure rates associated with MILCON appropriations that have expired or been canceled. The Army, the Air Force, and the Navy consistently expended over 90 percent of amounts appropriated in fiscal years 2005 through 2011. This appendix also provides supporting details on the active component’s execution of MILCON appropriations for fiscal years 2010 through 2016. Using Department of Defense (DOD) data, we identified two groups of MILCON projects: congressionally directed and other. “Congressionally directed” projects are those MILCON projects specifically identified in an appropriation act, explanatory statement, and/or committee reports accompanying the appropriation act for a specific fiscal year. “Other” projects refer to congressionally directed MILCON projects identified in an appropriation act, explanatory statement, and/or conference committee reports in a previous fiscal year. Overall, the active component obligated about 89 percent of its fiscal years 2010 through 2012 appropriations for congressionally directed projects whose appropriations expired on September 30, 2017. Unobligated and Unexpended Balances Tables 10 through 12 present detailed information on unexpended and unobligated balances for each military department of the active component’s MILCON appropriation for fiscal years 2005 through 2016, as reported by DOD as of September 30, 2016. Army Table 10 shows that for fiscal years 2005 through 2012, the Army expended almost all of its MILCON appropriations. Specifically, with the exception of fiscal year 2012, the Army expended at least 90 percent of its appropriations received each fiscal year for 2005 through 2011. Unexpended rates for amounts appropriated for fiscal years 2014 through 2016 vary and unobligated amounts for these years remain available for new obligations. Table 11 shows that, for fiscal years 2005 through 2013, the Air Force expended almost all of its MILCON appropriations. Specifically, the Air Force expended at least 95 percent of its appropriations received each year for fiscal years 2005 through 2011 and also in fiscal year 2013. Unexpended rates for amounts appropriated for fiscal years 2014 through 2016 vary and unobligated amounts for these years remain available for new obligations. Table 12 shows that for fiscal years 2005 through 2012, the Navy expended almost all of its MILCON appropriations. Specifically, the Navy expended at least 90 percent of its appropriations received each fiscal year for 2005 through 2011. Unexpended rates for amounts appropriated for fiscal years 2014 through 2016 vary and unobligated amounts for these years remain available for new obligations. Execution of Military Construction Appropriations Tables 13 through 15 provide detailed information on budget execution for each active duty military department’s MILCON appropriation for “congressionally directed” and “other” MILCON projects for fiscal years 2010 through 2016, as reported by DOD as of September 30, 2016. Army Table13 shows the obligations made by the Army for MILCON appropriations for fiscal years 2010 through 2016. We analyzed the obligations made during these appropriations’ period of availability for congressionally directed and other MILCON projects. For fiscal year 2010, using data in the table, we found that about 97.2 percent of obligations were for congressionally directed projects and 2.8 percent were for other projects, as discussed above. In fiscal year 2011, about 94 percent of obligations were for congressionally directed projects and 4.2 percent were for other projects; and in fiscal year 2012, about 86.5 percent of obligations were for congressionally directed projects and 7.2 percent were for other projects. Table 14 shows the obligations made by the Air Force for MILCON appropriations for fiscal years 2010 through 2016. We analyzed the obligations made during these appropriations’ period of availability for congressionally directed and other MILCON projects. For fiscal year 2010, using the data listed in the table, we found that 90.5 percent of obligations were for congressionally directed projects and 7.3 percent were for other projects, as discussed above. In fiscal year 2011, about 84.3 percent of obligations were for congressionally directed projects and 12.9 percent were for other projects; and in fiscal year 2012, about 87.5 percent of obligations were for congressionally directed projects and 9.0 percent were for other projects. Table 15 shows the obligations made by the Navy for MILCON appropriations for fiscal years 2010 through 2016. We analyzed the obligations made during these appropriations’ period of availability for congressionally directed and other MILCON projects. For fiscal year 2010, using data in the table, we found that 84.7 percent of obligations were for congressionally directed projects and 15.0 percent were for other projects, as discussed above. In fiscal year 2011, about 87.7 percent of obligations were for congressionally directed projects and 11.8 percent were for other projects; and in fiscal year 2012, about 85.5 percent of obligations were for congressionally directed projects and 13.4 percent for other projects. Appendix III: Comparison of Completed Military Construction Projects’ Initial Cost Estimates with Contract Award Amounts, Fiscal Years 2010 through 2016 This appendix provides information on our analysis of DOD’s estimated initial costs and contract award amounts of projects that had been initiated and completed during fiscal year 2010 through fiscal year 2016 by the active component. An official from the Office of the Assistant Secretary of Defense for Energy, Installations, and Environment told us that, to determine whether initial cost estimates were over- or underestimated, a comparison between initial Form 1391 estimates and contract award amounts would be a valid approach since contract award amounts are, in general, estimates of the same requirements identified on a Form 1391. The official also noted that supervision, inspection, overhead, and contingency costs included on a Form 1391 are not included in contract award amounts, which could create differences between the Form 1391 cost estimates and contract award prices. Because of this, we excluded the supervision, inspection, overhead, and contingency costs from the Form 1391 estimates in the table below to eliminate those differences. Form 1391 cost estimates may also vary from contract award amounts for reasons such as changes in project size or scope, changes in project characteristics, unexpectedly high or low contractor bids, or differences in expected building material costs, among other things. A negative percent change from the Form 1391 estimate to the contract award amount indicates the estimated project cost was overestimated and a positive percent change indicates the project was underestimated. We did not determine the precise reasons for any differences between estimated costs and contract award amounts. Table 16 lists information on 414 completed projects funded with military construction (MILCON) appropriations during fiscal year 2010 through fiscal year 2016 sorted by largest percentage overestimated to largest percentage underestimated. Appendix IV: Military Department Guidance for Developing Military Construction Cost Estimates The military departments of the active component have gone beyond the Unified Facilities Criteria and developed their own guidance for military construction (MILCON) that more closely aligns with the 12 steps needed for developing high-quality, reliable estimates. Table 17 describes the guidance developed by the military departments to align with those steps. Appendix V: Comments from the Department of Defense Appendix VI: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Maria Storts, Assistant Director; Bonita Anderson; Shawn Arbogast; Ronald Bergman; Brian Bothwell; Robert Brown; Farrah Graham; Mae Jones; Jennifer Leotta; Amie Lesser; Felicia Lopez; Carol Petersen; Vikki Porter; Steve Pruitt; and Karen Richey made key contributions to this report.

Why GAO Did This Study Between fiscal years 2005 and 2016, Congress annually appropriated between $2.5 to $9.6 billion in MILCON funding for the active component of the U.S. military to use for projects worldwide. Reliable project construction cost estimates are of great importance, since those estimates drive these appropriations. House Report 114-537 accompanying a proposed bill authorizing national defense activities for fiscal year 2017 included a provision for GAO to report on DOD's MILCON cost estimating. This report examines the extent to which (1) the active component obligated and expended the MILCON appropriations received during fiscal years 2005-2016, (2) the active component reprogrammed MILCON appropriations during fiscal years 2010 through 2016, and (3) DOD's MILCON cost estimates are reliable for selected projects and DOD's guidance for developing estimates fully incorporates the steps needed for developing reliable estimates. GAO analyzed the active components' MILCON execution data and reviewed DOD's guidance for cost estimating and compared it with the best practices identified in GAO's Cost Estimating and Assessment Guide . What GAO Found During fiscal years 2005 through 2016, Congress appropriated about $66 billion in military construction funds (MILCON) to the active duty Army, Navy, and Air Force (referred to as the active component) for projects. As of September 30, 2016, the active component had obligated all but about $5.1 billion and expended all but about $11 billion of those funds. Of the $5.1 billion remaining unobligated, about $4.6 billion was still available to be obligated because MILCON appropriations are generally available for new obligations for 5 years. According to Department of Defense (DOD) officials, available but unobligated amounts no longer needed may be either taken back by Congress or reprogrammed to other MILCON projects that the active component identifies as needing additional funding. During fiscal years 2010 through 2016, the active component reprogrammed about $1.6 billion in MILCON appropriations to fund emergency projects, projects that were authorized but did not receive specific appropriations, and projects needing additional funding. Of this amount, the Army reprogrammed about $789 million; the Navy, about $535 million; and the Air Force, about $295 million. DOD's guidance does not fully incorporate the steps needed for developing reliable estimates and the estimates for three projects that GAO reviewed were not reliable. Specifically, two of the three high-value projects GAO examined experienced a more than 30-percent increase from the initial cost estimates submitted to Congress. GAO determined that DOD cost estimators did not follow all the best practices associated with the four characteristics—comprehensive, well-documented, accurate, and credible—of a reliable estimate for these projects. GAO's Cost Estimating and Assessment Guide identifies 12 steps that, if used, are more likely to result in reliable and valid cost estimates. However, as shown below, DOD's construction guidance—the Unified Facilities Criteria—does not include all of these steps. Until DOD incorporates these steps, DOD and congressional decision-makers may not have reliable estimates to inform their decisions regarding appropriations and the oversight of projects. What GAO Recommends GAO recommends that DOD ensure that its cost estimating guidance fully incorporate the steps needed for developing reliable cost estimates. DOD partially concurred with GAO's recommendation and stated that it will issue revised cost guidance in fiscal year 2019 that more fully incorporates those steps that would benefit the military construction program.

gao_GAO-18-210

gao_GAO-18-210_0

Background Information security is a critical consideration for any organization that depends on information systems and computer networks to carry out its mission or business, and is especially important for government agencies, where maintaining the public’s trust is essential. Concerns about cyber threats to government systems and networks are well-founded, due to the dramatic increase in reports of security incidents, the ease of obtaining and using hacking tools, and advances in the sophistication and effectiveness of cyberattack technology, among other reasons. Without proper safeguards, systems are vulnerable to individuals and groups with malicious intent who can intrude and use their access to obtain or manipulate sensitive information, commit fraud, disrupt operations, or launch attacks against other computer systems and networks. We and federal inspectors general have reported extensively on information security deficiencies that place federal agencies at risk of disruption, fraud, or inappropriate disclosure of sensitive information. Accordingly, since 1997, we have designated federal information security as a government-wide high-risk area. This area was expanded to include the protection of critical cyber infrastructure in 2003 and protecting the privacy of personally identifiable information in 2015. Federal Law Establishes Security Requirements to Protect Federal Information and Systems The Federal Information Security Modernization Act of 2014 (FISMA) is intended to provide a comprehensive framework for ensuring the effectiveness of security controls over information resources that support federal operations and assets as well as the effective oversight of information security risks. FISMA assigns responsibility to the head of each agency to provide information security protections commensurate with the risk and magnitude of the harm resulting from unauthorized access, use, disclosure, disruption, modification, or destruction of information systems used or operated by an agency or on behalf of an agency. The law also delegates to the agency’s Chief Information Officer (or comparable official) the authority to ensure compliance with FISMA requirements. FISMA requires each agency to develop, document, and implement an agency-wide information security program to provide risk-based protections for the information and information systems that support the operations and assets of the agency, including those provided or managed by another entity. Such a program includes assessing risk; developing and implementing cost-effective security plans, policies, and procedures; developing plans for providing adequate information security for networks, facilities, and systems; providing security awareness and specialized training; testing and evaluating the effectiveness of controls; planning, implementing, evaluating, and documenting remedial actions to address information security deficiencies; developing and implementing procedures for detecting, reporting, and responding to security incidents; and ensuring continuity of operations. In addition, FISMA requires agencies to comply with NIST standards. Office of Management and Budget Provides Guidance to Agencies on Implementing FISMA In accordance with FISMA, the Office of Management and Budget (OMB) is responsible for the oversight of agencies’ information security policies and practices. OMB establishes requirements for federal information security programs and assigns agency responsibilities to fulfill the requirements of statutes such as FISMA. OMB requires agencies to oversee the implementation of security and privacy controls by contractors that collect, use, process, store, maintain, and disseminate federal information on behalf of a federal agency. For specific technical direction, OMB requires agencies to implement standards and guidelines established by NIST. NIST’s Framework for Critical Infrastructure Cybersecurity Establishes a Baseline for Protecting Critical Information Assets NIST has issued a suite of information security standards and guidelines, including Recommended Security Controls for Federal Information Systems and Organizations and the Framework for Improving Critical Infrastructure Cybersecurity. These documents collectively provide comprehensive guidance on developing and implementing information security programs to agencies and entities that perform work on their behalf. The framework serves as a baseline for protecting critical information assets. In response to Executive Order 13636, NIST issued the framework in February 2014. It is intended to help organizations apply the principles and best practices of risk management to improve the security and resilience of critical infrastructure. The framework outlines a risk- based approach to managing cybersecurity that is composed of three major parts: a framework core, profile, and implementation tiers. The framework core includes a list of functions, categories, subcategories, and informative references that describe specific cybersecurity activities identified as being in common across all critical infrastructure sectors. Additionally, the framework contains implementation tiers that provide context on how an organization views cybersecurity risk and the processes in place to manage that risk. Further, the framework provides guidance on documenting individual organizational profiles that describe how the functions, categories, and subcategories align with the business requirements, risk tolerance, and resources of the organization. According to NIST, the framework core represents a common set of activities for managing cybersecurity risk. The framework also states that, while it is not exhaustive, it is extensible, allowing organizations, sectors, and other entities to use subcategories and informative references that are cost-effective and efficient and that enable them to manage their cybersecurity risk. Table 1 lists the five functions and 22 categories of the framework core. Subsequent to the issuance of the Cybersecurity Framework, a May 2017 executive order required agencies to use the framework to manage cybersecurity risks. It outlined actions to enhance cybersecurity across federal agencies and critical infrastructure to improve the nation’s cyber posture and capabilities against cybersecurity threats to digital and physical security. In addition, the order directed agencies to develop plans to implement the framework within 90 days. The order required agencies to include in their plans: the status of planning, organizing, and submitting IT budget materials, as directed in the Fiscal Year 2018 IT Budget Capital Planning Guidance, that are aligned with the framework, the proposed internal management of cybersecurity risk using the updated metrics aligned to the framework, a timeline to map existing and planned capabilities with the framework the proposed use of the terminology and concepts in the framework to organize and communicate cybersecurity activities and outcomes. CMS Shares Medicare Fee-for- Service Beneficiary Data with Three Major Types of External Entities CMS shares Medicare beneficiary data with three major types of external entities: (1) Medicare Administrative Contractors (MACs), the contractors that provide the core processing and distribution functions that support the payment of Medicare Part A, Part B, and Durable Medical Equipment (DME) beneficiary claims on behalf of CMS, (2) research organizations (researchers), academic and non-profit entities that use Medicare beneficiary data to assist CMS in monitoring, managing, and improving Medicare programs or the services provided to beneficiaries, and (3) qualified public or private entities that use claims data on behalf of CMS to evaluate the performance of Medicare service providers and equipment suppliers. Medicare Administrative Contractors Access Fee- for-Service Beneficiary Data to Process Claims MACs process more than 1.2 billion claims for Medicare Fee-for-Service beneficiaries annually. To do so, they interact with more than 1.5 million health care providers enrolled in the Medicare Fee-for-Service program. In addition to claims processing, some of the specific functions that the MACs perform include customer service for beneficiaries and providers, financial and debt management, audit and appeals functions, and medical reviews. Each MAC contract covers a specific geographic area and a specific type of processing—either (1) Medicare Parts A and B claims or (2) DME claims for beneficiaries. Some MACs may hold multiple contracts and, thus, process multiple types of claims. In total, a network of eight MACs covers 16 multi-state jurisdictions, serving as the primary operational connection between the Medicare Fee-for-Service program and health care providers enrolled in the program. The geographic jurisdictions of the MACs that support Parts A and B and DME beneficiary claims are shown in figures 1 and 2, respectively. In order to collect, store, and process information needed to process claims and make benefits payments on behalf of CMS, MACs connect directly to CMS systems. Specifically, MACs connect to CMS’s Virtual Data Centers (VDCs) through its CMSNet telecommunications network. MACs process Medicare Fee-for-Service claims, which include beneficiaries’ PII and protected health information, through the VDCs. The VDCs consists of two large datacenters that are operated and managed by CMS that collectively serve as a platform for Medicare claims processing software systems. MACs use a combination of four CMS systems that operate within the VDCs to process claims. These systems and their functions are described in table 2. Health care providers submit Medicare fee-for-service claims to the MACs. The claims are reviewed to check if the claim is in a valid format, if the requestor is valid, and whether it is a duplicate. In addition, MACs process claims in the Fiscal Intermediary Shared Systems, Multi-Carrier System, VIPS Medicare System, and Common Working File. Processing includes adjudicating claims, checking whether the services are covered by Medicare, and determining the price that should be paid to the provider for the service. The links between external entities and CMS systems can take several different paths. Figure 3 shows how these entities are connected to CMS systems in order to obtain and use Medicare beneficiary data. Researchers Access Fee- for-Service Beneficiary Data to Assist CMS in Monitoring, Managing, and Improving Medicare Programs and Services Researchers use Medicare beneficiary data to study how healthcare services are provided to beneficiaries. Examples of research entities include universities and colleges, non-profit research institutes, and policy research organizations. CMS offers researchers a broad range of data on the Medicare program to support research on current and future spending, past and present enrollment, and claims, which can benefit the public through improved delivery of healthcare services. Research performed using this data may also assist CMS in monitoring, managing, and improving Medicare programs and services to beneficiaries. To obtain Medicare data from CMS, researchers must apply for access to a specific dataset, such as the Carrier file which includes claims for services provided by physicians and other non-institutional providers. In the application, the researcher provides information explaining how the data are to be used and stored, and CMS reviews and approves (or denies) the application. The researcher then enters a data use agreement with CMS for access to specific sets of Medicare beneficiary data, which are to be used only for stated research objectives. The data use agreement specifies which beneficiary data can be accessed, for what purpose, the duration of access, and data protection and confidentiality requirements. Unless the agreement authorizes the release of the data in accordance with CMS policy, it is not to be released by the researcher. As of October 2017, 195 research entities had received Medicare data. Researchers access Medicare beneficiary data in one of two ways. To gain access from their computers, they connect to CMS’s Chronic Conditions Warehouse/Virtual Research Data Center (CCW/VRDC) through a CMS-provided secure network connection. Within the CCW/VRDC, researchers are given access to an individually tailored computing environment containing only copies of the specific sets of beneficiary data they have been authorized to use. Researchers can then conduct their analysis on the data using software tools provided by CMS within this secure environment. Researchers can also access data by having it shipped to them in encrypted form through the U.S. mail. Once it has been received, researchers decrypt the data and load it into their own information systems for analysis. The data use agreements specify requirements for protecting beneficiary data obtained in this fashion. Qualified Entities Access Medicare Fee-for-Service Beneficiary Data to Evaluate the Performance of Service Providers and Equipment Suppliers Qualified entities use CMS claims data to assess the effectiveness of Medicare service providers and equipment suppliers. The Medicare Data Sharing for Performance Measurement Program, originally established to comply with the Patient Protection and Affordable Care Act, requires qualified entities to combine the Medicare data with claims data from sources other than Medicare to produce and publicly disseminate CMS- approved reports on provider and supplier performance with regard to measures of quality, efficiency, effectiveness, and resource use. Like researchers, after they have been approved to access data by CMS, qualified entities must enter into a data use agreement with CMS. The agreement specifies which beneficiary data can be accessed, for what purpose, the duration of access, and data protection and confidentiality requirements. A separate agreement is required for each qualified entity’s activity. The Medicare beneficiary data to be accessed are encrypted and can either be shipped to the qualified entity on an external hard drive or saved within the CCW/VRDC to be accessed through a Secure File Transfer System connection. Once it has received the electronic files, the qualified entity decrypts the files and analyzes the data on its own system(s). As of October 2017, ten organizations had received Medicare data as a qualified entity. Each entity is responsible for analyzing and reporting on provider performance for one or more specific geographic area. CMS Established Information Security Requirements that Align with Federal Guidance for Some, but Not All, External Entities CMS has developed requirements for implementing security controls that align with federal guidance for two of the three types of external entities that access Medicare Fee-for-Service data. Specifically, adherence to the requirements, which CMS defined using a risk-based process, is mandatory for MACs and qualified entities. However, CMS does not consider the requirements to be applicable to researchers because they are not CMS contractors. Without providing comprehensive, risk-based requirements for implementing security controls to all external entities that have access to Medicare beneficiary data, CMS increases the risk that external entities possessing CMS data may not have applied security controls that meet CMS standards. CMS Requirements for MACs and Qualified Entities Reflect a Risk- Based Assessment and Generally Align with the NIST Cybersecurity Framework To assist agencies in the selection of appropriate security controls, NIST developed the Cybersecurity Framework, which specifies controls that support the core security functions of identifying, detecting, preventing, responding to, and recovering from security incidents. Further, to ensure that controls are selected that achieve the security goals of the organization, NIST recommends that organizations use risk-based methods to tailor the selection of controls within this framework for implementation. According to NIST risk management guidance, the tailoring process includes identifying a baseline of security controls, assigning specific values to organization-defined security control parameters, such as password complexity, and supplementing baselines with additional controls and control enhancements. Once an agency has assessed security risks and identified appropriate controls to mitigate them, NIST recommends that the agency establish specific requirements for implementing those controls to ensure consistency both internally and externally to the agency. This is important in meeting the requirements of FISMA, which requires that a federal agency’s security efforts include information and systems provided or managed by another agency, contractor, or other source. Additionally, the Cybersecurity Framework recommends that contracts or other formal agreements abide by NIST guidance to provide a means to ensure privacy and security controls; it also states that contractors are to protect PII in the same manner as their customers. CMS developed minimum security requirements based on applicable federal guidance, for its own internal systems and for the systems operated by its contractors, such as MACs and qualified entities. These requirements are documented primarily in CMS’s Acceptable Risk Safeguards (ARS). CMS designed the ARS as a tailored selection of NIST controls reflecting FISMA requirements as well as the agency’s own policies, procedures, and guidance; other federal and non-federal guidance; and industry leading practices. According to the agency, the requirements in the ARS are intended to ensure that systems meet a minimum level of information security and privacy assurance and reflect the agency’s information systems security policy. CMS requires all employees, contractors, sub- contractors, and their respective facilities supporting agency business missions and performing work on behalf of the agency to observe this policy. Because MACs are CMS contractors, the agency requires them to align their security practices with the ARS as well as with broader federal guidance, including NIST’s catalog of recommended security controls, its minimum security standard for federal information systems, and OMB’s guidance on information management. Additionally, as part of the Qualified Entity Certification Program and consistent with NIST guidance, CMS requirements state that systems used by qualified entities to process Medicare beneficiary data have been assessed at a moderate impact level and accordingly are held to the ARS implementation guidance using the minimum controls specified for moderate risk systems. According to agency officials responsible for developing and maintaining the ARS, CMS used a risk-based process to select security controls to include in the requirements, thus ensuring that the ARS appropriately reflected agency needs and priorities. The process began with a review of baseline control requirements outlined in NIST guidance to ensure that all of those controls were reflected in the requirements. Then, the agency reviewed the rest of the NIST information security controls that were not included in the baseline and determined whether to include them in the ARS as “optional” controls. For example, the officials stated that certain controls appeared to apply primarily to national security systems and would not be needed for CMS applications. In all, the agency decided not to include 13 of the 165 controls specified in the NIST Cybersecurity Framework, none of which were designated by NIST as mandatory baseline controls. By undertaking this process of assessing the risk associated with each of the information security controls, the agency helped to ensure that its ARS reflects security requirements that are necessary and appropriate for its own systems and for systems operated by contractors on its behalf. A complete description of the NIST Cybersecurity Framework controls and how the ARS aligns with them can be found in appendix II. CMS Does Not Provide Security Guidance for Researchers While CMS requires MACs and qualified entities to implement security controls consistent with NIST guidance and provides additional guidance to ensure that those controls are consistent with CMS standards, it does not provide supplemental guidance tailored for researchers. Specifically, as part of its data use agreements with researchers, CMS includes a broad requirement to implement security and privacy protections that are consistent with NIST and OMB guidance. However, the agency has not provided risk-based guidance defining the minimum acceptable security controls that researchers should implement to protect Medicare beneficiary data. Nor has CMS provided guidance to researchers on how to select and implement specific security controls. According to CMS officials who oversee researcher access to CMS data, all researchers are required to prepare data management plans that outline their planned safeguards for protecting Medicare beneficiary data in their custody. In determining what controls to implement, however, they have only broad federal guidance, such as NIST’s catalog of controls, to use as a reference. The officials stated that CMS has not developed specific requirements based on an assessment of the risks associated with researcher functions that would define a minimum set of required safeguards. This is in contrast with the MACs and qualified entities, which have specific requirements based on the ARS that they are to implement to adequately protect data received from CMS. The lack of specific requirements does not affect all data that researchers access on behalf of CMS. In many cases, researchers access and process Medicare beneficiary data on systems operated by CMS and are not responsible for implementing the security controls for those systems. In such cases, the researchers access beneficiary data within a virtualized environment, called the CCW/VRDC, which allows CMS to monitor data retrieval and use. However, in other cases, CMS provides beneficiary data to researchers on external hard drives or other physical media that are outside of the Chronic Conditions Warehouse. In those cases, researchers receive Medicare beneficiary data that they transfer to and process on their own systems. These systems are secured according to individual researchers’ own policies and procedures, which may or may not be consistent with CMS requirements applied to other entities. CMS requirements tailored specifically for researchers could address topics such as password complexity, patch management, and encryption of sensitive data, all of which otherwise may be implemented inconsistently by different researchers. According to CMS officials responsible for overseeing researcher access to data, CMS does not require researchers to adhere to its Information Systems Security and Privacy Policy or to implement the controls specified in the ARS because researchers are not agency contractors. The CMS officials said it was not necessary for the agency to set specific security requirements for entities that do not have a contractual relationship with the agency. Additionally, these officials stated that they believe the lack of specific guidance gives the researchers more flexibility to independently assess their security risks and determine which controls to implement based on that assessment. However, by not providing guidance to researchers that includes security implementation requirements tailored to CMS-authorized uses of Medicare data, CMS cannot ensure that researchers implement security measures that are commensurate with the sensitivity of the data that is provided to them. As a result, there is an increased risk that sensitive PII and protected health information may be at risk of compromise. CMS Has Not Consistently Overseen the Implementation of Security Controls by External Entities CMS has established a program to oversee the MACs’ implementation of security and privacy protections over Medicare beneficiary data, but it does not consistently track low-risk weaknesses in the CMS FISMA Controls Tracking System. MACs are subject to two types of independent annual assessments that regularly identify weaknesses in their implementation of security controls. The assessments have identified several recurring categories of weaknesses; however, the agency does not track low-risk weaknesses that could be related to these recurring categories. Additionally, CMS has not established a corresponding program for overseeing the implementation of security controls by researchers and qualified entities. Without more consistently tracking identified issues at MACs and establishing effective oversight measures for researchers and qualified entities, CMS cannot fully ensure that the security of Medicare beneficiary data is being adequately protected. CMS Has Overseen Independent Assessments at the MACs, but Has Not Consistently Tracked Issues Identified by Those Assessments Requirements for agencies to oversee the implementation of security protections are established in law and federal guidance. For example, the NIST Cybersecurity Framework specifies that organizations should assess security controls to determine the extent to which the controls are implemented correctly, operating as intended, and producing the desired outcome. The framework states that, as part of the process for conducting security control assessments, organizations should track and monitor weaknesses and develop remedial actions. Further, according to the framework, the security assessment process is intended to provide feedback to organizations that can use the information to make risk- based adjustments to protections for their systems and networks. In addition, both FISMA and the Medicare Prescription Drug, Improvement, and Modernization Act (MMA) of 2003 set specific requirements for CMS oversight of the implementation of information security controls by the MACs. FISMA requires an annual independent evaluation of an agency’s information systems, including those provided or managed by contractors, to ensure compliance with NIST requirements. Further, OMB’s FISMA guidance specifies regular testing of all security controls with an agency-determined, risk-based subset to be tested annually. The MMA likewise requires the MACs to undergo an independent evaluation of their information security program on an annual basis. Specifically, an independent assessor is to annually test an appropriate subset of a contractor’s systems and assess compliance with federal requirements for information security policies, procedures, standards, and guidelines, as defined by OMB. MACs Undergo Two Types of Annual Assessments In order to meet the requirements of the MMA and FISMA, CMS established two separate annual information security assessment processes for the MACs. Specifically, to comply with MMA, CMS has overseen independent annual evaluations of these contractors since the law was enacted in 2003. CMS selected an independent assessor to perform all of the MMA assessments. The assessor first reviews documentation of the implementation of security controls by the contractor and then reviews technical security controls onsite at each MAC. In 2010, CMS expanded the MMA assessments into more technical areas and has included penetration testing as part of the assessments. In addition, agency officials that oversee the MMA assessments stated that CMS reviews contractor policies and procedures for configuration management twice a year and conducts an on-site review of the implementation of selected technical controls every June. In 2016, the independent assessor performed tests in nine categories of security controls at eight MAC datacenters. In total, these assessments reported 168 weaknesses, of which 53 were categorized as high or moderate risk and 115 were low-risk. Further, to comply with FISMA requirements that all controls are tested regularly, CMS requires MACs to test one-third of their system security controls annually. CMS determines the control families to be tested in any given year and rotates the selection each year so that all controls are tested by the end of the 3-year testing cycle. For the 2016 FISMA assessment, CMS selected 121 security requirements within 8 control families. The independent assessor is responsible for assessing the security controls and making recommendations on how to correct weaknesses and address identified vulnerabilities. To determine compliance with CMS requirements, controls are assessed against the minimum security requirements defined in the CMS ARS. According to CMS officials from the Medicare Contractor Management Group, the two annual assessment processes together ensure that sufficient testing is being conducted each year. For example, in any given year, the MMA assessments may cover different security controls than the FISMA assessments. In addition, the FISMA assessors may identify outstanding recommendations that were made from the prior year’s MMA assessment and provide a status update on progress made to address open recommendations. Corrective Actions and Milestones Have Not Always Been Tracked Consistently Tracking and remediation are key parts of an organization’s security program that help to ensure that identified issues are addressed promptly and effectively. CMS requires the MACs to develop corrective action plans to remediate most of the weaknesses identified by the MMA and FISMA assessments. CMS requires that these weaknesses, along with plans of action and milestones for correcting them, be captured and tracked in its CMS FISMA Controls Tracking System, which is an agency- wide system for tracking the remediation of identified weaknesses. The tracking system maintains the certification and accreditation documents for all MAC systems and manages plans of action and milestones, their remediation activities, and completion. CMS monitors the disposition of all issues captured in the CMS FISMA Controls Tracking System, which helps to ensure that the MACs take steps to address weaknesses within required time frames. However, because CMS does not routinely track low-risk weaknesses, it may not be ensuring that all weaknesses consistently receive appropriate management attention and timely remediation. Specifically, with regard to the MMA assessments, CMS requires MACs to develop a corrective action plan to remediate only high and medium-risk weaknesses, which are tracked using plans of action and milestones. CMS does not require the tracking of low-risk weaknesses, which are shown in the assessment reports as recommended improvements rather than weaknesses in need of correction. In certain cases, MMA assessments have classified weaknesses as low-risk, and they have not been tracked in the CMS FISMA Controls Tracking System, even though similar weaknesses were classified by other assessments as medium- or high-risk, and were tracked. In contrast to the MMA assessments, CMS requires that MACs track all weaknesses identified in FISMA assessments in the CMS FISMA Controls Tracking System. Examples of inconsistently classified weaknesses reported in the 2016 MMA assessments include (1) maintaining complete and up-to-date inventories of information system components and (2) ensuring that protections against malicious software are installed and kept up-to-date. Of the six assessments that reported that MACs did not have a complete and accurate listing of systems and devices supporting Medicare claims processing, three classified this weakness as medium-risk and created a plan of action and milestones, while the other three assessed a low-risk level and did not create a plan of action and milestones. Similarly, eight assessments reported that MACs either did not have malicious software protections installed or they were not up-to-date. Of these eight, CMS officials stated that three were classified as medium-risk and were tracked by CMS, while the other five were assigned a low-risk level and not tracked. The inventory and malicious software protection weaknesses that were tracked inconsistently are related to categories of weaknesses that have posed recurring challenges for the MACs in recent years. Since 2009, both the MMA and FISMA assessments have reported incomplete implementation of several types of high-risk security requirements across all the MACs. The weaknesses identified during these assessments— which generally involved configuration management, system security plans, and system inventories—have yet to be fully resolved. Table 3 describes these key categories of weaknesses. According to CMS officials, weaknesses identified in the annual MMA assessments may be ranked at different risk levels because the specific circumstances of each finding can vary. However, documentation of the specific weaknesses identified in the 2016 MMA assessment reports does not make clear why findings that are characterized in similar terms or have the same name may have been assigned different risk levels. CMS officials who oversee the information security testing at MACs stated that they are aware of the recurring areas of weaknesses identified in the annual assessments and have been taking actions to address them. For example, in 2009, CMS began requiring MACs to submit evidence that their configuration management programs complied with CMS requirements. According to the officials, since this program has been put into place, configuration management processes at the MACs have become more consistent and more thoroughly documented. Nevertheless, the 2016 FISMA assessments concluded that a MAC’s system security plan did not include procedures for testing changes made to their production environments, and the MAC was not tracking changes made to the production environments. According to the CMS officials, the fact that recurring issues such as these have not yet been fully resolved may be due to the root causes of the deficiencies not yet being addressed. Without more consistent tracking of identified issues through plans of action and milestones, it may be difficult for CMS to fully determine the extent to which security weaknesses identified during assessments of the MACs are remediated. Weaknesses that appear to be low-risk may be indicators of more significant underlying issues and, thus, may not be receiving appropriate management attention or prompt remediation, unnecessarily exposing Medicare beneficiary data to security risks. CMS Does Not Have Effective Oversight Processes and Procedures for Researchers and Qualified Entities While CMS has established assessment programs for MACs, the agency has much more limited security oversight mechanisms in place to ensure that qualified entities and researchers with access to Medicare beneficiary data implement appropriate security controls. CMS oversight processes and procedures for qualified entities and researchers consists primarily of reviewing the data protections that researchers and qualified entities describe in the data management plans they submit when requesting access to Medicare beneficiary data. According to CMS officials who review these plans, they may ask follow- up questions to obtain more information or make recommendations on how to better implement security safeguards in accordance with CMS requirements. However, no further reviews are conducted for any qualified entities or researchers. For example, CMS does not conduct on- site reviews of the implementation of security controls and does not collect or review evidence of whether the controls have been appropriately implemented. Further, it does not conduct or require any independent testing of security controls. As an additional check for qualified entities, instead of assessing their security controls, CMS assesses their responses to questions relating to 213 moderate-level data security controls from 26 control families set forth in the ARS. However, once the initial document review has been completed, CMS does not perform any in-person or document reviews of security controls that are in place unless the qualified entity reports a major change in its data security environment after initial approval. According to officials of the Office of Enterprise Data Analytics, which is responsible for overseeing access to Medicare data by researchers and qualified entities, CMS has, in the past, conducted remote and on-site reviews as a pilot project. These reviews examined selected researchers’ security controls, based on factors such as the use of data described in the researchers’ data management plans. According to these officials, the pilot project is no longer being conducted because funding for the program has stopped. The need to ensure that these entities have effectively implemented information security controls is demonstrated by data breaches that these organizations have reported. Of the 195 research entities that CMS has data use agreements with, six have suffered data breaches involving the loss of over 500 records containing PII covered under the Health Insurance Portability and Accountability Act of 1996, which they reported to the HHS Office of Civil Rights. These breaches included Internet-based intrusions into researcher systems as well as other IT- related incidents. According to CMS officials who oversee access to Medicare data for researchers and qualified entities, the data use agreement requires organizations to report any breach of PII or personal health information from the CMS data files to the agency. These officials also stated that the six organizations did not report any breaches to CMS and that they were unaware that the organizations had reported compromises. The officials noted that if the breaches did not involve PII or personal health information from CMS data files provided under a data use agreement, the organizations were not required to report this information to CMS. Further, these officials stated that the agency is currently revising its data management plan to include a requirement for organizations to fully disclose all breaches to the agency, which may impact whether or not to grant access to Medicare data for organizations that were breached. Given that, in the past, researchers’ systems have been successfully attacked, effective implementation of security controls is critical to reducing threats of compromise. However, without more robust oversight processes and procedures, CMS cannot determine whether qualified entities or researchers have implemented security controls appropriately and, thus, cannot ensure that the risks associated with their use of Medicare beneficiary data have been adequately mitigated. Conclusion CMS shares Medicare beneficiary data with external entities primarily for processing Medicare claims, supporting medical research, and evaluating the performance of Medicare service and equipment providers. CMS has set basic requirements for protecting the security of Medicare beneficiary data that it shares with MACs, qualified entities, and researchers. However, CMS has not required the documentation of low-risk weaknesses in the CMS FISMA Controls Tracking system so that CMS can track the MACs’ remediation of weaknesses that have been identified in recurring annual assessments. In addition, MACs and qualified entities are given guidance that generally aligns with federal guidance and is based on an assessment of risks specific to CMS to ensure that appropriate controls have been included. However, CMS has not provided guidance to researchers on how to select and implement specific security controls. Until CMS provides more comprehensive, risk- based guidance on implementing security controls to all of its external partners, there is an increased risk that researchers will not fully implement appropriate protections for Medicare beneficiary data. CMS has developed and implemented an oversight program for the MACs’ implementation of security controls based on two types of annual independent assessments, which together help ensure that sufficient testing is being conducted each year. However, CMS has not ensured that the MACs track and remediate identified weaknesses consistently, including weaknesses that have been identified in recurring annual assessments. Further, CMS has not established an oversight program for qualified entities and researchers to assess whether they are implementing security controls as they are required. Without more effective oversight programs in place, CMS lacks full assurance that external entities are appropriately implementing security protections for Medicare beneficiary data. Recommendations We are making three recommendations to the Administrator of the Centers for Medicare and Medicaid Services: Develop and distribute guidance for researchers defining minimum security controls and implementation guidance for those controls that is consistent with NIST guidance. (Recommendation 1) Develop processes and procedures to ensure that findings from all MAC assessments are classified consistently and tracked appropriately. (Recommendation 2) Develop processes and procedures to ensure that qualified entities and researchers have implemented information security controls effectively throughout their agreements with CMS. (Recommendation 3) Agency Comments and our Evaluation We received written comments on a draft of this report from HHS. In the comments (reprinted in appendix III), the department concurred with our three recommendations and discussed actions that the department has planned or taken. If fully and effectively implemented, the intended actions should help HHS to address weaknesses in processes and procedures for ensuring the protection of Medicare beneficiary data used by the department’s contractors. The department also provided technical comments, which we have incorporated in the report, as appropriate. We are sending copies of this report to the appropriate congressional committees, the Secretary of Health and Human Services, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-9342 or marinosn@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made major contributions to this report are listed in appendix IV. Appendix I: Objectives, Scope, and Methodology Our objectives were to (1) identify the major entities that collect, store, and process Medicare beneficiary data and that connect with Centers for Medicare and Medicaid Services (CMS) systems and networks; (2) determine whether requirements for the protection of Medicare beneficiary data align with federal guidance; and (3) assess the programs CMS has in place to oversee the implementation of security protections for Medicare beneficiary data. To address our first objective, we analyzed prior GAO reports and CMS documentation, such as CMS data maps and system documentation. Additionally, we conducted interviews with agency officials to identify major external entities that access Medicare beneficiary data, including with Medicare Administrative Contractors (MAC) and researchers. We analyzed the information obtained from CMS to describe the type of Medicare data each entity has access to and purposes for which such access is provided. Further, we analyzed agency agreements with external entities to describe external uses for the data CMS collects and distributes. Regarding our second objective, we analyzed CMS guidance, specifically its Acceptable Risk Safeguards (ARS), to determine baseline requirements for the protection of Medicare beneficiary data that have been established by CMS. To assess the completeness of this guidance, we compared the ARS to the National Institute of Standards and Technology’s (NIST) Cybersecurity Framework’s controls included in the “identify,” “protect,” “detect,” and “respond” categories. We did not include the “recover” category because it is more focused on data recovery than on the identification, protection, and detection capabilities necessary to prevent incidents. We compared the controls referenced by NIST to the controls that were documented in the ARS to identify controls that had not been included. We also interviewed CMS officials responsible for developing the ARS to determine the process that the agency uses to select controls. Additionally, to determine how CMS required external entities to implement security measures, we reviewed formal agreements that were entered into with those organizations. For the MACs, we analyzed contracts to determine CMS security requirements. For researchers and qualified entities, we reviewed the data use agreement templates to determine what requirements CMS specified for selecting and implementing security measures. To address our third objective, we analyzed system assessments performed by CMS and conducted interviews with CMS officials responsible for overseeing the security of Medicare beneficiary data provided to external entities. Specifically, we analyzed information security assessments to determine the nature and extent of reported findings, the disposition of assessment recommendations, and whether assessment results were being addressed in a timely fashion over the span of time that they have been conducted. For the MACs, we reviewed assessments performed in accordance with the Federal Information Security Management Act and the Medicare Prescription Drug, Improvement, and Modernization Act of 2003. For researchers and qualified entities, we obtained information from CMS about ongoing and previously performed assessment programs. Through interviews with relevant CMS officials, we obtained and analyzed information about the findings that were not resolved in a timely fashion and about the constraints that prevented the ongoing assessment of researchers and qualified entities. We conducted this performance audit from October 2016 to January 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Analysis of CMS Acceptable Risk Safeguards We compared the Centers for Medicaid and Medicare Services (CMS) Acceptable Risk Safeguards (ARS) with the National Institute of Standards and Technology (NIST) Cybersecurity Framework to determine the extent to which the ARS aligns with the framework. To do this, we compared the controls noted as informative references by the framework to the controls documented in the ARS. We did not assess the “Recover” category because it is more focused on data recovery than on the identification, protection, and detection capabilities necessary to prevent incidents. Appendix III: Comments from the Department of Health and Human Services Appendix IV: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, John De Ferrari (assistant director); Thomas Johnson (analyst-in-charge); Chris Businsky, Kavita Daitnarayan, Nancy Glover, Charles Hubbard III, Monica Perez-Nelson, and Richard Sayoc made key contributions to this report.

Why GAO Did This Study Recent data breaches have highlighted the importance of ensuring the security of health information, including Medicare beneficiary data. Such data are created, stored, and used by a wide variety of entities, such as health care providers, insurance companies, financial institutions, researchers, and others. GAO was asked to conduct a study of CMS efforts to protect Medicare beneficiary data accessed by external entities. GAO's objectives were to (1) identify the major external entities that collect, store, and process Medicare fee-for-service beneficiary data; (2) determine whether requirements for the protection of Medicare beneficiary data align with federal guidance; and (3) assess CMS oversight of the implementation of those requirements. GAO analyzed information about how external entities access data, reviewed CMS documentation on who they share data with, compared federal standards with CMS security requirements for external entities, and analyzed results of independent security reviews. GAO also interviewed CMS officials about their oversight activities. What GAO Found The Centers for Medicare and Medicaid Services (CMS) shares Medicare beneficiary data with three major types of external entities: (1) Medicare Administrative Contractors (MAC) that perform processing and distribution functions that support the payment of Medicare benefits; (2) research organizations (researchers) that use Medicare beneficiary data to study how health care services are provided to beneficiaries; and (3) qualified public or private entities that use claims data to evaluate the performance of Medicare service providers and equipment suppliers. CMS has developed requirements for implementing security controls that align with federal guidance for two of the three types of external entities that access Medicare beneficiary data. While CMS has developed guidance for MACs and qualified entities, it has not developed equivalent guidance for researchers. Researchers must adhere to broad governmentwide standards, but are not given guidance on which specific controls to implement. According to CMS, the lack of specific guidance gives the researchers more flexibility to independently assess their security risks and determine which controls are appropriate to implement; however, without providing comprehensive, risk-based security guidance to researchers, CMS increases the risk that external entities possessing agency data may not have applied security controls that meet CMS standards. Additionally, CMS has established an oversight program for the security of MAC data, but has not established a corresponding program to oversee security implementation by researchers and qualified entities. Without effective oversight measures in place for researchers and qualified entities, CMS cannot fully ensure that the security of Medicare beneficiary data is being adequately protected. Regarding MACs, although they are subject to two types of independent annual assessments, which have regularly identified weaknesses in their implementation of security controls, the weaknesses that have been assessed as low-risk have not been consistently tracked in the CMS finding tracking system. Without more consistent tracking of these low-risk weaknesses, it may be difficult for CMS to determine if all weaknesses are being addressed in a timely manner. Examples of categories of recurring weaknesses that have been identified during annual assessments are listed in the table. What GAO Recommends GAO recommends that CMS develop additional guidance for researchers on implementing security controls required by CMS, consistently track results of independent assessments, and provide oversight of researchers and qualified entities. CMS concurred with GAO's three recommendations and described actions it has planned or taken to address them.

gao_GAO-18-47

gao_GAO-18-47_0

Background Defense Support of Civil Authorities Under the Robert T. Stafford Disaster Relief and Emergency Assistance Act (the Stafford Act), when state capabilities and resources are overwhelmed and the President of the United States declares an emergency or disaster, the governor of an affected state can request assistance from the federal government for major disasters or emergencies. The Stafford Act aims to provide a means of assistance by the federal government to state and local governments in responding to a presidentially declared major disaster or emergency. A governor’s request for the President to declare a major disaster or emergency is required to be based on a finding that the situation is of such severity and magnitude that effective response is beyond the capabilities of the state and the affected local governments and that federal assistance is necessary. Additionally, under the Economy Act, a federal agency may request the support of another federal agency, including DOD, without a presidential declaration of a major disaster or an emergency. This act permits one federal agency to request goods and services from another federal agency provided that, among other things, the service is available and cannot be obtained more cheaply or conveniently by contract. In July 2016, the White House issued the Presidential Policy Directive on United States Cyber Incident Coordination (hereafter referred to as PPD- 41) to establish principles governing the federal government’s response to cyber incidents involving government or private sector entities. Subsequently, in December 2016, the Department of Homeland Security issued an updated National Cyber Incident Response Plan that outlines domestic cyber-incident response coordination and execution among federal; state, territorial, and local governments; and the private sector. Overall coordination of federal incident-management activities is generally the responsibility of the Department of Homeland Security. DOD supports the lead federal agency in the federal response to a major disaster or emergency. When authorized to provide support to civil authorities for domestic emergencies, DOD may provide capabilities and resources— such as military forces (including the National Guard under Title 10 and Title 32, U.S. Code), DOD civilians, and DOD contractors—through DSCA. DOD components can also provide support to civil authorities under separate authority. For example, under Executive Order 12333, the National Security Agency, as an element of the intelligence community, is authorized to provide technical assistance and cooperation to law enforcement and other civil authorities not precluded by applicable law. DOD Components with DSCA Responsibilities In an effort to facilitate DSCA across the nation and at all organizational levels, DOD has assigned responsibilities within the Office of the Secretary of Defense (such as the Assistant Secretary of Defense for Homeland Defense and Global Security); the Chairman of the Joint Chiefs of Staff; various combatant commanders, such as the NORTHCOM and the U.S. Pacific Command (PACOM) Commanders; and the Chief of the National Guard Bureau, among others. A combatant command is a unified or specified command with a broad continuing mission under a single commander established and designated by the President, through the Secretary of Defense and with the advice and assistance of the Chairman of the Joint Chiefs of Staff. DOD’s Assistant Secretary of Defense for Homeland Defense and Global Security is the principal civilian advisor responsible for homeland defense, DSCA, and cyber for the department. This official is to develop policies, conduct analysis, provide advice, and make recommendations on homeland defense, DSCA, emergency preparedness, and cyberspace operations within the department. The Chairman of the Joint Chiefs of Staff advises the Secretary of Defense on the effects of requests for DSCA on national security and identifies available resources for support in response to DSCA requests. NORTHCOM and PACOM provide support to civil authorities at the federal, state, and local levels, as directed. Further, CYBERCOM synchronizes the planning for cyberspace operations in coordination with other combatant commands, the military services, and other appropriate federal agencies. In August 2017, DOD initiated the process to elevate CYBERCOM from a subunified command to a unified combatant command. According to DOD, this elevation “will help to streamline command and control of time-sensitive cyberspace operations by consolidating them under a single commander with authorities commensurate with the importance of those operations and will ensure that critical cyberspace operations are adequately funded.” Additionally, a dual-status commander could serve as an intermediate link between the separate chains of command for state and federal forces and is intended to promote unity of effort between federal and state forces to facilitate a rapid response during major disasters and emergencies. DOD’s Section 1648 Report Addressed Some of the Statutorily Required Elements DOD did not develop a comprehensive plan; instead, the department submitted a collection of separate documents that addressed some, but not all six statutorily required elements (hereafter referred to as DOD’s Section 1648 report). Table 1 lists each of the required elements and shows our determination of the extent to which the elements were addressed in DOD’s Section 1648 report. DOD officials agreed that their submission was not a comprehensive plan. These officials told us they developed a report that they believed would address the required elements of the legislation and articulate the department’s comprehensive work to prepare for supporting civil authorities in response to a cyber incident with plans, policies, guidance, among other things. Specifically, DOD’s Section 1648 report is a collection of separate documents that, according to DOD, outline core federal, state, local, and private sector roles and responsibilities; summarize plans for coordination at all levels of government and across sectors in the event of a cyber incident; and prescribe the roles and responsibilities of the active and reserve components. DOD’s Report Fully Addressed Two of the Six Elements Required by the Statute As noted in the table above, DOD’s Section 1648 report addressed two of the six elements required by the statute–to provide (1) descriptions of the roles, responsibilities, and expectations of federal, state, and local authorities and (2) a description of legislative and administrative actions necessary to carry out its plan to support domestic cyber incident response efforts. Specifically, DOD’s Section 1648 included copies of the PPD-41 and the Department of Homeland Security’s National Cyber Incident Response Plan. Both of these documents provide general descriptions of the roles, responsibilities, and expectations of federal, state, and local authorities. For example, the National Cyber Incident Response Plan was developed to articulate the roles and responsibilities, capabilities, and coordinating structures that support how the nation responds to and recovers from significant cyber incidents posing risks to critical infrastructure. DOD’s Section 1648 report also included a description of administrative actions that the department believed were necessary to carry out its plan to support domestic cyber incident response efforts. Specifically, according to the report, DOD had drafted a directive type memorandum to provide supplementary policy guidance, assign responsibilities, and detailed procedures for providing defense support for cyber incident response. This memorandum was signed and issued by DOD subsequent to the department submitting its Section 1648 report to Congress. DOD officials also acknowledged that there were incorporating cyber into all aspects of policy, doctrine, and guidance. In the report, DOD stated that the department believed current authorities were sufficient and did not recommend any legislative actions. DOD’s Report Partially Addressed Three of Six Elements Required by the Statute but the Information Provided Was Incomplete DOD partially addressed three of the six elements required by the statute—to provide (1) descriptions of the roles, responsibilities, and expectations of the active and reserve components of the armed forces; (2) the department’s plans for coordination with heads of other federal agencies and state and local governments; and (3) a list of exercises previously conducted that are used in the formulation of the plan. DOD’s Section 1648 report includes a copy of DOD Directive 3025.18, Defense Support of Civil Authorities, which establishes DSCA policy and provides guidance for the execution and oversight of DSCA, as an appendix. This directive includes a section that identifies roles and responsibilities of DOD components such as the Joint Staff, the combatant commands, and the military departments, among others. However, we have previously reported that DOD’s guidance does not clearly define the department’s roles and responsibilities. For example, we found inconsistency on which combatant command would be designated the supported command and have the primary responsibility for providing support to civil authorities during a cyber incident. Consequently, as noted in appendix I, we recommended that DOD issue or update guidance that clarifies roles and responsibilities for relevant entities and officials to support civil authorities in a domestic cyber incident. However, key DOD documents such as DOD Directive 3025.18, DOD’s Section 1648 report, and the Directive Type Memorandum issued in June 2017 do not clarify roles and responsibilities of DOD components, liaisons, and personnel who DOD had previously assigned coordination roles and responsibilities for supporting civil authorities. As a result, there is still uncertainty about these roles and responsibilities within the department. For example, disagreement still exists among officials in the department regarding whether NORTHCOM and PACOM (as the geographic combatant commands) or CYBERCOM, which according to command officials maintains the department’s existing inventory of cyberspace command and control capabilities, is the supported command in a cyber incident requiring civil support. DOD officials acknowledged to us that there are a number of planning and guidance documents that need to be updated to clarify roles and responsibilities. Until DOD clarifies the roles and responsibilities of its key entities for cyber incidents, as we recommended, department leaders and components will continue to experience uncertainty about the roles and responsibilities of different components and commands in providing support to civil authorities in the event of a significant cyber incident. In an effort to describe the department’s plans for coordination with heads of other federal agencies and state and local governments, DOD’s Section 1648 report provided information on the department’s role in supporting a whole-of-government approach during a significant cyber incident. Specifically, DOD included copies of PPD-41, the Department of Homeland Security’s National Cyber Incident Response Plan, and DOD’s Department of Defense (DOD) Significant Cyber Incident Coordination Procedures. These documents recognize that the department coordinates with other federal agencies (and state and local governments, as appropriate) through the Cyber Response Group and the Cyber Unified Coordination Group that were consistent with PPD-41. According to PPD-41, the Cyber Unified Coordination Group is the primary method for coordinating between and among federal agencies responding to a significant cyber incident, as well as for integrating private sector partners into incident response efforts. While DOD’s Section 1648 report recognizes the role and value of these two groups, these groups have limited coordination opportunities with state and local governments. For example, the Cyber Response Group is a national-level policy coordination group composed of federal department and agencies (i.e., does not include state and local governments). Also, the Cyber Unified Coordination Group is an ad-hoc group that is convened in response to a significant cyber incident and will not include state and local governments unless it is required by the scope, nature, and facts of a particular significant cyber incident. In addition, the report did not identify any plans for coordinating with heads of other federal agencies and state and local governments, as required by the statute. DOD guidance and joint doctrine state that, among the defense coordinating officers’ multiple responsibilities, they are supposed to develop and promote relationships with federal, state, tribal, and local governmental and non-governmental organizations, and with private sector entities in the assigned Federal Emergency Management Agency (FEMA) region. However, DOD’s Section 1648 report did not identify how the defense coordinating officers, their supporting elements, or other DOD components that coordinate with civil authorities—including state and local governments—plan to coordinate in preparation to provide support of DSCA activities. We are not making a recommendation on this issue because we previously recommended that DOD issue or update guidance that clarifies roles and responsibilities for DOD components—such as for the defense coordinating officers and their supporting elements—to support civil authorities in response to a domestic cyber incident. DOD’s Section 1648 report also includes a list of cyber civil support exercises that DOD conducted over the last 3 years. However, this list was incomplete because DOD did not include all exercises where DOD components provided support to civil authorities in a cyber incident. For example, the report did not include NORTHCOM’s 2015 exercises—Vista Host and Vista Code. These two exercises examined planning assumptions, potential resource requirements, and roles and responsibilities associated with cyber-related defense support to civil authorities operations. By not including this information in this one-time report, DOD missed an opportunity to provide Congress more complete information about exercises that the department is conducting to prepare itself and commands to support civil authorities for a cyber incident within the United States. During our review of DOD’s Section 1648 report, we also found that the department had yet to conduct a command and control (i.e., operational- level) exercise focused on providing support to civil authorities in a cyber incident—a gap acknowledged by officials from NORTHCOM, PACOM, and CYBERCOM. According to these officials, the exercises identified in the Section 1648 report focused on strategic–level decisions (e.g., Cyber Guard 16 Legal and Policy table top exercises) or tactical-level actions (e.g., Cyber Guard 16). CYBERCOM officials told us that they believe that Cyber Guard is a tier 1 exercise. However, a 2015 DOD Cyber Strategy implementation document stated that while Cyber Guard is a valuable “whole-of-nation” scenario, its focus is much more tactical in nature and that the department needed another tier 1-level exercise. Similarly, officials from both DHS and DOD acknowledged that Cyber Guard was a tactical-level exercise. As previously discussed and identified in appendix I, we previously recommended that DOD conduct a tier 1 exercise to prepare its forces in the event of a disaster with cyber effects. CYBERCOM officials told us the command is currently planning an internal staff exercise to address our recommendation to exercise its forces at the operational-level of leadership. However, an internal staff exercise (i.e., an exercise that does not exercise command-and-control relationships with other combatant commanders) will not be consistent with DOD guidance that states tier 1 exercises are designed to prepare national-level organizations and combatant commanders and staff at the strategic and operational levels to integrate a diverse audience in highly complex environments. We maintain our position that Cyber Guard in its current form is not a tier 1 exercise that would enable the department to achieve its DOD Cyber Strategy goal of exercising its DSCA capabilities in support of the Department of Homeland Security (DHS) and other agencies, including state and local authorities. We continue to believe that DOD should conduct a tier 1 exercise to improve the department’s planning efforts to support civil authorities in a cyber incident. DOD’s Report Did Not Address One of the Elements Required by the Statute and DOD Has Not Ensured That Staff Are Trained DOD’s Section 1648 report did not address one of the six required elements—to provide a plan for internal DOD collective training activities that are integrated with exercises conducted with other agencies and state and local governments. Instead, the department provided a classified list of planned exercises for 2017 that, according to officials, have training value for cyber incident response. Officials from ODASD (HDI/ DSCA) and ODASD (Cyber Policy) told us that DOD does not train for DSCA. Rather, the department trains and exercises its forces to conduct military missions and can apply the knowledge and experience from these activities to support civil authorities when requested and approved. The officials emphasized that, while exercises generally test whether DOD forces have learned training, in the case of DSCA exercises are a key training tool. While exercises may have training value, DOD did not provide information on existing DSCA-related training efforts within the department—such as on NORTHCOM’s DSCA course offered to officials from DOD and other federal agencies. Specifically, according to NORTHCOM officials, the command’s DSCA course focuses on training senior military officers, DOD civilians, and their staff to ensure DOD’s readiness to support its homeland defense and civil support missions. The officials explained that this course introduces participants to national, state, local, and DOD statutes, directives, plans, command and control relationships, and capabilities with regard to disaster and emergency response. By not including this information in this one-time report, DOD missed an opportunity to provide Congress more complete information about training that the department is conducting to prepare itself and commands to support civil authorities for a cyber incident within the United States. In addition, during our review, we found that DOD had not met the training requirements outlined in PPD-41, which was included in DOD’s Section 1648 report. Specifically, the policy directive requires federal agencies, including DOD, to update cyber incident coordination training to incorporate the tenets of PPD-41 by December 2016 and to identify and maintain a cadre of personnel qualified and trained in the National Incident Management System and unified coordination to manage and respond to a significant cyber incident. According to the PPD-41, the overarching document guiding DOD’s Section 1648 report, these personnel would provide necessary expertise to support tasking and decision making by a Cyber Unified Coordination Group. In addition, DOD’s Significant Cyber Incident Coordination Procedures require the Chairman of the Joint Chiefs of Staff, through the National Military Command Center, to maintain a list of senior DOD officials from specified organizations that could represent DOD during a Cyber Unified Coordination Group and who are trained in the National Incident Management System. As of August 2017, DOD officials acknowledged the department had not updated its cyber incident coordination training to incorporate the tenets of PPD-41. Joint Staff officials told us they have staff qualified and trained in the National Incident Management System; however, the officials were unable to provide us a list of senior officials from DOD organizations that could participate in a Cyber Unified Coordination Group that had been trained in the National Incident Management System. An official from the Office of DOD Principal Cyber Advisor acknowledged the Joint Staff is not tracking personnel who have been qualified and trained in the National Incident Management System, as required by the DOD Significant Cyber Incident Coordination Procedures. Consequently, it is unclear whether senior DOD officials who may be asked to participate in a Cyber Unified Coordination Group will be trained in the National Incident Management System. Until DOD updates its cyber incident response training and maintains a list of senior DOD officials from organizations who could represent DOD during a Cyber Unified Coordination Group and who are trained in the National Incident Management System, the department will not be in compliance with PPD- 41 and may not have the personnel with expertise to manage and respond to a significant cyber incident. Conclusions DOD recognizes that a disruptive, manipulative, or destructive cyberattack could present a significant risk to U.S. economic and national security and that the department must be prepared to support civil authorities in all domains—including in cyberspace. While DOD addressed some of the required elements set forth in Section 1648, the report submitted does not highlight the full scope of the department’s planning and preparation efforts to support civil authorities in response to a cyber incident. We are not making recommendations on these issues because we have previously made recommendations in areas where the Section 1648 report did not contain complete information. However, without complying with the training requirements outlined in PPD-41 and the DOD Significant Cyber Incident Coordination Procedures, the department cannot reasonably ensure it has the personnel with expertise to manage and respond to a significant cyber incident. Taking action to improve the areas we have highlighted should help DOD sustain the progress it has already made. With the President’s decision to elevate CYBERCOM to a unified combatant command, such actions will also help as DOD continues to plan to support civil authorities in response to a cyber incident and where CYBERCOM has a significant role. Recommendations for Executive Action We are making the following two recommendations to DOD: The Assistant Secretary of Defense for Homeland Defense and Global Security, in coordination with the Chairman of the Joint Chiefs of Staff and other appropriate DOD components, should update the department’s cyber incident coordination training to incorporate the tenets of PPD-41. The Chairman of the Joint Chiefs of Staff should maintain a list of senior DOD officials from organizations that could represent DOD during a Cyber Unified Coordination Group and that are trained in the National Incident Management System. Agency Comments and Our Evaluation We provided a draft of our report to DOD for review and comment. In its written comments, DOD partially concurred with our first recommendation and concurred with the second. DOD’s written comments are reprinted in their entirety in appendix IV. DOD partially concurred with our recommendation to update the department's cyber incident coordination training to incorporate the tenets of PPD-41. In its response, DOD acknowledged the need to continue its emphasis on cyber incident coordination training and states that the department is wholly committed to updating the appropriate training as part of its formal after action reviews during each exercise and training event. DOD stated that it prepares for cyber incidents by exercising interagency roles and responsibilities, and command and control within a cyber threat scenario. While these exercises emphasize the development of comprehensive cyber incident response plans and seek to foster cyber incident coordination, DOD did not identify any specific exercise or training event in which the department will incorporate the tenets of PPD- 41. Accordingly, we continue to believe that our recommendation is warranted. As we reported and DOD acknowledged, Cyber Guard is a tactical-level exercise that would not fully incorporate all DOD components that would participate in a unified cyber response consistent with PPD-41. DOD would meet the intent of our recommendation by conducting one or more cyber incident exercises that incorporate the tenets of PPD-41 into command and control (i.e., operational-level) relationships across all relevant commands and not just across CYBERCOM. DOD concurred with our recommendation that the Joint Staff maintain a list of senior DOD officials from organizations who could represent DOD during a Cyber Unified Coordination Group and who are trained in the National Incident Management System. DOD stated that the Joint Staff will ensure that senior DOD personnel are familiar with the National Incident Management System, or advised by personnel that are, prior to representing the department during a Cyber Unified Coordination Group. The department also plans to re-emphasize these efforts as part of its onboarding process for newly assigned senior leaders, as appropriate. We are sending copies of this report to the appropriate congressional committees, the Secretary of Defense, the Assistant Secretary of Defense for Homeland Defense and Global Security, the Chairman of the Joint Chiefs of Staff, and the Secretary of Homeland Security. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-9971 or kirschbaumj@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. Key contributors to this report are listed in appendix V. Appendix I: Status of Three Recommendations from Our Recent Reports on Defense Cyber Civil Support During our review of the Department of Defense’s (DOD) Section 1648 report, we followed up on three recommendations from our recent reports that could improve the department’s planning and processes for supporting civil authorities in a cyber incident. Table 2 summarizes the status of these recommendations. Appendix II: Section 1648 (a) of the National Defense Authorization Act for Fiscal Year 2016 SEC. 1648. COMPREHENSIVE PLAN AND BIENNIAL EXERCISES ON RESPONDING TO CYBER ATTACKS. (a) COMPREHENSIVE PLAN OF DEPARTMENT OF DEFENSE TO SUPPORT CIVIL AUTHORITIES IN RESPONSE TO CYBER ATTACKS BY FOREIGN POWERS.— (1) PLAN REQUIRED.— (A) IN GENERAL.—Not later than 180 days after the date of the enactment of this Act, the Secretary of Defense shall develop a comprehensive plan for the United States Cyber Command to support civil authorities in responding to cyber attacks by foreign powers (as defined in section 101 of the Foreign Intelligence Surveillance Act of 1978 (50 U.S.C. 1801)) against the United States or a United States person. (B) ELEMENTS.—The plan required by subparagraph (A) shall include the following: (i) A plan for internal Department of Defense collective training activities that are integrated with exercises conducted with other agencies and State and local governments. (ii) Plans for coordination with the heads of other Federal agencies and State and local governments pursuant to the exercises required under clause (i). (iii) A list of any other exercises previously conducted that are used in the formulation of the plan required by subparagraph (A), such as Operation Noble Eagle. (iv) Descriptions of the roles, responsibilities, and expectations of Federal, State, and local authorities as the Secretary understands them. (v) Descriptions of the roles, responsibilities, and expectations of the active components and reserve components of the Armed Forces. (vi) A description of such legislative and administrative action as may be necessary to carry out the plan required by subparagraph (A). (2) COMPTROLLER GENERAL OF THE UNITED STATES REVIEW OF PLAN.—The Comptroller General of the United States shall review the plan developed under paragraph (1)(A). Appendix III: Objective, Scope, and Methodology Our objective was to determine the extent to which the Department of Defense’s (DOD) Section 1648 report submission addressed the statutorily required elements. To determine the extent to which DOD’s Section 1648 report addressed the statutorily required elements, we analyzed the text of DOD’s Section 1648 report. To conduct our analysis of DOD’s Section 1648 report, two of our analysts analyzed the text of the Section 1648 report and assessed the extent to which the report addressed the six elements required by the statute. The analysts assessed each element in the report as “fully addressed,” “partially addressed,” or “not addressed.” If the Section 1648 report addressed all aspects of the required element, the analysts determined that DOD had “fully addressed” the element. If the report addressed some aspects of a required element, but not all, the analysts determined that DOD had “partially addressed” the element. If the report did not address any aspects of a required element, the analysts determined that DOD “did not address” the element. A third independent analyst reviewed the initial determinations and assessed whether they were accurate. For further information, we met with relevant officials from DOD components—such as from the Office of the Assistant Secretary of Defense for Homeland Defense and Global Security, including the Office of the Deputy Assistant Secretary of Defense for Homeland Defense Integration and Defense Support of Civil Authorities and the Office of the Deputy Assistant Secretary of Defense for Cyber Policy; the Joint Staff; U.S. Northern Command (NORTHCOM); U.S. Pacific Command (PACOM); U.S. Cyber Command (CYBERCOM); and the National Guard Bureau. We also interviewed Department of Homeland Security officials to obtain clarifying and supporting information on the process by which the department plans and prepares for a cyber incident requiring civil support. In the cases in which the analysts determined that the plan did not address some aspects of a required element, they discussed their preliminary analyses with officials from the Office of the Assistant Secretary of Defense for Homeland Defense and Global Security to seek additional information. Additionally, DOD officials offered clarification regarding the Defense Support of Civil Authorities process, DOD roles and responsibilities in civil support, and information on ongoing initiatives. We conducted this performance audit from May 2017 to November 2017 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix IV: Comments from the Department of Defense Appendix V: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the individual named above, Tommy Baril (Assistant Director), Tracy Barnes, David Beardwood, Pamela Davidson, Ashley Houston, Gabrielle Matuzsan, and Spencer Tacktill made key contributions to this report. Related GAO Products Defense Civil Support: DOD Needs to Identify National Guard’s Cyber Capabilities and Address Challenges in Its Exercises. GAO-16-574. Washington, D.C.: September 6, 2016. Civil Support: DOD Needs to Clarify Its Roles and Responsibilities for Defense Support of Civil Authorities during Cyber Incidents. GAO-16-332. Washington, D.C.: April 4, 2016. Civil Support: DOD Is Taking Action to Strengthen Support of Civil Authorities. GAO-15-686T. Washington, D.C.: June 10, 2015. Homeland Defense: DOD Needs to Address Gaps in Homeland Defense and Civil Support Guidance. GAO-13-128. Washington, D.C.: October 24, 2012. Homeland Defense: DOD Can Enhance Efforts to Identify Capabilities to Support Civil Authorities during Disasters. GAO-10-386. Washington, D.C.: March 30, 2010. Homeland Defense: DOD Needs to Take Actions to Enhance Interagency Coordination for Its Homeland Defense and Civil Support Missions. GAO-10-364. Washington, D.C.: March 30, 2010.

Why GAO Did This Study The Presidential Policy Directive on United States Cyber Incident Coordination states that significant cyber incidents are occurring with increasing frequency impacting public and private infrastructure in the United States. Section 1648 of the National Defense Authorization Act for Fiscal Year 2016 included a provision that DOD develop a comprehensive plan for CYBERCOM to support civil authorities in responding to cyberattacks by foreign powers against the United States. Section 1648 also included a provision that GAO review DOD's plan. This review assesses the extent to which DOD's Section 1648 report addressed the statutorily required submission elements. To conduct this work, GAO assessed DOD's Section 1648 report against the elements outlined in the statute. GAO also discussed the Section 1648 report with DOD policy, Joint Chiefs of Staff, combatant commands, and military service officials. What GAO Found The Department of Defense (DOD) did not develop a comprehensive plan for U.S. Cyber Command (CYBERCOM); instead, the department submitted a report consisting of a collection of documents that fully addressed two of the six statutorily required elements; partially addressed three elements; and did not address the sixth element on DOD training activities. Legend: ○ Did not address: Submission does not include required element. GAO also found that, in addition to not addressing the training element in the report, DOD had not ensured that staff are trained as required by the Presidential Policy Directive on United States Cyber Incident Coordination or DOD's Significant Cyber Incident Coordination Procedures, which were included DOD's Section 1648 report. Taking action to improve these areas should help DOD sustain progress it has already made. With the President's decision to elevate CYBERCOM to a unified combatant command, such actions will also help as DOD continues to plan to support civil authorities in response to a cyber incident and where CYBERCOM has a significant role. What GAO Recommends GAO has previously recommended that DOD take actions on elements of the Section 1648 report that were partially addressed. GAO is making two new recommendations that DOD update cyber incident coordination training and maintain a list of officials trained in the National Incident Management System. DOD concurred with maintaining a list of trained officials and partially concurred on updating cyber training. GAO continues to believe the updating recommendation is warranted.

gao_GAO-18-574

gao_GAO-18-574_0

Background Health Care Quality Measures As we have previously reported in reviews of health care quality outside of the MHS, health care quality measures are standard, evidence-based metrics designed to assess the extent to which patients receive health care that increases the likelihood of desired health outcomes and are consistent with current professional knowledge. These measures may be used to assess the quality of care in various settings, including hospitals and physician offices. Health care quality measures are intended to (1) inform providers about opportunities for potential improvements in their delivery of care, (2) encourage or incentivize providers to consistently provide high quality care, and (3) inform consumers about which providers are most likely to deliver high quality care. There are broad categories of clinical quality measures that address various aspects of quality of care. See table 1 for a description of these broad categories of quality measures. The data used to calculate the results of health care quality measures can come from a number of different sources. Some measures often require detailed clinical information obtained from patient medical records, such as process measures that indicate whether timely and effective care was provided in a specific situation, for example, or whether stroke patients received clot-dissolving medication appropriately. Other measures are designed to use information on patient demographics and diagnoses that can be obtained from more readily accessible sources, such as claims data or other administrative data that have already been collected for other purposes such as billing. In addition, patients can be asked directly, usually through surveys, to report on their experiences receiving care. The MHS Structure and Administration of Direct and Purchased Care The MHS is a complex organization in which responsibility for the delivery of health care is primarily shared among the military services—Army, Navy, and Air Force—and the Defense Health Agency (DHA). The Army, Navy and Air Force medical commands report through their service chiefs to their respective military department Secretaries and then to the Secretary of Defense. DHA reports through the Office of the Assistant Secretary of Defense for Health Affairs and the Under Secretary of Defense for Personnel and Readiness to the Secretary of Defense. The Office of the Assistant Secretary of Defense for Health Affairs manages the Defense Health Program appropriation, which funds the medical and health care programs at the medical commands of the military services. As of fiscal year 2018, most of the MTFs, including military hospitals and clinics, were under the direction and control of the military services, which are responsible for staffing, training, and equipping those MTFs to meet mission requirements. DHA has responsibility for the managed care support contracts through which the MHS administers its purchased care, and DHA also administers several MTFs in the vicinity of Washington, DC. Figure 1 depicts the MHS organizational structure. Recently enacted changes will affect the administration of the MTFs in future years. Most notably, DOD will alter administration of the MTFs, shifting responsibility from the military services to DHA. Section 702 of the National Defense Authorization Act for Fiscal Year 2017 (NDAA 2017) directed DOD to give DHA responsibility for the administration of all MTFs, including budgetary matters, information technology, and health care administration and management. In the conference report for NDAA 2017, Congress stated its intention that the creation of a single agency responsible for all MTFs would improve and sustain readiness, reduce costs, and increase efficiency. DOD has since prepared a series of implementation plans as it works to develop the specific policies and procedures to enable this change to take effect starting October 1, 2018. The most recent plan issued by DOD in June 2018 envisions a 3-year transition to be completed October 1, 2021. For purchased care, DOD contracts with civilian health care contractors to manage its civilian providers on a regional basis. The primary responsibilities of these managed care support contractors include the following: developing civilian provider networks, which include hospitals and ensuring adequate access to health care; referring and authorizing beneficiaries to receive health care; processing health care claims; educating providers and beneficiaries; and conducting utilization management and quality management programs. There have been several generations of multi-year contracts since 1996. In July 2016, DOD awarded its fourth generation of managed care support contracts to two regional contractors, and on January 1, 2018, the MHS began health care delivery under these contracts. Selection of Measures to Assess Quality of Direct and Purchased Care According to our review of DOD documents, the MHS uses a structured process to select the measures on its dashboards that are used to assess the quality of direct and purchased care. Specifically, DOD documents state that the core direct care measures that are on the Core Dashboard are selected through the MHS’s performance management system called Partnership for Improvement (P4I), which began in 2015. The documents show that proposals for potential quality measures are developed by work groups that focus on different specialized areas, such as maternity care or mental health. These proposals are reviewed and approved by the Steering Committee for P4I, which develops the list of core quality measures for direct care. The Steering Committee then presents the list of core quality measures to a succession of governance bodies—each of which incorporates representation from the three military services plus DHA—for review and approval. DOD documents indicate that the MHS repeats this process annually as it decides which quality measures to add, drop, or modify for the coming fiscal year from the Core Dashboard. The DOD documents we reviewed lay out a parallel process that the MHS follows to select which purchased care quality measures will be tracked in the Purchased Care Dashboard. A work group that specializes on purchased care issues with representation of DHA and the three military services develops the proposed list of quality measures for the Purchased Care Dashboard. This list is then reviewed and approved by the same succession of governance bodies that decide on the Core Dashboard measures. Officials told us and DOD documents confirmed that the MHS and the purchased care contractors also track additional quality measures that are not included in the Core and Purchased Care Dashboards. For example, MHS clinicians who provide maternity care track a set of measures developed by the National Perinatal Information Center. Similarly, a number of military hospitals report on surgical quality measures to the National Surgical Quality Improvement Program. The MHS also conducts surveys of MHS beneficiaries from which it obtains data for patient experience measures for both direct and purchased care. Additionally, the MHS requires the managed care support contractors that administer the MHS’s networks of civilian providers for purchased care to monitor several different sets of quality measures or indicators, many of which focus on patient safety. These include patient safety indicators, hospital acquired conditions, and serious reportable events. They also analyze measures selected from Hospital Compare and the Healthcare Effectiveness Data and Information Set (HEDIS), some of which correspond to measures included in the Core and Purchased Care Dashboards. The Hospital Compare and Core Quality Measures Collaborative Measure Sets Adopted by Medicare and Private Health Insurers While health care systems in the United States can use a variety of measures to assess the quality of care, two of the most widely adopted sets of quality measures include the Hospital Compare measure set developed by Centers for Medicare & Medicaid Services (CMS) for inpatient care and the CQMC measure sets jointly developed by CMS and major private health insurers for outpatient care. Since 2005, CMS has collected results for individual hospitals on a specific list of health care quality measures that are posted on a website known as Hospital Compare. CMS does this to make comparable information on the quality of care provided by different hospitals publicly available. Hospital Compare currently covers more than 4,000 hospitals that participate in the Medicare program. These hospitals supply data to CMS for quality measures of inpatient and emergency department care. These data reflect the care provided to all patients treated at these hospitals, not just those covered by Medicare. Each year CMS goes through a formal process, including receiving input from experts and stakeholders, to review and revise the mix of quality measures that these hospitals are expected to report. The purpose of this review, according to CMS, is to ensure that the set of measures reported on Hospital Compare provides meaningful information for quality improvement while reducing unnecessary administrative burden. Initiated in 2014, the CQMC is a multi-stakeholder voluntary effort focused on quality measure alignment that has developed eight sets of measures for outpatient primary and specialty care, known as the CQMC measure sets. In developing the measure sets, CMS and private health insurers negotiate sets of core measures on which they agree to focus on measuring care quality for certain conditions. Physician specialty societies, employer groups, consumer groups, and regional collaboratives also participate in the negotiations. The CQMC measure sets have been adopted by CMS for Medicare and by 15 major private health insurers for commercial health plans. Additionally, section 728 of the NDAA 2017 directs the MHS to use, to the extent appropriate, these quality measures to assess the quality of direct and purchased care. CQMC documents show that the members of the CQMC intend to continually update these core measure sets as more meaningful measures are developed over time. CMS and the private health insurers plan to expand their application of these measures incrementally, as CMS conducts its annual reviews of Medicare’s quality measures and the insurers update or renew their contracts with different providers. The MHS Does Not Use a Common Set of Quality Measures for Direct and Purchased Care, and the Measures It Uses Assess a Limited Range of Quality Areas The MHS does not use a common set of measures on its Core and Purchased Care Dashboards to assess the quality of care provided through direct and purchased care. In addition, for both direct and purchased care, the MHS uses measures on its dashboards that assess a more limited range of quality care areas and medical conditions as compared to the Hospital Compare and CQMC measures adopted by Medicare and private health insurers. The MHS Does Not Use a Common Set of Quality Measures for Direct and Purchased Care Although the NDAA 2016 directed the MHS to align its quality measures for direct and purchased care, we found that as of March 31, 2018, the MHS used separate sets of measures on the Core and Purchased Care Dashboards to assess the quality of care delivered in direct and purchased care, respectively. To assess the quality of direct care, the MHS tracks 43 measures on its Core Dashboard, and to assess the quality of purchased care, the MHS tracks 18 measures on its Purchased Care Dashboard. The MHS tracks 8 measures that are the same for both dashboards, leaving 35 measures tracked only on the Core Dashboard for direct care and 10 measures tracked only on the Purchased Care Dashboard for purchased care. (See fig. 2.) According to MHS officials, since launching the P4I performance management system in 2015, the MHS has focused on making systematic improvements in the quality of care across the MTFs in direct care. As a result, the 43 measures they have chosen for the Core Dashboard reflect their priorities for quality improvement within direct care only. In the case of purchased care, MHS officials stated that requiring civilian providers to report on the same 43 measures that are used on the Core Dashboard for direct care would add burden, and the MHS had concerns that this would make civilian providers less likely to participate in purchased care. Instead, the MHS tracks 18 measures on the Purchased Care Dashboard that rely on information sources other than provider reporting, such as claims that the providers submit in the normal course of receiving payment for their services and surveys that the MHS conducts of its beneficiaries. MHS officials explained that they try to minimize the reporting burden for purchased care providers because for most of these civilian providers, eligible MHS beneficiaries represent only a small proportion of their patient population. We also found that for direct care, the MHS uses its quality measures on the Core Dashboard to assess the quality of care delivered to beneficiaries served by individual MTFs, such as hospitals or clinics. However, for purchased care, the MHS uses its quality measures on the Purchased Care Dashboard to assess the quality of care delivered to the beneficiary population served by each contractor’s network as a whole– not the quality of care delivered by individual civilian hospitals, clinicians, or other providers in the network. Specifically: In direct care, the MHS uses the 43 measures on the Core Dashboard to track the quality of care delivered by individual MTFs. For example, on a measure of central line-associated bloodstream infections, the MHS tracks the incidence of such infections by individual MTF and by military service (i.e., the incidence of such infections in Army, Navy and Air Force MTFs). In contrast, in purchased care, the MHS assesses information on the 18 measures on the Purchased Care Dashboard for all beneficiaries in each of the networks administered by the two managed care support contractors. For example, on a measure of the percentage of beneficiaries with diabetes who have their hemoglobin level tested annually, the MHS calculates an overall rate of hemoglobin testing across all the diabetic patients that receive care in each contractor’s network. The beneficiary population-level reporting on quality measures on the Purchased Care Dashboard reflects the nature of the MHS’s relationship with its managed care support contractors for purchased care. Under the terms of the contracts that the MHS has negotiated with the contractors that administer the networks of civilian providers to care for eligible beneficiaries, the contractors bear responsibility for ensuring the quality of care delivered by those providers. While the MHS requires the managed care support contractors to monitor different sets of quality measures or indicators, such as patient safety indicators, hospital acquired conditions, and serious reportable events to identify possible cases of individual patient harm and determine appropriate interventions, the contractors report this information in annual reports to the MHS for their network as a whole, as opposed to reporting on individual providers. Because the MHS largely uses separate measures for direct and purchased care on its dashboards and tracks the quality of care delivered by civilian providers in purchased care in the aggregate rather than individually, the MHS lacks the information it needs to make comparable assessments of the quality of care delivered across the MHS as a whole. This, in turn, limits the MHS’s ability to ensure it has the information needed to determine whether it is achieving the department’s overall strategic goals of providing high quality care across the MHS as a whole and ensuring that beneficiaries receive a consistent level of high quality care regardless of whether that care is delivered in direct or purchased care. Moreover, using a different set of quality of measures on the dashboards for direct and purchased care is inconsistent with section 730 of the NDAA 2016, which directs the MHS to align its measures for direct and purchased care so it can reduce performance variation across the MHS. MHS officials acknowledge in principle the value of using aligned measures to assess quality of care in direct and purchased care, but the officials cited a range of factors that pose challenges to achieving this objective, such as the large number of civilian providers and the lack of common health information technology systems. Based on our review, we found that one way the MHS could have a common set of quality measures for both direct and purchased care, without increasing the reporting burden on civilian providers, would be to use, as appropriate, Hospital Compare and CQMC quality measures. Notably, the MHS states on its website that almost all of the civilian hospitals that are in the contractors’ networks for purchased care already report information on the measures posted on the Hospital Compare website. As a result, there potentially would be no additional burden for these purchased care providers to report information on the Hospital Compare quality measures. Similarly, major health plans report that they have begun implementing the CQMC measure sets in their contracts with physicians, meaning that physicians participating in those plans already report information on CQMC outpatient quality measures. To the extent that those physicians are also in the MHS contractors’ networks for purchased care, the information the physicians report on the CQMC measures could be used by the MHS. We found the MHS is already using some Hospital Compare and CQMC measures for inpatient and outpatient care. There are a total of 76 measures that Medicare and private health insurers report to Hospital Compare and a total of 60 CQMC outpatient measures. Besides the measures used in the Core and Purchased Care Dashboards, MHS also collects 24 of 76 Hospital Compare measures and 10 of the 60 CQMC outpatient measures. For the most part, these measures are not part of the direct and purchased care dashboards that MHS leadership uses to assess the performance of direct and purchased care. Furthermore, MHS officials told us that they have no specific plans to increase the number of measures that the MHS uses from Hospital Compare for inpatient care delivered in its hospitals. In the case of outpatient care, our review of DOD documents shows that the MHS plans on expanding reporting to only 5 more CQMC quality measures, in large part to minimize its reporting burden. The MHS Uses Quality Measures that Assess a Limited Range of Quality Care Areas and Medical Conditions Compared to Measures Adopted by Medicare and Private Health Insurers We found that the measures the MHS uses on its Core and Purchased Care Dashboards to assess the quality of direct and purchased care address only a limited range of quality areas and medical conditions when compared with the Hospital Compare and CQMC measure sets that are adopted by Medicare and private health insurers. According to the National Quality Forum, which plays a central role in developing and annually reassessing the Hospital Compare measure set and also was consulted in the development of the CQMC measure sets, the measures used to assess quality of care should comprise an appropriate mix of recognized measure types, including outcome measures, process measures, experience of care measures, and cost and structure measures. These measures should cover a broad enough range of measure types and medical conditions so that they provide an accurate overall assessment of the quality of care patients receive. Based on our analysis, Table 2 below shows the limited range of measures on the Core and Purchased Care dashboards used by the MHS to assess inpatient care, as compared to the range of inpatient measures that Medicare hospitals report for Hospital Compare. In general, each of the five types of measures shown in the table below addresses different aspects of health care quality in hospital settings. For direct care, the MHS uses no more than one measure on its Core Dashboard for all of these five measure types except for “Outcome” measures; for purchased care, the MHS does not use any inpatient care measures on its Purchased Care Dashboard. Similarly, based on our analysis, Table 3 below shows the limited range of measures on the Core and Purchased Care Dashboards used by the MHS to assess outpatient care, as compared to the range of outpatient measures that are part of the CQMC measure sets adopted by Medicare and private health insurers. The MHS uses measures on its dashboards that assess fewer clinical focus areas and medical conditions as compared with those measures included in the CQMC measure sets. As with hospital care, the difference is greatest with respect to purchased care. The limitations we found in the quality measures used by the MHS—the relatively narrow range of measures as well as the relatively few measures used across direct and purchased care—reflect the MHS’s priorities in selecting quality measures. In short, the MHS focuses on the value and impact of implementing individual measures, but does not prioritize aligning the measures used across direct and purchased care or expanding the range of medical conditions and quality areas covered in the aggregate by the measures. The MHS’s annual assessment of quality measures focuses only the Core Dashboard measures. For each Core Dashboard measure for which a change is under consideration—such as dropping, modifying, or adding another quality measure to the Core Dashboard—MHS officials apply a standard set of criteria involving both the feasibility of collecting the data needed for that measure and the utility of that measure for addressing a strategic priority or promoting performance improvement. When asked about the potential value of increasing the number of Hospital Compare measures, MHS officials said they need to make a value-based determination of whether the benefits of obtaining results for any given Hospital Compare measure justified the costs of collecting and transmitting the data required for that measure. In discussions about potential measures for the Purchased Care Dashboard, MHS officials also focused on the characteristics of specific measures being considered for inclusion in the dashboard. Because the MHS does not prioritize expanding the range of medical conditions and quality areas covered by common measures across direct and purchased care, the measures the MHS uses provide DOD’s senior health care leadership with an incomplete picture of the quality of care across the MHS. As we have noted, the MHS has reported to the Congress that its DOD health care leaders rely on the Core and Purchased Care Dashboard measures to establish accountability throughout the MHS and identify areas where quality improvement is needed. However, the current approach may not lead to the selection of quality measures for the two dashboards that would enable MHS officials to identify the most critical quality of care issues in the MHS. The lack of that information, in turn, limits the ability of DOD’s senior health care leadership to target their performance improvement efforts most effectively in support of DOD’s overall strategic goals of providing high quality care across the MHS as a whole. The MHS Has Established Performance Standards and Related Corrective Action Requirements for Individual Providers in Direct Care but Not in Purchased Care The MHS has established performance standards in direct care related to the Core Dashboard measures and has corrective action requirements for MTFs that do not meet the standards. However, the MHS has not established performance standards related to the Purchased Care Dashboard measures for individual civilian providers in purchased care and therefore does not have related corrective action requirements for these providers. The MHS Has Established Direct Care Performance Standards and Related Corrective Action Requirements for Military Treatment Facilities As part of its P4I performance management system for direct care, the MHS has established specific performance standards that each MTF must meet in delivering quality care to MHS beneficiaries. These standards—some of which are under development—specify a minimum level of performance that each MTF should achieve related to the Core Dashboard quality measures tracked in direct care. For example, in the case of the HEDIS All Cause Readmission measure on the Core Dashboard, the MHS’s performance standard is that MTFs should have a rate of unplanned acute readmissions within 30 days of an initial hospital admission that is as good as or better than the national 75th percentile. This performance standard is based on the readmission rates that the National Committee for Quality Assurance, the lead entity for that measure, has observed across U.S. hospitals. During regularly recurring governance meetings throughout the year, MHS governance bodies review how MTFs have performed relative to the performance standards for the Core Dashboard measures. Our review found that during these meetings, the governance bodies generally do not examine the circumstances of MTFs that do not perform well on the performance standards related to the Core Dashboard measures. Consequently, DOD’s senior health care leadership within the governance bodies may receive limited information on the challenges faced by low-performing individual MTFs. However, during these meetings, officials from the military services and DHA highlight MTFs that are performing well on the established performance standards, and the officials share best practices and specific strategies used to achieve high performance. We also found that in direct care, the MHS requires MTFs that do not meet the MHS’s performance standards related to its Core Dashboard measures to take corrective actions to improve the quality of care they deliver. The military services—Army, Navy and Air Force—and DHA have been responsible for implementing this requirement. For example, Navy officials explained that they periodically review information collected on the MHS’s Core Dashboard quality measures to analyze areas where MTFs do not meet established performance standards tied to these measures and to oversee MTFs’ efforts to correct these deficiencies. Officials told us that each of the services exercises its discretion to independently develop and implement the corrective actions that the service determines best address the performance issues identified through the use of the MHS’s quality measures. For example, to help reduce the number of Central Line-Associated Bloodstream Infections (CLABSI), the Army began financially awarding MTFs that performed well on the CLABSI measure, whereas the Air Force developed a toolkit to help providers prevent CLABSI. As the MHS moves to transfer administration of the MTFs from the individual military services to DHA as directed by section 702 of the NDAA 2017, the approach for assessing performance and implementing corrective actions is likely to change. The MHS’s recently issued implementation plan as of June 2018 outlines some alterations to the current performance assessment process. Specifically, MTFs will create and submit a performance plan that will be reviewed and approved by DHA. DHA will host monthly review sessions with MTFs to track performance on the plan. MTFs will be evaluated using a set of measures aligned to the Quadruple Aim that will include many but not all of the Core Dashboard measures. The MHS Has Not Established Performance Standards and Related Corrective Action Requirements for Individual Civilian Providers in Purchased Care The MHS has not established performance standards related to the 18 Purchased Care Dashboard measures for individual civilian hospitals, clinicians, or other providers in purchased care. Instead, the MHS has established performance standards related to the 18 Purchased Care Dashboard measures that MHS officials use to track the performance of each of the two managed care support contractors. According to MHS officials, the MHS does not require the contractors to ensure that each individual hospital, physician, or other provider in these networks meets the performance standards related to the Purchased Care Dashboard measures. For example, in the case of a measure on the use of imaging for low back pain, the MHS has set a performance standard for each managed care support contractor, one that aims at avoiding excessive imaging across the beneficiary population in the contractor’s network. However, officials told us that the information that the MHS collects on the measure—the number of beneficiaries in each of the contractors’ networks who receive imaging services for low back pain—does not indicate the extent to which each individual civilian provider in the contractor networks meets or fails to meet the performance standard. Thus, the information the MHS obtains on the quality measure and its related performance standard does not identify which hospitals, clinicians, or other providers need to improve their performance in order for all beneficiaries to receive the expected level of care quality that the performance standard represents. Because the MHS has not established performance standards related to the Purchased Care dashboard measures for individual civilian hospitals, clinicians, or other providers in purchased care, there are no related requirements for corrective action. Instead, the MHS requires its managed care support contractors to undertake other activities to promote improved quality of care across civilian providers in their networks. These include investigations of quality issues, focused reviews, analyses of Hospital Compare data, and value-based purchasing pilots, as discussed further below. However, our review found that these efforts are not applied comprehensively across all individual purchased care providers. Investigations of Quality Issues. One approach the MHS uses to promote improved quality of care across purchased care providers is to direct its managed care support contractors to investigate whether individual beneficiaries have experienced what the MHS refers to as a quality issue. Potential quality issues are defined by the MHS as any instance when there are indications that a purchased care provider has deviated from what the managed care support contractors deem acceptable standards of professional practice. The contractors can identify these potential quality issues through beneficiary complaints; analyses of patient safety indicators, hospital acquired conditions, and serious reportable events; or by the MHS or contractor staff. Once potential quality issues are identified, they are investigated by a clinician, who reviews the patient’s complete medical record. Based on the clinician’s review of the patient’s medical records, the clinician verifies whether or not a quality issue has occurred and, if so, assigns the quality issue a severity level. To address the quality issue, the managed care support contractors may take a range of steps, including educating the provider, monitoring the provider, notifying the appropriate state or federal bodies, and removing the provider from the MHS’s purchased care provider network. In practice, however, MHS officials said and documents we reviewed show that providers are rarely removed from the network. For example, MHS officials reported that one contractor estimated that one provider was removed from its network over quality issues every 1 to 2 years. Focused Reviews. Another way the MHS uses its managed care support contractors to promote improved quality of care across purchased care providers is through focused reviews. During these reviews, the managed care support contractors review the medical records for a selected patient population to determine the extent to which a specified quality concern is a widespread problem. For example, in 2015 one contractor reviewed the medical records of 96 beneficiaries to determine the frequency of obstetric trauma, an injury related to vaginal deliveries. If a focused review determines that there is a widespread quality problem, the contractor may implement a quality improvement initiative designed to prompt all of its network providers to address that concern, as opposed to targeting specific providers. Analyses of Hospital Compare Data. The MHS also requires the contractors to conduct an annual examination of the performance of hospitals in their networks on the different quality measures reported on Medicare’s Hospital Compare. However, the managed care support contractors have considerable flexibility in deciding how to structure these analyses and how to follow-up on results. Consequently, the two managed care support contractors have adopted different analytical approaches to define and identify hospitals with relatively low performance. For example, the managed care support contractors chose to examine different quality measures and use different criteria to identify hospitals with relatively low performance. In their most recent annual reports issued during 2017, both managed care support contractors indicated that they were considering contacting the lower performing hospitals to prompt remedial action, but because no action had yet occurred, the reports leave open what steps were ultimately taken and how these hospitals responded. Nonetheless, these activities suggest that the managed care support contractors have the ability to use Hospital Compare to analyze and address individual provider performance on a standard set of quality measures. However, the MHS has not specified how this process should proceed, leaving it to the managed care support contractors to decide what and how much they will do in conducting these analyses of individual hospitals. Value-Based Purchasing Pilots. The MHS has recently begun to test different approaches to incentivize purchased care providers to deliver high quality care through several value-based purchasing pilots. For example, in February 2018 the MHS launched a maternity care pilot that pays providers more for better performance on specified quality measures. The pilot also implements a ‘steerage model’ approach that identifies higher performing providers in directories provided to patients by indicating providers as “Gold Stork” or “Silver Stork.” These pilots may provide the MHS another way to influence the quality of care provided by certain subsets of its purchased care providers. MHS officials stated that although DOD has not arrived at specific goals, it plans to expand these pilots to cover around 20 to 25 percent of its purchased care services by 2020. The use of performance standards and corrective action requirements for individual hospitals, clinicians, or other providers who serve MHS beneficiaries is consistent with federal internal control standards for monitoring, which state that management should establish monitoring activities, evaluate the results, and remediate any deficiencies. While the MHS has established performance standards related to its Core Dashboard measures in direct care and has corrective action requirements for MTFs that do not meet those standards, it has not done so for individual civilian hospitals, clinicians, or other providers in purchased care related to its Purchased Care Dashboard measures. Additionally, if the MHS aligned quality measures on the Core and Purchased Care Dashboards at the provider level, the MHS could require its managed care support contractors to monitor the performance of individual civilian providers relative to set performance standards comparable to the ones that the MHS has established for MTFs. This approach would allow the MHS to determine the extent of performance variability, both among individual civilian providers and across MTFs and individual civilian providers. By not establishing consistent performance standards at the provider-level for direct and purchased care and requiring corrective action requirements to ensure that these standards are met by providers in both direct and purchased care, the MHS is limited in its ability to address variation in the quality of care delivered. This further limits the MHS’s ability to ensure that it is achieving the department’s overall strategic goals of providing high quality care across the MHS as whole and ensuring that beneficiaries receive a consistent level of high quality care regardless of whether that care is delivered in direct or purchased care. Conclusions Congress directed DOD to reduce variation in the quality of care beneficiaries receive through the MHS. DOD has taken important steps towards this goal by identifying a set of core measures that DOD senior health care leadership use to assess quality of care in direct care and another set of measures that they use to assess quality in purchased care. DOD health care leaders rely on these measures on their Core and Purchased Care Dashboards to establish accountability throughout the MHS and identify areas where quality improvement is needed. However, with few exceptions, the MHS uses different measures on its Core and Purchased Care Dashboards to assess the quality of direct and purchased care, making it difficult to determine the extent to which it is ensuring consistent quality across the MHS as a whole. Furthermore, for both direct and purchased care, the MHS uses measures on its dashboards that assess a limited range of quality areas and medical conditions when compared to the widely used quality measure sets adopted by Medicare and private insurers. Without using a broader range of available quality measures available—measures that many purchased care providers already report to CMS and private health insurers—DOD is missing an opportunity to better target the most critical quality of care issues in the MHS. The limitations we identified in the MHS’s Core and Purchased Care Dashboard quality measures reflect the fact that in its annual measure selection process, the MHS does not prioritize aligning the quality measures across direct and purchased care and expanding the range of measures it uses across the two systems of care. Finally, our review shows that while DOD has established performance standards for the core measures in direct care and corrective action requirements for MTFs that do not meet these standards, DOD has not done so for individual purchased care providers. Notably, DOD does not set clear expectations that individual purchased care providers should meet the performance standards related to the quality measures on the Purchased Care Dashboard. Performance standards and related corrective action requirements are critical for holding both MTFs and individual civilian providers accountable for providing quality care. Without consistent standards and related corrective action requirements across the MHS, DOD is limited in its ability to ensure that beneficiaries consistently receive high quality care, regardless of whether they receive that care in the direct or purchased care systems. Recommendations for Executive Action We are making two recommendations to the Assistant Secretary of Defense for Health Affairs. As MHS governing bodies conduct their recurring reviews of quality measures selected for MHS’s Core Dashboard and Purchased Care Dashboards, the Assistant Secretary of Defense for Health Affairs should direct those bodies to prioritize, as appropriate, the selection of measures that apply to both direct and purchased care at the provider level and that expand the range of quality measure types and medical conditions that are assessed. (Recommendation 1) The Assistant Secretary of Defense for Health Affairs should establish, as appropriate, performance standards related to the Purchased Care Dashboard measures that are consistent with the MHS’s performance standards for direct care; ensure they are applied to individual purchased care providers; and take steps, such as amending its managed care support contracts, if necessary, to require corrective actions to be taken when providers do not meet those standards. (Recommendation 2) Agency Comments and Our Evaluation We provided a draft of this report to DOD for review, and DOD provided written comments, which are reprinted in appendix I. In its written comments, DOD concurred with both of our recommendations. With regards to the first recommendation, DOD stated that it plans to enhance the process for selecting quality measures that apply to both direct and purchased care; optimize use of data on Hospital Compare to expand the types and medical conditions evaluated; augment their governance reporting structure so that senior leadership can review quality measures included on the Core Dashboard and Purchased Care Dashboard; and implement the CQMC measure sets for outpatient care. Additionally, DOD stated that it has efforts underway to create a library of all quality measures used across direct and purchased care. With regards to the second recommendation, DOD acknowledged the need to strengthen accountability for meeting performance standards that apply to both direct and purchased care providers. It also agreed that measures of individual provider performance in purchased care should be augmented and consistent with measures in direct care, where possible. DOD noted, however, that because it works through managed care support contractors for purchased care, it can hold the contractors accountable for meeting performance standards but cannot currently take action against individual providers based solely on performance. Instead, DOD stated that rather than taking a corrective action approach, it plans to expand its value-based purchasing efforts and incentivize providers that meet and exceed certain quality standards. This raises concerns, as DOD’s current plans to expand its value-based purchasing efforts would only be applicable for between 20 and 25 percent of the services MHS beneficiaries receive from purchased care providers by 2020, as we noted in our report. Without having all providers managed consistently and subject to prompt remediation of deficiencies, DOD is missing an opportunity to improve the quality of purchased care, and it increases the risk that not all beneficiaries will receive a consistent level of high quality care across the MHS. Acknowledging DOD’s comment that it cannot currently take action against individual providers based solely on performance, we have modified our recommendation to clarify that DOD should take the steps it determines are necessary, such as amending its managed care support contracts, to institute corrective action requirements for purchased care providers. We are sending copies of this report to the Secretary of Defense and appropriate congressional committees. The report is also available at no charge on GAO’s website at http://www.gao.gov. If you or your staff has any questions regarding this report, please contact me at (202) 512-7114 or silass@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix II. Appendix I: Comments from the Department of Defense Appendix II: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Rashmi Agarwal, Assistant Director; Eric Peterson Analyst-in-Charge; Muriel Brown; Shaunessye Curry; Michael Erb; Krister Friday; Jacquelyn Hamilton; and Colbie Holderness made key contributions to this report. Related GAO Products Defense Health Reform: Steps Taken to Plan the Transfer of the Administration of the Military Treatment Facilities to the Defense Health Agency, but Work Remains to Finalize the Plan. GAO-17-791R. Washington, D.C.: Sep 29, 2017. Health Care Quality: HHS Should Set Priorities and Comprehensively Plan Its Efforts to Better Align Health Quality Measures. GAO-17-5. Washington, D.C.: Oct 13, 2016. VA Health Care Quality: VA Should Improve the Information It Publicly Reports on the Quality of Care at Its Medical Facilities. GAO-17-741. Washington, D.C.: Sep 29, 2017.

Why GAO Did This Study The National Defense Authorization Act for fiscal year 2016 contains provisions for GAO to review DOD's plans to (1) improve the experience of beneficiaries who receive care through military hospitals and clinics or from civilian providers and (2) reduce variation in the quality of care. In this report, GAO examines (1) measures DOD uses to assess the quality of direct and purchased care, and (2) the extent to which DOD has established performance standards related to the measures and corrective action requirements for providers who do not meet those standards. GAO reviewed the measures in DOD's Core Dashboard for direct care and Purchased Care Dashboard for purchased care. It also reviewed DOD documents and reports to Congress, and interviewed MHS officials, including officials from the Army, Navy, and Air Force. GAO also compared the quality measures DOD uses to those used in Medicare and by private insurers, which have been vetted by multiple stakeholders. GAO assessed DOD's use of performance standards and corrective action requirements in the context of federal internal control standards. What GAO Found The National Defense Authorization Act for fiscal year 2016 (NDAA 2016) directed the Department of Defense (DOD) to align its measures of health care quality used in the Military Health System (MHS) to improve beneficiary experience and reduce variation in the quality of care. GAO reviewed the quality measures DOD identified in March 2017 in response to the NDAA 2016; DOD senior leadership tracks these measures on dashboards to gauge progress on MHS strategic goals. GAO found that DOD does not use a common set of measures on its dashboards to assess the quality of care provided by either military hospitals and clinics—known as direct care—or networks of civilian hospitals and other providers, known as purchased care. (See figure.) As a result, DOD's senior leadership has limited information on the extent to which MHS beneficiaries receive consistently high quality care across the MHS. Furthermore, for both direct and purchased care, DOD uses measures on its dashboards that track a limited range of quality care areas and medical conditions compared to the measures adopted by Medicare and by private health insurers. For example, whereas civilian hospitals report to Medicare information on 11 measures of patients' self-reported experience in hospitals, Military hospitals report only 1 such measure. By using a limited range of quality measures, DOD may not detect key quality issues. Further, when selecting quality measures, the MHS does not prioritize using common measures across direct and purchased care or expanding the range of measures it uses. GAO also found that for direct care DOD has established performance standards and corrective action requirements for military hospitals or clinics that do not meet those standards in direct care. The performance standards indicate the level of performance providers should meet on the various quality measures DOD tracks on its dashboards, and the corrective action requirements instruct providers to take steps to improve care. However, for purchased care, DOD has not established similar performance standards for individual providers. Without consistent performance standards and corrective action requirements, DOD is limited in its ability to address variation in the quality of care delivered and help ensure that its beneficiaries receive consistent high quality care across the MHS. What GAO Recommends The MHS should (1) prioritize, as appropriate, selecting quality measures common for both direct and purchased care that expand the range of quality areas covered by the measures and (2) establish consistent performance standards and corrective action requirements for direct and purchased care providers. DOD concurred with both recommendations.

gao_GAO-19-4

gao_GAO-19-4_0

Background Statutory Provisions for Entering into IGSAs The IGSA statute (10 U.S.C. § 2679) authorizes such agreements based on a determination that the agreement will serve the best interests of the department by creating efficiencies or economies of scale, including by reducing costs, or by enhancing mission effectiveness. The law also states that IGSAs are not subject to other provisions of law governing the award of federal government contracts for goods and services. In addition, IGSAs may be entered into on a sole source basis with a state or local government and may use wage rates normally paid by that state or local government. At the same time, there are limitations on the use of IGSAs. Specifically, any installation services obtained through an IGSA must already be provided by the state or local government for its own use, and any contract awarded by the federal government or by a state or local government pursuant to an IGSA must be awarded competitively. In addition, IGSAs cannot be used to circumvent the requirements of Office of Management and Budget Circular A-76, which governs competitions to determine whether commercial activities should be performed by government employees or by private contractors. Finally, IGSAs are statutorily limited to a term of no more than 10 years, but the statute does not preclude their renewal after the initial agreement period ends. Process for Developing, Approving, and Implementing IGSAs The military services each have a process for developing, approving, and implementing IGSAs. These processes generally begin with meetings between installation and state or local government officials to discuss services the installation requires that the state or local government could provide. If there is agreement that an IGSA could be beneficial to both parties, installation officials put together an IGSA proposal for obtaining the service from the state or local government. Proposals are required to include a business case analysis showing the proposed IGSA is expected to provide a financial or nonfinancial benefit. For example, Army Regulation 5-9, Installation Agreements, states that an Army installation must submit a proposal and a cost benefit analysis that demonstrates the IGSA will bring financial benefits. Similarly, a memorandum from the Assistant Secretary of the Navy for Energy, Installations, and Environment requires that Navy and Marine Corps installations include material describing the IGSA’s risks and benefits, including financial benefits and enhanced mission effectiveness. Air Force guidance requires that IGSA proposals include a business case analysis and meet the purpose of 10 U.S.C. § 2679 by either bringing financial benefits or enhancing mission effectiveness. Once an IGSA proposal is complete, installation commanders either approve it or submit it to a higher command for review and approval, according to certain dollar thresholds. Figure 1 shows the office within each military service that reviews the proposed IGSA for approval, based on the dollar thresholds. Once a proposed IGSA is approved, installation officials draft the agreement in coordination with state or local government officials. When finalized, representatives of the military service and the state or local government sign the agreement. For Army, Navy, and Marine Corps IGSAs, the installation commander has authority to sign the IGSA, while Air Force IGSAs must be signed by an installation contracting officer. The IGSA is then implemented. Military Services Have Benefitted from Selected IGSAs but Are Not Fully Monitoring the Benefits of Implemented IGSAs Military Services Have Realized Financial and Nonfinancial Benefits from Selected IGSAs Our analysis of a sample of 8 implemented IGSAs, and interviews with officials about these and other IGSAs, found that the military services have realized financial (i.e., cost savings and cost avoidances) and nonfinancial (e.g., enhanced mission effectiveness) benefits from these agreements. Cost Savings In 5 of the 8 IGSAs in our sample, we found that the actual cost of each IGSA during its first year of implementation was lower than the expected cost of obtaining the installation service through an alternative contract, as shown in the respective business case analysis for each IGSA. Table 1 provides each installation’s estimated cost for obtaining the installation service through an alternative contract and the estimated cost for obtaining the installation service through an IGSA; the actual cost paid by the installation to the local government for the first year of each IGSA, based on our analysis of monthly invoices; and our calculation of the estimated realized cost savings achieved from using the IGSA, relative to the alternative. Overall, we found that the estimated cost savings realized by these 5 IGSAs totaled about $2.4 million during the first year of implementation. For example: Fort Polk realized an estimated $1.9 million in cost savings by implementing an IGSA for waste removal with its local government. Specifically, installation officials estimated that a private contract would have cost the installation about $4.5 million from June 2017 through May 2018, while we found that the implemented IGSA cost about $2.6 million for the same period. Moody Air Force Base realized an estimated $270,000 in savings in fiscal year 2017 by implementing an IGSA for water and wastewater treatment. In the IGSA’s business case analysis, officials estimated that continuing to obtain this service from their existing contractor would have cost about $642,000, while the implemented IGSA cost was about $372,000, according to our analysis. Marine Corps Logistics Base Barstow realized an estimated $68,000 in cost savings during the first 9 months of its IGSA for water testing and analysis with the local government. Specifically, installation officials estimated that renewing the previous contract would have cost about $153,000 for 1 year, or about $115,000 for 9 months, while we found that the actual cost under the implemented IGSA was about $47,000 for 9 months. Cost Avoidances The other 3 IGSAs in our sample resulted in cost avoidances, according to installation officials. First, officials at Fort Sill, Oklahoma, told us that the Army Medical Command and the Army Public Health Command had previously provided stray animal control without cost to the installation. When this arrangement ended, Fort Sill had to find an alternative. Officials stated that implementing an IGSA with the city of Lawton, Oklahoma, allowed the installation to avoid the higher costs of a private contractor or of renovating facilities and hiring civilians to perform these duties. Second, Fort Bragg, North Carolina, implemented an IGSA for stray animal control with Cumberland County, North Carolina, that, according to its proposal documentation, allowed the installation to avoid the cost of replacing its stray animal control facility, which was inadequate and sub-standard. Finally, after 2 years with no contract in place, Fort Bragg implemented an IGSA with the city of Fayetteville, North Carolina, for maintenance services at the Airborne and Special Operations Museum that allowed the installation to avoid the overhead costs and fee involved in securing the services through a contract with the Army Corps of Engineers. Nonfinancial Benefits According to officials from all four services, achieving financial benefits has been a primary purpose for utilizing IGSAs, but IGSAs can also provide nonfinancial benefits—such as enhanced mission effectiveness and readiness, reduced administrative time, and greater flexibility. Enhanced mission effectiveness and readiness. Military service officials cited examples of IGSAs that led to enhanced mission effectiveness and readiness. For example, according to its IGSA proposal package, Fort Polk was using military personnel to conduct grounds maintenance, which was contrary to the Army’s guidance that military personnel, while at Fort Polk, should be training for their mission. Officials told us that once the IGSA was implemented military personnel were no longer assigned to grounds maintenance duty, thus potentially enhancing mission effectiveness. In addition, an official in the Army Partnerships Office told us that the IGSA at the Presidio of Monterey, California, for various installation services enabled the installation to obtain a work order for flood damage to a satellite component within a matter of minutes. Direct contact between installation officials and the local government, he stated, provides a quicker response time and has a significant impact on installation readiness. Reduced administrative time and greater flexibility. Installation officials stated that IGSAs had reduced the time personnel spent on managing the services being provided. For example, Marine Corps Logistics Base Barstow officials stated that the installation’s IGSA with the city of Barstow, California, for water testing and analysis had eliminated the time that installation personnel had to use to manage the previous contract. At the time of our review, they were considering further IGSAs, such as one for tree-trimming, that they said would likely not bring cost savings, but that would provide flexibility and ease of managing due to reduced administrative time and regular communication with city officials. Similarly, officials at Fort Polk and Fort Bragg stated that managing their IGSAs is easier than managing other contracts for services, as they can make any needed changes to the IGSA by working directly with the local government. Other benefits. Installation officials also cited benefits such as improved relations with the local government, better quality of service, and the local community’s stronger commitment to working with the installation, compared with contractors. For example, Moody Air Force Base officials noted that the installation’s IGSA for water and wastewater treatment has been positive because the local government cares about the overall good of the installation, due to its importance to the community. Military Services Are Not Fully Monitoring the Benefits of Implemented IGSAs As part of the approval process for IGSAs, the military services collect information on IGSAs’ potential expected benefits, which are estimated prior to IGSA implementation. However, once IGSAs are implemented, the services do not fully monitor whether these IGSAs are resulting in actual financial and nonfinancial benefits. Standards for Internal Control in the Federal Government states that management should design processes, and document them in policy, to obtain relevant, accurate information that it can use to evaluate the entity’s performance in achieving key objectives and make informed decisions about any needed changes. The standards also state that management communicates such information throughout the entity to support achieving those key objectives. Following are descriptions of the status of each military service’s approach and plans for monitoring the benefits of implemented IGSAs. Army headquarters collects data on the expected financial benefits of IGSAs, based on information provided in the IGSAs’ business case analyses. These data reflect the financial benefits that the installations expect to achieve by using IGSAs, which are estimated prior to IGSA implementation. Army headquarters does not, however, currently monitor whether financial or nonfinancial benefits are actually realized from IGSAs after implementation. Army officials told us in May 2018 that they were drafting guidance that likely will assign responsibility for tracking both the realized financial and nonfinancial benefits of IGSAs, on an annual basis, to the Army’s four land-holding commands. They noted, however, that they have not yet decided what specifically to track or finalized a process for monitoring IGSA benefits and evaluating program performance, but stated that their goal was to have a process in place by the end of 2018. Navy headquarters collected information on the expected benefits of the IGSAs it has thus far approved. In addition, in May 2018, the Navy Installations Command chose 12 high-priority IGSA opportunities identified by its regional commands to focus on for implementation and monitoring. According to the Navy Installations Command official who oversees the Navy’s IGSA efforts, this effort is in the very early stages. The official also stated that the expected financial benefits for these 12 will likely be tracked by the Navy Installations Command, but any monitoring of realized financial benefits after the IGSAs are implemented would be left to the regional commands. On the other hand, the official stated that nonfinancial benefits are very subjective and the Navy has not yet determined what information will be collected. Marine Corps headquarters officials stated that they collect information on the expected benefits of IGSAs, but they are not currently monitoring the actual performance of implemented IGSAs because few are in place and existing IGSAs are less than 2 years old. They added that the Marine Corps plans to establish a process to track and analyze the realized financial benefits of IGSAs, but the headquarters official with lead responsibility for IGSAs did not provide a timeline for doing so. He added that the process will likely task the regional installation commands with tracking cost savings, with headquarters officials collecting and maintaining consolidated regional data. In addition, he did not indicate that the Marine Corps plans to monitor whether nonfinancial benefits are realized by implemented IGSAs. In addition to collecting data on the expected benefits of IGSAs prior to their implementation, Air Force headquarters has taken some steps to monitor the benefits realized by the 8 implemented IGSAs it had in place as of July 2018. Specifically, Air Force Community Partnership Program officials have taken the initiative to request information at the beginning of each fiscal year from Air Force installations on any actual financial benefits realized from their implemented IGSAs, including cost savings and cost avoidance. However, officials stated that submitting information is voluntary for installations, and some installations do not always provide timely information. For example, two installations with IGSAs in place beginning in fiscal year 2015 did not provide information in response to the fiscal years 2016 and 2017 data requests. Officials with the partnerships office also noted that they plan to monitor nonfinancial benefits to use for lessons learned and program talking points, but that information on any nonfinancial benefits from implemented IGSAs was currently being collected anecdotally. The military services generally are not monitoring whether all of their IGSAs are bringing financial and nonfinancial benefits because they have not established formal processes to obtain this information and documented them in their policies or procedures, as called for in Standards for Internal Control in the Federal Government. Specifically, the Army, Navy, and Marine Corps IGSA policies do not include processes for monitoring the degree to which expected benefits from IGSAs were actually realized after implementation. The military services also differ in regard to the types of benefits they plan to monitor in the future. In addition, the informal process used by the Air Force to collect some data on realized IGSA benefits is not documented in Air Force policy or procedures. Officials from all four services stated that they are still in the early stages of developing their IGSA processes. In addition, officials from the Army and the Air Force told us that they believed that it may be premature to monitor IGSA performance because the authorization to use IGSAs has only been in use for 3 years and only a limited number of IGSAs have been approved. We recognize the use of IGSAs is relatively new, but developing and documenting formal processes to collect and monitor information on the benefits realized through implemented IGSAs now, as the services continue to refine their IGSA programs, could assist the services in at least two ways. First, it would provide the services with information they could use to assess the performance of IGSAs in comparison with the expected benefits outlined in the IGSAs’ business case analyses. An accurate assessment of actual performance would provide decision makers with important context when reviewing individual IGSAs for possible renewal, and could inform the services’ decisions on developing and implementing similar agreements in other locations. Second, developing formal processes to monitor the performance of implemented IGSAs would provide information that the military services could communicate internally to their installations as part of their outreach efforts to increase awareness of and, when beneficial, expand the use of IGSAs as a means of achieving financial benefits or enhancing mission effectiveness. Military Services Have Supported IGSA Use but Not Fully Monitored Whether Installations Are Evaluating IGSA Opportunities Military Services Have Developed Approaches for Supporting Installations’ Use of IGSAs The military services have developed various approaches for supporting their installations’ use of IGSAs. These include issuing policies on the use of IGSAs; issuing procedures and templates for IGSA development and approval; and providing headquarters-level support, such as facilitating meetings between installation and state and local government officials. The following are descriptions of these approaches for supporting installations’ use of IGSAs. Policies. The services have issued IGSA policies that, among other things, either direct their installations to evaluate opportunities for using IGSAs to obtain installation services or to implement mutually beneficial partnerships that include IGSAs. Army Installation Management Command policy states that installations are to explore opportunities to enter into IGSAs with state or local governments, and it directs installations to review current, soon-to-expire installation support contracts for possible transition to an IGSA. Similarly, Navy Installations Command and Marine Corps Installations Command policies direct regional commands and installations to investigate and identify existing and potential services that could be provided by the state or local governments surrounding their installations. While Air Force policy does not specifically direct its installations to evaluate opportunities for using IGSAs, it does direct Air Force installations to implement mutually beneficial partnerships with their local communities. Procedures and templates. Each of the military services has issued procedures for its installations to follow in order to develop, obtain approval for, and implement IGSAs. For example, the Marine Corps has issued an IGSA handbook that provides information on the roles and responsibilities of installation and headquarters officials in the IGSA process. The handbook also outlines a process installations can follow to develop an IGSA, which includes identifying a need that an IGSA could address, meeting with potential state and local partners, developing a draft and final IGSA, and signing and implementing the IGSA. Additionally, the Army and the Air Force have provided templates of required documents to help installations through the IGSA approval process. For example, Army Installation Management Command’s IGSA procedures include templates and examples of documents installations are to develop, such as a memorandum from the installation commander describing the IGSA proposal; a business case analysis that demonstrates the benefits of the proposed IGSA; and the IGSA document to be signed by the installation and the state or local government. Similar to the Army and the Air Force, one of the Navy’s regional commands has included templates along with their region-specific IGSA procedures, and a Navy Installation Command official told us that the Navy may adapt those procedures and templates for Navy-wide use. Headquarters support. The Army and the Air Force have established partnership offices within their headquarters that serve as resources to support installations interested in using IGSAs. Support includes facilitating meetings between installation and state and local government officials to identify IGSA opportunities. For example, the Army facilitated such meetings at Fort Polk in late 2016, during which officials identified the IGSA opportunity between Fort Polk and Vernon Parish, Louisiana, for waste removal that was implemented in June 2017. The Air Force partnership office also provides periodic training on IGSAs and other topics, as well as a website with various resources for installations to use in developing IGSAs. Marine Corps Installations Command officials stated that they provide headquarters support to installations for developing IGSAs—to include facilitating meetings between installation and local and state officials—but as a collateral duty to other responsibilities. Navy Installation Command officials said that their regions and installations are to take the lead on IGSA development but that they have offered assistance to regions, as needed. Service officials added that representatives from each military service meet quarterly to discuss their IGSA programs, including best practices and lessons learned. Military Services Have Not Fully Monitored Whether Installations Are Evaluating Opportunities to Use IGSAs Officials from all four military services told us that they are not fully monitoring whether all of the services’ installations are complying with their respective service policies to evaluate opportunities to use IGSAs to reduce costs or enhance mission effectiveness. Army headquarters officials told us that their efforts to date have been focused on raising awareness of IGSAs at installations and removing any obstacles that prevent IGSAs from being approved. However, Army officials stated that they currently do not monitor whether Army installations are evaluating opportunities to use IGSAs, but they said that installations may need greater encouragement from higher headquarters to use IGSAs. Thus, Army officials said they are planning to revise Army IGSA policy to include a process for obtaining information from installation officials on whether they evaluated expiring contracts for transition to IGSAs, as well as any reasons for not doing so, and expect it to be complete by the end of 2018. Additionally, Army officials said they plan to review installation contracts for waste removal services to determine whether IGSAs can be used instead, and that additional installation services will be identified for review in the future. The Navy Installations Command has, as discussed previously, collected a list of IGSA opportunities from the Navy’s regional commands and plans to focus on implementing 12 of them, according to the Navy Installation Command official who oversees the Navy’s IGSA efforts. However, the official said that the Navy Installations Command does not know how the regions identified these IGSA opportunities, and it has not directed the Navy regions to monitor whether each of their installations are evaluating opportunities to use IGSAs going forward. The Navy official said that asking each installation whether it identified any IGSA opportunities would be a fair question in order to avoid missing any potential IGSA opportunities. Marine Corps Installation Command officials said they monitor the efforts of installations that are already in the process of developing an IGSA or that have already implemented an IGSA, but they do not monitor the efforts of other installations in the Marine Corps to identify IGSA opportunities. However, a Marine Corps Installation Command official said that such monitoring could help expand the use of IGSAs in the Marine Corps. Officials in the Air Force partnerships office told us that beginning in fiscal year 2018 they had begun to monitor whether some of their installations are evaluating IGSA opportunities for certain installation services that are needed at all Air Force installations—specifically, waste management, grounds maintenance, and pavement maintenance. Air Force officials stated that they are in the process of contacting installations that have volunteered for the Air Force’s community partnership program—which includes most, but not all, installations for active-duty personnel—to determine whether they have evaluated IGSAs as a means to obtain these services. As of July 2018, the military services had approved 45 IGSAs at 33 installations (see app. I). Opportunities for more IGSAs—and thus opportunities to achieve more financial and nonfinancial benefits similar to those we found in our analysis of 8 selected IGSAs—may exist at the services’ installations, including their more than 160 active-duty installations. Recognizing this potential, the services have directed their installations to evaluate IGSA opportunities or to implement mutually beneficial partnerships with local communities, which can include IGSAs. However, the military services do not know the extent to which their installations are evaluating opportunities for IGSAs because service IGSA policies and procedures do not include a process for monitoring whether these evaluations are occurring or for obtaining information on the outcome of any such evaluations. Standards for Internal Control in the Federal Government states that management should design control activities to achieve objectives, such as monitoring actual performance and comparing it with established goals and objectives. Additionally, those standards state that management should implement those control activities by, for example, documenting responsibilities in policies. Army, Navy, and Marine Corps IGSA policies and procedures do not include a process for monitoring whether installations are complying with service directives to evaluate IGSA opportunities, or for obtaining information on the outcome of those evaluations. Additionally, the process that the Air Force is currently using to monitor whether some of its installations are evaluating opportunities to use IGSAs for specific types of installation services is not documented in Air Force policy or procedures. As a result, it is uncertain whether these and any other monitoring efforts will continue beyond the current leadership of the Air Force partnerships office. Without establishing, implementing, and documenting a process to monitor whether installations are evaluating opportunities to use IGSAs and obtain information on the outcome of those evaluations, which may also identify challenges that could hamper the ability to use IGSAs, the military services do not fully know whether their installations are conducting these evaluations, and thus may be missing opportunities to reduce costs or enhance mission effectiveness. Military Services Have Identified Statutory, Review Time, and Financial Incentive Challenges to Using IGSAs Statutory-Related Challenges to Using IGSAs Air Force and Army officials identified instances in which they did not implement an IGSA because of provisions in the IGSA statute on the term limit for IGSAs—which was originally 5 years and is currently 10 years— and on the prohibition against contracting for services that are designated for federal civilians to provide. Term limits. Buckley Air Force Base, Colorado, and Fairchild Air Force Base, Washington, did not use IGSAs to obtain firing range services because of the IGSA term limit, according to Air Force headquarters and installation officials. In both cases, the installations were considering using IGSAs in which local governments would construct new firing ranges that would be shared by the installation and those local governments. Air Force officials told us that in each case the local governments planned to fund the new construction costs with municipal bonds; however, the repayment periods for those bonds would have been longer than the IGSA term limit, and thus the Air Force would not have been able to sign an IGSA that would have covered the entire term of the repayment periods. For example, an official at Fairchild Air Force Base told us that the local government would not sign an IGSA with a term limit of fewer than 20 years because the local government wants to ensure they receive sufficient funding to repay their bond. As a result, that official from Fairchild Air Force Base told us that the Air Force has continued to use its existing firing range for training, but it needs to be replaced because of ventilation problems and limitations on the types of weapons that can be fired at the range. At Buckley Air Force Base, an official told us that the installation received military construction appropriation funding in fiscal year 2017 to build a new firing range at a cost of $10.5 million—approximately $2 million more than the estimated cost of the IGSA. Air Force officials added that they had discussed increasing the IGSA term limit with Members of Congress. Prohibition against contracting for services designated for federal civilians. According to Army officials, two Army installations— Aberdeen Proving Ground, Maryland, and Fort Leonard Wood, Missouri—decided not to use IGSAs for grounds maintenance because of legal concerns regarding the IGSA statute’s prohibition on using IGSAs to circumvent the requirements of Office of Management and Budget Circular A-76 regarding public-private competitions. According to those requirements, public-private competitions must be performed to determine if government personnel should perform commercial activities that are required by an agency. Further, 10 U.S.C. § 2461 states that no function of DOD that is performed by civilian employees may be converted to performance by a contractor unless based on a public-private competition that follows a detailed list of requirements under that statute. Currently, however, DOD is prohibited from conducting such competitions. Army officials told us that Aberdeen Proving Ground submitted an IGSA proposal in 2016 for grounds maintenance that they expected to result in a cost savings of approximately $1 million annually. However, those services had previously been provided by temporary Army civilian employees. Because of this and based on the Army’s interpretation of the IGSA statute, Army officials said the IGSA proposal was not approved. Additionally, Army officials told us that Fort Leonard Wood also considered using an IGSA for grounds maintenance services in 2017 because it had unfilled civilian positions and was using military personnel instead, which took those personnel away from their primary mission. However, officials said that the installation did not submit an IGSA proposal because officials did not think it would be approved, due to the existing civilian positions. Army officials told us they worked with the Office of the Secretary Defense to try to address some of the legal concerns within the Army regarding these types of IGSA proposals. Specifically, language was included in a May 2018 memorandum from the Assistant Secretary of Defense for Manpower and Reserve Affairs stating that even though DOD is prohibited from conducting Circular A-76 public-private competitions, this does not preclude the use of an IGSA as long as the IGSA is not used to circumvent Circular A-76 requirements. Although the memorandum does not provide any further details, an official with the Army Partnerships Office stated that the memorandum may provide more support for the use of IGSAs during internal legal reviews of IGSA proposals and could result in additional IGSAs being approved. If not, Army officials plan to communicate to Congress the effects of the current language in the IGSA statute and make any appropriate recommendations to address those effects. Review Time-Related Challenges to Using IGSAs Officials from each of the 6 installations we met with during our review told us that the length of time to review and approve IGSAs was a challenge, in part due to the multiple levels of review required before an IGSA is approved. For example: Marine Corps Logistics Base Barstow officials told us that their IGSA proposal for water testing and analysis took approximately 1 year to be reviewed and approved—first at the installation level, then at Marine Corps Installation Command-West (a regional command), and finally at Marine Corps Installation Command headquarters. As a result, officials said they had to continue to pay their contractor for an additional year to perform those services, which they estimate cost them approximately $80,000 more than if the IGSA had been approved and in place. The officials added that IGSAs are a new way to obtain installation services within the Marine Corps, and this IGSA was the Marine Corps’ first, which likely contributed to the long review time. In July 2017, Fort Polk submitted an IGSA proposal for both facility maintenance and repair services and also grounds maintenance services, which would be provided by a local government. However, Fort Polk officials said that approval of the proposal was delayed at Army headquarters because there was concern by those headquarters officials about replacing the existing AbilityOne contractor at Fort Polk, which was providing facility maintenance and repair services for the installation. As a result, Fort Polk re-submitted an IGSA proposal only for the grounds maintenance services, and this narrower IGSA was approved in March 2018—8 months after the original IGSA proposal was submitted. The military services have delegated responsibility to approve IGSAs to lower levels, which could decrease the review and approval time for IGSAs. For example, in January 2018 the Air Force delegated approval authority to installation commanders for IGSAs that cost less than $15 million over a 10-year time frame—with the exception of any IGSAs that obtain installation services currently obtained from an AbilityOne contractor. Financial Incentive- Related Challenges to Using IGSAs An installation may lack a financial incentive to use IGSAs because that installation’s military service may choose to use any realized cost savings for service-level priorities elsewhere. As discussed earlier, we found that 5 installations in our sample realized cost savings from their implemented IGSAs. Three of those installations—Fort Polk, Luke Air Force Base, and Marine Corps Logistics Base Barstow—were able to retain those savings to apply to other installation needs that were not funded, according to installation officials. For example, Fort Polk officials stated that they were able to reallocate savings from the waste removal IGSA to repair landing strips at the installation, and Luke Air Force Base officials told us that the ability to retain IGSA cost savings was an incentive for them to put in the effort to implement an IGSA. Officials at the other 2 installations with IGSA cost savings—Moody Air Force Base and Fort Bragg—told us that those savings were retained by the installations’ higher headquarters. The military services are at various stages in deciding how IGSA cost savings are to be used, according to service officials. Air Force officials said they are considering letting their installations retain IGSA cost savings to incentivize the use of IGSAs. Army officials stated that they do not yet have a policy on using IGSA savings, but their commands are responsible for contributing resources to supporting readiness, which may include the use of IGSA cost savings. Marine Corps officials similarly told us that they have not yet developed a policy, but added that IGSA cost savings will be retained within Marine Corps Installation Command. Finally, Navy officials told us that they have not yet considered a policy that would allow installations to retain any cost savings, although they added that Navy Installations Command does not intend to recoup any IGSA cost savings achieved by an installation or a region. Conclusions DOD budgets about $25 billion annually to operate and support its installations, and our analysis shows that IGSAs have provided opportunities for the military services to reduce some of those costs. However, the services could improve the visibility they have over the performance of IGSAs after implementation. Specifically, developing processes to monitor any benefits being realized from implemented IGSAs and documenting these processes in policies or procedures would enhance the military services’ ability to evaluate the performance of these agreements and provide lessons learned that could inform their efforts to encourage greater use of IGSAs. In addition, the military services have already taken steps to direct and facilitate the use of IGSAs. However, without a process to monitor whether their installations are evaluating opportunities to use IGSAs and obtain explanations of the outcomes of such evaluations, the military services do not have visibility over whether their installations are considering the use of IGSAs, as directed in guidance. Consequently, the services may be missing opportunities to reduce costs or enhance mission effectiveness. Furthermore, by documenting their processes in policies or procedures, the military services will increase the likelihood that such oversight will endure beyond the initiatives of current leadership and officials. Taking these actions would support the military services’ oversight of IGSAs and could potentially expand interest in and the use of IGSAs. Recommendations for Executive Action We are making the following eight recommendations to DOD: The Secretary of the Army should (a) finalize and implement a process to collect and monitor information on the extent to which all implemented IGSAs have resulted in financial and nonfinancial benefits and (b) complete documentation of that process in Army IGSA policy or procedures. (Recommendation 1) The Secretary of the Navy should (a) establish and implement a process to collect and monitor information on the extent to which all implemented IGSAs have resulted in financial and nonfinancial benefits and (b) document that process in Navy IGSA policy or procedures. (Recommendation 2) The Commandant of the Marine Corps should (a) establish and implement a process to collect and monitor information on the extent to which all implemented IGSAs have resulted in financial and nonfinancial benefits and (b) document that process in Marine Corps IGSA policy or procedures. (Recommendation 3) The Secretary of the Air Force should (a) establish and implement a formal process to collect and monitor information on the extent to which all implemented IGSAs have resulted in financial and nonfinancial benefits and (b) document that process in Air Force IGSA policy or procedures. (Recommendation 4) The Secretary of the Army should (a) finalize and implement a process to monitor whether Army installations are evaluating opportunities for using IGSAs and to obtain explanations from installations on the outcome of their evaluations and (b) complete documentation of that process in Army IGSA policy or procedures. (Recommendation 5) The Secretary of the Navy should (a) establish and implement a process to monitor whether Navy installations are evaluating opportunities for using IGSAs and to obtain explanations from installations on the outcome of their evaluations and (b) document that process in Navy IGSA policy or procedures. (Recommendation 6) The Commandant of the Marine Corps should (a) establish and implement a process to monitor whether Marine Corps installations are evaluating opportunities for using IGSAs and to obtain explanations from installations on the outcome of their evaluations and (b) document that process in Marine Corps IGSA policy or procedures. (Recommendation 7) The Secretary of the Air Force should document in Air Force IGSA policy or procedures its process for monitoring whether Air Force installations are evaluating opportunities for using IGSAs. (Recommendation 8) Agency Comments and Our Evaluation We provided a draft of this report to DOD for comment. DOD provided written comments, which are reproduced in appendix III. DOD concurred with six recommendations and non-concurred with two recommendations, but DOD’s response indicates that the department plans to implement all of the actions we recommend. DOD concurred with our six recommendations to the Army, the Navy, and the Air Force. DOD did not concur with our two recommendations to the Marine Corps, stating that the Marine Corps is one of two military services within the Department of the Navy and that the recommendations are unnecessary. While we understand that the Marine Corps is within the Department of the Navy, we made recommendations to the Marine Corps because we learned during the course of our review that the Marine Corps had developed service-specific IGSA processes. For our two recommendations to the Navy, DOD stated that the Deputy Assistant Secretary of the Navy (Installations and Facilities) will issue policy by November 30, 2018, directing the Chief of Naval Operations and the Commandant of the Marine Corps to implement our recommendations. We believe that implementing these actions will meet the intent of our recommendations to the Marine Corps. We are sending copies of this report to the appropriate congressional committees; the Secretary of Defense; the Secretaries of the Army, the Navy, and the Air Force; and the Commandant of the Marine Corps. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-4523 or leporeb@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. Key contributors to this report are listed in appendix IV. Appendix I: Military Services’ Intergovernmental Support Agreements Approved as of July 2018 Table 2 shows the military service, installation, state or local government, and type of installation service for each of the 45 intergovernmental support agreements that have been approved within the military services as of July 25, 2018. Appendix II: Organizations We Met with During This Review We met with officials from the following offices, installations, and local governments during this review. Unless otherwise specified, these organizations are located in or near Washington, D.C. Appendix III: Comments from the Department of Defense Appendix IV: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, individuals who made key contributions to this report include Maria Storts (Assistant Director), Whitney Allen, Vincent Buquicchio, Michele Fejfar, Mae Jones, Amie Lesser, Geoffrey Peck, Ophelia Robinson, Jack Wang, and Erik Wilkins- McKee.

Why GAO Did This Study The Department of Defense (DOD) budgets about $25 billion annually to operate and support its installations. GAO has designated DOD support infrastructure management as a high-risk area since 1997, in part because DOD has needed to reduce its installation support costs. In 2013, Congress authorized the military services to enter into IGSAs with local and state governments to receive installation services, if an agreement will provide financial benefits or enhance mission effectiveness. As of July 2018, the military services had approved 45 IGSAs at 33 installations. In this report, GAO, among other objectives, evaluated the extent to which the military services have (1) realized and monitored the benefits from IGSAs and (2) supported the use of IGSAs and monitored whether installations are evaluating opportunities to use IGSAs. GAO reviewed the IGSA statute and policies and procedures; evaluated a nongeneralizable sample of 8 IGSAs, selected based on factors including the military service involved, the amount of expected financial benefits, and the length of time in place; compared the services' processes and actions against standards for internal control; and interviewed service, installation, and local government officials. What GAO Found Based on analysis of 8 selected intergovernmental support agreements (IGSAs) and interviews with officials, GAO found that the military services have realized financial and nonfinancial benefits from using IGSAs with local or state governments to obtain installation services such as waste removal, grounds maintenance, and stray animal control. Financial benefits. Of the 8 selected IGSAs, 5 resulted in cost savings, in which the actual cost of each IGSA during its first year was lower than the expected cost of a contract the installation had previously used to obtain the installation service. For example, Moody Air Force Base realized an estimated cost savings of $270,000 by using an IGSA for water and wastewater treatment services, versus continuing to obtain this service via contract. Installation officials stated that the other 3 selected IGSAs resulted in cost avoidances, in which the installations used the IGSAs to obtain a service they were not previously paying for at a lower cost than other alternatives. Nonfinancial benefits. According to officials from all four services, IGSAs have provided nonfinancial benefits such as enhanced mission effectiveness and readiness, reduced administrative time, and improved relationships with local communities. However, the military services are not fully monitoring benefits being realized from implemented IGSAs because they have not established formal processes to do so. For example, Navy and Marine Corps officials stated that they are not monitoring the financial and nonfinancial performance of implemented IGSAs in part because they are in the early stages of using IGSAs. The Air Force monitors some information on realized IGSA financial benefits, but this information is not complete because reporting by installations is voluntary. Developing and documenting processes to monitor any realized benefits of implemented IGSAs would provide the services with useful information on IGSA performance as they make decisions on devoting resources to developing and implementing these agreements in other locations. The military services have developed various approaches for supporting the use of IGSAs to reduce costs or enhance mission effectiveness. For example, the services have issued policies and procedures for their installations to follow in order to develop, obtain approval for, and implement IGSAs. However, officials from each of the military services told us they are not fully monitoring whether installations are evaluating opportunities to use IGSAs. For example, Army policy states that installations are to review current, soon-to-expire contracts for possible transition to an IGSA, but Army officials said they are not yet monitoring whether installations are doing so. Without a process in place to monitor whether installations are evaluating opportunities to use IGSAs, the military services do not know the extent to which this is occurring and thus may be missing opportunities to further reduce costs or enhance mission effectiveness. What GAO Recommends GAO is making eight recommendations to monitor both the benefits realized from implemented IGSAs and whether installations are evaluating IGSA opportunities. DOD concurred with six recommendations and non-concurred with two, but plans to implement them all.

gao_GAO-18-75

gao_GAO-18-75_0

Background The F-35 Lightning II program, also known as the Joint Strike Fighter program, is a joint, multinational acquisition intended to develop and field a family of next-generation strike fighter aircraft for the U.S. Air Force, Navy, and Marine Corps (hereinafter referred to as the services), eight international partners, and foreign military sales customers. There are three F-35 variants and each will be a multi-role, stealthy strike aircraft replacement for or complement to legacy fighter aircraft, as seen in figure 1. F-35 Milestones and Stakeholders DOD initiated the F-35 program in October 2001, and it is nearing the end of system development and preparing for operational testing. DOD has also been concurrently fielding and operating a growing fleet of aircraft as part of low-rate initial production. As of August 2017, 253 aircraft have been fielded and are flying from nine locations in the United States and three international locations. The Marine Corps and Air Force declared initial operational capability in 2015 and 2016, respectively, and the Navy is scheduled to declare initial operational capability in 2018. In 2019, DOD plans to begin full-rate production of the aircraft. See figure 2 for a timeline of major events and anticipated fleet growth in the F-35 program. By full-rate production, DOD would generally be required to establish adequate sustainment and support systems for the F-35. Per DOD guidance for weapon system acquisitions, these sustainment and support systems should be defined in a support concept that is incorporated into a sustainment strategy. For the F-35, this concept should comprise the necessary plans to conduct operations, maintenance, and sustainment throughout the system’s life cycle, with the F-35 Life Cycle Sustainment Plan serving as the principal document governing F-35 sustainment. According to F-35 operational requirements, this concept must provide warfighting and peacetime capability with the lowest cost of ownership, and all variants must be able to deploy rapidly, sustain high mission reliability, and sustain a high sortie-generation rate. Sustainment for the F-35 aircraft is a large and complex undertaking with many stakeholders. The F-35 Joint Program Office is responsible for managing and overseeing the support functions required to field and maintain the readiness and operational capability of the F-35 aircraft across the enterprise. The F-35 program currently relies heavily on contractors to provide sustainment support and has two product support integrators. As the product support integrator for the aircraft system, Lockheed Martin is charged with integrating sustainment support for the system, including that for the F-35 supply chain, depot maintenance, and pilot and maintainer training, as well as providing engineering and technical support. Currently, DOD is contracting for sustainment support with Lockheed Martin largely through annual contracts, and according to F-35 program officials, plans to transition to 5-year, fixed-price, performance-based sustainment contracts in 2020. DOD has established a Hybrid Product Support Integrator organization— a collaboration of government and contractor organizations tasked with managing product support to meet the F-35 strategy and performance outcomes. This organization was initially established in 2016 as a part of the F-35 Joint Program Office, and is expected to be fully implemented by 2019. According to program officials, the establishment of the Hybrid Product Support Integrator is an acknowledgement that DOD needs to take a more significant role in providing sustainment support for the F-35. In addition, the U.S. Air Force, Navy, and Marine Corps have each established an F-35 integration office or cell focused on how the services will operate and afford the F-35, among other things. The F-35 Global Support Solution DOD is planning to meet the sustainment requirements of its F-35 customers by providing a common, global support solution. As part of this common solution, participants share critical aspects of sustainment support, some of which are discussed below, and which are in various stages of implementation to support the growing fleet. Depot maintenance: The F-35 sustainment strategy has a two-level maintenance concept, consisting of organizational-level maintenance performed by squadron-level personnel, and depot-level maintenance. Depot-level maintenance includes structural repair, software upgrades, engine system overhaul and repair, component repair, and other activities that require specialized skills, facilities, or tooling to conduct the repairs. DOD is establishing modification and repair capabilities at six military service depots in the United States and additional repair facilities overseas. Supply chain: All F-35 customers, including the U.S. military services and international partners, share a global pool of spare parts, which is managed by Lockheed Martin. According to program officials, these pooled assets are unique to the F-35 and include consumable and repairable spare parts for the airframe, support equipment, pilot flight equipment, and training devices. The services and international partners can also purchase packages of spare parts that are tailored to their individual deployment and shipboard operational requirements. Training: Currently, the F-35 program is conducting pilot and maintainer training at Eglin Air Force Base, Luke Air Force Base, Marine Corps Air Station Beaufort, and Naval Air Station Lemoore. The F-35 program’s training system includes pilot and maintenance training devices and courseware that are tailored to multiple variants and services. Infrastructure: F-35 customers are responsible for setting up their own F-35 facilities—hangars, training facilities, and depots, among other things—and the program office works with them in a supporting role. Sustainment infrastructure requirements to support the F-35 are defined in a series of facility requirement documents that are updated and provided to all customers annually. F-35 Costs and Technical Characteristics Many of the costs of F-35 sustainment—also known as operating and support costs—are allocated across the military services and international partners based upon a number of factors, including the number of aircraft that each customer owns and their operational requirements. Such operating and support costs consist of sustainment costs incurred from the initial system deployment through the end-of-system operations, and they include all costs of operating, maintaining, and supporting a fielded system. The Office of the Director for Cost Assessment and Program Evaluation develops independent cost estimates for F-35 operating and support costs, which are reported in DOD’s annual F-35 Selected Acquisition Report as the official operating and support cost estimates for the program. Additionally, the program office develops an annual estimate for the operating and support costs of maintaining and supporting the F- 35 over its 60-year life cycle, which can differ from the estimate conducted by the Office of the Director for Cost Assessment and Program Evaluation, due in part to differences in assumptions between the two estimates. Additionally, there are numerous factors that will affect life- cycle operating and support costs for the F-35, including aspects of the F- 35 program that are still maturing. These include the following: Reliability and maintainability: Reliability and maintainability data measure aircraft performance to determine how often the aircraft experiences failures and how much time it takes to repair those failures. These data are monitored through a series of metrics that measure the intended performance of the aircraft in meeting its requirements as it progresses toward maturity at a cumulative 200,000 flight hours, with at least 75,000 flight hours each for the F- 35A and F-35B, and 50,000 flight hours for the F-35C. Reliability and maintainability drive sortie-generation rates and the size of the logistics footprint for the F-35, as well as inform program operating and support costs, which are tied to the performance of the system at maturity. Technical data: Technical data for weapon systems include the details necessary to ensure the adequacy of performance, as well as instruction, maintenance, and other actions needed to support weapon systems. Technical data constitute an important part of a weapon system program, such as the F-35. Identifying technical data needs, costs, and ownership are essential for DOD to effectively consider and maximize competition for future product support of F-35 sustainment. DOD Is Currently Sustaining More Than 250 F-35 Aircraft, but Insufficient Planning Has Led to Significant Challenges That Pose Risk to Its Growing Fleet DOD Is Sustaining More Than 250 F-35 Aircraft, but Faces Significant Challenges That Are Affecting Readiness DOD has currently fielded and is sustaining more than 250 F-35 aircraft, and the number is expected to triple by the end of 2021 and keep growing as the program moves into full-rate production. DOD has also supported significant F-35 milestones such as the initial operational capability declarations of the Marine Corps and Air Force in 2015 and 2016, respectively, and the transfer of an operational squadron to Japan in early 2017. As a fifth generation aircraft, the F-35 is intended to improve situational awareness through sensor fusion and will enhance the ability of legacy aircraft to conduct various missions while flying together with it. The F-35 was also designed with increased stealth capabilities, the capacity to carry weapons internally instead of externally to reduce drag and enable stealth, and advanced sensor systems. In particular, the aircraft is designed to execute missions in high-threat areas, requiring fewer support assets and possessing a greater survivability rate as compared with fourth generation aircraft such as the Air Force’s F-16s and the Navy’s and Marine Corps’ F/A-18s. Squadron officials at multiple F-35 locations that we visited expressed enthusiasm for the unique capabilities of the aircraft, such as the increased situational awareness that the F-35 provides pilots relative to legacy aircraft and the relative ease with which pilots are able to learn how to employ its tactical capabilities. They also noted improvement in the performance of the aircraft as it has been continuously developed. However, DOD is facing several key sustainment challenges that pose risks to its ability to meet current and future warfighter readiness requirements, and these could limit the ability of the military services to fully leverage the capabilities of the aircraft. Table 1 summarizes these challenges, which are largely attributable to insufficient planning, as discussed in more detail below. Repair capacity: DOD does not have enough capacity to repair F-35 aircraft parts because the establishment of repair capabilities at the military depots is 6 years behind schedule. There are many different components of the F-35 aircraft that DOD plans to repair at the six military depots within the United States, as documented in an F-35 Depot Implementation Plan. Repair capabilities at the military depots were originally planned to be completed by 2016, but program officials told us that some capabilities have now been delayed until 2022. Program officials in part attributed these delays to the military services not providing enough funding for depot requirements; however, service officials told us that the program office did not clearly identify some depot requirements in a timely manner necessary for the services to fund those requirements. In addition, DOD did not plan for and fund the stocks of material needed to repair parts at the depots—referred to as “lay-in material.” Program officials said that they had incorrectly assumed that lay-in material would be included as part of the contracts for establishing repair capabilities at the military depots. As a result, DOD has had to fund and negotiate additional contracts with the prime contractor for the lay-in material. Currently, moreover, due in part to the late identification of requirements and funding, the lay-in material to support repairs for more than a dozen different aircraft components is not expected to be delivered to the depots until months—or in some cases, years—after the technical capabilities to conduct the repairs have been established. As seen in figure 3, for certain F-35 parts, these delays have resulted in repair times that are significantly longer than those the program had projected, leading to repair backlogs. According to prime contractor officials, because of these capacity shortfalls, DOD is currently relying on the original equipment manufacturers to repair parts, but the capacity of these manufacturers is already strained by requirements to produce the parts needed to support aircraft production. Program officials said that establishing the depot repair capabilities is now the F-35 Joint Program Office Product Support Manager’s top priority. As such, the program is working to implement several different initiatives to accelerate the development of repair capabilities, including trying to better align lay-in material requirements with the activation of repair capabilities, prioritizing the establishment of certain repair capabilities to align with the readiness requirements of the fleet, and looking at options to decrease the amount of time that it takes to establish repair capabilities for each component line. However, program officials said that plans are still preliminary, and that they are unsure how much funding will be available to implement these initiatives. Spare parts: DOD is experiencing shortages of spare parts in the F-35 supply chain, resulting in lower than expected readiness. From January through August 7, 2017, the prime contractor reported that the average percentage of time that F-35 aircraft were unable to fly because they were awaiting parts was about 22 percent—more than double that of DOD’s objective of 10 percent, as seen in figure 4. According to program office and contractor officials, the shortages of spare parts are due in part to the delays in the establishment of depot repair capabilities, incomplete plans and funding that did not account for the long lead time for parts, insufficient amounts of service funding, and poor reliability of certain parts. For instance, 19 percent of F-35 parts have a lead time of more than 2 years. The 2 to 3 years that it takes to procure these parts includes both a lengthy period for contracting and a period for the production of the parts once contracts have been established. However, program office and military service officials told us that the timing of prior service funding authorizations and contract awards did not account for this long lead time to procure parts, resulting in parts that were late to meet the military services’ operational needs. According to DOD officials, the parts within the F-35 global pool of spare parts are unique to the F-35 system and generally cannot be obtained from other sources. The program office and prime contractor have identified steps needed to increase the availability of spare parts to prevent these challenges from worsening as the number of aircraft in the fleet grows, such as improving the production and repair capacity of suppliers and aligning the timing of the military services’ funding authorizations with the required lead time for parts. However, according to DOD documentation, planned funding and contract awards for fiscal years 2018 and 2019 are still forecasted to be later than needed to meet demand for new parts, and the program’s ability to accelerate this timeline is uncertain. Thus, parts shortages are expected to continue for several years and may worsen if DOD and the contractor are not able to fully implement these actions. DOD Has Not Fully Defined Future Technical Data Needs, and Some Technical Data Are Immature DOD has not fully defined all of the technical data it needs from the prime contractor to maximize the potential for future competition of contracts among providers for sustainment requirements, nor does it know the associated costs of these data. In 2014, we recommended that the program office develop a long-term Intellectual Property Strategy to include the identification of all critical technical data needs and their associated costs. As of September 2017, the program has taken some steps to develop an Intellectual Property Strategy, but it has not identified all critical needs and their associated costs. Program officials said that they are currently working with the prime contractor to develop a list of technical data requirements. Program officials said that once this effort is complete, DOD will be in position to begin prioritizing and negotiating for specific data rights that the program needs to facilitate its sustainment plans. Officials acknowledged, however, that there is risk associated with efforts to obtain required technical data rights for F-35 sustainment to promote increased competition because the contractors may not be willing to provide these rights, or the costs may be too high. They also told us that the program office deals with such risks on a case-by-case basis, and that if a data right needed by the program office to implement the sustainment strategy cannot be obtained, then plans will have to be adjusted accordingly. Program officials said that, in some cases, they will likely have to make legal claims against the prime contractor’s technical data rights assertions, based on government funding of such products. Moreover, the technical data needed to repair F-35 aircraft, such as maintenance instructions, are still not fully developed. According to contractor officials, the contractor and DOD have developed and verified more than 84 percent of the unit-level technical data needed to address known maintenance requirements, such as instructions for how to replace specific parts on the aircraft. However, according to program and contractor officials, the technical data needed for maintainers to troubleshoot issues with the aircraft are lagging behind planned development. Such data are intended to help maintainers when the source of a maintenance issue is unclear, by providing guidance on the actions needed to isolate the most likely problems. In the absence of troubleshooting instructions, maintainers sometimes incorrectly identify what needs to be fixed on the aircraft. For instance, officials from one squadron said that the troubleshooting data are sometimes insufficient to pinpoint the issue with the aircraft, which can lead the maintainer to remove a component, order a new part from the contractor, and subsequently find that the new part does not fix the issue—a scenario that is both inefficient and costly. According to program and contractor officials, the immaturity of technical data for troubleshooting maintenance issues could be contributing to the high rate of parts that the F-35 squadrons are sending to the depots for repair that do not actually need to be repaired, resulting in inefficiencies at the depots. For example, officials at one depot we visited said that 68 percent of the parts they receive from F-35 squadrons do not need to be repaired and that the process for testing such parts usually takes nearly 10 hours to complete, which is both inefficient and can add to repair backlogs. DOD Has Not Developed a Plan for Intermediate-level Maintenance Capabilities The Navy and Marine Corps require intermediate-level maintenance capabilities for shipboard deployments because it is more difficult and time-consuming to obtain spare parts, or to send parts to the depots for repair, when onboard a ship. DOD has been conducting analyses to support the requirement and has recently identified the initial intermediate-level repair capabilities that it plans to implement, including select avionics, support equipment, and hydraulic repairs. These decisions will trigger other requirements and related costs that must be planned for—such as for personnel, technical data, support equipment, and updates to policies governing the maintenance of spare parts— before the capability can be implemented. For example, program officials told us that once determinations are made about intermediate-level maintenance, the program will have to develop a plan that specifies what technical data rights are needed, and when, to facilitate intermediate-level maintenance, and will then have to negotiate with the contractor to obtain those technical data rights. In August 2017, the program office identified new funding requirements for DOD to implement initial intermediate-level maintenance capabilities for fiscal years 2019 through 2023. However, these requirements are not currently funded in DOD’s budget, leaving a projected shortfall of $267 million over this time period. Because a funded plan for intermediate-level maintenance is not yet in place, the Marine Corps will not have the desired level of intermediate- level maintenance capabilities for its initial shipboard deployments planned for 2018. Accordingly, it will be highly reliant on the currently challenged F-35 supply chain and depot repair capabilities for support, and will likely experience degraded readiness. In addition, without such a plan, it is unclear whether such capabilities will be available to support the Navy’s first planned F-35 shipboard deployments in 2021. DOD Faces Delays in Required ALIS Development, and Its Development Plan Is Not Fully Funded Central to F-35 sustainment is the Autonomic Logistics Information System (ALIS)—a complex system supporting operations, mission planning, supply-chain management, maintenance, and other processes. However, ALIS is in continuous development, with planned updates that support required sustainment capabilities for years to come. For example, future versions of ALIS are intended to improve data collection and reporting, and to provide capabilities to support intermediate-level maintenance. Historically, ALIS has experienced delays. For instance, an ALIS version that was initially planned to be completed for testing in 2010, is now being tested in 2017. In 2016 we found that DOD did not have a plan to ensure that ALIS was fully functional as key program milestones approached, and we recommended that DOD develop a plan to prioritize and address ALIS risks. Since that time, the program office has implemented this recommendation through the development of an ALIS Technical Roadmap to plan for these requirements. However, emerging requirements, such as to address cyber security vulnerabilities and system obsolescence, will likely lead to changes in the Roadmap that could further delay the date when these sustainment capabilities are provided. Furthermore, the requirements and associated timelines for ALIS development that are identified in this plan may not be realistic because the requirements are not fully funded in upcoming service budgets, resulting in additional risks to the program’s plan. DOD’s Sustainment Plans Do Not Fully Include Key Requirements, Associated Timelines, and Aligned Funding, but Some Initial Steps Are Being Taken As discussed above, DOD’s challenges are due in large part to sustainment plans that do not fully include key requirements, associated timelines, and aligned (that is, timely and sufficient) funding to support those requirements. F-35 program stakeholders have long recognized the program’s need for more comprehensive and detailed planning documents to identify the key activities and decision points necessary to establish sustainment capabilities and guide the F-35 sustainment strategy. For instance, in 2009 an Independent Logistics Assessment team recommended, among other things, that DOD develop a program- wide integrated master schedule that includes key governmental activities and tasks necessary to establish F-35 logistics capabilities required through full-rate production, but the program did not develop such a tool. In 2014 the program office identified the need to establish a road map with clear decision points to prepare the F-35 enterprise for long-term sustainment. Finally, in December 2016 the Under Secretary of Defense for Acquisition, Technology, and Logistics directed the program office to submit an integrated master schedule for the deployment of global F-35 sustainment capabilities by January 2017, which is not yet completed. Program officials said that they are now developing an integrated master schedule, and that this schedule will incorporate major sustainment milestones required to implement the program’s sustainment strategy. DOD is also updating sustainment strategy documents, including the F-35 Life Cycle Sustainment Plan and Acquisition Strategy, to include an Intellectual Property Strategy. However, the timeframes for completion of these documents are uncertain, in part due to ongoing DOD efforts to refine its follow-on modernization plans for the F-35, which will affect the sustainment plans. Thus, the scope and the degree to which these updates will address the challenges that DOD is facing are unclear. For instance, an Office of the Secretary of Defense official charged with reviewing these plans said that there is still significant work to be done by the military services and the program office to identify and align sustainment requirements with funding in order to support the fiscal year 2019 budget process, which will ultimately be necessary to inform these plans. Military service headquarters officials told us that, as customers of the program, they need to better understand from the program office when sustainment capabilities—such as military depots—will be established, and when associated funding is needed to support that schedule. In August 2017, the program office identified some specific funding requirements for the military services, beyond what they have already budgeted for F-35 sustainment, which are needed to address some of the sustainment challenges discussed above—including spare parts shortages, gaps in depot lay-in material, and ALIS development. While this is a positive step by the program office, it also demonstrates that DOD faces a funding shortage of approximately $1.5 billion between fiscal years 2018 and 2023 for F-35 sustainment, as well as significant readiness risks associated with this lack of alignment between requirements and funding. The different elements of F-35 sustainment support are highly integrated, and challenges or delays in one area can significantly affect outcomes in other areas. For example, the delays in established repair capacity at the depots constitute one of the reasons why the program has an insufficient supply of spare parts. Procurement decisions can also significantly affect sustainment outcomes. The Air Force and Marine Corps are considering an acceleration of their purchases of F-35 aircraft, thus creating more demand on the already strained sustainment enterprise, for which DOD has not always provided timely funding (for example, funding for spare parts). Our prior work on acquisition program management has identified a number of key program management practices that can improve program outcomes if implemented, such as clearly establishing well-defined requirements, developing realistic cost estimates and schedules, and securing stable funding that matches resources to requirements. As DOD prepares for the growth of the fleet and attempts to address existing sustainment challenges, its effort to develop an integrated master schedule is a positive step. Such a schedule, if comprehensive and realistic, could be a critical tool to guide the revision of DOD’s sustainment plans to better ensure that the plans that form the basis of its strategy are sufficient to meet warfighter requirements. Ultimately, however, without plans that include all key requirements and decision points with aligned funding, the F-35 program will likely face continual challenges in providing timely sustainment support to the warfighter, and may have difficulties in fully implementing its F-35 sustainment strategy in time to meet the needs of a growing fleet. Further, as the services consider accelerating their purchases of F-35 aircraft, DOD risks purchasing aircraft that the program and the services are not ready to sustain. DOD Is Testing Agreements with the Contractor but May Not Be Well Positioned to Enter into Multi-year, Performance-based Sustainment Contracts by 2020 DOD Is Testing Performance-based Agreements to Incentivize the Prime Contractor DOD is conducting pilot—or trial—performance-based agreements with the prime contractor as a part of its annual cost-reimbursable sustainment contracts, in order to test metrics and performance-management processes. According to F-35 program officials, DOD plans to transition to multi-year, fixed-price, performance-based contracts in fiscal year 2020. Performance-based logistics is a support strategy that emphasizes performance in contracts, rather than delivery of goods and services, and payment is related to the degree to which performance meets contracted standards. In 2012, the Under Secretary of Defense for Acquisition, Technology, and Logistics directed an increased use of performance- based logistics agreements, stating that such agreements can yield significant cost and performance benefits if effectively implemented. DOD has developed a series of performance objectives to provide insight into the level of sustainment support that the prime contractor is providing to the military services. From these objectives, DOD has selected three system-level metrics, listed below, to incentivize the contractor under the pilot performance-based agreements: Air Vehicle Availability (AVA): measures the percentage of total time during which aircraft are safe to fly, available for use, and able to perform at least one tasked mission; Full Mission Capable (FMC): measures the percentage of time during which aircraft are fully capable of accomplishing all tasked missions; Mission Effectiveness (ME): measures the extent to which the F-35 components and mission systems affected the successful completion of each assigned mission. In these pilot agreements, DOD and the contractor together negotiated minimum and objective targets against which the performance of the aircraft—and the support provided by the contractor to enable that performance—is measured. For fiscal years 2016 and 2017, these agreements were 1-year, cost-reimbursable contracts with potential incentives for the contractor based on assessed performance of the aircraft across the three system-level metrics. According to F-35 program officials and documentation we reviewed, DOD plans to establish a 2-year contract for fiscal years 2018 through 2019, with select elements that are performance-based, in preparation to transition to a 5- year, fixed-price, performance-based contract for the 2020—2024 time period. Program officials said that this 5-year contract is planned to include 2 base years and 3 pre-negotiated option years. DOD Has Not Achieved Most of Its Performance Targets for the Pilot Agreements and May Not Be Using the Appropriate Metrics to Achieve Desired Outcomes DOD Has Not Achieved Most of Its Performance Targets DOD did not achieve most of the performance targets that it set for the pilot performance-based agreements for the 2016 sustainment contract. Subsequently, DOD negotiated lower targets for some metrics in the 2017 sustainment contract. As of June 2017, DOD was meeting several of the minimum targets established in the 2017 sustainment contract, but none of the objective targets. According to program and contractor officials, the failure to meet these targets is largely due to the sustainment challenges that we discussed previously in this report. For example, the limited availability of spare parts within the F-35 supply chain is contributing to lower than expected AVA and FMC rates. Figure 5 below shows the actual fleet performance results for the 2016 and 2017 (through June 2017) pilot performance-based agreements. The 2016 pilot performance- based agreement began in March 2016 and spanned a 10-month period, through December 2016. The 2017 agreement began in March 2017, and program officials said that it is expected to continue through February 2018. Furthermore, the performance targets established in the sustainment contracts for the pilot performance-based agreements are lower than the desired aircraft performance targets that the services have identified for their aircraft. As part of the pilot performance-based agreements, each of the military services has established individual agreements with the program office that identify their respective required levels of minimum and objective aircraft performance for their units, across key metrics. Program officials said that while they try to meet the services’ performance requirements when negotiating the contracts, the agreements with the services are not binding. The performance targets that have been negotiated on the sustainment contracts are generally lower than those required by the services. For instance, the Marine Corps established a minimum performance target for non-deployed units of 60 percent FMC aircraft for 2017, but the minimum target established in the contract for that same metric was 14 percent. Similarly, the Air Force identified a minimum performance target for non-deployed units of 65 percent AVA, but the minimum target established in the contract for that same metric was 52 percent. Program officials said that the costs of meeting the services’ performance requirements would be too high given the current supply chain challenges across the fleet. Figure 6 shows the differences between the performance targets required by the Marine Corps and those that DOD was able to negotiate under the pilot performance-based agreement in 2017. DOD may not be using the appropriate metrics under the pilot performance-based agreements to achieve desired outcomes. DOD guidance states that optimal performance-based contracts use objective, measurable, and manageable metrics that accurately assess the support provider’s performance against the delivery of targeted warfighter outcomes. It also defines ideal metrics as those that are, among other things: (1) reflective of processes over which the contractor has control, and (2) able to motivate desired behavior. We found the following: The contractor does not have full control over the performance outcomes for which it is paid: The system-level metrics that the prime contractor is being assessed against are not fully reflective of processes over which the contractor has control, because actions that the F-35 squadrons take when maintaining or operating the aircraft affect the metric outcomes being measured. For example, a contractor official at one site that we visited cited an instance when a military service maintainer towed an aircraft into a hangar and broke a surface panel, resulting in the aircraft not being able to fly for 60 days because there was no surface panel replacement available in the supply chain. Thus, the contractor could be held accountable for a lack of performance that the customer created. Conversely, to keep aircraft flying, military service maintainers have taken actions that mask contractor failures to provide support—for example, cannibalizing parts from other aircraft at rates significantly higher than DOD intends, based on data provided by the prime contractor and shown in figure 7. Because the contractor does not fully control the outcomes for which it is being assessed, prime contractor and military service officials said that contentious negotiations have occurred at times about how to assign responsibility for performance. This ultimately makes it difficult for DOD to hold the contractor accountable. Further, one of the three system-level metrics—Mission Effectiveness—is assessed by pilots subjectively after each flight. Some pilots and service officials whom we spoke to said that different pilots may make differing determinations about the effectiveness of the mission, which could affect the measured performance outcomes. DOD has established performance review groups to review and reconcile data in instances where the contractor does not believe that it should be held responsible for certain metric outcomes, but this process requires both DOD and the contractor to make subjective determinations about the root causes of particular performance failures in order to determine whether the contractor or the military services are to blame. Figure 8 shows how this reconciliation process can result in adjustments to the measured performance data when assessing the level of support provided by the contractor. Under the pilot performance-based agreements, the reconciled data points serve as the basis for calculating contractor incentive fees. Additionally, DOD is working to implement agreements that define lower- level metrics for which the military services will be held responsible, such as defining how long it should take for maintainers to conduct maintenance, but these agreements have not yet been fully implemented. Ultimately, service officials told us that the complexity of these adjudication efforts indicates that DOD may not be holding the contractor accountable for the appropriate metrics. Current metrics may not motivate the desired behaviors from all stakeholders. The current metrics may not consistently motivate the necessary behaviors from all stakeholders to either achieve desired warfighter outcomes or meet the current metrics on contract. For example, DOD has established AVA as its primary metric, and it provides greater incentive fees to the contractor for meeting the AVA targets as compared with the other two metrics. However, Marine Corps and Navy officials told us that FMC aircraft are more important for operational deployments, as they represent aircraft that are ready for war. DOD’s performance-based logistics guidance states that it is important to exercise caution when selecting a combination of metrics, to ensure that they do not create undesirable conflicts. The achievement of the AVA and FMC metrics may at times be in conflict with one another. For instance, according to contractor and program officials, an aircraft is still considered to be available if its low observable—or stealth—systems are not working, but for it to be considered a fully-mission capable aircraft, a military service would have to ground the aircraft for several days to repair the low observable system. Contractor officials have also expressed concern that the metrics they are being paid for may not be as important to the services as other factors—such as achievement of flying hours or the ability to train pilots—and that this could affect whether the services will take all necessary actions to meet the targets for which the contractor is paid. Officials from a training unit we visited said that they were focused on training pilots, not on achieving the metric targets identified in the contract. This unit was able to exceed its required flight hours to support pilot training in April 2017, even though the performance of its aircraft fell well below desired Marine Corps performance levels for AVA and FMC. Program office and contractor officials noted that pilot performance-based agreements were put in place to gather lessons learned and ensure that DOD has the appropriate metrics before entering into 5-year, fixed-price contracts. However, contractor officials said that the performance review process does not include a step to review how the metrics are driving behaviors or to determine whether DOD has the appropriate metrics in place, and they suggested that a more robust effort to consider lessons learned from the pilot agreements is needed. Service officials have suggested that incentivizing simpler metrics that focus on individual aspects of F-35 sustainment for which the contractor has more control— such as supply chain responsiveness or depot-level repair—instead of system-level performance metrics may be more appropriate. Without reexamining the metrics to ensure that they are objectively measurable, reflective of processes that the contractor can control, and able to motivate desired behaviors, DOD may not be well positioned to accurately assess contractor performance or achieve optimal outcomes across future performance-based sustainment contracts that will likely cost tens of billions of dollars. DOD Does Not Yet Have Full Information on F-35 Sustainment Costs or Technical Aircraft Characteristics DOD does not yet have full information on F-35 sustainment costs or technical characteristics such as reliability and maintainability, and this could pose risks to its ability to effectively negotiate 5-year, fixed-price performance-based contracts with the prime contractor by 2020. Although DOD has fielded more than 250 aircraft, the aircraft system remains immature. DOD has established a target for system maturity of 200,000 total flight hours, with minimum flight hours for each variant. DOD reached 100,000 total F-35 flight hours in July 2017, and it does not expect to reach its maturity targets for all variants until fiscal year 2024. Specifically, we found that DOD does not have full visibility into the actual costs for some key sustainment requirements that are considered cost- drivers within the program, such as the actual costs of parts and repairs. Given the immaturity of the system, DOD has relied on projected parts reliability and pricing to formulate cost estimates, but officials said that actual costs are needed to improve both their confidence in the estimates and their understanding of how cost is related to performance. There is potential for the actual costs of sustainment requirements to change significantly from initial projections. For instance, the costs of initial spare parts over the life cycle increased by $447 million in the program’s estimate from the 2014 estimate to the 2015 estimate, due largely to increases in unit prices from those initially projected. According to program officials, their understanding of actual costs is limited in part because of the immaturity of the system. Program officials said that they are taking steps to obtain more actual cost information as the aircraft matures, and to determine how much repairs should cost, in order to better position themselves for contract negotiations. However, in addition to system immaturity, program officials said that they are experiencing challenges in obtaining important details about existing cost data needed to inform their cost models from the contractor, such as the costs of the individual parts and repairs that the contractor purchases from its suppliers. Further, we found that there are a number of technical aspects of the aircraft that are immature or uncertain. While the F-35 is meeting expectations for some measurements of reliability and maintainability, other measurements are still lagging behind operational requirements. For example, aircraft are experiencing failures that result in the loss of a capability to perform a mission-essential function at more than twice the rate expected across all variants. Mean repair times for critical components that fail are also more than twice as long as the operational requirements dictate. Additionally, the significant software releases required to complete F-35 system development—referred to as Block 3F—are planned to be tested and released in 2017. However in April 2017 we reported that the program’s schedule for completion of Block 3F and associated testing would likely be delayed due in part to software issues and system instability. Additionally, as of June 2017, the DOD Office of the Director for Operational Test and Evaluation predicted that required initial operational test and evaluation for Block 3F would likely not begin until late 2018 or early 2019. According to operational testing officials, such software releases can lead to different reliability and maintainability issues than were previously known, as the aircraft becomes capable of flying at higher speeds and altitudes. According to these officials, there would be inherent risk in signing a fixed-price, performance-based contract before the reliability and maintainability data for Block 3F are more fully known, as those data will influence how much aircraft performance should cost at maturity. DOD guidance states that in order for performance-based arrangements to be effective, the government must clearly understand program requirements, costs, and technical characteristics; and that systems should achieve a level of maturity and design stability. Program officials said they believe that DOD can gain sufficient knowledge of the costs and technical characteristics of the aircraft prior to 2020, and that they will seek to write options into the multi-year, performance-based contract if there are still risks that need to be mitigated. However, program officials said that the program office has not established criteria addressing the extent of the cost and technical data that it will require prior to entering into the planned agreements. While the program still has a few years until that date, program officials said that the process to develop this contract is expected to begin in late 2017. In April 2017 we reported on the risks of moving forward with additional F-35 program development before DOD has a full understanding of the aircraft’s baseline Block 3F capabilities, specifically citing difficulties in presenting a sound business case for soliciting contractor proposals without such knowledge. The program office could face similar challenges preparing for a fixed-price, performance-based sustainment contract amid existing uncertainty. Without a full understanding of F-35 costs and technical characteristics at maturity, DOD may not be well positioned to accurately determine how much fleet performance should cost over a 5-year, fixed-price, performance-based contract, and thus may be at risk of overpaying the contractor while not receiving the expected level of sustainment support. DOD Has Taken Some Actions Aimed at Reducing F-35 Sustainment Costs, but These Costs Continue to Rise and Are Not Fully Transparent to the Military Services DOD Has Undertaken Some Initiatives to Reduce Rising F-35 Sustainment Costs but Has Not Established Affordability Constraints Based on the Military Services’ Budgets DOD has taken some actions to try to reduce estimated sustainment costs for F-35 operating and support, which, according to the program office’s fiscal year 2016 cost estimate, are projected to cost $1.06 trillion in then-year dollars (see figure 9 below). For example, the program office has established a Cost War Room to identify and implement cost- reduction initiatives with the goal of reducing the program office’s 2012 operating and support cost estimate by 30 percent by 2022. These initiatives include updating assumptions about fuel usage, among others. According to program documentation, such efforts are projected to result in a cost avoidance of $60.7 billion. The program office also has an effort targeted at improving reliability and maintainability of F-35 components. As of May 2017, the program office had completed 38 improvement projects that are expected to result in $1.7 billion in operating and support cost avoidance. However, at the same time, the projected operating and support costs estimated by the program office have increased from fiscal year 2012 to fiscal year 2016, due to an increase in projected flying hours, an extension of the aircraft’s life cycle from 56 to 60 years, and refinements to the cost models, among other factors. Figure 9 shows the increase to the program office’s life cycle operating and support cost estimate since fiscal year 2012. In addition, DOD has not established affordability constraints for the F-35 program that are linked to the military services’ budgets, as we recommended in September 2014. In our prior work, we found that the program’s affordability targets may not be reflective of what the services can actually afford because it did not use the military services’ budgets to establish the targets. At that time, the annual F-35 operating and support costs were estimated to be considerably higher than the combined annual costs of several legacy aircraft, and according to DOD officials, the sustainment strategy was not affordable. We recommended that DOD establish affordability targets linked to the services’ budgets, because without such targets DOD cannot be sure whether the cost savings they are pursuing will lead to an affordable sustainment strategy. The department concurred with this recommendation but has not taken specific action on it at the program level. We made this a priority recommendation for DOD in July 2017. The Senate Armed Services Committee also directed DOD to provide it with a plan for improving the transparency and affordability of the F-35 sustainment strategy, to include identifying affordability constraints linked to, and informed by, the military services’ budgets. The Marine Corps has recently taken steps to develop budget-based affordability targets for their portion of F-35 sustainment costs. The Marine Corps identified the need to reduce steady-state sustainment costs per aircraft by at least 20 percent through cost modeling efforts and budget analysis, and Marine Corps officials said they believe that such a reduction would make the program affordable for the Marine Corps. Marine Corps officials stated that to achieve such reductions, they are exploring options to reduce costs—such as transitioning maintenance tasks from depots to operational units, and revising sustainment support personnel requirements—in coordination with the program office and prime contractor. The program office could use this service target to inform the establishment of program-level affordability constraints. As previously discussed, the program is experiencing sustainment challenges due in part to some requirements not being fully funded, and this could present a continued risk going forward if sustainment for the F- 35 is not affordable within the services’ budgets. Program officials also told us that if the services cannot fully fund sustainment requirements, DOD will have to prioritize funding and defer requirements to later years. However, given the F-35’s global sustainment strategy of providing support across the military services and the international partners through shared pools of funding, a single customer that cannot fully fund requirements may affect the ability of DOD and the contractor to provide adequate sustainment support across the global F-35 fleet. Actual F-35 Sustainment Costs Are Not Fully Transparent to the Military Services F-35 actual sustainment costs that are being charged by the program office to the military services, as well as the capabilities associated with those costs, are not fully transparent to the services. In addition to estimating projected costs for F-35 sustainment over the aircraft’s life cycle as described above, the program office also calculates the actual F- 35 sustainment costs that will be charged to the military services on an annual basis. To determine these actual sustainment costs, the military services first submit their F-35 sustainment capability requirements to the program office for approval. The program office approves requirements as a basis for its annual life-cycle operating and support cost estimate, which is used to provide each of the military services with an estimate for their respective portion of F-35 sustainment costs to support the services’ budget planning process. The program office then negotiates with the prime contractor the level of support the contractor will provide to meet service sustainment requirements. It is at this point that the program office informs the services of the actual costs that they will be charged for contracted sustainment. According to program officials, the contracted level of support may not include all the requirements initially submitted by the military services for a given contract period, and the associated costs of the contract services may not align with initial estimates given to the military services, because support is negotiated between the program office and the prime contractor. Air Force, Navy, and Marine Corps officials told us that they do not fully understand how the actual costs that they are charged by the program office for F-35 sustainment are clearly linked to the capabilities that they are receiving. They cited issues related to unexplained cost increases, difficulty in tracking their requirements to the contracts, and concerns about how to track their dollars to shared pools of sustainment assets, as discussed in detail below. Unexplained or unexpected growth in actual sustainment costs: Service headquarters officials cited concerns about unexplained or unexpected growth in sustainment costs, particularly between the cost estimate that they were quoted for budget planning purposes and what they are actually charged by the program office in the budget execution year. For example, according to program documentation, the Marine Corps was initially given a funding requirement for fiscal year 2017 sustainment support of $293 million, which then increased to $364 million in the execution year, largely due to increases in contractor personnel costs. Marine Corps officials said that the reasons behind this growth in personnel costs were not clearly substantiated for the Marine Corps by the program office. In order to afford these increased costs for sustainment support, Marine Corps officials said that the Marine Corps had to reduce its planned flying hours. In another instance, documents provided by the Navy show that the program office increased the cost of the Navy’s and Marine Corps’ combined spare parts requirements for fiscal year 2017 from an original estimate of $261 million to $402 million over the course of the execution year. In addition, service officials told us that they sometimes become aware of the growth in sustainment costs late in the services’ budgeting process, making it difficult for them to find additional funding for such changes. Tracking requirements to negotiated contract services and costs: Officials from two of the services told us that they have had difficulty in tracking their respective services’ requirements to the costs being charged by the program and the capabilities that are negotiated on the contract. For instance, Air Force officials stated that the Air Force specified a desired performance level for AVA of 65 percent to the program office as a minimum target for its squadrons, but ultimately the program office contracted for a target of 52 percent. Air Force officials said they were not aware of this change until after the contract was negotiated. Similarly, Navy officials also told us that the program office does not notify the Navy of changes from the estimated costs to the actual contract costs or the requirements that are included during negotiations for sustainment contracts, even when the requirements differ from what the Navy intended. As a result, officials said that the services often have limited visibility into the support that the contractor will provide along with the actual costs for which the services are responsible, until after the contract is signed. Shared pools of F-35 sustainment assets: These transparency concerns are complicated by the fact that the services are paying into shared pools for F-35 sustainment, and the costs they are being charged for some requirements—such as for spare parts—cannot be directly tracked to an item that the services own or support that is specifically provided to an individual service. Service officials said that the funds they have contributed to the shared sustainment support have not resulted in the expected sustainment support. Specifically, Air Force officials questioned why key performance points in the program—such as depot repair capabilities and supply availability— are lagging by several years in some instances, and said that they need better accounting from the program office on how the money the Air Force has contributed to the program has been spent, and why those funds have not resulted in improved performance. Furthermore, Air Force officials raised questions about whether all program participants are paying for their required shares of F-35 sustainment costs, and said that they have not been able to obtain such information from the program office. This lack of transparency is due in part to insufficient communication between the program office and the services, particularly as requirements and costs change. Program officials have acknowledged that the program office has not always provided the services with the level of detail and clarity around costs that the services would like, but said that recently the program has been more focused on communicating with the military services. Program officials also told us that the services are free to contact the program office should they have any concerns regarding F-35 sustainment costs and how they are shared. However, given the consistent concerns expressed to us across the services, it appears that this level of dialogue has not been adequate to facilitate the services’ understanding of sustainment costs. Two of the services have requested organizations external to the F-35 program to conduct reviews of the program to better understand their respective portions of F-35 sustainment costs and, in some cases, identify potential opportunities for cost savings. While these studies will likely provide valuable information to the services and the program office, they also add costs to an already expensive weapon system. For example, according to program officials, the contract for the study requested by the Marine Corps has cost the program office at least $2.7 million. Further, reliance on one-time studies by external organizations to help program participants understand their F- 35 sustainment costs and associated capabilities is not a practical substitute for the effective communication needed in a program of this magnitude. F-35 program guidance has emphasized the need to ensure that costs are transparent to stakeholders. Further, our prior work examining programs with multiple governmental customers found that when customers understand how costs and underlying assumptions are determined, they can better anticipate potential changes to those assumptions, identify their effects on costs, and incorporate that information into their budget plans. Without better communication on the relationship between the costs and the associated capabilities delivered for F-35 sustainment support, the military services may not be able to appropriately plan for sustainment costs over the life cycle of the F-35 or to make affordability trade-offs between requirements, as they try to prioritize funding within their budgets. Conclusions DOD’s F-35 program is at a critical juncture. With aircraft development nearing completion within the next few years, DOD must now shift its attention and resources to sustaining the growing F-35 fleet. While production accelerates, DOD’s reactive approach to planning for and funding the capabilities needed to sustain the F-35 has resulted in significant readiness challenges—including multi-year delays in establishing repair capabilities and spare parts shortages. There is little doubt that the F-35 brings unique capabilities to the U.S. military, but without revising sustainment plans to include the key requirements and decision points needed to fully implement the F-35 sustainment strategy, and without aligned funding plans to meet those requirements, DOD is at risk of being unable to leverage the capabilities of the aircraft it has recently purchased. Furthermore, until it improves its plans, DOD faces a larger uncertainty as to whether it can successfully sustain a rapidly expanding fleet. DOD’s plan to enter into multi-year, performance-based contracts with the prime contractor has the potential to produce cost savings and other benefits. However, important lessons are emerging from its pilot agreements with the contractor that are intended to inform the upcoming multi-year contract negotiations. To date, DOD has not achieved the desired aircraft performance under the pilot agreements, but it continues to move quickly toward negotiating longer-term contracts—which are likely to cost tens of billions of dollars—by 2020. Without examining whether it has the appropriate metrics to incentivize the contractor or a sufficient understanding of the actual costs and technical characteristics of the aircraft before entering into multi-year, performance-based contracts, DOD could find itself overpaying for sustainment support that is not sufficient to meet warfighter requirements. Finally, on a broader level, DOD’s projected costs to sustain the F-35 fleet over its life cycle have risen since 2012 despite the department’s concerted efforts to reduce costs. Already the most expensive weapon system in DOD’s history, these rising costs are particularly concerning because the military services do not fully understand what they are paying for. This puts them in a precarious position as they consider critical trade-offs that might make F-35 sustainment more affordable. Without improving communication with the services to help them better understand how the sustainment costs they are being charged relate to the capabilities that they receive, the services may not be able to effectively budget for the F-35 over the long term. Recommendations for Executive Action We are making the following four recommendations to DOD. The Under Secretary of Defense for Acquisition, Technology, and Logistics, in coordination with the F-35 Program Executive Officer, should revise sustainment plans to ensure that they include the key requirements and decision points needed to fully implement the F-35 sustainment strategy and aligned funding plans to meet those requirements. (Recommendation 1) The Under Secretary of Defense for Acquisition, Technology, and Logistics, in coordination with the F-35 Program Executive Officer, should re-examine the metrics that it will use to hold the contractor accountable under the fixed-price, performance-based contracts to ensure that such metrics are objectively measurable, are fully reflective of processes over which the contractor has control, and drive desired behaviors by all stakeholders. (Recommendation 2) The Under Secretary of Defense for Acquisition, Technology, and Logistics, in coordination with the F-35 Program Executive Officer, should, prior to entering into multi-year, fixed-price, performance- based contracts, ensure that DOD has sufficient knowledge of the actual costs of sustainment and technical characteristics of the aircraft after baseline development is complete and the system reaches maturity. (Recommendation 3) The Under Secretary of Defense for Acquisition, Technology, and Logistics, in coordination with the F-35 Program Executive Officer, should take steps to improve communication with the services and provide more information about how the F-35 sustainment costs they are being charged relate to the capabilities received. (Recommendation 4) Agency Comments We provided a draft of this report to DOD for review and comment. In its written comments, reproduced in appendix II, DOD concurred with our recommendations and identified actions that it would take in response. We are sending copies of this report to appropriate congressional committees; the Secretary of Defense; the Under Secretary of Defense for Acquisition, Technology, and Logistics; the F-35 Program Executive Officer; the Secretaries of the Air Force and Navy; and the Commandant of the Marine Corps. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-5431 or russellc@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. Staff members making key contributions to this report are listed in appendix III. Appendix I: Scope and Methodology For each of our objectives, we reviewed relevant sustainment plans, guidance, and program documentation, and collected information by interviewing officials from the Office of the Assistant Secretary of Defense (Logistics and Materiel Readiness), the F-35 Joint Program Office, the U.S. Air Force, the U.S. Navy, the U.S. Marine Corps, and the prime contractor, Lockheed Martin. To interview officials and observe F-35 operations, maintenance, and training, we conducted visits to two F-35 operational locations—Hill Air Force Base, Utah, and Marine Corps Air Station Iwakuni, Japan; two F-35 training locations—Eglin Air Force Base, Florida, which also includes a Navy F-35 training squadron, and Marine Corps Air Station Beaufort, South Carolina; and two F-35 maintenance depots—Ogden Air Logistics Complex, Utah, and Fleet Readiness Center Southeast, Florida. A full listing of organizations with whom we met is provided later in this appendix. We also gathered various data related to F-35 sustainment, such as supply chain and repair data and aircraft performance data. To determine the reliability of these data, we collected information on how the data were collected, managed, and used through a questionnaire and interviews with cognizant Department of Defense (DOD) officials and the prime contractor. In our assessment, we identified some limitations in the way that certain data are collected and reported, such as data related to aircraft performance (Air Vehicle Availability, Full Mission Capable, and Mission Effectiveness metrics), data related to aircraft that are not mission capable due to supply issues, and parts cannibalization rates that could potentially result in inaccuracies. However, these data come from the program’s data systems of record, and are the same data used by the program office and prime contractor to monitor the health of the supply chain and assess aircraft performance against contract requirements and program objectives. As such, they are the best source of data available to provide information on the progress and challenges within the program. We determined that these data presented in our findings are sufficiently reliable for the way in which we report them. Specifically, the parts cannibalization rates are consistent with the trends observed across other key data elements within the program, and with the testimonial evidence provided to us by the units with whom we met during our review, and are sufficiently reliable to report as a data trend relative to program objectives. All other performance data presented in our report are sufficiently reliable to present as specific data points, in order to describe the status of sustainment requirements and measured aircraft performance across key metrics as reported by the prime contractor and DOD. To assess the status of DOD’s efforts to sustain the F-35 fleet and any challenges it has faced, we reviewed DOD and contractor plans, briefings, and schedules to determine the current status of key requirements and decision points necessary to establish F-35 sustainment capabilities, such as depot and other maintenance capabilities, the supply chain, technical data, and development of key software systems, among other things, and spoke with cognizant officials about these issues. We also compared actual data obtained about F-35 repair and supply chain capabilities with DOD’s objectives for these capabilities to identify areas of challenge for the program. Specifically, we obtained data on aircraft that were not mission capable due to supply issues from January 2017 through August 7, 2017 and average repair times as of May 2017, in order to provide the most recently available information about the health of the supply chain. As discussed above, we determined that these data are sufficiently reliable to present as specific data points. In addition, we identified key acquisition program management practices that can improve program outcomes if implemented—such as clearly establishing well-defined requirements, developing realistic cost estimates and schedules, and securing stable funding that matches resources to requirements—and assessed DOD’s sustainment planning efforts against these criteria. To assess the extent to which DOD is positioned to enter into multi-year performance-based F-35 sustainment contracts, we reviewed documentation related to DOD’s pilot—or trial— performance-based agreements for F-35 sustainment, such as sustainment contracts, readiness data provided by the military services, metric taxonomies, and agreements between the program office and the military services that identify the services’ desired performance targets. We also obtained aircraft performance data from the Sustainment Performance Management System for the 2016 pilot performance-based agreement (March 2016 – December 2016) and the 2017 pilot performance-based agreement (March 2017 – June 2017) to the extent available at the time we completed our audit work. As discussed above, we determined that these data are sufficiently reliable to present as specific data points. These time periods are the only time periods for which the program office has assessed contractor performance under these pilot arrangements. We also reviewed performance-management guidance and processes and interviewed officials to determine how performance data are being collected and assessed. In addition, we reviewed aircraft maturity, reliability, and maintainability data, and documentation related to cost- visibility issues, and we spoke with cognizant officials about these issues to determine DOD’s level of understanding of the costs and technical characteristics that will affect future sustainment support. In addition, we obtained cannibalization data from March 2016 to March 2017 in order to review and report recent trends in cannibalization rates over a time in which the program has introduced a significant amount of aircraft to the fleet. As discussed above, we determined that these data are sufficiently reliable to present as trend data relative to the program objective. Further, we reviewed DOD guidance and best practices related to performance- based agreements to identify attributes of ideal performance metrics and effective performance-based agreements. We then compared these attributes with the information described above to determine whether DOD has the appropriate metrics to achieve desired outcomes and the necessary information to effectively negotiate multi-year, fixed-price, performance-based contracts with the prime contractor by 2020, as planned. To assess the progress, if any, DOD has made toward reducing F-35 sustainment costs, and the extent to which costs are transparent to the military services, we reviewed F-35 Joint Program Office sustainment- cost estimates from fiscal year 2012 to fiscal year 2016 in order to identify changes to the estimate since the program’s sustainment cost baseline was established in 2012; documentation related to program office and service cost-reduction efforts; sustainment contracts; F-35 cost-sharing rules; and budget documentation from both the program office and the military services. The fiscal year 2016 sustainment-cost estimate is the most current cost estimate conducted by the program office. In addition, we interviewed cognizant officials from the F-35 Joint Program Office and military services to discuss how the program office informs the military services of F-35 sustainment costs, and the degree to which the services understand these costs and the sustainment capabilities provided for those costs. We compared this information with program guidance and with key operating principles for programs that involve multiple governmental customers identified in our prior work in order to assess the transparency of F-35 sustainment costs for the military services. We conducted this performance audit from October 2016 to October 2017 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. We met with officials from the following Department of Defense (DOD) and contractor organizations during our review. We selected these organizations based on their oversight, planning, and execution roles in support of F-35 sustainment and operations. DOD Organizations Other Organizations Appendix II: Comments from the Department of Defense Appendix III: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Alissa Czyz (Assistant Director), Vincent Buquicchio, Kasea Hamar, Jeff Hubbard, Amie Lesser, Sean Manzano, Carol Petersen, Clarice Ransom, Michael Silver, Maria Staunton, Cheryl Weissman, and Delia Zee made key contributions to this report. Related GAO Products F-35 Joint Strike Fighter: DOD’s Proposed Follow-on Modernization Acquisition Strategy Reflects an Incremental Approach Although Plans Are Not Yet Finalized. GAO-17-690R. Washington, D.C.: August 8, 2017. F-35 Joint Strike Fighter: DOD Needs to Complete Developmental Testing Before Making Significant New Investments. GAO-17-351. Washington, D.C.: April 24, 2017. F-35 Joint Strike Fighter: Continued Oversight Needed as Program Plans to Begin Development of New Capabilities. GAO-16-390. Washington, D.C.: April 14, 2016. F-35 Sustainment: DOD Needs a Plan to Address Risks Related to Its Central Logistics System. GAO-16-439. Washington, D.C.: April 14, 2016. F-35 Joint Strike Fighter: Preliminary Observations on Program Progress. GAO-16-489T. Washington, D.C.: March 23, 2016. F-35 Joint Strike Fighter: Assessment Needed to Address Affordability Challenges. GAO-15-364. Washington, D.C.: April 14, 2015. F-35 Sustainment: Need for Affordable Strategy, Greater Attention to Risks, and Improved Cost Estimates. GAO-14-778. Washington, D.C.: September 23, 2014. F-35 Joint Strike Fighter: Slower Than Expected Progress in Software Testing May Limit Initial Warfighting Capabilities. GAO-14-468T. Washington, D.C.: March 26, 2014. F-35 Joint Strike Fighter: Problems Completing Software Testing May Hinder Delivery of Expected Warfighting Capabilities. GAO-14-322. Washington, D.C.: March 24, 2014. F-35 Joint Strike Fighter: Restructuring Has Improved the Program, but Affordability Challenges and Other Risks Remain. GAO-13-690T. Washington, D.C.: June 19, 2013. F-35 Joint Strike Fighter: Program Has Improved in Some Areas, but Affordability Challenges and Other Risks Remain. GAO-13-500T. Washington, D.C.: April 17, 2013. F-35 Joint Strike Fighter: Current Outlook Is Improved, but Long-Term Affordability Is a Major Concern. GAO-13-309. Washington, D.C.: March 11, 2013. Joint Strike Fighter: DOD Actions Needed to Further Enhance Restructuring and Address Affordability Risks. GAO-12-437. Washington, D.C.: June 14, 2012. Joint Strike Fighter: Restructuring Added Resources and Reduced Risk, but Concurrency Is Still a Major Concern. GAO-12-525T. Washington, D.C.: March 20, 2012. Joint Strike Fighter: Implications of Program Restructuring and Other Recent Developments on Key Aspects of DOD’s Prior Alternate Engine Analyses. GAO-11-903R. Washington, D.C.: September 14, 2011. Joint Strike Fighter: Restructuring Places Program on Firmer Footing, but Progress Is Still Lagging. GAO-11-677T. Washington, D.C.: May 19, 2011. Joint Strike Fighter: Restructuring Places Program on Firmer Footing, but Progress Still Lags. GAO-11-325. Washington, D.C.: April 7, 2011. Joint Strike Fighter: Restructuring Should Improve Outcomes, but Progress Is Still Lagging Overall. GAO-11-450T. Washington, D.C.: March 15, 2011.

Why GAO Did This Study The F-35 aircraft represents the future of tactical aviation for the U.S. military, and is DOD's most expensive weapon system, with sustainment costs alone estimated at more than $1 trillion over a 60-year life cycle. As the F-35 program approaches full-rate production, DOD is working to deliver an affordable sustainment strategy that is able to meet the needs of the military services. This strategy is being tested as DOD stands up military depots, trains personnel, and supports its first operational squadrons—with plans to establish multi-year, performance-based contracts by 2020. The National Defense Authorization Act for fiscal year 2017 includes a provision for GAO to review the F-35 program's sustainment support structure. This report assesses (1) the status of DOD's efforts to sustain the F-35 fleet and any challenges it has faced; (2) the extent to which DOD is positioned to enter into multi-year, performance-based F-35 sustainment contracts; and (3) the progress, if any, DOD has made toward reducing F-35 sustainment costs and the extent to which costs are transparent. GAO reviewed DOD and contractor documentation, analyzed data, and interviewed relevant officials. What GAO Found The Department of Defense (DOD) is sustaining over 250 F-35 aircraft (F-35) and plans to triple the fleet by the end of 2021, but is facing sustainment challenges that are affecting warfighter readiness (see table). These challenges are largely the result of sustainment plans that do not fully include key requirements or aligned (timely and sufficient) funding. DOD is taking steps to address some challenges, but without more comprehensive plans and aligned funding, DOD risks being unable to fully leverage the F-35's capabilities and sustain a rapidly expanding fleet. DOD's plan to enter into multi-year, performance-based F-35 sustainment contracts with the prime contractor has the potential to produce costs savings and other benefits, but DOD may not be well positioned to enter into such contracts by 2020. To date, DOD has not yet achieved its desired aircraft performance under pilot (i.e., trial) performance-based agreements with the prime contractor. In addition, the level of performance DOD has contracted for is generally below what the services desire (see figure 2 for Marine Corps example). Also, the three performance metrics DOD is using to incentivize the contractor under these pilot agreements may not be the appropriate metrics to achieve desired outcomes, in part because they are not fully reflective of processes for which the contractor has control. This can make it difficult for DOD to hold the contractor accountable. Further, due to system immaturity, DOD does not have full information on F-35 sustainment costs and technical characteristics such as reliability and maintainability, which could hinder its ability to effectively negotiate performance-based contracts with the contractor by 2020. Without examining whether it has the appropriate metrics to incentivize the contractor or a full understanding of the actual costs and technical characteristics of the aircraft before entering into multi-year, performance-based contracts, DOD risks overpaying the contractor for sustainment support that does not meet warfighter requirements. DOD has taken actions to reduce F-35 sustainment costs, but estimated life cycle costs have increased and are not fully transparent to the military services (see figure 3). Specifically, the services do not fully understand how the costs they are being charged by the program office are linked to the capabilities they are receiving, citing unexplained cost increases and difficulty in tracking their requirements to contracts. For example, the Marine Corps received an initial funding requirement for fiscal year 2017 sustainment of $293 million, which then increased to $364 million in the execution year. This lack of transparency is due in part to insufficient communication between the program office and the services, and it puts the services in a difficult position as they consider critical trade-offs that may make F-35 sustainment more affordable. Without improving communication with the services about the costs they are being charged, the services may not be able to effectively budget for long-term sustainment. What GAO Recommends GAO recommends that DOD revise sustainment plans, re-examine metrics and ensure that it has sufficient knowledge of costs and technical characteristics before entering into performance-based contracts, and improve communication with the services about sustainment costs. DOD concurred with these recommendations.

gao_GAO-18-219

gao_GAO-18-219_0

Background Inter-American Organizations The U.S. government engages with the governments of other countries in the Western Hemisphere through various inter-American organizations including the OAS, PAHO, IICA, and PAIGH. According to State, the OAS is the primary inter-American political forum through which the United States engages with other countries in the Western Hemisphere to promote democracy, human rights, security, and development. PAHO serves as the Regional Office for the Americas of the World Health Organization, the United Nations agency on health. IICA supports agricultural development and rural well-being through technical cooperation and the execution of agricultural projects throughout the hemisphere. PAIGH specializes in regional cartography, geography, history, and geophysics and has facilitated the settlement of regional border disputes. According to U.S. agency officials, the organizations’ regional knowledge and technical expertise make them effective implementing partners for projects serving U.S. national interests and priorities throughout the hemisphere. U.S. Contributions to Inter- American Organizations Member states collectively finance these organizations by providing assessed and voluntary contributions (see table 1). For each organization, its member states’ assessed contributions are intended to finance the organization’s regular budgets, which generally cover the organization’s day-to-day operating expenses, such as facilities and salaries. Member states also finance certain OAS, PAHO, and IICA activities and projects through voluntary contributions. According to U.S. officials, the United States provides voluntary contributions to the OAS, PAHO, and IICA primarily through assistance agreements for specific projects from State, USAID, HHS, and USDA. Organizations’ Oversight of Their Funds The Institute of Internal Auditors (IIA) provides the framework for international organizations to oversee funds such as the assessed contributions provided by member states to OAS, PAHO, IICA, and PAIGH. The institute’s authoritative guidance, International Standards for the Professional Practice of Internal Auditing, includes mandatory performance standards that describe the nature of internal audit activities and provide criteria for evaluating these activities. Organizations are required to subscribe to these IIA standards, according to PAIGH officials and OAS, PAHO, and IICA documents. U.S. Agencies Oversee the Execution of Their Assistance Agreements Assistance agreements are a critical tool the U.S. government uses to achieve important national objectives. As we have previously reported, effective oversight and internal control are important to provide reasonable assurance to federal managers and taxpayers that assistance agreements are awarded properly, recipients are eligible, and federal funds are used as intended and in accordance with applicable laws and regulations. State, USAID, HHS, and USDA oversee funds provided to OAS, PAHO, and IICA through assistance agreements using monitoring activities such as financial and performance reports. Within each of these agencies, various bureaus and offices are responsible for awarding and managing assistance agreements to these inter-American organizations, including State’s Office of Weapons Removal and Abatement in the Bureau of Political-Military Affairs, USAID’s Office of U.S. Foreign Disaster Assistance, HHS’s Centers for Disease Control and Prevention and its Food and Drug Administration, and USDA’s Animal and Plant Health Inspection Service. The documentation of these monitoring activities as called for by federal standards for internal control enables the agencies to determine the effectiveness of the agreement activity. The Strategic Goals of the Four Inter- American Organizations Are Predominantly Aligned with U.S. Agencies’ Strategic Goals We found that the strategic goals of the four inter-American organizations are predominantly aligned with the high-level strategic goals for the Western Hemisphere documented by State, USAID, HHS, and USDA. According to officials, the agencies all consider U.S. strategic goals when deciding which projects to fund at OAS, PAHO, and IICA. State, USAID, HHS, and USDA have goals for foreign assistance to the Western Hemisphere, as shown in table 2. For example, four of the five goals in State and USAID’s Joint Strategy correspond with goals at the OAS, IICA, and PAIGH. U.S. agencies, on an ongoing basis, evaluate each inter- American organization to ensure U.S. and organization goals are aligned. Officials from all four agencies provided examples of how they help to ensure alignment of U.S. strategic goals when funding projects at OAS, PAHO, and IICA. State: According to State officials, State created an Annual Performance and Budget Review process in 2014 specifically to review entities, such as the OAS, that receive voluntary contributions funded through the International Organization and Programs account. This process examines performance of State-funded activities relative to those activities from the previous year and the extent to which the activities advance U.S. priorities and objectives. State officials further noted that the Annual Performance and Budget Review helps inform State’s decision-making on what to include in the following year’s budget request. For example, during the 2016 review for the OAS Development Assistance Program, State reported the program’s significant activities, funds expended, and achievements such as training government officials on successful small business policies in the United States. USAID: According to USAID officials, USAID’s project design and approval policies and procedures ensure that all USAID-funded activities are linked to applicable U.S. and USAID strategies. USAID’s agency guidance requires, at a minimum, that each project or activity must be formally approved in writing by the relevant Mission Director or Principal Officer for a given program. Officials stated that this approval memo and supporting documentation address a number of planning considerations, including how the proposed activity aligns with broader strategies. Furthermore, officials stated that USAID’s lawyers review project approval documentation prior to final approval and verify that the activity complies with all applicable statutes, regulations, and policies. HHS: According to HHS officials, HHS engages with PAHO on its Biennial Work Plans, which are operational planning instruments that PAHO uses to identify activities that it can implement within each of its member states. HHS officials noted that they use PAHO’s Biennial Work Plan to strengthen U.S. approaches on issues of common concern and to advance U.S. priorities within the region. According to HHS officials, proposals for technical cooperation projects are required to correspond to one of the technical priorities in PAHO’s strategic plan for 2014–2019 and to be aligned with the HHS global strategy and U.S. priorities. USDA: USDA officials said that they compare the U.S. strategic goals with IICA’s goals and objectives when they formulate project proposals with IICA to ensure that the projects are aligned with U.S. priorities for the region. Additionally, USDA officials told us that USDA helped shape and influence IICA’s recent 10-year strategic plan, ensuring that IICA’s strategic objectives were closely aligned with U.S. strategic goals. OAS, PAHO, IICA, and PAIGH Have Established Oversight Mechanisms OAS, PAHO, IICA, and PAIGH have established mechanisms for overseeing their use of assessed and voluntary contributions, such as external auditors and internal audit boards as required by IIA standards. State and USDA have directly supported these oversight mechanisms. The Four Organizations Have Oversight Mechanisms as Required by the IIA OAS, PAHO, IICA, and PAIGH have oversight mechanisms, as shown in table 3. The four organizations follow the internal control standards of the IIA, codified in the International Standards for the Professional Practice of Internal Auditing, according to PAIGH officials and OAS, PAHO, and IICA documents. All four organizations have internal and external auditors, as required by these standards. Furthermore, OAS, PAHO, and IICA have additional oversight mechanisms, such as anti-fraud policies and program evaluation processes. The officials we interviewed from State, USAID, HHS, and USDA expressed confidence in the four organizations’ management of their assessed and voluntary contributions. All four organizations document the status of their financial and internal control activities in audit reports posted on their public websites. For example, the OAS Office of the Inspector General’s April 2017 Annual Report included an update on its five ongoing audits and investigations. The report also outlined progress made against prior recommendations. U.S. Agencies Support Oversight Mechanisms at OAS, PAHO, and IICA U.S. agency officials support budget and administrative subcommittees in three of the four organizations and promote the participation of U.S. experts on independent audit committees, as shown in table 4. For example, according to officials, State plays a significant role in promoting policies on oversight and accountability at the four organizations through formal engagement in deliberations and decision-making of each organization’s governing body and through informal engagement with other member states and the secretariat by recommending best practices in governance, management, and oversight. State and USDA are also directly involved in implementing some of the additional oversight mechanisms at the organizations. For example, a USDA official serves as a member of IICA’s Audit Review Committee. Additionally, an IICA official told us the United States was involved in defining IICA’s Convention and Rules of Procedure for its governing bodies, which established the requirement for internal and external auditing. According to State officials, the United States led efforts to strengthen oversight at several of the organizations under review in recent years, such as advocating for the creation of an ethics officer position at PAHO, proposing language to strengthen the authority and independence of the OAS’s Office of the Inspector General, and encouraging the creation of audit committees at both organizations. In addition, State has played a lead role in supporting the ongoing reform of the OAS administration, which includes improved oversight and accountability, according to officials from the OAS and the U.S. Mission to the OAS. Two of Four U.S. Agencies Did Not Include All Key Monitoring Provisions in the Agreements We Reviewed We reviewed 12 selected assistance agreements that the four U.S. agencies awarded to OAS, PAHO, and IICA that were active during calendar years 2014 through 2016, and found that two agencies did not consistently include all key monitoring provisions in their agreements. While HHS and USAID implemented applicable guidance by including all key monitoring provisions in their agreements, USDA and State did not do so. USDA and State agency officials did not explain why USDA and State did not include these monitoring provisions in their agreements. However, State has since taken corrective action to ensure that they are included in future agreements, according to State officials. Agency Guidance Calls for Agencies to Conduct Monitoring Activities; Internal Control Standards State That Agencies Should Document These Requirements with Provisions in the Agreements Applicable agency guidance calls for agencies to conduct monitoring activities as part of their oversight of their agreements. Each of the four agencies has established agency-specific guidance that outlines the monitoring activities for assistance agreements. In some cases, the agency-specific guidance may mandate additional monitoring activities beyond those called for in applicable federal regulations, such as risk assessments. For example, State’s guidance calls for the creation of a monitoring plan. Federal standards for internal control require that agencies include in agreements all key provisions delineating the parties’ responsibilities. For the 12 agreements we reviewed, the number of total key monitoring provisions per agreement varied—including within one agency—depending on when the agency issued and updated its guidance relative to when the agreements were approved. Federal standards for internal control call for agencies to document internal controls, transactions, and significant events. Specifically, internal control standards state that agency management should include internal control activities (e.g., monitoring activities) in policies or directives for transactions such as assistance agreements. HHS and USAID Included All Key Monitoring Provisions in Their Assistance Agreements, but USDA and State Did Not For the 12 assistance agreements we reviewed, USDA and State did not include provisions implementing 6 of the 55 total (11 percent) applicable monitoring activities required by applicable guidance to carry out required monitoring activities (see table 5). State took corrective action in 2015 by issuing a standard operating procedure. USDA: USDA did not include 4 of the 13 key monitoring provisions implementing the applicable guidance for the three USDA agreements we reviewed (see table 6). Two of the agreements and supporting documentation each included all four key applicable monitoring provisions. However, Agreement 2 in the table did not include 4 of the 5 monitoring provisions in the agreement or work plan, which documents the monitoring provisions. The agreement partially included performance goals, because it included objectives for the agreement’s activities, but did not include time frames to complete all of the activities. The USDA grant official did not explain why the work plan did not adhere to applicable federal regulations when it was drafted and approved. State: State did not include 2 of the 21 key monitoring provisions implementing the applicable guidance for the three State agreements we reviewed (see table 7). Two of the agreements and supporting documentation we reviewed included the 7 monitoring provisions implementing the requirements in the applicable agency guidance. However, one agreement awarded in 2012 did not include 2 of the provisions: a risk assessment and a monitoring plan. That office that awarded this agreement took corrective action in 2015 by issuing a standard operating procedure requiring that risk assessments and monitoring plans accompany its grants and cooperative agreements. USAID: USAID included both key monitoring provisions implementing the applicable guidance for the three USAID agreements we reviewed (see table 8). USAID’s Automated Directives System 308, Standard Provisions for Cost-Type Awards to Public International Organizations contains two key monitoring provisions for agreements. USAID incorporated the monitoring provisions nearly verbatim into the agreements we reviewed, using templates from this guidance for required terms and conditions. HHS: HHS included the 15 monitoring provisions implementing the applicable guidance for the three HHS agreements we reviewed (see table 9). None of the U.S. Agencies Had Full Documentation of Monitoring Activities Called for by All of Their Assistance Agreements That We Reviewed None of the agencies provided us with full documentation to demonstrate their adherence to the required monitoring activities called for in all of their agreements that we reviewed, including the previously mentioned key monitoring provisions that we reviewed. State and HHS have taken corrective actions to address the gaps we found in documentation for the agreements we reviewed. Agency officials told us that they use these monitoring documents, such as financial and progress reports, to inform future budgetary and programmatic decisions. Therefore, they may lack information needed to make such decisions if they do not have access to complete monitoring documentation. Assistance Agreements Include Requirements for Monitoring Activities; Internal Control Standards Say Agencies Should Document These Activities According to federal standards for internal control, each agency is to include key monitoring provisions as part of its agreements. In the individual assistance agreements, the agencies specify the requirements to fulfill these activities, such as requiring financial reports on a quarterly basis or including specific information in performance reports. Grants officers at times, if they deem it necessary or appropriate, include additional monitoring provisions requiring activities beyond those required by the applicable guidance, such as site visits. Federal standards for internal control call for agency management to design monitoring activities, such as financial and performance reporting, so that all transactions are completely and accurately recorded. Recording these activities maintains their relevance and value to management in controlling operations and making decisions. Without access to complete monitoring documentation, the agencies risk weakening the effectiveness of these controls. U.S. Agencies Did Not Have Full Documentation of Monitoring Activities for Most of the Agreements We Reviewed None of the four U.S. agencies had full documentation of all of the monitoring activities required by their agreements we reviewed (see table 10). The agencies did not have full documentation of monitoring activities for 9 of the 12 agreements we reviewed. For the 42 monitoring activities identified across all of the individual agreements, the four agencies did not have full documentation of 18 of the activities (43 percent). However, State took corrective action in May 2017 to address its gaps in documentation, and according to HHS officials, the Food and Drug Administration addressed its gap in documentation by implementing its agreement monitoring program in fiscal year 2018. USDA did not have full documentation of any of the 10 monitoring activities we identified (see table 11). USDA demonstrated that it had partially documented 2 of the 10 monitoring activities (20 percent) by providing us with some, but not all, quarterly performance reports. For one of the agreements, USDA had no documentation of the monitoring activities for that agreement. For its other two agreements, USDA did not have full documentation of the required monitoring activities. USDA officials did not explain why they did not have full documentation. Without full documentation of the required monitoring activities, USDA may not have the information it needs to make appropriate budgetary and programmatic decisions. USAID did not have full documentation of 2 of the 11 total monitoring activities (18 percent) we identified across the three agreements we reviewed (see table 12). USAID had partial documentation of those 2 monitoring activities. For example, USAID provided us with some, but not all, records such as financial reports required by the terms of the monitoring activities in the agreements. According to USAID officials, the agencies’ lack of complete monitoring documentation was in part due to agency officials not following some of their agency’s requirements for managing agreement documents, such as placing all documents in a shared document management system. For example, for one of the agreements we reviewed, USAID officials stated that they stored some agreement documentation electronically—such as modifications, correspondence with the agreement recipient, and quarterly financial reports—but primarily maintained paper files. USAID officials told us they use the monitoring documents of these agreements, such as financial and progress reports, to inform future budgetary and programmatic decisions. For example, according to USAID officials, USAID uses monitoring documents to identify and address potential project delays or other “red flags.” For one of the agreements we reviewed, USAID officials stated these monitoring reports also assist them in determining whether to award additional funds and establish new indicators in subsequent agreements. Without full documentation of the required monitoring activities, USAID may not have the information it needs to make appropriate budgetary and programmatic decisions. State State did not have full documentation for 5 of the 16 monitoring activities (31 percent) we identified across the three agreements we reviewed (see table 13). However, State had partial documentation of 4 of those 5 monitoring activities. For example, State had some, but not all, records such as standard reporting metrics, required by the terms of the monitoring activities in one of the agreements. State did not have documentation of one of the monitoring activities (site visits). According to State officials, the agency’s lack of complete monitoring documentation was in part due to agency officials not following some of the agency’s requirements for managing agreement documents, such as placing all documents in a shared document management system. For example, according to State officials, for one of the agreements we reviewed, the grants officer mistakenly had saved site visit reports and similar documents to personal folders because the officer did not know how to use State’s grant document storage system. As a result, neither the current grants officer nor other State officials could retrieve these documents. In May 2017, after awarding the agreements we reviewed, State took corrective action by issuing the Federal Assistance Directive to establish internal guidance, policies, and procedures for all domestic and overseas grant-making bureaus, offices, and posts within the department when administering federal financial assistance. The directive notes that State implemented a grant management system for domestic and overseas grants to resolve its “significant deficiency in the management of Federal financial assistance.” In addition, the directive indicates that officials from State’s Bureau of Administration, Office of the Procurement Executive, Federal Assistance Division will evaluate compliance with risk assessment requirements and review documentation for selected agreements each fiscal year. One of the stated purposes of these reviews is to mitigate risk by strengthening management and oversight of awards, including grants. According to a State Office of Inspector General report, State should complete the full deployment of this system for overseas grants in fiscal year 2019. HHS did not have full documentation of 1 of the 5 applicable monitoring activities (20 percent) we identified across the three agreements we reviewed (see table 14). HHS had partial documentation of the semiannual progress report activity for one of its agreements, required by the terms of its agreement. HHS officials did not explain why they did not have full documentation for this monitoring activity. HHS had full documentation of all applicable monitoring activities for the other two agreements we reviewed. According to agency officials, the Food and Drug Administration, which administered one of the HHS agreements we reviewed, has taken corrective action to evaluate its agreement documentation and address deficiencies. According to HHS officials, the Food and Drug Administration developed a pilot program intended to provide an additional layer of oversight to ensure adherence to the terms of each agreement. Under the pilot, officials said, a grant monitoring specialist reviews the agreement documentation and monitoring reports to identify agreements that need additional assistance. According to HHS officials, the Food and Drug Administration implemented this program in fiscal year 2018 and will eventually include all Food and Drug Administration agreements. Conclusions U.S. assistance agreements for projects with inter-American organizations further U.S. strategic goals in the Western Hemisphere, but State, HHS, USAID, and USDA did not consistently include all key monitoring provisions as part of their assistance agreements or demonstrate that they had full documentation of monitoring activities for the agreements we reviewed. Of these four agencies, USAID and USDA have not taken corrective actions. Monitoring the implementation of U.S. assistance agreements and fully documenting the results of such monitoring are key management controls to help ensure that U.S. agreement recipients use federal funds appropriately and effectively. The agencies risk weakening the effectiveness of these controls by not including all key monitoring provisions called for by applicable agency guidance. Further, if the agencies do not have full documentation of the agreements’ required monitoring activities, they may not be able to effectively manage federally funded projects that support U.S. strategic goals. In addition, agencies may not have all the information they need to make budgetary and programmatic decisions. Recommendations for Executive Action We are making a total of three recommendations: one to USAID and two to USDA. The USAID Administrator should ensure that USAID officials have full documentation of required monitoring activities in agreements with inter- American organizations. (Recommendation 1) The Secretary of Agriculture should ensure that USDA includes all key monitoring provisions specified by applicable guidance as part of agreements with inter-American organizations. (Recommendation 2) The Secretary of Agriculture should ensure that USDA officials have full documentation of required monitoring activities in agreements with inter- American organizations. (Recommendation 3) Agency Comments We provided a draft of this report for comment to State, USAID, HHS, and USDA. USDA concurred with our recommendations in an e-mail. In its written comments, reproduced in appendix IV, USAID stated that it has policies, procedures, and training in place for the officials who manage these agreements. In response to our recommendations, USAID stated that it will issue an agency notice to remind all such officials of these responsibilities, including the requirement to maintain complete files for each agreement. State and HHS did not provide formal comments. They did provide technical comments, which we incorporated as appropriate. As agreed with your offices, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies of this report to the appropriate congressional committees, the Secretary of State, the Administrator of the U.S. Agency for International Development, the Secretary of Health and Human Services, and the Secretary of Agriculture. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have questions about this report, please contact me at (202) 512-9601, or melitot@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix V. Appendix I: Objectives, Scope, and Methodology Congressional requesters asked us to review several issues related to the Organization of American States (OAS), the Pan American Health Organization (PAHO), the Inter-American Institute for Cooperation on Agriculture (IICA), and the Pan-American Institute of Geography and History (PAIGH). In this report, we (1) assess the extent to which the organizations’ strategic goals align with those of U.S. agencies; (2) examine how the organizations oversee the use of their funds and the extent to which U.S. agencies have supported those efforts; (3) assess the extent to which U.S. agencies included key monitoring provisions as part of their assistance agreements; and (4) assess the extent to which U.S. agencies had documentation of monitoring activities, including those called for by these provisions. To address the first objective, we gathered documentation and interviewed officials from the four U.S. agencies and the four organizations to determine the U.S. strategic goals for foreign assistance to the Western Hemisphere and the goals of the four organizations. According to Department of State (State) and U.S. Agency for International Development (USAID) officials, the strategic document that underpins their foreign assistance priorities for the region is The Department of State’s Bureau of Western Hemisphere Affairs’ and USAID’s Bureau for Latin American and Caribbean Affairs’ Joint Regional Strategy. Department of Health and Human Services (HHS) officials said that HHS’s relevant strategic document is The Global Strategy of the U.S. Department of Health and Human Services. U.S. Department of Agriculture’s (USDA) strategic goals for foreign assistance, according to officials, are outlined in the United States Department of Agriculture Strategic Plan FY2015–2018. The OAS outlined its strategic goals in the Comprehensive Strategic Plan of the Organization, adopted on October 31, 2016. PAHO’s goals are laid out in Strategic Plan of the Pan American Health Organization 2014–2019. IICA’s strategic document is the IICA 2010–2020 Strategic Plan, which took effect in October 2010. PAIGH’s strategic document is the Declaration and Guide for the Pan American Agenda 2010-2020. We compared the strategic goals articulated by the four organizations against U.S. strategic goals to assess the extent to which the organizations’ goals contribute to U.S. interests in the region. We then interviewed officials from the four agencies and reviewed relevant documentation on efforts they undertake to ensure that U.S.-funded activities align with U.S. strategic goals. To address the second objective, we reviewed documentation of the organizations’ internal control mechanisms and confirmed our findings with the organizations. We identified mechanisms to include policies, directives, rules, practices, and organizational structures that can have an oversight role in the use of the organizations’ funds. We also interviewed officials from State, USAID, HHS, and USDA to discuss their support of these mechanisms. To address the third objective, we identified 60 active assistance agreements that these agencies oversaw with OAS, PAHO, and IICA during calendar years 2014 through 2016 and selected a nongeneralizable sample of 12 agreements, three each from State, USAID, HHS, and USDA. To determine which agreements we would review for each agency, we selected the three agreements with the lowest, median, and highest dollar value. If any of an agency’s agreements supported the same country or activity or were for one-time projects such as seminars, we selected the next appropriate agreement based on dollar value. For these selected agreements, we then identified the applicable agency guidance for monitoring activities in the agreements, which we define as all documents related to each agreement provided to us by the agencies, such as monitoring reports. The number of key monitoring provisions varied—even within each agency—depending on when agency guidance was issued and updated relative to when the agreements were approved. USDA did not have applicable internal agency-specific guidance for monitoring of assistance agreements at the time it awarded the agreements we reviewed; thus, with USDA’s input, we used the applicable sections of the Code of Federal Regulations, which together have five key monitoring provisions for agreements. However, USDA approved two of the agreements in 2012 and the third agreement in 2016, and this third agreement was subject to an amended version of the Code of Federal Regulations, which added an additional provision for performance goals. State’s four applicable grants policy directives have seven key monitoring provisions for agreements that were applicable at the time the agreements we reviewed were approved. USAID’s Standard Provisions for Cost-Type Awards to Public International Organizations (PIOs): A Mandatory Reference to ADS Chapter 308 has two key monitoring provisions for agreements: audits and records, and the organization’s adherence to their rules. HHS’s grants policy has five key monitoring provisions for grant documentation. We identified key monitoring provisions for agencies to include as part of agreements to ensure oversight of the use of funds, such as financial and progress reports. For the 12 agreements in our sample, we analyzed the assistance agreements from the four agencies, and then determined the extent to which the agencies’ agreements included key monitoring provisions implementing monitoring activities called for by applicable agency guidance. We did not include subsequent amendments to these 12 agreements in our review of key monitoring provisions. We interviewed officials from State, USAID, HHS, and USDA (1) to confirm we were applying the appropriate federal or agency guidance and (2) to discuss instances in which the agreements did not include key monitoring provisions. To address the fourth objective, we reviewed the 12 selected assistance agreements and guidance to identify specific required monitoring activities, such as financial and program reports, site visits, and other forms of oversight. The agreements specify the requirements for these activities such as requiring financial reports on a quarterly basis. For these 12 agreements, we reviewed all the documentation provided to us by the agencies, then determined the extent to which the agencies had full documentation of key monitoring activities as specified in the assistance agreements, including those called for by key monitoring provisions, as well as those called for by guidance when the monitoring provisions were absent. We did not include subsequent amendments to these 12 agreements in our review of monitoring activities. We interviewed officials from State, USAID, HHS, and USDA to discuss instances in which the agency did not have full documentation of key monitoring activities. We also discussed how State, USAID, and HHS agency officials manage their agreement documentation and use the information in the agreements’ required monitoring documentation. We conducted this performance audit from July 2016 to December 2017 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Financial Oversight Mechanisms of Four Inter-American Organizations The Organization of American States (OAS), Pan American Health Organization (PAHO), the Inter-American Institute for Cooperation on Agriculture (IICA), and the Pan-American Institute on Geography and History (PAIGH) have established mechanisms for overseeing their use of funds. Tables 15–18 show the mechanisms (oversight policies and oversight committees and organizations) for each of these inter-American organizations, as confirmed by the organizations’ officials. Appendix III: Key Monitoring Provisions Implementing Federal Regulations or Agency Guidance To oversee the execution of their agreements, the Department of State (State), the U.S. Agency for International Development (USAID), the Department of Health and Human Services (HHS), and U.S. Department of Agriculture (USDA) are to conduct monitoring activities called for by applicable federal regulations or agency guidance and document these provisions in assistance agreements as called for by federal standards for internal control. We identified key monitoring provisions implementing the applicable agency guidance for State, USAID, HHS, and the applicable regulations for USDA, as shown in table 19. For both the agency guidance and the federal regulations, those listed are the ones that were in effect when the agreements in our sample were approved. Some of the agency guidance and regulations have since been amended or superseded. Appendix IV: Comments from the U.S. Agency for International Development Appendix V: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Pierre Toureille (Assistant Director), Julia Jebo Grant (Analyst-in-Charge), Leslie Stubbs, and Paul Sturm, Alana Miller, and Shirley Min made key contributions to this report. In addition, David Dayton, Martin de Alteriis, Neil Doherty, Jeff Isaacs, and Alex Welsh provided technical assistance.

Why GAO Did This Study The United States is a member of the OAS, PAHO, IICA, and PAIGH, which promote democracy, health care, agricultural development, and scientific exchange. GAO was asked to review U.S. assistance to these four organizations. In this report, GAO (1) assesses the extent to which the organizations' strategic goals align with those of U.S. agencies; (2) examines how the organizations oversee the use of their funds and the extent to which U.S. agencies have supported those efforts; (3) assesses the extent to which U.S. agencies included key monitoring provisions as part of assistance agreements; and (4) assesses the extent to which U.S. agencies had documentation of monitoring activities, including those called for by these provisions. GAO analyzed documents and interviewed officials from State, USAID, HHS, USDA, and the organizations. GAO also analyzed a nongeneralizable sample of 12 of the 60 assistance agreements that were awarded by State, USAID, HHS, and USDA to OAS, PAHO, and IICA and were active during calendar years 2014 through 2016. For each agency, GAO selected three agreements with the lowest, median, and highest dollar value. What GAO Found GAO found that strategic goals of the Organization of American States (OAS), the Pan American Health Organization (PAHO), the Inter-American Institute for Cooperation on Agriculture (IICA), and the Pan-American Institute of Geography and History (PAIGH) are predominantly aligned with the strategic goals of the Department of State (State), the U.S. Agency for International Development (USAID), the Department of Health and Human Services (HHS), and the U.S. Department of Agriculture (USDA). For example, IICA's strategic goals of a productive agricultural sector, enhancing agricultural development, and food security are aligned with USDA's foreign assistance goals. State, USAID, HHS, and USDA fund activities in the form of assistance agreements (e.g., grants and cooperative agreements) with OAS, PAHO, and IICA, which in 2016 totaled $32 million. According to agency officials, the agencies employ mechanisms to ensure that these agreements align with U.S. strategic goals. OAS, PAHO, IICA, and PAIGH have established mechanisms for overseeing their use of funds, such as external auditors, internal audit boards, and anti-fraud and ethics policies. State and USDA have directly supported these mechanisms. For example, State engaged in the selection process for OAS's Inspector General. GAO's review of 12 selected assistance agreements found that USDA included no financial or performance monitoring provisions in one of its agreements and that State did not include two key monitoring provisions in one of its agreements, called for by applicable guidance. GAO found that the remaining 10 agreements it reviewed contained all key monitoring provisions and that State has since taken corrective action. GAO found that U.S. agencies did not have full documentation of 18 of the 42 monitoring activities required by the 12 assistance agreements GAO reviewed (see table). For example, USDA did not have full documentation, such as for financial reports, of any of its 10 required monitoring activities and USDA officials did not explain their lack of documentation. USAID officials explained that their lack of full documentation was due, in part, to grant officers not always following their document management policies. State and HHS have since taken corrective action. If an agency does not have full documentation of monitoring activities, it may lack information needed to make appropriate budgetary and programmatic decisions. What GAO Recommends GAO recommends that (1) USDA ensure inclusion of all monitoring provisions as part of agreements and (2) USAID and USDA ensure full documentation of monitoring activities. USDA and USAID concurred with GAO's recommendations.

gao_GAO-19-87

gao_GAO-19-87_0

Background Indian tribes and nations are recognized as “distinct, independent political communities” that are part of the unique political structure of layered sovereigns and internal governments that comprise the U.S. system of government. Tribal powers of self-government are recognized by the Constitution, legislation, treaties, judicial decision, and administrative practice. Tribal governments have many of the same responsibilities as state and local governments. However, tribes are generally unable to establish a strong tax base structured around the property taxes and income taxes typically available to state and local governments, according to a 2016 joint report from the Native Nations Institute and the Harvard Project on American Indian Economic Development and a 2003 report from the U.S. Commission on Civil Rights. For example, the reports found that tribes are unable to levy property taxes on some of their lands because of the legal status of the land. In addition, most tribes have a limited land base. Tribes generally do not levy income taxes because many tribal communities have disproportionately high levels of unemployment and a lack of employment opportunities. To the degree that they are able, some tribes use sales and excise taxes, but these do not generally generate enough revenue to fully support tribal governments. Therefore, some tribes rely on a combination of federal funds and economic development initiatives as fundamental sources of financial support for the government programs and services provided to their communities. According to Cohen’s Handbook of Federal Indian Law, “federal services to Indians were never mere gratuities. Instead, they were provided in exchange for cessions of land and rights, and to achieve distinctly federal purposes.” Generally, the programs that provide basic tribal services are supported through tribal priority allocation (TPA) funds that Congress appropriates. TPA funds are used to provide a wide variety of services to tribal communities—either through BIA-administered programs or self- determination contracts and self-governance compacts—and all federally recognized tribes are eligible to receive those funds. Bureau of Indian Affairs BIA, through its 12 regional offices and more than 80 agency offices, administers programs that provide services and funding to tribes. For example, BIA programs include social services, natural resources management, economic development, law enforcement and detention services, tribal court administration, implementation of land and water claim settlements, repair and maintenance of roads and bridges, repair of structural deficiencies on high hazard dams, land consolidation activities, and electric utilities. In some cases, a BIA agency office may serve one tribe, and in other cases, a BIA agency office may administer programs on behalf of more than one tribe. For example, BIA’s Central California Agency administers programs to 56 tribes, the largest multi-tribal field office in the contiguous 48 states. These programs may also be administered by tribal governments under a self-determination contract or self-governance compact. BIA is responsible for administering self-determination contracts, including negotiating and approving each contract and its associated annual funding agreement and disbursing funds to the tribes. For instance, under its procedures, BIA is to provide tribes that are interested in pursuing a self-determination contract with key information about the program and available funding. ISDEAA transfers control over programs to tribes, but as stated in Cohen’s Handbook of Federal Indian Law, “financial responsibility remains with the federal government.” ISDEAA provides that tribes who decide to administer federal programs are to receive the same funds that would have been provided had the federal government operated the programs. BIA identifies the amount of funds available to a tribe under a self- determination contract or self-governance compact for the administration of a federal program. In general, the most basic process for calculating the program amount is as follows: The program amount equals the total amount of funds Interior used to operate a program minus residual funds. Residual funds are the funds necessary for the federal government to carry out residual functions. Residual functions are inherently federal functions that only federal employees’ may perform if all tribes were to assume responsibilities for all programs that ISDEAA permits. Inherently federal functions are not defined in Title I or Title IV of the ISDEAA, and a 1994 Solicitor of the Interior memo reports that inherently federal functions are to be determined on a case-by-case basis when they fall outside certain defined categories. BIA officials told us the basic calculation is most likely to be used when a tribe is served by an agency office that only serves one tribe and the tribe took over administration of a program from that agency office. In cases where the total amount of funds BIA used to operate a program serves more than one tribe, additional data and factors may be included in the methodology to calculate the amount of funds available to administer the program. This is needed to ensure BIA can continue to provide services to the tribes that did not take over administration of the program. A BIA official told us that some regions and agency offices may divide the total amount by the number of tribes served, as shown in the following example: The program amount equals the total amount of funds Interior used to operate a program minus residual funds divided by the number of tribes served by the program. In other cases, regions and agency offices may include additional data to weight the calculation, such as tribal population or tribal land acres. When a tribe elects to pursue a self-determination contract, BIA is to meet with tribal officials to discuss and negotiate the terms of the contract, including what functions will be retained by BIA, the annual funding amount, and terms for the frequency of disbursing funds—that is, disbursed in a single lump sum or other intervals, such as quarterly payments. According to Interior budget officials responsible for BIA’s budget, after Interior receives its appropriations, departmental budget officials determine how to distribute any changes between the Administration’s budget and the final budget among BIA offices that deliver direct services to tribes and to tribes that contract the services through self-determination contracts. According to Interior budget officials, they calculate changes in the budget amounts for each contract after consulting BIA program officials and based on statutory requirements, historical percentages, or other distribution factors. After the budget calculations are completed, Interior officials transfer funds to BIA regional offices to distribute to BIA agency offices and tribes. An awarding official in the regional or agency office then provides contracting tribes an updated annual funding agreement that identifies the amount of funds for that fiscal year. Self-Determination Contracts and Self- Governance Compacts ISDEAA authorizes federally-recognized tribes to assume administration of certain federal programs and functions that were previously managed by the federal government. It is Interior policy to facilitate tribal administration of programs through self-determination contracts and remove obstacles that hinder tribal autonomy and flexibility to administer such programs. Under Title I of ISDEAA, an interested tribe may request by tribal resolution to enter into a self-determination contract with BIA. ISDEAA requires the parties to such contracts to negotiate annual funding agreements and determine the frequency and timing of payments under the contract. Payments may occur throughout the fiscal year in accordance with terms identified in the annual funding agreements as Interior’s Indian Affairs Office of Budget and Performance Management makes appropriated funds available. Under Title IV of ISDEAA, an interested tribe may request to enter into a self-governance compact. Under the law, to be eligible for participation in self-governance compacting, a tribe must, among other things, demonstrate financial stability and management capability, which can be evidenced by participating in a self-determination contract for at least 3 years with no material audit exceptions. Interior’s Office of Self- Governance (OSG) is responsible for administering self-governance compacts and funding agreements for Interior programs. OSG assists tribes that want to enter into self-governance compacts by providing training, determining eligibility, participating in negotiations with the tribes and Interior agencies to identify the amount of funds that will be included in the self-governance compacts, and approving tribes to participate in self-governance. In addition, tribes with self-governance compacts negotiate annual funding agreements with OSG rather than BIA. OSG is also responsible for transferring funds from Interior to tribes with a self- governance compact, ensuring audit compliance, and processing waivers of BIA regulations. Further, OSG is responsible for preparing an annual report to Congress on the costs and benefits of self-governance. As of fiscal year 2016, OSG has entered into self-governance compacts that cover 47 percent of federally recognized tribes (267 tribes). For additional information on the differences between self-determination contracts and self-governance compacts, see table 1. Several Factors, Including Certain Federal Actions, Can Affect Tribes’ Use of Self-Determination Contracts, Self- Governance Compacts, and the HEARTH Act Several factors, including federal agencies’ processes and actions can affect tribes’ use of mechanisms that further tribal self-government such as self-determination contracts, self-governance compacts, or leasing authority under the HEARTH Act that further tribal self-government. Some of these factors, such as federal training and resources, can help tribes develop the tribal capacity needed to take over administration of federal programs and thereby facilitate tribes’ use of these mechanisms. In contrast, other factors, specifically federal processes and actions, can hinder or delay tribes’ use of these mechanisms. Some of these processes include: (1) BIA’s approach for sharing information with tribes, (2) Interior’s process to disburse funds, and (3) Interior’s process to review proposed tribal leasing regulations submitted under the HEARTH Act. In addition, the adequacy of federal resources needed to administer a program is a factor that can affect tribes’ use of self-determination contracts and self-governance compacts, according to several tribal stakeholders and federal officials we spoke with, government reports, our prior reports, and other articles we reviewed. Tribal Capacity is a Key Factor That Can Facilitate Tribes’ Use of Self- Determination Contracts, Self-Governance Compacts, and Authority under the HEARTH Act The capacity of a tribal government to administer a federal program or manage its resources is a key factor that can affect a tribe’s decision to enter into a self-determination contract or self-governance compact, or to use the authority available under the HEARTH Act, according to some reports we reviewed. For example, the Harvard Project on American Indian Economic Development found that successful tribal assertions of sovereignty and self-government are backed by capable institutions of governance that contribute to tribal capacity. According to federal officials and agency training documents we reviewed, Interior has contributed to the capacity of tribal governments by increasing tribes’ knowledge about self-governance compacting and the HEARTH Act. For example, Interior’s OSG provides opportunities for tribes to learn about self-governance compacting and build capacity by partnering with a non-profit organization to conduct training events, including an annual week-long training program. In addition, BIA offered several training sessions in 2014 and 2015 on the HEARTH Act to educate tribes on the benefits of developing tribal leasing regulations. Furthermore, Interior’s Office of Indian Energy and Economic Development administers a grant program, Tribal Energy Development Capacity, intended to help tribes build the capacity to enter into a tribal energy resource agreement (TERA) or develop leasing regulations under the HEARTH Act. Some tribal stakeholders identified the EPA’s Indian Environmental General Assistance Program (GAP) as a model for a federal program that helped their tribes build the capacity needed to administer environmental programs from EPA. These tribal stakeholders also told us this capacity benefitted the tribes as they sought to take over similar programs from Interior. Some tribal stakeholders told us the GAP program is effective in assisting tribal governments build capacity because it is designed to provide consistent funding over multiple years. According to reports we reviewed that discuss building tribal capacity, effective capacity building efforts should both provide for sustained, consistent funding over time, since developing capacity can be an ongoing effort that may take longer than 1 year to achieve and facilitate a tribe’s ability to develop a program that is responsive to each tribe’s unique conditions and priorities. Factors That Can Hinder Tribes’ Use of Self- Determination Contracts, Self-Governance Compacts, and Authority under the HEARTH Act We found that several factors can hinder tribes’ ability to use self- determination contracts, self-governance compacts, or leasing authorities under the HEARTH Act, including: (1) BIA’s approach for sharing key information with tribes seeking to develop a program using a self- determination contract, (2) Interior’s process to disburse funds to tribes associated with self-determination contracts and self-governance compacts, (3) Interior’s review of tribal leasing regulations submitted under the HEARTH Act, and (4) BIA’s management and maintenance of federal programs that tribes may pursue to take over under a self- determination contract. BIA’s Approach for Sharing Key Information with Tribes According to several tribal stakeholders, BIA’s approach for sharing key information with tribes does not always ensure that tribes have the information they need to design programs under self-determination contracts prior to negotiations. As a result, this has been a factor that has hindered or delayed tribes’ use of self-determination contracts for administering programs. Interior guidance and policy call for BIA to provide tribes information that includes, among other things, calculations BIA uses to identify the amount of funds available to a tribe if it takes over administration of a program. In accordance with Interior’s policy, BIA should provide tribes with the information necessary to design programs those tribes would like to administer under a self-determination contract to meet the needs of their communities consistent with their diverse demographic, geographic, economic, cultural, health, social, religious, and institutional needs. Also in accordance with Interior guidance, when a tribe requests to enter into a self-determination contract with Interior, BIA should disclose information to the tribe that identifies the amount of program funding available, the methodology used to identify available amounts, the process used to arrive at available amounts, an identification of the amount of funding retained by BIA, and any other information useful to understand how contract amounts were calculated. Moreover, Interior regulations call for BIA to provide to tribes, for the negotiation of annual funding agreements for self-governance compacts, a brief justification as to why specific functions have been determined inherently federal. However, according to several tribal stakeholders, they do not receive this information, including calculations BIA uses to identify the amount of funds available to tribes, prior to negotiating their self-determination contracts. Some BIA regional and agency office officials we interviewed told us they do not generally provide information to tribes prior to negotiating the terms of a self-determination contract because the determinations of inherently federal functions and the amount of funding the bureau would retain to perform such functions generally occurs during meetings with BIA and the tribe. A tribal stakeholder told us that without documentation on funding calculations and methodologies, tribes are at a disadvantage and have little basis to negotiate during these meetings. A tribal stakeholder told us that, in practice, the negotiation generally consists of BIA informing the tribe of the amount of funds to request in its proposal and what federal functions BIA will retain without any documentation to support its determination of inherently federal functions or the resources to be made available to the tribe to administer a program using a self- determination contract. BIA’s approach is not consistent with Interior’s policy of sharing information so tribes can develop programs. By developing a process that results in BIA’s regional and agency offices providing tribes with documentation on calculations and methodologies to identify resources available to administer a program using a self-determination contract, BIA would be adhering to Interior’s policy and have greater assurance that tribes have the information they need to design the programs that they would like to pursue under a self-determination contract. In addition, BIA guidance states the bureau will ensure functional consistency in the determination of inherently federal functions when the Central Office and all regional offices are compiling that information for negotiating annual funding agreements with tribes. We found examples that suggest that BIA has not consistently determined whether programs and functions are inherently federal, which can affect some tribes’ use of self-determination contracts. For example, a BIA official in one regional office told us that the region had previously decided all functions associated with the Land Titles and Records Offices were inherently federal and told tribes that BIA would not approve a self-determination contract for those functions. However, other BIA regional offices did not consider the functions of the Land Titles and Records Offices as inherently federal, and some tribes in those regions had taken over administration of those functions. BIA does not have a process that results in consistent determinations of inherently federal functions and does not provide tribes with information on its prior determinations. A BIA official told us that determinations of inherently federal functions are made on a case-by-case basis because each tribe and its circumstances are unique. However, this approach does not provide BIA leadership with reasonable assurance of functional consistency throughout the bureau in the determination of inherently federal functions—consistent with bureau guidance. By developing a process that results in consistent determinations of inherently federal functions, BIA could have greater assurance that these decisions are being made appropriately across the agency. BIA could also increase transparency in the process by providing tribes with documentation on activities and functions previously determined to be inherently federal and the basis for making these determinations. Interior’s Process to Disburse Funds According to tribal stakeholders we spoke with, Interior’s process to disburse funds associated with the tribes’ self-determination contracts and self-governance compacts is a factor that hinders expansion of self- determination contracts or self-governance compacts. Several tribal stakeholders and federal officials we interviewed said that the process does not ensure that tribes receive funds within the time frame specified in ISDEAA’s Model Agreement for self-determination contracts or as agreed to by Interior and the tribes in their annual funding agreements. Two tribal stakeholders stated that in prior years, funds were disbursed several weeks or months after Interior received its apportionment from the Office of Management and Budget. We were unable to determine the extent to which Interior disburses funds for self-determination contracts within the time frame agreed to in a self- determination contract because Interior does not systematically track the disbursement of funds from the date it received its appropriations through the date that it made funds available to tribes and does not compare its actual performance to expected performance. Not tracking this information and comparing actual performance to expected performance is contrary to federal internal control standards, which state that agency management should design control activities to achieve objectives and respond to risks, such as by comparing actual performance to planned or expected results and analyzing significant differences. This is not a new issue for Interior. Specifically, in 2015, an Interior contractor reported on an evaluation of Interior’s process for disbursing funds and identified opportunities for improvement. Consistent with our findings, the report also found that, among other things, the process used by Interior to disburse funds is a manual process that does not include a real-time tracking mechanism. Without such a mechanism, the report found that officials must spend time trying to determine the status of documents and finding misplaced or lost documents. For example, the report found that in fiscal year 2014, Interior had more than 6,000 scanned documents that required up to 6 signatures each, for a total of up to 36,000 signatures, to disburse funds including funds to tribes for self-determination contracts or self-governance compacts. To finalize these documents, the report estimates that 600 hours of staff time were spent scanning, uploading, and printing the documents. Several Interior officials told us they conduct monitoring activities within a specific BIA region or BIA agency office, such as tracking disbursement information through an Excel spreadsheet, but these activities were not part of a systematic process. An Interior official told us there are no plans to develop a real-time tracking mechanism. Interior officials we interviewed cited several reasons why some funds associated with self-determination contracts and self-governance compacts were not disbursed in accordance with time frames outlined in its Model Agreement or negotiated in funding agreements. The reasons include the following: Interior’s financial data management system. Interior officials told us that the agency’s financial data management system is used for all of Interior and is not equipped for the unique aspects of self- determination contracts and self-governance compacts—making it difficult to properly track and monitor key actions. Prior use of an inefficient process. An Interior official told us that prior to fiscal year 2017, BIA used several spreadsheets to coordinate TPA information for distributions. The official stated that these spreadsheets were over 15 years old, and they made the process inefficient and time-consuming. To distribute funds, officials would use one spreadsheet to gather information and another to summarize the amounts by functional area and region. The official stated that BIA updated the process in fiscal year 2018 and does not expect it to delay funding in the future. Staff shortages in key positions. BIA officials in several regions told us they are experiencing staff shortages in key positions that are responsible for the transfer of funds from BIA to tribal governments, such as awarding officials. Interior officials said the Office of Self Governance also needs additional awarding officials with only one awarding official for self-governance compacts. Interior officials stated that the challenges from staff shortages are compounded by Congress’ use of continuing resolutions that result in BIA repeating its fund distribution process multiple times in a single year. Delays in receiving tribal signatures. Interior officials we interviewed told us that they have experienced delays disbursing funds to a tribe because they must wait for tribal officials to sign documents before funds may be disbursed. When funds are not disbursed in a timely manner, a tribal stakeholder told us that tribes may have to use funds from their general revenue accounts to cover expenses for federal programs or seek other sources, such as loans, to cover program expenses. According to several tribal stakeholders, when a tribe has to use its own funds for the administration of programs—even temporarily—it can adversely affect the tribe in various ways, including lost opportunities to use tribal funds for improving the tribes’ economic conditions, reducing other services provided to tribal communities, and furloughing tribal government employees. In addition, several tribal stakeholders told us that the timeliness of disbursements for self-determination contracts is a factor they consider when deciding whether to take over additional programs under a self-determination contract. The tribal stakeholders said that the tribe must consider if it is able to use tribal funds or willing to obtain a loan to fund a program when the federal government is late disbursing funds. Without establishing a process for tracking and monitoring the disbursement of funds associated with self-determination contracts and self-governance compacts, Interior will not have reasonable assurance it disburses funds in a systematic way or in accordance with agreed upon time frames. Interior’s Process for Reviewing Proposed Tribal Leasing Regulations Interior has not clearly documented its process for reviewing proposed tribal leasing regulations with timeframes associated with each step of the process. The process can often be lengthy and time consuming; according to tribal stakeholders, this can be a factor that hinders the tribes’ ability to make decisions about the use of tribal resources. Under the HEARTH Act, tribes are to submit proposed leasing regulations for Interior’s review and approval before a tribe can approve leases for the use of tribal lands, and Interior’s review is to be completed within 120 days after the dates on which the tribal regulations are submitted to the agency. Interior officials told us they interpret the statutory review time frame requirements of the HEARTH Act as applying only to the agency’s review to ensure tribes incorporated all changes identified in prior reviews. Specifically, Interior officials told us the agency does not consider the statutory time frame to begin until it has received a final version of the proposed tribal leasing regulations. These officials described the final version of proposed tribal leasing regulations as regulations that have already undergone review by BIA and Interior’s Solicitor’s office, have been revised by the tribe, and have been resubmitted for additional review by BIA and the Solicitor’s office. This process can be repeated multiple times before Interior considers the tribe’s proposed leasing regulations to be final. In contrast, a tribal stakeholder told us that Interior’s interpretation of how to measure the time frame is inconsistent with the tribe’s interpretation of the statutory time frame. The tribal stakeholder told us that a tribe considers its leasing regulations initially submitted to Interior as final, although the tribe understands that BIA and the Solicitor’s office may request revisions. Some tribal stakeholders told us that because Interior is not considering the 120 days as a time frame from first submission until approval, tribes do not know when to expect a final decision on draft tribal regulations. We found that some of this confusion could be attributed to the fact that Interior has not clearly documented its review process to include established time frames associated with each step of the process. Under federal standards for internal control, management should design control activities, such as clearly documenting internal control in management directives, administrative policies, or operating manuals. The HEARTH Act seeks to expand tribal self-government and promote economic development by shifting the authority for leasing from the Secretary to the tribes. By developing a clearly documented review process that includes established time frames for each step in the process for reviewing proposed tribal leasing regulations submitted under the HEARTH Act, Interior can better ensure that it is eliminating uncertainty and better communicating the process for approval to the tribes. We also found that the approval process can be lengthy in some cases. Our review of 42 tribal leasing regulations submitted to Interior for review from 2012 through 2017 for which BIA provided us with data on the date the tribe submitted the regulations to Interior and the date of Interior’s approval found that 4 of the 42 leasing regulations were approved within 120 days. For the other 38 proposed regulations, the time from when the tribe submitted the regulations to Interior to when the agency approved the regulations ranged from 134 days to 980 days. Half of the 42 proposed regulations were under review by Interior for a year or longer, with 5 of the 21 under review for more than 2 years. Interior’s review was generally not continuous during the entire period; instead, these time periods included review by BIA and the Solicitor’s office and the time spent by the tribe revising its leasing regulations in response to Interior’s review. Tribal stakeholders also shared with us several examples that illustrate Interior’s lengthy review process for tribal governments’ use of the HEARTH Act. For example, in one case, Interior received a tribe’s leasing regulations for review and approval in May 2015. Interior approved the tribe’s leasing regulations and published the decision in the Federal Register in April 2018—more than 2 years later. Officials representing this tribe told us they considered the leasing regulations initially submitted on May 18, 2015, as final, though they understood that Interior could request revisions. These officials explained that the tribe has its own extensive process and procedures for lawmaking and developed its leasing regulations consistent with its Constitution, Legislative Procedures Act, and Administrative Rulemaking Procedure, which take into account comments from tribal members and tribal agencies and includes a judicial review, legislative analysis, fiscal impact review, and adoption by the tribe’s elected business committee. Tribal stakeholders told us that after each communication with BIA about the leasing regulations, they believed the regulations were satisfactory for approval. For example, the tribe received preliminary approval from BIA in October 2016. Then, tribal stakeholders said in August 2017—nearly 10 months later—the tribe received correspondence from BIA stating that the tribe needed to add several additional provisions, including language regarding Indian irrigation projects and districts even though the tribe does not have any irrigation projects or districts within its boundaries. Additional correspondence took place between the tribe and Interior, resulting in final approval in January 2018. Tribal stakeholders told us that the lengthy review delayed the tribe’s ability to implement leasing regulations and delayed the tribe’s ability to make decisions about the use of tribal resources. In another case, Interior received tribal leasing regulations for review and approval on January 17, 2014. The tribe stated in documentation submitted to Interior that it was seeking increased decision-making authority under the HEARTH Act because it had finalized various construction agreements and needed to approve surface leases for an economic development project. During the time that the tribe’s leasing regulations were under review at Interior, BIA asked the tribe to submit multiple versions of its leasing regulations. According to BIA documents, the bureau took approximately 2 months to transfer the tribe’s regulations from BIA headquarters to a regional office for its review. Once the regional office received the tribal leasing regulations, the office conducted its review over a 3-month period and provided comments to BIA’s headquarters. BIA’s data show that headquarters sent the tribe’s leasing regulations to the Solicitor’s office nearly 5 months after it received the tribal leasing regulations. Over the next couple of years, Interior requested the tribe make changes to its leasing regulations three more times and resubmit revised versions for review. On March 3, 2016—more than 2 years after receiving the tribe’s leasing regulations—Interior documented that it had “one small change” it would like the tribe to make to the regulations. The tribe made the requested change and resubmitted the leasing regulations to Interior via certified mail, which showed receipt at Interior on July 1, 2016. Interior approved the tribal leasing regulations on October 7, 2016—more than 3 months after the tribe submitted regulations with the “small change.” Interior approved the tribe’s leasing regulations and published the decision in the Federal Register in October 2016—more than 2 years after Interior first received the tribe’s leasing regulations. Interior officials told us there was not a single reason for the lengthy review times. In some cases, Interior officials said the review times were long because the BIA official responsible for managing Interior’s review had left the bureau. In other cases, Interior officials told us they were short-staffed in the Office of the Solicitor and the legal review took longer than anticipated. However, they acknowledged that the uncertainty associated with how long Interior’s review will take can make it difficult for tribes to plan and execute economic development projects. For example, a BIA official told us that a tribe was unable to pursue an economic development opportunity because of the time it took for Interior to complete the process to review the tribe’s regulations. In contrast, a timely review of a tribe’s proposed leasing regulations can positively affect tribal control and decision making. For example, a tribal stakeholder said after several months waiting for BIA to approve a surface lease needed for a tribe to develop a wind farm, the tribe decided to pursue authority under the HEARTH Act so that it could review and approve the lease without waiting for BIA’s review of the surface lease. Interior reviewed and approved the tribe’s leasing regulations submitted under the HEARTH Act authority in 31 days. According to the tribal stakeholder, the timely review and approval of the tribe’s leasing regulations allowed the tribe to review and approve the surface lease needed for construction of the wind farm to commence before the expiration of tax credits—a key component that made the project feasible. BIA’s Past Management of Federal Programs Past mismanagement of federal programs under the administration of BIA is a factor that can affect tribes’ decisions whether to take over federal programs through self-determination contracts, according to several tribal stakeholders and BIA officials. As documented in a 2003 report by the U.S. Commission on Civil Rights, decades of general mismanagement of infrastructure and programs under BIA’s administration can hinder a tribes’ use of self-determination contracts. In 1999, BIA reported to Congress that funds provided under self-determination must be used not only for current operations but also “to repair 150 years of general neglect” of Indian programs. In these cases, taking over programs with long-standing neglect is a liability that some tribes are not willing to assume. For example, a tribal stakeholder told us that its BIA agency office neglected tribal land records for many years. As a result, the tribe is reluctant to assume the liability associated with administering a real estate program without accurate property records. In another example, BIA operates an irrigation project that provides electric utility service to two tribes. Both tribes have taken over certain functions associated with the utility service provided to their communities through self-determination contracts, and both tribes have expressed interest in expanding the functions they administer. However, BIA and tribal officials said that concerns over infrastructure that needs to be repaired or replaced and the liability associated with rights-of-way have deterred both tribes from taking over the remaining functions of the utility. For example, many utility poles on the project’s transmission lines are more than 50 years old and are in need of replacement, and the project has over 1,500 miles of transmission lines and 2,000 miles of distribution lines. According to a BIA document, these lines might have been extended without receiving a formal right-of-way. The report states that “many of ’s rights-of-way are unperfected and there are no supporting documents evidencing a legal right-of-way.” According to tribal stakeholders these kinds of uncertainties are significant factors they must consider in their decisions related to self- governance of BIA programs. Adequacy of Resources Affects Tribes’ Use of Self- Determination Contracts and Self- Governance Compacts The adequacy of resources is a long-standing concern that has been a factor affecting tribal participation in self-determination contracts and self- governance compacts, according to several tribal stakeholders and federal officials we interviewed, government reports, our prior reports, and articles we reviewed. Specifically, a lack of adequate resources has been a long-standing concern that can limit the number of programs tribes take over using self-determination contracts and self-governance compacts. For example, the U.S. Commission on Civil Rights 2003 report noted that the authority tribes have to take over the administration of federal programs is useful to the extent that adequate funds are made available to the tribes to operate the program. According to Interior officials, Interior does not have an estimate on the extent to which it can provide adequate resources to tribes that want to administer federal programs. For one program, BIA estimated in a report to Congress that the dollars BIA expended in fiscal year 2013 for BIA and tribes to operate detention and corrections centers fund about forty percent of the estimated operating needs. Faced with funding shortfalls from the BIA budget to administer federal programs under federal self-determination contracts or self-governance compacts, many tribal stakeholders told us that they supplement federal funding. Officials from one tribe told us that the tribe has supplemented all the programs it has taken over from BIA. For example, the tribe reported that the Land Titles and Records Program has a shortfall of about $300,000 annually; the Law Enforcement program with about $564,000 annually; and the Probate Program with about $129,000 annually. Officials from the tribe told us that tribes may rely on revenues generated from economic development or tax revenue to supplement federal dollars for programs they have taken over from the federal government. However, tribal stakeholders we interviewed told us that not all tribes are in a position to supplement additional federal programs because of limited economic development opportunities and tax revenue; therefore, those tribes may not have the option to take over additional federal programs. According to a tribal stakeholder, dual-taxation—when both a tribe and state tax the same non-tribal members and businesses on tribal land— can significantly limit a tribe’s tax revenue because tribes must reduce or eliminate their taxes to stay competitive and attract business and enterprise to their lands. Furthermore, the funds tribes may use to supplement federal programs are needed to fund other governmental services and activities, which place tribal leaders in the position of deciding whether to use funds to provide governmental services not funded by the federal government or to increase self-governance by administering additional federal programs. As we have previously reported, when tribes supplement the federal program they take over, it diverts funds away from other economic development opportunities and other government functions and services they provide to their communities and citizens. Lastly, several tribal stakeholders told us that not receiving adequate resources from the federal government to administer federal programs makes them reluctant to administer additional federal programs because they believe BIA has a better chance than the tribe to obtain additional resources that can be used to supplement program shortfalls. This is, in part, because they believe that BIA agency offices and regional offices have access to funding sources that are not available to tribes and because BIA does not always make tribes aware of funds that are available. For example, the Department of the Interior’s Self-Governance Advisory Committee reported in 2015 that the distribution of year-end funds is entirely within the discretion of the local awarding official and that not all tribes are notified that these funds are available. Conclusions Interior has taken steps to assist tribes pursuing tribal self-government by providing training opportunities focused on self-governance compacts and the use of the HEARTH Act to help increase tribal capacity. However, several factors have continued to hinder tribes’ use of these mechanisms to further tribal self-government. First, BIA’s approach for sharing key information with tribes when tribes seek to administer a program using a self-determination contract does not provide the tribes with the information they need to understand how the self-determination contract amounts were calculated. As a result, tribal leaders are at a disadvantage in making sound decisions regarding the feasibility of taking over the administration of federal programs. Second, BIA does not have a process that results in consistent determinations of inherently federal functions and does not provide tribes with information on its prior determinations. By developing a process that results in consistent determinations of inherently federal functions, BIA could have greater assurance that such determinations are being made appropriately across the agency and BIA could increase the transparency of the process by providing tribes with documentation on activities and functions previously determined to be inherently federal and the basis for the determinations. Third, Interior does not have an effective process for tracking and monitoring the disbursement of funds associated with tribes’ self- determination contracts and self-governance compacts or as agreed to with the tribes. Without establishing an effective tracking and monitoring process, Interior does not have reasonable assurance that it is disbursing funds in accordance with ISDEAA or time frames agreed to with the tribes. Lastly, Interior has not documented its process to include established time frames associated with each step of the process to review proposed tribal leasing regulations submitted under the authority provided by the HEARTH Act. This has resulted in lengthy review times—in some cases, multiple years. By developing a clearly documented review process that includes established time frames for each step in the process for reviewing proposed tribal leasing regulations submitted under the HEARTH Act, Interior can better ensure that it is eliminating uncertainty in the process to approve tribal leasing regulations. Recommendations for Executive Action We are making the following four recommendations to Interior: The Assistant Secretary of Indian Affairs should develop a process so that all regional and agency offices consistently provide tribes with documentation on calculations and methodologies to identify resources available to administer a program using a self- determination contract. (Recommendation 1) The Assistant Secretary of Indian Affairs should develop a process that results in consistent determinations for inherently federal functions and to provide documentation to tribes on specific activities and functions determined to be inherently federal. (Recommendation 2) The Assistant Secretary of Indian Affairs should establish a process to track and monitor the disbursement of funds associated with self- determination contracts and self-governance compacts. (Recommendation 3) The Assistant Secretary of Indian Affairs should coordinate with the Office of Solicitor and BIA to develop a clearly documented process with established time frames for each step in the process for reviewing proposed tribal leasing regulations submitted under the HEARTH Act. (Recommendation 4) Agency Comments We provided a draft of this report to Interior for comment. In its comments reproduced in appendix II, Interior generally concurred with our recommendations. Interior also provided technical comments, which we incorporated as appropriate. As agreed with your office, unless you publicly announce the contents of this report earlier, we plan no further distribution until 28 days from the report date. At that time, we will send copies to the appropriate congressional committees, the Secretary of the Interior, the Assistant Secretary of Indian Affairs, and other interested parties. In addition, the report will be available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-3841 or ruscof@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix III. Appendix I: Scope and Methodology For this report, we reviewed a range of reports, articles, conference proceedings, congressional testimony, and other publications from federal and tribal governments, academics, and nonprofit organizations. These publications included general background information related to tribal self- government and tribes’ use of self-determination contracts, self- governance compacts, and the HEARTH Act, as well as historical perspectives, successes and challenges, and identified some factors that can affect a tribe’s decision to use one of these mechanisms. We identified these articles and publications by searching various Web-based databases, such as ProQuest, Scopus, DIALOG, Academic OneFile, JSTOR, and Lexis to identify existing studies from articles, peer-reviewed and other journals, including law review journals, and government and academic publications. We searched terms such as tribal sovereignty, self-governance, self-determination, and capacity, as well as relevant acts or program names. We also asked tribal stakeholders that we interviewed to recommend additional reports, congressional testimony, and other articles on the topic. We did not set specific time frames for the search, and identified more than 50 articles from 1982 to 2017. We examined summary-level information about the literature identified in our search and identified a few of the articles as directly related to our report. These five publications are identified throughout this report. Other articles provided beneficial context and historical information but did not contribute to us identifying factors to include in this report. We reviewed relevant laws and regulations including the Indian Self- Determination and Education Assistance Act of 1975 (ISDEAA), as amended and Helping Expedite and Advance Responsible Tribal Home Ownership Act of 2012 (HEARTH Act). We also reviewed Interior’s policy manual, Interior’s procedures handbook for contracting under Title I of ISDEAA, the Interior Solicitor’s opinions on inherently federal functions, and other guidance documents. We reviewed Interior reports and audits related to self-determination contracts, self-governance compacts, and the Hearth Act, including Interior budget justification reports and evaluations of tribes’ performance with trust programs administered under a self-governance compact. ISDEAA also allows tribal governments to take over administration of certain programs from the Department of Health and Human Service’s Indian Health Service. For this review, we focused on tribes’ use of self-determination contracts and self- governance compacts to administer Bureau of Indian Affairs (BIA) programs. To determine tribes’ use of self-determination contracts, we obtained data from Indian Affairs’ Office of Chief Financial Officer for all current contracts as of November 2017. The data provided included the contract number, the tribe or tribal organization with the contract, and the program included in the contract. To assess the reliability of the data, we consulted with knowledgeable federal officials and found examples in one of our prior reports that generally supported the data we obtained from the Office of Chief Financial Officer. To determine tribes’ use of self- governance compacts, we reviewed data that Interior provides to Congress in annual reports that cover tribal use of self-governance compacts. To assess the reliability of the data, we consulted with Interior’s Office of Self Governance officials and tribal stakeholders and compared information provided to us from Interior with information obtained from the Tribal Self-Governance Communication and Education Consortium. We determined that the data were sufficiently reliable for the purpose of our report. To obtain a better understanding of the information found in self- determination contracts, we requested BIA provide information from self- determination contract files. We requested contract files that would represent a range of BIA regions and programs. We also sought to use this information to identify examples from the contract file where BIA documented the amount of program funding available to the tribe and retained by BIA, and the methodology BIA used to identify available amounts. Through our review of several contract files, we were able to corroborate information from BIA officials and tribal stakeholders, who told us that BIA does not systematically document the amount of program funding available to the tribe and retained by BIA and the methodology BIA used to identify available amounts. The findings from the contract reviews are not generalizable to those we did not request and obtain. We also collected information from 9 BIA regions on the number of retrocessions (tribes that voluntarily turned back administration of a program to BIA), reassumptions (programs where BIA took back administration from a tribe because of noncompliance with contract requirements), and declinations (programs that tribes requested to take over administration but BIA declined) from 2012 through 2017. BIA does not have a centralized data system to collect this information and through consultations with knowledgeable federal officials, we determined that each of BIA’s regions was in the best position to provide us with this information. To determine tribal participation with the HEARTH Act and the extent to which Interior’s review is consistent with the Act, we collected data from BIA on the number of tribes that have submitted leasing regulations for BIA’s review, and the number of tribal leasing regulations BIA approved under the HEARTH Act. In most cases, Interior provided an internal checklist that included, among other things, the dates tribes submitted information and dates of Interior responses. We used this information to identify the amount of time associated with BIA’s review of tribal leasing regulations. In some cases, we also gathered information from tribes. We determined that the data were sufficiently reliable for the purposes of this report. We interviewed federal officials from Interior’s Office of Solicitor, Indian Affair’s Office of Self Governance, Office of the Chief Financial Officer, and Office of Budget and Performance Management. Within BIA, we met with officials from Office of Trust Services, the Office of Indian Services and interviewed or received written responses from regional officials in all 12 BIA regions. Through these interactions we asked officials to identify processes associated with tribes entering into, negotiating, and administering federal programs under a self-determination contract or self-governance compact. We also discussed processes associated with Interior’s disbursement of funds agreed upon in contracts and compacts. In addition, we discussed processes for tribes to submit leasing regulations to BIA and for BIA’s review of tribal leasing regulations. We compared the information collected through discussions with federal officials and federal documents with Interior guidance documents and Standards for Internal Control in the Federal Government. We also discussed the use of self-determination contracts and self-governance compacts with Interior’s Bureau of Land Management and Bureau of Reclamation, and interviewed officials from the Environmental Protection Agency to discuss tribes’ use of existing authorities to administer environmental programs and the agency’s efforts to build tribal capacity. To identify factors that can affect a tribe’s decision to use self- determination contracts, self-governance compacts, and the HEARTH Act—as well as tribes’ experience with these mechanisms—we interviewed leaders and officials from 29 federally recognized Indian tribes and nations, the Department of the Interior Self-Governance Advisory Committee, and non-profits representing tribal interests, such as the National Congress of American Indians (NCAI) and the Native Governance Center. The key factors we included in this report are those that were most frequently mentioned and that are specifically related to federal government policies and processes. During the review, we identified factors that tribes may consider but that are not related to the federal government; because these factors were outside of the scope of this review, we did not include them in our report. We selected Indian tribes and nations to ensure a representation of tribes with a range of experience using self-determination contracts and self-governance compacts, tribal size, and geographic location. We also selected tribes to ensure we had representation from tribes that developed leasing regulations under the HEARTH Act and those that have elected to not yet develop or submit leasing regulations under the HEARTH Act. We also met with representatives from tribal consortia, such as the Coalition of Large Tribes; the Great Plains Tribal Chairman’s Association; the Department of the Interior Tribal Self-Governance Advisory Committee; the United South and Eastern Tribes; and the United Indian Nations of Oklahoma, Kansas, and Texas to gather additional perspectives on factors that can affect tribal participation. To encourage increased participation and perspectives from tribal leaders and officials, we provided opportunities for tribes to contact us for individual discussions by requesting that tribal consortia, as well as NCAI, include information about our review in their newsletters or other correspondence with tribal stakeholders. As a result of these efforts, several additional tribes contacted us to share information about their experiences. For the purposes of this review, we refer to tribal leaders, tribal government officials, and representatives from tribal consortia as tribal stakeholders. Throughout the report, we use the following categories to quantify statements made by stakeholders: “some” is defined as two to five entities and “several” is defined as six to 10 entities. Because each of the federally recognized tribes and nations are unique, the information obtained in our discussions with tribal stakeholders is not generalizable, but provides examples of tribes’ experiences with self-determination contracts, self-governance compacts, and the HEARTH Act. It is possible we did not identify all of the factors that can affect a tribe’s decision to use self-determination contracts, self-governance compacts, or the HEARTH Act and there may be other factors we did not present. We conducted this performance audit from February 2017 to January 2019 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Comments from the Department of the Interior Appendix III: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Christine Kehr (Assistant Director); Jay Spaan (Analyst in Charge); John Delicath, William Gerard, Cindy Gilbert, Greg Marchand, Dan Purdy, Vasiliki Theodoropoulos, and Leigh White made key contributions to this report.

Why GAO Did This Study For more than 4 decades, federal Indian policy has promoted tribal self-government—the practical exercise of Indian tribes and nations' inherent sovereign authority. Under ISDEAA, federally recognized tribes may request to enter into self-determination contracts and self-governance compacts with Interior, transferring the administration of federal programs to the tribe. Under the HEARTH Act, tribes may issue certain leases on their lands without Interior approval if such leases are executed under approved tribal regulations. GAO was asked to evaluate issues related to tribal self-government. This report examines factors affecting tribes' use of self-determination contracts, self-governance compacts, and tribal leasing authority under the HEARTH Act. GAO reviewed key legislation and regulations, relevant literature, federal and tribal documents; analyzed agency data; and interviewed federal officials at 12 BIA regional offices, 29 tribes that used at least one of these mechanisms, and 7 tribal organizations. What GAO Found GAO found that various factors can affect tribes' use of self-determination contracts and self-governance compacts under the Indian Self-Determination and Education Assistance Act of 1975 (ISDEAA), as amended, and tribal leasing under the Helping Expedite and Advance Responsible Tribal Home Ownership Act of 2012 (HEARTH Act). A key factor that helps tribes use these self-governance mechanisms is tribal government capacity to administer a federal program or manage these resources. Federal efforts that have helped build this capacity have included training, such as that offered by the Bureau of Indian Affairs (BIA) in 2014 and 2015 to educate tribes on the benefits of developing tribal leasing regulations under the HEARTH Act. In contrast, GAO found that other factors can hinder tribes' use of these mechanisms including: Inadequate Information Sharing. The Department of the Interior's (Interior) policy and guidance states that tribes should be provided necessary information to design programs they would like to self-administer, such as the amount of funding available to the tribes for the programs and the amount retained by Interior for inherently federal functions. However, according to several tribal stakeholders and some BIA regional officials GAO spoke to, some of this information is not made available to the tribes prior to self-determination contract negotiations, such as information on funding calculations and determinations of inherently federal functions. Without this information, according to a tribal stakeholder, tribes may be at a disadvantage when negotiating with BIA and designing programs for self-determination contracts. Delays in Disbursing Funds. According to tribal stakeholders, Interior's process does not ensure that funds associated with their self-determination contracts and self-governance compacts are disbursed in a timely manner. These funding delays can therefore be a factor that hinders their use of self-government mechanisms. Some tribal stakeholders said that disbursement delays have ranged from weeks to months. GAO was unable to determine the extent to which Interior disburses funds in accordance with ISDEAA or within agreed-upon time frames with the tribes, because Interior does not systematically track and monitor the disbursement of these funds. Lengthy Review of Proposed Tribal Leasing Regulations. Interior does not have a clearly documented process for reviewing proposed tribal leasing regulations submitted under the HEARTH Act with identified time frames associated with each step of the process. As a result, tribal stakeholders told GAO that they are uncertain about how long the process will take and how it aligns with the 120 day requirement in the Act. According to tribal stakeholders and GAO's analysis of proposed regulations submitted from 2012 through 2017, Interior's review process has resulted in lengthy review times—in some cases, multiple years. Some tribal officials told GAO that Interior's lengthy review process had delayed the tribe's ability to make decisions about the use of their resources. By developing a clearly documented process that includes established time frames for each step in the review, Interior can help eliminate uncertainty and improve the transparency of the review process for the tribes. What GAO Recommends GAO is making four recommendations, including that Interior develop processes to share how it makes funding and inherently federal function determinations with tribes, to track and monitor the disbursement of funds within agreed upon time frames, and for the review of proposed tribal leasing regulations including review time frames. Interior concurred with GAO's recommendations.

gao_GAO-18-415

gao_GAO-18-415_0

Background The FSM and the RMI are independent countries located about 3,000 miles southwest of Hawaii (see fig. 1). The FSM is a federation of four semiautonomous states—Chuuk, Kosrae, Pohnpei, and Yap—whose population and income vary widely. Chuuk, the largest state by population, has the lowest per capita gross domestic product (GDP). Overall, the FSM had a 2016 population of approximately 102,000 and a GDP per capita of about $3,200. The RMI’s 2016 population was approximately 54,000 with a GDP per capita of about $3,600. The RMI’s most recent census, in 2011, found that approximately three-quarters of the population lived in Majuro, the nation’s capital, and on the island of Ebeye in the Kwajalein Atoll. Table 1 shows the FSM’s, FSM states’, and RMI’s estimated population and annual GDP per capita in fiscal year 2016. FSM and RMI Government Revenues The FSM states maintain considerable authority, relative to the FSM national government, to allocate U.S. assistance and implement budgetary policies. While the United States provides compact sector grants directly to the FSM national government, a large portion of these grants is passed through and provided to the four FSM states. The states also receive other U.S. program grants that have been passed through from the national government but may also receive grants directly from U.S. agencies. Overall, FSM public sector revenue sources include U.S. compact and program grants; grants from other countries; taxation, including taxation of foreign corporations domiciled in the FSM; and Parties to the Nauru Agreement fishing fees charged to vessels operating in its waters. In addition to maintaining departmental budgets, both the FSM national government and the FSM states have government-owned enterprises and component units, such as public utilities and port authorities, whose operations are supported by public funds. Some of these component units also receive U.S. compact sector grants or other U.S. grants passed through the FSM national or state governments or directly from U.S. agencies. According to Graduate School USA, the FSM’s public sector accounted for about 53 percent of all employment in the FSM in fiscal year 2016. The RMI government is responsible for allocating U.S. assistance in that country, though the RMI’s 24 local governments exercise local government authority. RMI public sector revenue sources include U.S. compact and program grants, grants from other countries, ship and corporate registry earnings, and Parties to the Nauru Agreement fishing fees. The RMI government also has state-owned enterprises and component units whose operations are supported by public funds. Some of these component units receive U.S. compact sector grants or other U.S. grants passed through the RMI government or directly from U.S. agencies. According to Graduate School USA, in fiscal year 2016, RMI’s public sector accounted for approximately 48 percent of all employment in the RMI. The U.S. Army Garrison–Kwajalein Atoll, located near Ebeye island, also provides a significant source of employment for Marshallese. In September 2017, U.S. Army Garrison-Kwajalein Atoll officials estimated that approximately 1,100 Marshallese were employed at the garrison. Compact of Free Association (1986–2003) U.S. relations with the FSM and the RMI began during World War II, when the United States ended Japanese occupation of the region. Beginning in 1947, the United States administered the region under a United Nations trusteeship. During the 1940s and 1950s, the RMI was the site of 67 U.S. nuclear weapons tests on or near Bikini and Enewetak Atolls. The four states of the FSM voted in a 1978 referendum to become an independent nation, while the RMI established a constitutional government and declared itself a republic in 1979. Under the trusteeship agreement, both newly formed nations remained subject to the authority of the United States until 1986. In 1986, following a period of negotiations, the United States entered into a compact of free association with the FSM and the RMI that provided for economic assistance to the two countries, secured U.S. defense rights, and allowed FSM and RMI citizens to migrate to the United States. The compact provided a framework for the United States and the two countries to work toward achieving the following three main goals: (1) establish self-government for the FSM and the RMI, (2) ensure certain national security rights for all of the parties, and (3) assist the FSM and the RMI in their efforts to advance economic development and self- sufficiency. The compact’s third goal was to be accomplished primarily through U.S. direct financial assistance to the FSM and the RMI. Under the original compact, the FSM and the RMI used funds for general government operations; capital projects, such as building roads and investing in businesses; debt payments; and targeted sectors, such as energy and communications. The FSM concentrated much of its spending on government operations at both national and state levels, while the RMI emphasized capital spending. While the original compact set out specific obligations for reporting and consultations regarding the use of compact funds, the FSM, RMI, and U.S. governments provided little accountability over compact expenditures and did not ensure that funds were spent effectively or efficiently. Amended Compacts of Free Association (2004– Present) In 2003, following a period of negotiations, the United States approved separate amended compacts with the FSM and the RMI that went into effect on June 25, 2004, and May 1, 2004, respectively. Compact Grants and Trust Fund Contributions The amended compacts’ implementing legislation authorized and appropriated direct financial assistance to the FSM and the RMI in fiscal years 2004 through 2023, with the base amounts decreasing in most years. The annual decrements in assistance are added to the amounts deposited in the trust funds established under the amended compacts for the two nations. Earnings from the compact trust funds are intended to provide an annual source of revenue after the scheduled end of compact sector grants at the end of fiscal year 2023. Both the compact sector grants and trust fund contributions are partially adjusted for inflation each fiscal year. Appendix II provides additional information on the base and inflation-adjusted amounts of U.S. compact sector grants and trust fund contributions in fiscal years 2004 through 2023. The amended compacts and associated fiscal procedures agreements require that compact sector grants support the countries in six core sectors—education, health, infrastructure, environment, private sector development, and public sector capacity building—with the education and health sectors having priority. These grants are described in section 211(a) of each compact and are referred to as compact sector grants or 211(a) grants. Section 211(b) of the RMI compact further states that the RMI must target a specified amount of grants to Ebeye and other Marshallese communities within Kwajalein Atoll. The RMI MUORA states that the Kwajalein-related funds provided to the RMI in the compacts shall be provided through fiscal year 2023 “and thereafter for as long as this agreement remains in effect.” Compact Trust Fund Management and Implementation The amended compacts and their subsidiary trust fund agreements provided that each trust fund is to be managed by a compact trust fund committee. Each compact trust fund committee includes representatives from both the United States and the respective country, but the United States is required by the terms of the trust fund agreements to hold the majority of votes on each committee. The Director of Interior’s Office of Insular Affairs serves as the chair of each committee. Trust fund committee responsibilities include overseeing fund operation, supervision, and management; investing and distributing the fund’s resources; and concluding agreements with any other contributors and other organizations. As part of this oversight, the committees are to establish an investment and distribution policy. The committees are also to determine fiscal procedures to be used in implementing the trust fund agreements based on the fiscal procedures used for compact grant administration unless otherwise agreed by the parties to the agreement. The trust fund agreements between the United States and the FSM and the RMI allow for the agreements to be amended in writing at any time, with mutual consent of the governments. However, the U.S. legislation implementing the amended compacts requires that any amendment, change, or termination of all or any part of the compact trust fund agreements shall not enter into force until incorporated into an act of Congress. According to the trust fund agreements, each trust fund committee is to appoint a trustee and an independent auditor. Each committee has retained an Executive Administrator to manage the daily operations of the trust fund. In addition, the committee has the authority to appoint 1 or more investment advisers and may enter into a separate agreement with 1 or more money managers. The investment policy statement for each fund guides the fund’s investment strategy and portfolio. Compact Trust Fund Structure The compact trust fund agreements state that no funds, other than specified trust fund administrative expenses, may be distributed from the compact trust fund prior to October 1, 2023. From fiscal year 2024 onward, the maximum allowed disbursement from each compact trust fund is the amount of the fiscal year 2023 annual grant assistance, as defined by the trust fund agreement, with full adjustment for inflation. In addition, the trust fund committees may approve additional amounts for special needs. The RMI compact trust fund agreement excludes from the calculation of the allowed disbursement the amount of the Kwajalein- related assistance defined in section 211(b) of the RMI compact. Although the compact trust fund agreements state the maximum allowable disbursement level, they do not establish or guarantee a minimum disbursement level. Each country’s compact trust fund consists of three interrelated accounts: the “A” account, the “B” account, and the “C” account. The A account is the trust fund’s corpus and contains the initial, and any additional, U.S. and FSM or RMI contributions; contributions from other countries; and investment earnings. No funds, other than specified trust fund administrative expenses, may be disbursed from the A account. The B account is the trust fund’s disbursement account and becomes active in fiscal year 2023. All income earned in 2023 will be deposited in the B account for possible disbursement in 2024. Each subsequent year’s investment income will similarly be deposited into the B account for possible disbursement the following year. If there is no investment income, no funds will be deposited in the B account for possible disbursement the following year. The C account is the trust fund’s buffer account. Through 2022, any annual income exceeding 6 percent of the fund balance is deposited in the C account. From 2023 onward, if annual income from the A account is less than the previous year’s disbursement, adjusted for inflation, the C account may be tapped to address the shortfall. After 2023, any funds in the B account in excess of the amount approved for disbursement the following fiscal year are to be used to replenish the C account as needed, up to the maximum size of the account. The size of the C account is capped at three times the amount of the estimated annual grant assistance in 2023, including estimated inflation. If there are no funds in the C account, and no prior year investment income in the B account, no funds will be available for disbursement to the countries the following year. Figure 2 shows the compact trust fund account structure and associated rules. According to the U.S. trust fund agreements with the FSM and the RMI, contributions from other donors are permitted. In May 2005, Taiwan and the RMI reached an agreement that Taiwan will contribute a total of $40 million to the RMI’s compact trust fund A account between 2004 and 2023. A “D” account may also be established to hold any contributions by the FSM and the RMI governments of revenue or income from unanticipated sources. According to the trust fund agreements, the D account must be a separate account, not mixed with the rest of the trust fund. Only the RMI has a D account, governed in part by an agreement between Taiwan and the RMI. Compact Accountability, Management Structures, and Reporting The amended compacts’ implementing legislation and their subsidiary fiscal procedures agreements established committees to oversee compact grants to each country—the Joint Economic Management Committee (JEMCO) for the FSM and the Joint Economic Management and Financial Accountability Committee (JEMFAC) for the RMI. Each five- member committee comprises three representatives from the U.S. government and two representatives from the corresponding country, with the Director of Interior’s Office of Insular Affairs serving as the chair. JEMCO’s and JEMFAC’s designated roles and responsibilities include the following: reviewing the budget and development plans from each of the governments; approving grant allocations and performance objectives; attaching terms and conditions to any or all annual grant awards to improve program performance and fiscal accountability; evaluating progress, management problems, and any shifts in priorities in each sector; and reviewing audits called for in the compacts. JEMCO and JEMFAC can require that terms and conditions be attached to any and all annual compact sector grant awards to improve program performance and fiscal accountability. Under the fiscal procedures agreements governing the amended compacts, the Office of Insular Affairs is responsible for using financial reports to monitor each country’s budget and fiscal performance and for using performance reports submitted by the countries to evaluate sector grant performance. The FSM and the RMI also must adhere to specific fiscal control and accounting procedures and are required to submit annual audit reports, within the meaning of the Single Audit Act as amended. The FSM and RMI compacts require each country to develop multiyear plans that are strategic in nature and continuously reviewed and updated through the annual budget process and that address the assistance for the defined sectors. In 2013, we recommended that Interior, as Chair of JEMCO and JEMFAC, ensure that the FSM and the RMI complete plans to address the impact of declining compact sector grants (in this report, decrement management plans). In November 2013, the FSM finalized its decrement management plan for fiscal years 2014 through 2023; the plan indicated that a similar planning process is to be repeated in 3-year intervals. In September 2014, the RMI finalized its decrement management plan for fiscal years 2015 through 2023; the plan similarly stated that a comprehensive planning process to address the ongoing decrement may proceed on a 3-year update schedule. Each decrement management plan includes commitments for budget reductions in the national governments and, in the FSM, the state governments, as well as plans to undertake actions such as tax reform. Programs and Services Provided in Compact-Related Agreements The amended compacts’ implementing legislation incorporates by reference related agreements extending programs and services to the FSM and RMI. The programs and services agreement with each country identifies the following programs and services as being available to each country: U.S. postal services, weather services, civil aviation, disaster preparedness and response, and telecommunications. Each programs and services agreement extends for 20 years from the compact’s entry into force. Therefore, the agreement with the FSM ends on June 24, 2024, and the agreement with the RMI ends on April 30, 2024. Programs Authorized by U.S. Legislation The amended compacts’ implementing legislation (Pub. L. No. 108-188) and other U.S. legislation authorize other U.S. grants, programs, and services for the FSM and RMI. Pub. L. No. 108-188 authorized an annual supplemental education grant (SEG) for the FSM and RMI in fiscal years 2005 through 2023, to be awarded in place of grants formerly awarded to the countries under several U.S. education, health, and labor programs. The FSM and RMI are not eligible for the programs replaced by the SEG during these years. Unlike the compact sector grants, the amended compacts’ implementing legislation authorized the SEG but did not appropriate funds for it. Funding for the SEG is appropriated annually to the U.S. Department of Education (Education) and transferred to Interior for disbursement. Other provisions of the amended compacts’ implementing legislation, as well as other U.S. law, make the FSM and RMI eligible for a number of additional programs. Other federal departments are responsible for the administration and oversight of their respective programs in the FSM and RMI. The FSM and RMI Continue to Rely on U.S. Grants and Programs That End in 2023 Compact sector grants and the SEG, each of which end in 2023, continue to support a substantial portion of government expenditures in the FSM and RMI. In the FSM, compact sector grants and the SEG support about one-third of all government expenditures. The four FSM states rely on these grants to a greater extent than the FSM national government does. In the RMI, compact sector grants and the SEG support about one- quarter of all government expenditures. The end of the compacts’ programs and services agreements in 2024 would also require the FSM and RMI to bear additional costs to provide services currently provided by the United States as part of the Agreements. Appendix IV provides a detailed summary of programs and services we identified that have been provided through the amended compacts, the amended compacts’ implementing legislation, compact-related agreements, and other provisions of U.S. law, as well as their status in the FSM and RMI after 2023. U.S. Compact Grants and Other Grants Continue to Provide Substantial Support to the FSM and RMI Budgets U.S. Grants Scheduled to End in 2023 Support About One- Third of Total FSM Government Expenditures The FSM national and state governments overall continue to rely on U.S. support for program expenditures. Compact sector grants, the SEG, and other U.S. grants supported almost half of FSM national and state government expenditures in fiscal year 2016. Compact sector and supplemental education grants that end in 2023 supported approximately one-third of total FSM national and state government expenditures in fiscal year 2016, while other U.S. grants supported an additional 15 percent of total FSM government expenditures (see fig. 3). Compact sector and supplemental education grants that end in 2023 support a larger proportion of FSM state governments’ expenditures than of the FSM national government’s expenditures. In fiscal year 2016, compact sector grants and the SEG supported 8 percent of national government expenditures but supported 50 percent or more of each state’s government expenditures. Among the FSM states, Chuuk—both the largest state and the state with the lowest per capita income in the FSM—has the highest percentage of its expenditures supported by U.S. grants. (See table 2 for a summary of FSM national and state government expenditures supported by compact sector grants and the SEG, and by other U.S. grants.) Compact sector grants and the SEG support an even higher proportion of FSM states’ health and education expenditures. See app. III for a summary of the role of compact funds in the FSM health and education sectors. The RMI continues to rely on U.S. support for program expenditures. Compact sector and supplemental education grants that end in 2023 supported approximately 25 percent of the RMI’s $123.5 million in government expenditures in fiscal year 2016, while other U.S. grants supported an additional 8 percent. Compact Kwajalein-related grants that do not end in 2023 supported an additional 3 percent (see fig. 4). Compact sector grants and the SEG support an even higher proportion of RMI health and education expenditures. See app. III for a summary of the role of compact funds in the RMI health and education sectors. End of the Programs and Services Agreements Would Also Affect FSM and RMI Budgets FSM and RMI budgets would be affected if the countries were to assume responsibility for providing some additional programs and services currently provided by the United States. Current U.S. law enables U.S. agencies to continue providing some programs and services now provided under the agreements after they end in 2024. However, under current law, some programs and services provided in the programs and services agreements will end and would require the FSM and RMI to bear additional costs. See appendix IV for a summary of the status of programs and services provided under the programs and services agreements after the agreements end. Compact Trust Funds Face Continuing Risks That Trust Fund Committees Have Not Yet Addressed Previous studies of the FSM and RMI compact trust funds, including a review we conducted in 2007, found that after fiscal year 2023 the funds are unlikely to provide maximum annual disbursements, may provide no disbursements at all in some years, and are unlikely to sustain the funds’ fiscal year 2023 value. Our updated projections for the compact trust funds show similar outlooks. Several potential strategies could improve the compact trust funds’ outlook; some of these strategies could be implemented under the current trust fund agreements, while other strategies may require changing the trust fund agreements. The compact trust fund committees have not yet prepared distribution policies, required by the trust fund agreements, that could assist the countries in planning for the transition to trust fund income. In addition, the committees have not established fiscal procedures for oversight of compact trust fund disbursements as required by the trust fund agreements. Further, the trust fund committees have not yet addressed a potential misalignment between the timing of their annual calculation of the amounts available to disburse and the FSM’s and RMI’s budget timelines, potentially complicating each country’s planning and management. Previous Studies of Compact Trust Funds Found Increasing Risks to Disbursements and Sustainability under Current Rules Previous studies of the compact trust funds have found that some yearly disbursements from the funds after 2023 are likely to fall short of the inflation-adjusted amount of annual grant assistance in 2023 and that the funds may provide no disbursement at all in some years. Our 2007 analysis of the compact trust funds projected a wide range of potential balances and found that the funds’ capacity to provide the maximum allowable disbursement would likely decrease over time. In addition, our analysis showed an increasing likelihood that the trust funds would exhaust the C account and be unable to provide any disbursements in the latter years of our projection. Other analyses have similarly found risks of low or zero disbursements and risks to sustainability. Graduate School USA has prepared an annual series of economic reports on each country, including analyses of their compact trust funds. In 2015, an Asian Development Bank report separately analyzed the trust funds. The International Monetary Fund projected the status of the trust funds as part of its biennial FSM and RMI consultations. Updated Projections Show Continuing Risks to Compact Trust Fund Disbursements and Sustainability Our updated projections for the FSM and RMI compact trust funds after 2023 indicate a continued likelihood that, given their balance at the end of fiscal year 2017 and current compact trust fund rules—the baseline scenario—the funds will be unable to provide maximum disbursements (equal to the inflation- adjusted amount of annual grant assistance in 2023) in some years; unable to provide any disbursement at all in some years, with the likelihood of zero disbursement in a given year increasing over time; and unable to maintain the inflation-adjusted value of the compact trust fund after fiscal year 2023. The compact trust funds’ C account—designed as a buffer to protect disbursements from the B account in years when the funds do not earn enough to fund the disbursement—could be exhausted by a series of years with low or negative annual returns. Since current rules do not allow disbursements from the compact trust fund corpus (the A account), exhaustion of the C account would result in zero disbursement in years when fund returns are zero or negative. Thus, there may be no funds available to disburse even if the funds’ A accounts have a balance. As a result of low or zero disbursements, the countries could face economic and fiscal shocks and significant challenges in planning programs and budgets. FSM Compact Trust Fund Projections Our model projects that, given the baseline scenario and a 6 percent net return, the FSM compact trust fund will experience declining disbursements relative to the maximum allowable disbursements; an increasing chance of zero disbursements; and a declining likelihood of maintaining its 2023 balance. See appendix I for a full description of our methodology and appendix V for the baseline results with alternative net returns. Projected disbursements. We project that the FSM compact trust fund will, on average, be able to provide disbursements equal to 82 percent of the maximum allowable disbursement—the inflation- adjusted amount of 2023 annual grant assistance—in its first decade of disbursements. The likely average disbursement falls to 49 percent of the maximum in the next decade and falls further in subsequent decades. In addition, the amount available for disbursement may fluctuate substantially from year to year. Depending on the compact trust fund’s performance in the previous year, disbursements may be higher or lower than the average amount if the balance in the C account is not sufficient to provide additional disbursements. Likelihood of providing zero disbursement. We project a 41 percent likelihood that the FSM compact trust fund will be unable to disburse any funds in 1 or more years during the first decade of trust fund disbursements. This likelihood increases over time, rising to 92 percent in fiscal years 2054 through 2063. Likelihood of maintaining inflation-adjusted 2023 balance. We project a 13 percent likelihood that the FSM compact trust fund will maintain or exceed its inflation-adjusted fiscal year 2023 value in fiscal year 2033. This likelihood decreases in later years. Figure 5 shows our projections of the FSM compact trust fund’s average disbursements as a percentage of maximum disbursement, the likelihood of 1 or more years of zero disbursement, and the likelihood of the fund’s maintaining its inflation-adjusted fiscal year 2023 balance given the baseline scenario and a 6 percent net return. The FSM also maintains its own trust fund separate from the compact trust fund (see app. VI for additional information). We did not independently project the FSM Trust Fund’s future balance or potential disbursements after 2023. RMI Compact Trust Fund Projections Our model projects that, given the baseline scenario and a 6 percent net return, the RMI compact trust fund will experience declining disbursements relative to the maximum allowable disbursements; an increasing chance of zero disbursements; and a declining likelihood of sustaining its 2023 balance. Projected disbursements. We project that the RMI compact trust fund will, on average, be able to provide disbursements nearly equal to the inflation-adjusted amount of 2023 annual grant assistance as defined by the trust fund agreement—the maximum allowable—in its first decade of disbursements. However, the projected disbursements as a percentage of the maximum disbursements decline by about 10 percentage points in each subsequent decade. In addition, the amount available to disburse may fluctuate substantially from year to year. Depending on the compact trust fund’s performance in the previous year, disbursements may be higher or lower than the average amount if the balance in the C account is not sufficient to provide additional disbursements. Likelihood of providing zero disbursement. We project a 15 percent likelihood that the RMI compact trust fund will be unable to disburse any funds in 1 or more years during the first decade of trust fund disbursements. This likelihood increases over time, rising to 56 percent in fiscal years 2054 through 2063. Likelihood of maintaining inflation-adjusted 2023 balance. We project a 41 percent likelihood that the RMI compact trust fund will maintain or exceed its inflation-adjusted fiscal year 2023 value in fiscal year 2033. This likelihood decreases in later years. Figure 6 shows our projections of the RMI compact trust fund’s average disbursements as a percentage of maximum disbursement, its likelihood of 1 or more years of zero disbursement, and its likelihood of maintaining its inflation-adjusted fiscal year 2023 balance given the baseline scenario and a 6 percent net return. The RMI also maintains its own D account separate from the compact trust fund (see app. VI for additional information). We did not independently project the D account balance or potential disbursements from the D account after 2023. Reducing Disbursements, Making Additional Contributions, and Changing Disbursement Policies Would Each Affect the Outlook of the Compact Trust Funds We conducted a series of simulations to determine the likely effects of potential strategies for improving the outlook of the FSM and RMI compact trust funds. Prior studies by Graduate School USA, the Asian Development Bank, and the International Monetary Fund examined the effects of three general approaches for improving the trust funds’ outlooks: (1) reducing planned disbursements from the funds, (2) making additional contributions to the funds, and (3) changing the compact trust fund disbursement policies. These prior studies included strategies that would require changing the trust fund agreements to permit disbursements from the A account. To isolate the impact of individual changes on compact trust fund balance and disbursements, we developed and analyzed five potential strategies based on the approaches examined in the prior studies. 1. Annual disbursements are reduced below the maximum allowable disbursement. 2. Additional annual contributions are made to the trust fund in fiscal years 2018 through 2023. 3. The trust fund agreement disbursement policies are modified to limit the annual disbursement to a fixed percentage of the fund’s moving average balance over the previous 3 years, up to the maximum disbursement amount defined by the current trust fund agreement. 4. The trust fund agreement disbursement policies are modified to reduce the amount of the annual disbursement if the compact trust fund’s moving average balance over the previous 5 years is lower than a primary target amount. 5. The trust fund agreement disbursement policies are modified to set the target disbursement as 2.1 percent of the compact trust fund’s balance in fiscal year 2024. The disbursement amount is further decreased if the fund’s moving average balance over the previous 5 years is lower than the primary target balance. Implementing either of the first two potential strategies would not require any changes to disbursement provisions in the existing trust fund agreement, but implementing any of the remaining three strategies may require such changes. In strategies 3, 4, and 5, we analyzed strategies that would permit disbursement from the A account. Disbursing from the A account would require changing the compact trust fund agreements. The agreements can be amended in writing at any time, with mutual consent of the governments. However, the U.S. legislation implementing the amended compacts requires that any amendment, change, or termination of all or any part of the compact trust fund agreements shall not enter into force until incorporated into an act of Congress. All of the potential strategies we analyzed would reduce or eliminate the risk of the compact trust funds experiencing years of zero disbursement. However, all of the potential strategies would require the countries to exchange a near-term reduction in resources for more predictable and sustainable disbursements in the longer term. Appendix VII presents the detailed results of our analysis. Compact Trust Fund Committees Have Not Addressed Issues Related to Distribution Policies, Fiscal Procedures, and Disbursement Timing Trust Fund Committees Have Not Developed Distribution Policies Required by the Compact Trust Fund Agreements Under the compact trust fund agreements, each trust fund committee must develop a distribution policy, with the intent that compact trust fund disbursements will provide an annual source of revenue to the FSM and RMI after fiscal year 2023. The trust fund committees could use distribution policies to address risks to each fund’s sustainability. For example, the committees have the discretion to disburse an amount below the established maximum. Our analysis of potential strategies for improving the funds’ outlook shows that reducing the size of disbursements would improve each compact trust fund’s long-term sustainability. According to interviews with, and documents provided by, the trust funds’ administrator, the committees reviewed presentations in 2016, 2017, and early 2018 from the authors of previous studies and fund managers regarding the likely status of the trust funds after 2023 and have also reviewed options for addressing risks to the trust funds’ disbursements and sustainability, including changes to disbursement provisions in the compact trust fund agreements. However, as of January 2018, according to the trust funds’ administrator, neither committee had developed a distribution policy. Without a distribution policy that provides information about the size of expected disbursements, the FSM and RMI are hampered in their current and ongoing efforts to plan for the potential reduction in U.S. compact assistance after 2023. Trust Fund Committees Have Not Established Fiscal Procedures Required by Compact Trust Fund Agreements The compact trust fund committees have not yet established fiscal procedures for compact trust fund disbursements after fiscal year 2023. Each trust fund agreement requires the respective committee to determine the fiscal procedures to be used in implementing the trust fund agreement. The committees are to base their procedures on the compact fiscal procedures agreements, which define the membership and duties of the JEMCO and JEMFAC and single audit report requirements, among other things, unless the parties to the trust fund agreement agree to adopt different fiscal procedures. No compact trust fund disbursements are to be made unless the committee has established such trust fund fiscal procedures. U.S., FSM, and RMI officials are aware of the need to determine the fiscal procedures that will govern oversight of compact trust fund disbursements. Issues related to future oversight of compact trust fund disbursements have been raised for discussion with U.S. representatives on JEMCO and JEMFAC. However, according to an RMI representative on the compact trust fund committee, that committee has not discussed fiscal procedures for the compact trust fund disbursements. In addition, FSM officials noted that they were unsure whether the JEMCO or the compact trust fund committees would approve specific projects. Without fiscal procedures in place, the trust fund committees will not be able to provide disbursements and the United States, the FSM, and the RMI will lack clear guidance to ensure oversight for trust fund disbursements. Trust Fund Committees Have Not Addressed Issues Related to Disbursement Timing The timing of the trust fund committees’ calculation of the amounts available for annual disbursement to the FSM and the RMI does not align with the countries’ budget and planning timelines. The amounts available for disbursement in a given fiscal year cannot be determined until each fund’s returns have been determined at the end of the prior year. Further, if the disbursement amounts are calculated from audited fund returns as determined by annual audits required by the trust fund agreements, the amounts may not be determined until as late as March 31, 6 months into the fiscal year for which the disbursement is to be provided. However, both the FSM and the RMI government budget cycles are completed before the annual amounts available for disbursement will be known. As a result, the FSM and RMI would have to budget without knowing the amount to be disbursed, complicating their annual budget and planning processes. See figures 7 and 8 for the FSM and RMI budget timelines for fiscal year 2024, based on their current budget calendars, relative to the dates when the compact trust fund disbursement amounts will be determined on the basis of the funds’ unaudited end-of-fiscal-year balances and of their audited balances. Standards for Internal Control in the Federal Government—which is applicable to the U.S. government but can be adopted as a best practice by nongovernmental entities—states that management should use quality information to achieve the entity’s objectives. For example, as part of using quality information, the entity obtains relevant data from reliable internal and external sources in a timely manner based on the identified information requirements. Given the FSM’s and RMI’s current budget processes, the FSM and RMI will not have accurate and timely information on the amounts that will be available for annual disbursements for each fiscal year. The FSM Secretary of Finance and Administration, a member of the compact trust fund committee, indicated that she is aware of the discrepancy between the timing of the trust fund disbursement calculations and dates in the FSM’s budget and planning cycle and stated that the FSM would raise the issue of this discrepancy as part of its planning for the transition to relying on compact trust fund disbursements. One of the RMI’s representatives on the compact trust fund committee stated that the timing of the disbursement calculations was a challenge and would complicate RMI planning and management. Each trust fund committee received a briefing in 2016 from the trust funds’ administrator that discussed issues associated with the timing of the disbursement calculations. However, as of January 2018, the committees had not determined how they would address this issue. FSM and RMI Decrement Plans Were Not Implemented Because of Increased Revenues, but Each Country Has Begun New Planning Efforts The FSM and RMI did not implement planned budget reductions to address decreasing compact sector grants because of increasing revenue from other sources. FSM officials stated that they did not implement their plan’s planned budgetary reductions due to increasing revenues for the state and national governments. The RMI also did not implement budget reductions but used increased revenue, particularly from fishing fees, to offset the decrement in compact sector grants. FSM and RMI strategic plans in the key sectors of education and health focus on strategic goals and priorities rather than addressing the effect of the 2023 transition on health and education budgets. However, FSM and RMI infrastructure plans discuss funding requirements and potential alternative funding sources. The FSM, the RMI, and the United States have each established bodies to plan to address issues related to the 2023 transition to trust fund income. Previous Decrement Management Plans Were Not Implemented Because of Growth in Revenues FSM Long-Term Fiscal Framework The FSM has not implemented budget reductions scheduled in its decrement management plan, the FSM-Wide Long-Term Fiscal Framework (Long-Term Fiscal Framework). The FSM’s plan included a firm commitment for a 6 percent reduction in real terms in FSM state expenditures in fiscal year 2014. Two additional 6 percent expenditure reductions were planned for fiscal years 2017 and 2020, but these were contingent reductions that would not be implemented if the FSM states received offsetting revenue to address the reductions. According to FSM national government officials, revenue increases, including growth in revenue from fishing fees, have enabled the FSM to avoid implementing the 2017 contingent 6 percent expenditure reductions, and the further reductions in fiscal year 2020 are not likely to be implemented. FSM officials cited multiple reasons for not implementing the planned reductions: Increasing revenue to the state and national governments. The FSM’s Long-Term Fiscal Framework included a plan to increase the proportion of compact sector grant funding distributed among the FSM states and reduce the proportion retained by the national government. This change in the FSM’s internal compact grant distribution formula reduced the amount of the decrement in compact sector grants received by the states that would have otherwise occurred. The FSM national government’s revenue from fishing fees has increased rapidly in recent years, allowing it to use this revenue in place of compact sector grants. Effect of inflation adjustments on compact sector grants. According to FSM officials, because of inflation adjustments, the nominal value of the compact sector grants has not significantly declined. As a result, the FSM government questions the need for expenditure reductions. In addition to scheduling budget reductions, the FSM’s Long-Term Fiscal Framework included plans to implement unified tax reform measures, which also have not been implemented. However, plans to reduce the national government’s share of compact sector grants and to use surpluses to mitigate the effect of fiscal reforms were implemented. (See app. VIII for a summary of the FSM’s planned actions and their implementation.) As of January 2018, the FSM had not updated the Long- Term Fiscal Framework but had included information updates in its annual budget submittal. RMI Decrement Management Plan The RMI government has not implemented budget reductions scheduled in its decrement management plan. The RMI’s decrement management plan divided proposed budgetary reductions into three periods: fiscal years 2016 through 2017, fiscal years 2018 through 2020, and fiscal years 2021 through 2023. Only the reductions in the first period were to be considered binding, with adjustments in the later periods subject to review during the next 3-year planning cycle. According to RMI government officials, significant growth in fishing fee revenue and growth in ship registry and income tax revenue has minimized the initially anticipated impact of the compact decrements, thereby reducing the need to implement expenditure reductions. RMI officials noted that it expected to continue to use its own revenue in place of compact funds in fiscal years 2019 through 2023. In addition to scheduling the budget reductions, the RMI decrement management plan includes plans to implement new taxes, program fishing fees into the annual budget, reduce subsidies to state-owned enterprises, and reduce compensation to Majuro landowners for the use of their land for utilities. The RMI has programmed a portion of its fishing fee surplus into the annual budget in each fiscal year from 2015 to 2017 but has not implemented other planned actions. (See app. VIII for a summary of planned actions and their implementation.) As of January 2018, the RMI government had not updated its plan and did not intend to do so, according to RMI officials. However, the officials stated that the government has incorporated elements of the plan, particularly its expenditure analysis, into the RMI’s medium term budget and investment framework, a planning and budgeting document submitted to JEMFAC in August 2017. In comments on a draft of this report, the RMI stated that it is developing a long-term fiscal framework in addition to the medium term budget and investment framework. According to the RMI, the long-term fiscal framework will have a 10-year outlook through 2028 and take into account compact decrements and anticipated resources from the compact trust fund and other sources. FSM and RMI Have Developed Plans for Health, Education, and Infrastructure FSM Sector Plans FSM national and state infrastructure plans provide specific budgetary information to address the fiscal year 2023 transition from compact sector grants to trust fund income, such as funding requirements and sources of funding for planned infrastructure projects in fiscal years 2016 through 2025. The FSM national and state health and education plans generally focus on the national and state health and education departments’ strategic goals and priorities rather than discussing budget changes or new revenue generation strategies to address the possibility of reduced resources after 2023. In addition to preparing sector strategic development plans, the FSM national and state governments issued the 2023 Action Plan in 2014, designed to address fiscal and economic challenges before and after compact sector grant funding ends in fiscal year 2023. In contrast to the FSM Long-Term Fiscal Framework, which committed to specific expenditure reductions and government actions prior to fiscal year 2023, the 2023 Action Plan includes an economic growth strategy that seeks to boost private sector development. The plan addresses economic growth strategies and improved performance in key economic sectors such as tourism, agriculture, and fisheries and identifies the need for the FSM national and state governments to limit expenditure growth in the medium and long terms. RMI Sector Plans The RMI’s infrastructure plan addresses the scheduled cessation of compact sector grant assistance in fiscal year 2023 through a review of potential future budgets, while the RMI’s education and health plans outline strategic goals and priorities. Similar to the FSM’s infrastructure plans, the RMI National Infrastructure Plan reviews budget information to address the fiscal year 2023 transition, such as planned infrastructure investments and potential alternative funding sources for fiscal years 2017 through 2026. The RMI’s national education and health plans primarily focus on goals and objectives to address key challenge areas in health and education over the next few fiscal years and do not discuss specific budget changes for the transition in 2023. FSM, RMI, and U.S. Planning Groups Have Been Formed to Prepare for Transition to Trust Fund Income Both the FSM and the RMI have formed planning committees and charged them with planning for the fiscal year 2023 transition from compact sector grants to compact trust fund income. In addition, the U.S. Department of State (State) has organized a U.S. interagency planning group to help coordinate U.S. policy related to the transition. FSM Joint Compact Review and Planning Committee In 2016, the FSM national government established a Joint Compact Review and Planning Committee to coordinate FSM planning for the transition from compact sector grants to trust fund income in 2023. The committee is mandated to, among other things, set goals in anticipation of the end of compact grants, develop strategies and alternatives, identify financial assistance sources, analyze economic information, and provide periodic reports to the FSM Congress. The committee first met in May 2017. In September 2017, the committee hired an Executive Director, who in turn hired an economist and Executive Secretary prior to the committee’s February 2018 meeting. As of January 2018, according to FSM officials, the committee had not produced any publicly available products but had collected information from various FSM government agencies. RMI Compact Review Commission According to the RMI Office of Compact Implementation, the RMI established the Compact Review Commission in late 2016 to plan for the fiscal year 2023 transition from compact sector grants to trust fund income. According to the Office of Compact Implementation, the commission is mandated to review the compact and make recommendations to the cabinet regarding priorities to be addressed for the fiscal year 2023 transition. Specific priorities may include the status of federal programs that will expire in fiscal year 2023, the adequacy of the compact trust fund to provide needed revenue, and other issues relevant to the cessation of compact grant assistance. In January 2018, the RMI Presidential Cabinet appointed a Compact Review Commission Coordinating Committee, consisting of the RMI Ambassador to the United States, the Director of the RMI Office of Compact Implementation, the Secretary of Finance, a private sector representative, and a legal adviser, and directed it to coordinate the commission’s meetings, actions, and reporting. Ongoing U.S. Interagency Working Group State began holding regular meetings of the Interagency Working Group on the Freely Associated States in February 2017 to provide guidance and oversight for policy concerning the Compacts of Free Association and to coordinate U.S. policy in light of the fiscal year 2023 transition. The group met monthly through the rest of 2017, except in November. The monthly meetings have focused individually on the FSM and RMI, as well as addressed cross-cutting issues such as donor coordination. For example, in March and July 2017, the group’s monthly meetings focused on the FSM and included participation by the U.S. Ambassador and the FSM Ambassador, respectively. Similarly, in April and June 2017, the group’s monthly meetings focused on the RMI and included participation by the U.S. and RMI ambassadors, respectively. According to State officials, the meetings will continue indefinitely on a monthly basis. Conclusions The U.S. compacts of free association with the FSM and the RMI provided a framework for the United States and the two countries to work toward, among other things, the goal of assisting the FSM and the RMI in their efforts to achieve economic development and self-sufficiency. The end of U.S. compact sector grants in fiscal year 2023 and the beginning of disbursements from the compact trust funds in fiscal year 2024 will mark a key transition in these ongoing efforts, and the FSM and RMI are currently preparing plans for addressing issues associated with the transition to compact trust fund income. The countries’ transition to relying on income from the compact trust funds will likely require significant budgetary choices. However, lacking the trust fund distribution policies required under the trust fund agreements, the FSM and RMI are hampered in their efforts to plan for the potential reduction in U.S. compact assistance after 2023. In addition, without the required fiscal procedures governing trust fund actions after 2023, the trust fund committees will be unable to make disbursements and the United States, the FSM, and the RMI will not have assurance of necessary oversight of trust fund disbursements. Finally, without alignment between the timing of the trust fund committees’ annual calculation of the amounts available for disbursement and the countries’ annual budget cycles, the FSM and RMI will have to plan their budgets for each fiscal year without knowing the amount of the disbursements from the compact trust funds. Recommendations for Executive Action We are making the following six recommendations to Interior: The Secretary of the Interior should ensure that the Director of the Office of Insular Affairs, as Chairman of the FSM compact trust fund committee, works with other members of the committee to develop a distribution policy for the FSM compact trust fund, as required by the compact trust fund agreement, that takes into account potential strategies that could address risks to the fund’s ability to provide a source of income after fiscal year 2023. (Recommendation 1) The Secretary of the Interior should ensure that the Director of the Office of Insular Affairs, as Chairman of the FSM compact trust fund committee and of the FSM Joint Economic Management Committee, works with other members of the committees to develop the fiscal procedures required by the compact trust fund agreement. (Recommendation 2) The Secretary of the Interior should ensure that the Director of the Office of Insular Affairs, as Chairman of the FSM compact trust fund committee, works with other members of the committee to address the timing of the calculation of compact trust fund disbursements. (Recommendation 3) The Secretary of the Interior should ensure that the Director of the Office of Insular Affairs, as Chairman of the RMI compact trust fund committee, works with other members of the committee to develop a distribution policy for the RMI compact trust fund, as required by the compact trust fund agreement, that takes into account potential strategies that could address risks to the fund’s ability to provide a source of income after fiscal year 2023. (Recommendation 4) The Secretary of the Interior should ensure that the Director of the Office of Insular Affairs, as Chairman of the RMI compact trust fund committee and of the RMI Joint Economic Management and Financial Accountability Committee, works with other members of the committees to develop the fiscal procedures required by the compact trust fund agreement. (Recommendation 5) The Secretary of the Interior should ensure that the Director of the Office of Insular Affairs, as Chairman of the RMI compact trust fund committee, works with other members of the committee to address the timing of the calculation of compact trust fund disbursements. (Recommendation 6) Agency Comments and Our Evaluation We provided a draft of this report to the Departments of Agriculture, Commerce, Defense, Education, Energy, Health and Human Services, the Interior, Labor, State, the Treasury, and Transportation; the Federal Communications Commission; the Federal Deposit Insurance Corporation; the Federal Emergency Management Agency; the National Science Foundation; USAID; and the U.S. Postal Service, as well as to the FSM and RMI. We also provided copies of the draft to the administrator of each compact trust fund and to Graduate School USA for their technical review. The Department of the Interior, the U.S. Postal Service, and the FSM and RMI provided official comments, which are reproduced in appendixes IX through XII with, where relevant, our responses. The Departments of Agriculture, Education, Health and Human Services, Labor, State, and Transportation; the Federal Deposit Insurance Corporation; USAID; the RMI; the trust funds’ administrator, and Graduate School USA provided technical comments, which we incorporated as appropriate. The following summarizes the official comments from Interior, the U.S. Postal Service, the FSM, and the RMI, and our responses. Interior concurred with our recommendations and stated that discussions to address them are ongoing within the trust fund committees. In addition, Interior stated that a working group comprising staff from Interior’s Office of Insular Affairs and the Department of State’s Office of Australia, New Zealand and Pacific Island Affairs will present recommended actions related to our recommendations to the trust fund committees in 2018. The U.S. Postal Service stated that, in general, the report includes helpful information on the compact obligations regarding postal services provided to the FSM and RMI. However, the U.S. Postal Service also provided additional information on the reimbursement shortfall for its services since 2002 in the freely associated states. The U.S. Postal Service stated that it recommends that, upon expiration of the programs and services agreements, the FSM and RMI be treated as international postal origin and destination points. The FSM concurred with our recommendations to Interior. In addition, the FSM stated that the programs and services provided by U.S. agencies were essential to the FSM and should continue to the greatest extent possible after 2023. The FSM would like to work with U.S. officials to ensure timely approval of continuing these programs and services. The FSM also noted that we had reported the potential for the FSM compact trust fund to not provide disbursements sufficient to cover the estimated value of expiring federal services in 2002, prior to the signing of the amended compact. Further, the FSM provided additional information regarding its Long-Term Fiscal Framework and summarized ongoing public sector and tax reform efforts and its own contributions to the FSM Trust Fund. The RMI concurred with our recommendations to Interior and provided additional comments regarding the recommendations. The RMI asserted that, absent accountability issues, the maximum annual disbursement amount should be disbursed from the compact trust fund. However, as our report notes, the compact trust fund agreements state the maximum allowable disbursement level and do not establish or guarantee a minimum disbursement level. The RMI also stated that it would prefer that future accountability procedures be based on a new agreement rather than a reshaping of the current fiscal procedures agreement. In addition, the RMI raised the issue of compensation under the tax and trade provision of the original compact as well as the effect of delays in investing the RMI compact trust fund on its current value. We discuss the tax and trade provisions in Appendix VII of our report. The RMI also recommended that amendments to the trust fund agreement should not require action by the U.S. Congress. As our report notes, the U.S. legislation implementing the amended compacts requires that any amendment, change, or termination of all or any part of the compact trust fund agreements shall not enter into force until incorporated into an act of Congress. Finally, the RMI noted that programs and services provided through the amended compacts' implementing legislation (Pub. L. No. 108-188) and the compact programs and services agreement were essential and that the RMI could not replace them by using its own resources. We are sending copies of this report to the appropriate congressional committees and to the Departments of Agriculture, Commerce, Defense, Education, Energy, Health and Human Services, the Interior, Labor, State, the Treasury, and Transportation; the Federal Communications Commission; the Federal Deposit Insurance Corporation; the Federal Emergency Management Agency; the National Science Foundation; USAID; and the U.S. Postal Service, as well as the President of the FSM and the President of the RMI. In addition, the report will be available at no charge on GAO’s website at http://www.gao.gov. If you or your staff have questions about this report, please contact me at (202) 512-3149 or gootnickd@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made major contributions to this report are listed in appendix XIII. Appendix I: Objectives, Scope, and Methodology We were asked to review issues related to the Federated States of Micronesia’s (FSM) and Republic of the Marshall Islands (RMI) transition from compact grant assistance to relying on income from the compact trust funds. This report examines (1) the use and role of federal funds and programs in the FSM and RMI budgets, (2) projected compact trust fund disbursements and potential strategies to address risks to those disbursements, and (3) FSM and RMI efforts to prepare for the scheduled decrements in compact grant funding and the transition to relying on compact trust fund income. Federal Funds and Programs To identify the use and role of federal funds and programs, we reviewed relevant documents and interviewed knowledgeable U.S., FSM, and RMI officials during our site visits to the RMI in July 2017 and in the FSM in July and August 2017. We reviewed U.S. law; the amended compacts and associated programs and services agreements and military use and operating rights agreements with each country; each country’s government and component unit single audit reports for fiscal years 2012 through 2016; and U.S. Region IX reports for fiscal years 2015 and 2016. We analyzed expenditure and funding data in FSM and RMI single audit reports, including their Schedule of Expenditures of Federal Awards, to identify the sources of funds expended by the FSM national and state governments, the RMI national government, and their component units and calculated federal funds as a percentage of each entity’s total resources. We reviewed the single audit reports and found that the auditors did not express any qualified or adverse opinions regarding the information they used to prepare the audits’ Schedule of Expenditures of Federal Awards, which lists the amount and use of federal grants. We concluded that these data are sufficiently reliable for estimating the role of federal programs in the FSM and RMI budgets. To identify the FSM and RMI national government component units and FSM state government component units, we reviewed the websites of, and audit reports from, the FSM Office of the National Public Auditor and the RMI Office of the Auditor-General and confirmed the list of component units we identified with FSM and RMI officials. We also discussed the uses of federal funds in the countries with FSM national and state government officials, RMI government officials, and FSM and RMI component unit representatives during our site visits to the countries. Our portrayal of the role of federal funds in the government and component unit budgets does not capture the value of any noncash goods and services that do not appear in the single audit reports. In addition, it does not capture benefits that some programs provide to individuals, such as U.S. Department of Agriculture rural housing loans and Federal Deposit Insurance Corporation insurance that benefits depositors at the Bank of the Federated States of Micronesia. To determine the legal status of U.S. programs, services, and grants after fiscal year 2023, we analyzed the amended compacts, the compact- related agreements, and U.S. law governing the programs, services, and grants that we identified to determine whether, under current law, they would still be available to the FSM and RMI after the end of that fiscal year. For the programs and services agreement with each country, we reviewed the status of programs and services when the agreements end in fiscal year 2024. Our legal analysis included programs, services, and grants that we identified from the compacts, the amended compacts’ implementing legislation, the military use and operating rights agreements, and the programs and services agreements. We also included in our legal analysis the programs (1) that we identified through the single audit reports and Region IX reports and (2) that were not already identified through our review of the compacts, the amended compacts’ implementing legislation, and the compact-related agreements; and (3) that the single audit reports showed as having expenditures above $200,000 in any year in fiscal years 2012 through 2016 or the Region IX reports identified as providing more than $200,000 in federal funding in fiscal years 2015 or 2016. We prepared an initial list of federal programs based on our review. We then provided our list of programs to the FSM, the RMI, and the U.S. Departments of State and the Interior for their review and updated the list on the basis of information they provided. We prepared a preliminary analysis of the post-2023 status of the programs and funding sources we identified and asked officials of the relevant U.S. agencies to review and comment on the accuracy of the list. As part of this analysis, we contacted officials from the Departments of Agriculture, Commerce, Defense, Education, Energy, Health and Human Services, the Interior, Labor, State, and Transportation; the Federal Communications Commission; the Federal Deposit Insurance Corporation; the Federal Emergency Management Agency; the National Science Foundation; the U.S. Agency for International Development; and the U.S. Postal Service. We incorporated into our analysis the comments that these officials provided, and we again asked for their review of our analysis before we completed our draft report. Our conclusions are based on a review of current law. Therefore, any changes in the applicable law subsequent to our report but before 2023 may affect the FSM’s and RMI’s eligibility for U.S. programs and funding. In addition, the availability of programs depends on appropriations made for that purpose. Although we took multiple steps to validate our list of programs with the FSM and RMI and the relevant U.S. agencies, our analysis may not have captured all U.S. grants and programs provided in the FSM and RMI. Compact Trust Funds To examine projected compact trust fund disbursements and actions to address risks, we reviewed previous studies of the compact trust funds; the U.S.-FSM and U.S.-RMI compact trust fund agreements; and other governance and reporting documents such as investment policy statements, presentations to the compact trust fund committees, audits, and annual reports. We also interviewed FSM and RMI officials, compact trust fund committee members, authors of the previous studies, and the funds’ administrator, investment advisers and money managers. To project the compact trust funds’ likely income at their current value and under current trust fund rules (i.e., the baseline scenario), we built a Monte Carlo simulation model and performed 10,000 trial runs of projected returns and disbursements over a four-decade time period, using random values for key variables. We used the following key assumptions in our compact trust fund analysis: Compact trust fund balance. We used the unaudited FSM and RMI fiscal year 2017 year-end compact trust fund balances. C account balance. We estimated the C account balance on the basis of the unaudited FSM and RMI fiscal year 2017 year-end balances. To assess the reliability of the unaudited balances, we reviewed the previous years’ audits and confirmed with the trust funds’ Administrator that previous years’ audits had not resulted in any significant differences between the preliminary balances and the final audited balances. We concluded that the unaudited balances were sufficiently reliable as a basis for our projections of future trust fund performance. The March 2018 audited fund balances, released after we completed our analysis, were within $5 of the unaudited fund balances. Amount of future compact trust fund contributions. We based the amounts of future annual U.S. contributions to both trust funds on the inflation-adjusted amounts estimated in the U.S. Department of the Interior’s (Interior) Office of Insular Affairs’ Budget Justifications and Performance Information, Fiscal Year 2018. For the RMI, we assumed that Taiwan would continue to contribute $2.4 million per year to the RMI’s A account each year through 2023 in keeping with Taiwan’s May 2005 agreement with the RMI. Estimated annual grant assistance for fiscal year 2023. We based our estimates of fiscal year 2023 assistance on the inflation-adjusted amounts estimated by the Office of Insular Affairs. The office estimated that the FSM would receive $82 million in annual grant assistance in fiscal year 2023 and that the RMI would receive $36 million, including Kwajalein-related assistance. In keeping with the RMI compact trust fund agreement, we excluded from our analysis grants provided to the RMI under compact section 211(b) for Kwajalein-related assistance, resulting in an estimated $27 million in grant assistance to the RMI under compact section 211 in fiscal year 2023. The actual amount of annual grant assistance in fiscal year 2023 will depend on actual inflation rates in the years preceding 2023. Different assumptions about the inflation rates will result in different estimates of the amount of fiscal year 2023 annual grant assistance. Net rate of return. In the baseline scenario, we present our results based on a 6 percent rate of return after fees are deducted. To select and assess the reasonability of this projected net rate of return, we reviewed the capital market assumptions and projections used by the money managers for the compact trust funds as well as historical market rates of return. However, because projecting the funds’ long- term performance using the current portfolio and economic assumptions has limitations, we also conducted our analyses using different nominal values for the net returns—5 percent, 7 percent, and 8 percent—in each case using a standard deviation of 13 percent. These results are presented in appendix V. We assumed a normal distribution, but we tested the same baseline analyses with a t- distribution and found that a t-distribution did not substantially affect the results. Inflation rate after fiscal year 2023. We applied the 2 percent long- term inflation rate projected by the Congressional Budget Office. To further analyze actions that could address risks to the compact trust funds, we modeled alternative strategies for managing the funds that were analyzed by previous studies of the compact trust funds. We identified previous studies through a literature search and by interviewing cognizant agency and trust fund officials. On the basis of this review, we developed five potential strategies that are representative of the approaches identified in previous studies. These five strategies are examples of many possible strategies, including varying amounts of disbursement reductions, additional contributions, and methods of calculating annual disbursements. We are not recommending any specific strategy. To provide additional information about potential outcomes, we also analyzed another four strategies that assumed a lower amount of additional trust fund contributions, lower disbursement reductions, or a lower percentage of the compact trust fund balance that could be withdrawn (see app. VII). To help ensure that we had appropriately reproduced the methods used in previous studies, we shared our preliminary results for strategy 4, which modeled the Moving Adjustment Rule, and strategy 5, which modeled the Sustainability Adjustment for Enhanced Reliability (SAFER), with the Graduate School USA representatives who had initially prepared these potential strategies. We analyzed each strategy separately to isolate the impact of individual changes in the strategy on compact trust fund balance and disbursements. However, in practice, these individual changes could occur in combination with each other. We again performed the Monte Carlo analysis, using the same key assumptions as in the baseline scenario, to determine the likely effects, relative to the baseline, of five potential strategies representing three approaches: (1) reducing annual compact trust fund disbursements; (2) making additional contributions; (3) and changing the disbursement policies, including strategies that would require changing the trust fund agreements to permit disbursements from the A account. We present the results of this analysis with a 6 percent net return, a standard deviation of 13 percent, and a normal distribution and tested the results with 5 percent, 7 percent, and 8 percent net returns (see app. VII for further details). To summarize and compare our simulation results for the baseline and alternate scenarios, we analyzed the average disbursements in nominal dollars, the average disbursements in comparison with maximum disbursements, the likelihood of 1 or more years with zero disbursement, and the likelihood that the trust funds will maintain their inflation-adjusted value after fiscal year 2023. We calculated the average disbursement in the given time periods by averaging simulated disbursements over 10-year periods (averaging first over 10 years and then over 10,000 simulated cases). We calculated the average disbursement as a percentage of the maximum allowable disbursement by averaging the ratio of each simulated disbursement to the maximum inflation-adjusted allowable disbursement in the given period (averaging first over 10 years and then over 10,000 simulated cases). We calculated the likelihood of zero disbursement by counting cases with 1 or more years of zero disbursement among the 10,000 simulated cases in each 10-year period. We calculated the likelihood that the fund balance will maintain its inflation-adjusted fiscal year 2023 value by counting simulation cases where the simulated balance exceeds or equals its projected inflation- adjusted 2023 balance in the given year. We report the disbursement results averaged by decade for the first 40 years of compact trust fund disbursements—fiscal years 2024 through 2033, fiscal years 2034 through 2043, fiscal years 2044 through 2053, and fiscal years 2054 through 2063—to summarize the overall trend in disbursements. However, depending on market volatility, disbursements during these decades are likely to fluctuate from year to year. While the projected per-decade averages can show long-term trends in the funds’ disbursements and sustainability and provide a comparison of the likely effects of the potential strategies we analyzed, the projected averages do not provide information about the volatility of changes in annual disbursement. We compare the compact trust funds’ projected value with the projected inflation-adjusted fiscal year 2023 value through 2063 in 10- year increments beginning in fiscal years 2033. To document the status of the FSM Trust Fund and the RMI’s D account and their potential use to supplement FSM and RMI resources after 2023, we reviewed information about the FSM laws establishing the FSM Trust Fund, FSM economic reports, and the RMI-Taiwan agreement regarding the D account. We also interviewed FSM and RMI officials. We did not independently verify the FSM’s projections of the future size of, and disbursements from, its trust fund. The information on foreign law or on foreign government operations in this report is not the product of our original analysis, but is derived from interviews and secondary sources. FSM and RMI Plans To examine FSM and RMI efforts to prepare for the scheduled compact grant decrements, we reviewed each country’s decrement management plans to determine the FSM’s and RMI’s planned budget reductions and other actions. We then reviewed the FSM’s and RMI’s single audit reports and budget documents and interviewed FSM and RMI officials to determine whether the planned reductions had been implemented. We compared the planned actions to current legislation, single audit reports, or recent reports that discussed the status of FSM and RMI economic and financial reforms. In addition, we interviewed Interior, FSM, and RMI officials to determine whether FSM and RMI decrement management plans had been revisited or updated, why the plans were or were not adhered to, and whether the countries planned any future updates to the plans. We also conducted interviews with U.S. officials from the Department of State, a representative of Graduate School USA, and representatives of the World Bank and the International Monetary Fund regarding each country’s previous and current planning efforts. To assess whether the FSM and RMI strategic plans for the key sectors of health, education, and infrastructure addressed the 2023 transition from compact grants and other U.S. assistance to compact trust fund income, we first obtained the relevant plans from department heads in each key sector of the FSM and RMI national governments and FSM state governments and confirmed our identification of the documents with FSM and RMI officials. We reviewed the plans to determine whether they included any discussion of budget projections, economic or financial reforms, alternative funding sources or other revenue generation strategies, and expenditure cuts or saving strategies for periods before and after fiscal year 2023. We also reviewed the FSM’s 2023 Action Plan and the RMI’s updated Medium Term Budget and Investment Framework to determine whether these documents discussed budget changes to address the 2023 transition. Through our interviews with U.S., FSM, and RMI officials, we also learned about other ongoing planning efforts to address the 2023 transition: the U.S. Interagency Working Group on the Freely Associated States, the FSM Joint Compact Review and Planning Committee, and the RMI Compact Review Commission. Following our interviews, we reviewed and summarized documentation related to the working group’s purpose, meetings, and membership. We also contacted FSM and RMI committee members and officials to obtain additional information on the mandate, membership, and status of the FSM and RMI committees. The information contained in this report on foreign law or on foreign government operations is not the product of our original analysis, but is derived from interviews and secondary sources. We conducted this performance audit from March 2017 to May 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: U.S. Compact Sector Grants and Trust Fund Contributions, 2004 through 2023 The amended compacts’ implementing legislation authorized and appropriated direct financial assistance to the FSM and the RMI in fiscal years 2004 through 2023, and provided for partial inflation adjustment of the base amount of compact sector grants and trust fund contributions each year. The base amount is partially inflation-adjusted by the percentage that equals two-thirds of the percentage change in the U.S. gross domestic product implicit price deflator, or 5 percent, whichever is less in any 1 year, using the beginning of 2004 as a base. As the base amount of compact sector grants decreases, the trust fund contributions generally increase by an equivalent amount. Figure 9 shows the amount of compact sector grants and trust fund contributions each fiscal year from 2004 through 2023. The cumulative inflation adjustment largely offsets the amount of the decrement, resulting in a relatively steady nominal amount of compact sector grants after inflation adjustments (see fig. 10). However, because the inflation adjustment is not equal to full inflation, the value of compact sector grants continues to decline in real terms. Appendix III: Compact Grants Supporting Health and Education in the FSM and RMI U.S. grants that end in 2023 play a significant role in the budgets of the FSM states and the RMI in the health and education sectors. The colleges of both countries have also relied on compact sector grants but rely even more on Pell grants to support their operation. FSM states rely on U.S. grants scheduled to end in 2023 for the majority of their health and education expenditures. In fiscal year 2016, compact sector grants and the SEG supported 60 percent or more of health expenditures and 82 percent or more of state education expenditures. Table 3 shows the states’ health and education expenditures of compact sector grants and the supplemental education grant (SEG) in fiscal year 2016. In fiscal years 2012 through 2016, compact sector grants and the SEG supported 56 to 99 percent of FSM states’ health expenditures and 82 to 100 percent of FSM states’ education expenditures. Total expenditures (dollars) Amount (dollars) U.S. compact sector, supplemental education, and other grants also supported approximately 76 percent of the average $21 million in annual expenditures of the College of Micronesia–FSM, an FSM government component unit, in fiscal years 2012 through 2016. Compact sector grants and the SEG, each of which end in 2023, supported approximately 15 percent of the college’s annual expenditures. Pell grants, which provide support for education expenses for qualifying students, supported more than half of the college’s annual expenditures. College officials told us that the college would be unable to operate without Pell grants. According to officials from the U.S. Department of Education, the college will remain eligible after 2023 to receive Pell grants that benefit its students as long as such grants are available to institutions and students in the United States (see app. IV). The RMI relies on U.S. grants scheduled to end in 2023 for health and education expenditures. In fiscal year 2016, compact sector and supplemental education grants scheduled to end in 2023 supported approximately 25 percent of RMI health expenditures and approximately 59 percent of RMI education expenditures (see table 4). Kwajalein-related grants increased these percentages to 32 percent for health and 66 percent for education. In total, in fiscal years 2012 through 2016, compact sector grants and the SEG supported approximately 58 percent of RMI education expenditures and 29 percent of health expenditures. During this period, the percentage of education expenditures supported by compact sector and supplemental education grants scheduled to end in 2023 remained relatively steady and the percentage of health expenditures decreased slightly. U.S. compact sector, supplemental education, and other grants also supported approximately half of the average $11.9 million in annual expenditures of the College of the Marshall Islands, an RMI government component unit, in fiscal years 2012 through 2016. Compact sector grants and the SEG, each of which end in 2023, supported approximately 8 percent of the college’s annual expenditures. Pell grants supported about 39 percent of the college’s expenditures. According to officials from the U.S. Department of Education, the college will remain eligible after 2023 to receive Pell grants that benefit its students, as long as such grants are available to institutions and students in the United States (see app. IV). Also in fiscal years 2012 through 2016, compact sector and supplemental education grants scheduled to end in 2023 supported about half of the expenditures of the RMI government component unit, the Marshall Islands Scholarship, Grant, and Loan Board, which provides financial assistance for educational and training opportunities. Kwajalein- related compact grants that do not end in 2023 supported an additional 13 percent of the board’s expenditures. Appendix IV: Status of U.S. Grants and Programs in the FSM and RMI After 2023 The amended compacts, compact-related agreements, the amended compacts’ implementing legislation, and other U.S. laws provide grants or eligibility for U.S. programs and services for the FSM and RMI. The amended compacts provided compact sector, Kwajalein-related, and audit grants. Under current law, compact sector and audit grants are each scheduled to end in 2023, but the RMI military use and operating rights agreement (MUORA) extended the time frame of Kwajalein-related compact grants for as long as the agreement is in effect. The amended compacts’ implementing legislation provided additional grants, including authorizing a supplemental education grant (SEG), and identified several specific U.S. programs as available to the FSM and RMI. Under current law, the additional grants end in 2023 but the statutory authorizations for some programs identified in Pub. L. No. 108-188 provide for the continued eligibility of the FSM and RMI to receive benefits under the programs. However, after fiscal year 2023, the FSM and RMI will no longer be eligible under current U.S. law for some programs that the SEG replaced. The compact-related programs and services agreements with each country identify additional programs and services that the United States makes available to the FSM and RMI. While these agreements will end in 2024, under current law, some U.S. agencies may continue to provide programs and services similar to those provided in the agreement under other authorities. Based on the status of current law, the FSM’s and RMI’s eligibility for other programs we identified that have been provided under other current U.S. laws will not change after fiscal year 2023. Compact Sector and Audit Grants End in 2023, but Kwajalein-Related Grants for the RMI Will Continue Under current law, compact sector grants provided to the FSM and the RMI under their compact sections 211(a) are scheduled to end in 2023. However, the RMI is scheduled to continue to receive $7.2 million, partially inflation adjusted, related to the U.S. military base in Kwajalein Atoll and provided under section 211(b) of its compact. Under the terms of the RMI MUORA, the United States agreed to provide these Kwajalein- related grants for as long as the MUORA is in effect. The MUORA continues until 2066 and may be extended at the discretion of the United States until 2086. The amended RMI compact provides for $18 million, partially inflation adjusted, in annual payments to the RMI government to compensate for impacts from the U.S. Army Garrison–Kwajalein Atoll. These payments will continue for as long as the MUORA is in effect. Annual compact grants of up to $500,000 (not inflation adjusted) to each country to pay for required annual audits of compact grants are scheduled to end in 2023. See table 5 for a summary of compact sector, Kwajalein- related, and audit grants. FSM and RMI are No Longer Eligible for Many Programs Replaced by the Supplemental Education Grant The supplemental education grant (SEG) authorized by the amended compacts’ implementing legislation is scheduled to end in fiscal year 2023 and, under current law, FSM and RMI eligibility for most programs that the SEG replaced will not resume after fiscal year 2023. Absent changes to current law, the FSM and RMI will not be eligible after fiscal year 2023 for the following programs that the SEG replaced during fiscal years 2005 through 2023: U.S. elementary and secondary education grant programs, adult education and literacy programs, career and technical education programs, job training programs, and Head Start early education programs. However, under other provisions of current law, qualifying individuals in the FSM and RMI will be eligible after fiscal year 2023 for undergraduate education grants and work-study programs that the SEG replaced. See table 6. Some Programs and Services in the Programs and Services Agreement Will End, while Others May Continue under Other Authorities Although the programs and services agreements with the FSM and RMI will end in fiscal year 2024, current U.S. law enables U.S. agencies to continue providing some programs and services now provided under the agreements. No current provisions of U.S. law will enable the Federal Emergency Management Agency (FEMA) to provide disaster response funding or enable the Federal Deposit Insurance Corporation to provide deposit insurance or the U.S. Postal Service to provide services to the FSM and RMI after the agreements end. However, the National Weather Service, the U.S. Department of Transportation’s (DOT) Federal Aviation Administration (FAA), and the U.S. Agency for International Development (USAID) could, under other legal authorities, provide services similar to those they now provide under the programs and services agreements. National Weather Service. The programs and services agreements authorize the National Weather Service to fund the operations of weather stations in the FSM and RMI, which it can continue to fund after the end of the Agreements under other authorities, according to Department of Commerce officials. Federal Aviation Administration. The programs and services agreements authorize DOT’s FAA to provide technical assistance in the FSM and RMI, which it can continue to provide after the end of the Agreements under other provisions of current U.S. law. However, DOT officials stated that FAA would require new bilateral agreements with the FSM and the RMI in order for the countries to continue to receive the civil aviation safety services that FAA currently provides under the programs and services agreements. The FAA would also seek reimbursement for any technical assistance it provides to the FSM and RMI. With regard to the civil aviation economic services of the programs and services agreements, DOT officials stated that, while the FSM and RMI could voluntarily decide to allow U.S. air carriers to continue operations in the FSM and RMI, new bilateral agreements would be needed to assure that result. U.S. Agency for International Development. Following a U.S. presidential disaster declaration, FEMA provides the funding for disaster relief and reconstruction, which is programmed through USAID. Under current law, FEMA funds will no longer be available for this purpose once the agreements end; however, USAID will be able to provide foreign disaster assistance funding to the FSM and RMI under the same terms as it provides this assistance to other countries. After the programs and services agreements end, FEMA will be able to support disaster relief efforts only if USAID or the countries request it to do so on a reimbursable basis. In addition, according to State and Interior officials, telecommunications- related services that the two agencies provide to the FSM and RMI under the programs and services agreements will continue as long as the FSM and RMI provide appropriate authorization for such services. Table 7 shows the status after fiscal year 2024 of programs and services currently provided to the FSM and the RMI under the agreements. Programs Identified in the Amended Compacts’ Implementing Legislation Generally Continue after Fiscal Year 2023 Additional grants provided to the FSM and the RMI under the amended compacts’ implementing legislation will end in fiscal year 2023, but the countries’ eligibility for programs now provided under that legislation will generally continue under current U.S. law. Grants provided under the amended compacts’ implementing legislation for (1) judicial training in the FSM and the RMI, and (2) agricultural and planting programs on the RMI’s nuclear-affected Enewetak Atoll are scheduled to end. However, under current U.S. law, legal authorities permitting the operation of other programs would remain available to the FSM and RMI after fiscal year 2023. Eligibility under these legal authorities continues either because the amended compacts’ implementing legislation does not specify an ending date or because other provisions in current U.S. law make the FSM and RMI eligible for the program. Programs provided in the amended compacts’ implementing legislation include U.S. Department of Agriculture Rural Utilities Service grant and loan programs; U.S. Department of Education Pell grants for higher education and grants under Part B of the Individuals with Disabilities Education Act for children with disabilities; programs for nuclear-affected areas in the RMI; and additional programs provided by the Departments of Commerce and Labor as well as law enforcement assistance provided by the U.S. Postal Service. See table 8 for a summary of the programs identified in the amended compacts’ implementing legislation and their status as of the end of fiscal year 2023. Programs Identified in Other Legislation Generally Continue after Fiscal Year 2023 In addition to being eligible for the programs provided through the compact, its associated agreements, and the amended compacts’ implementing legislation, the FSM and RMI are also eligible for a number of programs under other provisions of current U.S. law. The FSM and RMI have each received funds from the U.S. Department of Agriculture for forestry and rural housing programs, multiple Health and Human Services public health program grants, Interior technical assistance and historic preservation programs, and the DOT FAA airport improvement program, among others. Under current U.S. law, the legal authorities permitting the provision of these programs in the FSM and RMI would not necessarily change after 2023. Table 9 shows the FSM’s and RMI’s eligibility for these additional grants and programs under current law after fiscal year 2023. Appendix V: Compact Trust Fund Baseline Outcomes Calculated with Varying Return Assumptions In order to the test the sensitivity of our compact trust fund projections to assumptions about the future rate of return, we also performed our Monte Carlo analysis using alternate rates of return. We projected the compact trust fund disbursements and balance under current compact trust fund rules on the basis of a 6 percent net return and also estimated the trust fund on the basis of 5 percent, 7 percent, and 8 percent net returns. Higher rates of return would improve the outlook for each compact trust fund. However, even with higher rates of return, our analysis shows a high likelihood that available compact trust fund disbursements will not reach an amount equivalent to maximum disbursements permitted by the compact trust fund agreement (i.e., the inflation-adjusted amount of fiscal year 2023 annual grant assistance, as defined by the trust fund agreements), a continuing risk of zero disbursements, and a decreasing likelihood that the fund will maintain or exceed its inflation-adjusted balance in fiscal year 2023. See tables 10 and 11 for our projections of FSM and RMI compact trust fund disbursements, likelihood of 1 or more years with zero disbursement, and likelihood of maintaining or exceeding its inflation-adjusted fiscal year 2023 value. Appendix VI: FSM and RMI Country Trust Funds The FSM and RMI each maintain their own country trust funds separate from the compact trust funds. These country trust funds are also available to provide a source of revenue after compact grants end at the end of fiscal year 2023. We did not independently project the future balance or potential disbursements from the FSM Trust Fund after 2023. FSM Trust Fund The FSM maintains its own trust fund, separate from the compact trust fund, which can provide additional resources after fiscal year 2023 to offset a reduction in resources relative to those made available as of fiscal year 2023. The FSM Trust Fund, established in 1999, has grown rapidly in recent years. In fiscal years 2012 through 2017, the FSM appropriated a total of $73.3 million for contributions to its trust fund. In addition, in 2015, the FSM changed its tax law to allocate 20 percent of revenue collected by the states to state subaccounts within the FSM Trust Fund. Along with investment gains, these appropriations and contributions of tax revenue have increased the FSM Trust Fund’s balance from $8 million at the end of fiscal year 2011 to $115 million as of the end of fiscal year 2017. As of 2017, the FSM proposed to continue adding $10 million annually from national government surpluses into its trust fund, with the aim of achieving a balance of $250 million by fiscal year 2023 and $10 million in annual disbursements. However, as of early 2018, according to FSM officials, the FSM planned to add $15 million per year to the FSM Trust Fund and projected that the fund would have a balance of $275 million by the end of fiscal year 2023. However, like the compact trust fund, the full balance of the FSM Trust Fund is not available for disbursement. Under current FSM law, funds in the FSM Trust Fund may not be withdrawn until fiscal year 2024. In addition, according to FSM officials, the FSM can withdraw only the fund’s earnings and cannot withdraw the inflation-adjusted value of the FSM Trust Fund corpus. RMI D Account The RMI also maintains its own trust fund—the compact trust fund’s D account. Although managed alongside the compact trust fund, the D account is not subject to the same disbursement provisions as the compact trust fund’s A, B, and C accounts. Instead, disbursements from the D account are subject to the provisions of the agreement between Taiwan and the RMI under which Taiwan contributed the $10 million that the RMI used to establish the D account. According to the terms of this agreement, the RMI may withdraw income after consultation with Taiwan but may not withdraw funds from the D account’s $10 million corpus. At the end of fiscal year 2017, the D account had a balance of $15.1 million, with $5.1 million potentially available for use by the RMI. Appendix VII: Potential Trust Fund Strategies and Model Results We conducted a series of simulations to determine the likely effects of potential strategies for improving the outlook of the FSM and RMI compact trust funds. Prior studies by Graduate School USA, the Asian Development Bank, and the International Monetary Fund examined the effects of three general approaches for improving the trust funds’ outlooks: (1) reducing planned disbursements from the funds, (2) making additional contributions to the funds, and (3) changing the compact trust fund disbursement policies. To isolate the impact of individual changes on the compact trust fund balance and disbursements, we developed and analyzed five potential strategies based on those examined in the previous studies. Reduced disbursements and additional contributions could occur without changes to the trust fund agreement, but changes to the disbursement policies may require changing the agreements. In strategies 3, 4, and, 5, we analyzed strategies that would permit disbursement from the A account. Disbursing from the A account would require changing the compact trust fund agreements. Table 12 shows the 5 potential strategies we analyzed. We analyzed two potential strategies that could be implemented without changes to the trust fund agreements: reductions in the amount of disbursements and additional contributions to the trust funds. Strategy 1: Annual disbursements are reduced below the maximum allowable disbursement. We analyzed the likely effects of reducing disbursements to an amount 30 percent below the maximum disbursement, relative to the baseline scenario, for both the FSM and the RMI compact trust funds. For the FSM, the average size of the disbursements would be lower in the first 10 years of our projection, fiscal years 2024 through 2033, but greater in later years. For the RMI, the average disbursement size would remain lower than the disbursement amounts we projected using the baseline scenario. Disbursement amounts would remain volatile from year to year if the balance in the C account is not sufficient to provide additional disbursements. For both countries, the risk of zero disbursements would be reduced, but not eliminated, in each decade. For both countries, the likelihood that the funds would maintain or exceed their inflation-adjusted fiscal year 2023 value after fiscal year 2023 would be higher in each decade. Reductions in annual disbursements could be effected by the compact trust fund committees at their discretion, without changes to the compact trust fund agreements. However, reductions in annual disbursements below the maximum amount would require each country to permanently adjust to having fewer resources for their budgets and economies than the compact grants provided. Strategy 2: Additional annual contributions are made to the trust fund in fiscal years 2018 through 2023. We analyzed the likely effects of additional contributions equivalent to 5 percent of each country’s fiscal year 2016 GDP, relative to the baseline scenarios, for both the FSM and the RMI compact trust funds. The average size of the disbursements would be greater. Disbursement amounts would remain volatile from year to year if the balance in the C account is not sufficient to provide additional disbursements. The risk of zero disbursements would be reduced but not eliminated. The likelihood that the funds would maintain or exceed their inflation- adjusted fiscal year 2023 value after fiscal year 2023 would be higher. Additional contributions to the FSM or RMI trust funds could be accepted at the discretion of compact trust fund committees, without changes to the compact trust agreements. However, unless the compact trust fund committees could identify other donors for these contributions, the countries would have to choose to reprogram existing revenues from other uses into compact trust fund contributions. The addition of funds from other donors would have no negative impact on the trust funds’ outlook if other conditions remained unchanged. Potential Strategies That Would Permit Disbursement from the A Account We analyzed three additional potential strategies that would involve calculating annual disbursements as a percentage of the FSM and RMI compact trust funds’ balance and which would permit disbursement from the A account. Disbursing from the A account would require changing the compact trust fund agreements, necessitating negotiation and agreement between the United States and each country and statutory enactment by the U.S. Congress. In strategy 3, disbursements are calculated as a fixed percentage of the funds’ moving average balance over the previous 3 years. In strategies 4 and 5, disbursements are calculated on the basis of the funds’ moving average balance over the previous 5 years as well as the committees’ determination of the target size for the funds’ balance or disbursements. All three potential strategies would require the FSM and the RMI to exchange a reduction in resources for more predictable disbursements in the longer term. Strategy 3: The annual disbursement is set as a fixed percentage of the fund’s moving average balance over the previous 3 years, up to the maximum disbursement amount defined by the current trust fund agreement. We analyzed the likely effects of limiting annual disbursements to 5 percent of the moving average balance over the previous 3 years, relative to the baseline scenario for the FSM and the RMI compact trust funds. In earlier years, average disbursements from the compact trust funds would be smaller than those in the baseline scenario; in later years, average disbursements would exceed those in the baseline scenario. For the FSM, the average disbursement would start to exceed that in the baseline scenario in the second decade after disbursements begin (fiscal years 2034-2043). For the RMI, the average disbursement would start to exceed that in the baseline scenario in the fourth decade after disbursements begin (fiscal years 2054-2063). Disbursement amounts would be less volatile from year to year than the volatility that could be experienced in the baseline scenario when the balance in the C account is not sufficient to provide additional disbursements. The risk of zero disbursements would be eliminated. The likelihood that the funds would maintain or exceed their inflation- adjusted fiscal year 2023 value after that year would be higher than in the baseline scenario. Strategy 4: The amount of the annual disbursement is reduced if the compact trust fund’s moving average balance over the previous 5 years is lower than a primary target amount. We analyzed the likely effects of implementing this strategy, relative to the baseline scenario for the FSM and the RMI compact trust funds. In the FSM, the average disbursement would be lower than that in the baseline scenario in earlier years but higher than that in the baseline scenario in the fourth decade after disbursements begin (i.e., fiscal years 2054-2063). In the RMI, the average disbursement would be lower than that in the baseline scenario in earlier years but would equal that in the baseline scenario in the fourth decade after disbursements begin (i.e., fiscal years 2054-2063). Disbursement amounts would be less volatile from year to year than the volatility that could be experienced in the baseline scenario when the balance in the C account is not sufficient to provide additional disbursements. The risk of zero disbursements would be greatly reduced but not eliminated. In the FSM, the risk would be 55 percentage points lower than in the baseline scenario in the fourth decade after disbursements begin (i.e., fiscal years 2054-2063). In the RMI, the risk would be less than 5 percent in each decade. The likelihood that the funds would maintain or exceed their inflation- adjusted fiscal year 2023 value after that year would be higher than in the baseline scenario. Strategy 5: The target disbursement is set as 2.1 percent of the compact trust fund’s balance in fiscal year 2024. The disbursement amount is further decreased if the fund’s moving average balance over the previous 5 years is lower than the primary target balance. Our analysis projected the following effects of implementing this strategy relative to the baseline scenario for the FSM and the RMI compact trust funds: For both countries, the average disbursement would be smaller than that in the baseline scenario in the first 3 decades after disbursements begin (i.e., fiscal years 2024-2053) but would exceed that in the baseline scenario in the fourth decade. Disbursement amounts would be less volatile from year to year than the volatility that could be experienced in the baseline scenario between 2024 and 2063 when the balance in the C account is not sufficient to provide additional disbursements. The risk of zero disbursements would be almost eliminated. The likelihood that the funds would maintain or exceed their inflation- adjusted fiscal year 2023 value would be much higher. Figures 11 through 16 compare projected compact trust fund disbursements and fund balances in the baseline scenario with projected disbursements and fund balances for the five selected potential strategies for improving the trust funds’ outlook. The amounts of disbursement reductions and additional contributions varied among the strategies examined in prior studies. To provide additional information about potential trust fund outcomes, we analyzed another four examples of the selected strategies that assumed a lower amount of additional trust fund contributions, lower disbursement reductions, or a lower percentage of the compact trust fund balance that could be withdrawn. Tables 13 and 14 show the results for all 9 analyses. Appendix VIII: Status of Other Planned Actions in the FSM and RMI Decrement Management Plans In addition to planning budget reductions in the FSM Long-Term Fiscal Framework (its decrement management plan) and the RMI Decrement Management Plan, the FSM and RMI planned other actions such as tax reforms and subsidy reductions to address the scheduled decrement in compact sector grants. The FSM implemented two of three planned actions and the RMI implemented one of four planned actions. The FSM did not implement unified tax reform measures but implemented a change in the formula for sharing compact sector grants with the FSM states and using planned surpluses to mitigate the effects of fiscal reforms. The RMI did not implement planned new taxes, reductions in subsidies to state- owned enterprises, or reductions in payments to Majuro landowners for the use of their land for utilities. The RMI did program a portion of its fishing fee surplus into the annual budget. As of January 2018, the FSM national government had implemented two of three actions that the FSM Long-Term Fiscal Framework indicated the FSM would take in addition to budget reductions. 1. Implementing unified tax reform measures Not implemented. The FSM Long-Term Fiscal Framework states that substantial effort and progress has been made towards comprehensive tax and revenue reform and that the FSM national and state governments anticipated that the Long-Term Fiscal Framework process would provide further impetus towards tax reform. However, according to FSM officials, two FSM states (Pohnpei and Yap) did not approve the Unified Revenue Act. According to FSM officials, the FSM is currently considering other models for tax reform and plans to revisit the issue in the future. 2. Reducing the national government’s share of compact grants and reallocating it to the FSM states Implemented. According to the Long-Term Fiscal Framework, FSM Public Law 18-12 reduced the national government’s share of fiscal year 2014 compact grants from 10 percent to 5 percent, with the amount of the reduction passed along to the FSM states. In May 2014, FSM Public Law 18-57 further reduced the national government’s share of compact grants to 0 percent and increased the amount of compact grants allocated to the state governments, according to the FSM. 3. Using planned surpluses for actions such as possible contributions to activities that mitigate the effects of fiscal reforms, the FSM’s compact trust fund, retiring debt, or reform costs. Implemented. The FSM national government has made additional trust fund contributions but, according to FSM officials, has made a policy decision to make these contributions to the FSM Trust Fund instead of the compact trust fund. As of January 2018, the RMI national government had implemented one of four other actions that its decrement management plan indicated it would take. 1. Implementing a value-added tax and net profits tax in 2017 Not implemented. Officials from the RMI Economic Policy, Planning, and Statistics Office and Ministry of Foreign Affairs confirmed that tax reform has not been implemented due to political challenges. However, a tax task force has been established to revisit tax revenue reforms. 2. Programming 80 percent of unallocated Marshall Islands Marine Resources Authority fishing fee surplus into the annual budget in fiscal year 2015 and using the remaining 20 percent to develop the fishing industry. Implemented. The RMI programmed a portion of its fishing fees into the annual budget in fiscal years 2015 through 2017—$15.8 million in fiscal year 2015, $26.3 million in fiscal year 2016, and $40 million in fiscal year 2017. Although fishing fees were programmed into the budget, according to RMI’s Office of Compact Implementation, the formula allocating 80 percent of fishing fee revenue into the annual budget and the remaining 20 percent to develop the fishing industry is part of proposed RMI legislation but has not become law. 3. Reducing state-owned enterprise subsidies by 10 percent in fiscal years 2016 and 2018. Not implemented. The RMI national government did not reduce the total amount of state-owned enterprise subsidies by 10 percent in fiscal years 2016 as committed in the 2014 decrement management plan. Audit reports for state-owned enterprises in fiscal years 2015 and 2016 indicate that, while the RMI reduced subsidy amounts for some state-owned enterprises, other subsidy amounts increased and overall subsidies were higher in both fiscal years 2015 and 2016 than in fiscal year 2014. See table 15. for the government’s use of their land for utilities by 20 percent in fiscal years 2016, 2018, and 2021. Not implemented. The RMI national government has not reduced government transfers to Majuro landowners to compensate for the government’s use of their land for utilities due to political challenges, according to RMI officials. RMI Ministry of Finance officials stated that, as of January 2018, there had been no reductions in government transfers to Majuro landowners. According to RMI government officials, the total rent payment bill has in fact increased as utilities in Majuro have expanded. Appendix IX: Comments from the Department of the Interior Appendix X: Comments from the U.S. Postal Service Now on page 68. Appendix XI: Comments from the Federated States of Micronesia GAO Comments 1. The FSM refers to our testimony in 2002 regarding the potential for the FSM compact trust fund to not provide funds sufficient to cover the estimated value of expiring federal services as early as 2002. 2. The FSM includes a graphic showing the effect of the partial inflation adjustments and the decrement in the compact sector grants. We include a similar portrayal of this analysis in figure 10 in this report. 3. The FSM states that the amount of the decrement in compact sector grants that is used for annual contributions to the FSM compact trust fund should be recorded as an FSM contribution to the fund. However, Section 215 of the FSM compact refers to the annually decreasing amounts provided to the compact trust fund as set forth in Section 216 of the FSM compact as United States contributions to the compact trust fund. Appendix XII: Comments from the Republic of the Marshall Islands GAO Comments 1. The RMI states that, absent accountability issues, the maximum annual disbursement amount should be disbursed from the compact trust fund. However, as our report notes, although the compact trust fund agreements state the maximum allowable disbursement level, they do not establish or guarantee a minimum disbursement level. 2. The RMI states that the 2-year delay in investing the compact trust fund will result in a compounded total loss of $33.6 million by the end of fiscal year 2023. Our 2007 analysis of the trust funds included information about the delays in establishing the trust funds. We did not update our 2007 analysis of the loss in income due to the delay in investing the compact trust fund for this report. 3. The RMI notes that the amended compacts' implementing legislation extended several important federal programs. Appendix IV of this report presents our conclusions, based on our analysis of current law, that the RMI will remain eligible as of the end of fiscal year 2023 for special education programs and for some programs replaced by the supplemental education grant. Appendix XIII: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Emil Friberg (Assistant Director), Ming Chen, Neil Doherty, Mark Dowling, Reid Lowe, Moon Parks, Shaundra Patterson, and Michael Simon made key contributions to this report. Justin Fisher, Jeff Isaacs, Julie Hirshen, Risto Laboski, Courtney LaFountain, and Jeffery Malcolm provided technical assistance.

Why GAO Did This Study In 2003, the United States approved amended compacts of free association with the FSM and RMI, providing a total of $3.6 billion in economic assistance in fiscal years 2004 through 2023 and access to several U.S. programs and services. Compact grant funding, overseen by the Department of the Interior, generally decreases annually. However, the amount of the annual decrease in grants is added to the annual U.S. contributions to the compact trust funds, managed by joint U.S.-FSM and U.S.-RMI trust fund committees. Trust fund earnings are intended to provide a source of income after compact grants end in 2023, but GAO and others have previously found that the trust funds may not provide sustainable income. GAO was asked to examine preparations for the transition in 2023. This report examines (1) the use and role of U.S. funds and programs in FSM and RMI budgets, (2) projected trust fund disbursements and potential strategies to address risks to those disbursements, and (3) FSM and RMI plans to prepare for grant decreases and the transition to trust fund income. GAO reviewed compact agreements, audit reports, and U.S. law; modeled trust fund performance under existing conditions and using potential strategies; and reviewed FSM and RMI plans. GAO visited each country and interviewed FSM, RMI, and U.S. officials. What GAO Found The Federated States of Micronesia (FSM) and the Republic of the Marshall Islands (RMI) continue to rely on U.S. grants and programs, including several that are scheduled to end in 2023. U.S. compact sector and supplemental education grants, both scheduled to end in 2023, support a third of the FSM's and a quarter of the RMI's expenditures. Agreements providing U.S. aviation, disaster relief, postal, weather, and other programs and services are scheduled to end in 2024, but some agencies may provide programs and services similar to those in the agreements under other authorities. FSM and RMI eligibility for some other U.S. grants and programs is expected to continue after 2023. Disbursements from the compact trust funds face risks that the trust fund committees have not addressed. GAO found that the trust funds are increasingly likely to provide no annual disbursements in some years and to not sustain their value. Potential strategies such as reduced trust fund disbursements or additional contributions from the countries or other sources could help address these risks. Changing the trust fund disbursement policies could also address these risks but may require revising the trust fund agreements with each country. However, the trust fund committees have not prepared distribution policies, required by the agreements, which could assist the countries in planning for the 2023 transition to trust fund income. The committees also have not prepared the required fiscal procedures for oversight of the disbursements or addressed differences between the timing of their annual determination of the disbursement amounts and the FSM's and RMI's annual budget cycles. The FSM and RMI did not implement planned budget reductions to address decreasing compact grants owing to increased revenues from other sources that offset the grant decreases. Current FSM and RMI infrastructure plans address the 2023 transition, while health and education plans focus on strategic goals. Both countries have established new compact planning committees to identify future challenges and develop plans for the 2023 transition to trust fund income. What GAO Recommends GAO recommends that Interior work with the compact trust fund committees to develop distribution policies and fiscal procedures for the funds and to address disbursement timing. Interior concurred with the recommendations.

gao_GAO-18-119

gao_GAO-18-119_0

Background Border Patrol Organization Border Patrol has divided geographic responsibility for the southwest border among nine sectors, as shown in figure 1. Each sector has a varying number of stations, which serve as bases of operation for agents, and agents are responsible for patrolling within defined geographic areas—known as areas of responsibility. Border Patrol uses a variety of land-based surveillance technologies under the Southwest Border Technology Plan to assist its efforts to secure the border by interdicting illicit cross-border activity and apprehending individuals attempting to cross the border illegally. Border Patrol is responsible for planning, acquiring, and deploying that technology along the southwest border. Border Patrol’s PMOD executes the acquisition and procurement of Border Patrol systems, supplies, and services, including current and planned technology deployments along the southwest border, which was previously conducted by CBP’s Office of Technology Innovation and Acquisition. CBP has an Office of Acquisitions that performs oversight. Southwest Border Technology Plan As noted above, the 2014 Southwest Border Technology Plan incorporated the 2011 Arizona Technology Plan and included plans to extend land-based surveillance technology deployments beyond Arizona to the remainder of the southwest border, beginning with selected areas in Texas and California. Border Patrol developed the Southwest Border Technology Plan using a two-step process. First, the Homeland Security Studies and Analysis Institute conducted an analysis of alternatives, which analyzed five technology options in 13 representative areas along the southwest border, identified the types of environmental conditions under which a given technology option might be more effective or less effective, and provided a general overview of the cost and effectiveness tradeoffs between the technologies. For example, the analysis of alternatives noted that IFTs are potentially effective if vegetation is sufficiently sparse and terrain is flat or rolling, such as in the Ajo station area of responsibility (see figure 2). However, according to Border Patrol officials, the IFT’s radar capabilities may not be suited for urban environments, where illegal crossers and narcotics traffickers can blend in with the legitimate traffic. In some of these locations, such as the Nogales port of entry, Border Patrol has determined that the RVSS is more effective. Second, Border Patrol developed a technology deployment plan that identified the types and quantities of each technology needed for each sector. To develop this plan, Border Patrol officials reviewed the results of the analysis of alternatives and considered each sector’s operational conditions, including patterns of traffic, terrain, infrastructure, weather, available resources, and challenges. For example, Border Patrol selected MSC units for Arizona’s Tucson and El Centro sectors, but not for Texas’s Rio Grande Valley sector because the radar was less effective in the dense vegetation of south Texas, an example of which is shown in figure 3. Figure 4 shows the border surveillance technology systems included in the Southwest Border Technology Plan. DHS’s Acquisition Life Cycle Border Patrol follows DHS’s acquisition policy to acquire planned technologies under the Southwest Border Technology Plan. DHS’s overall policy for acquisition management is outlined in Acquisition Management Directive 102-01 and its associated Instructional Manual 102-01-001. DHS’s Under Secretary for Management (USM) is currently designated as the department’s Chief Acquisition Officer and, as such, is responsible for managing the implementation of the department’s acquisition policies and acting as the acquisition decision authority for the department’s largest acquisition programs. Within DHS, the USM is supported by the Office of Program Accountability and Risk Management (PARM), which is responsible for overseeing the acquisition process and assessing the status of acquisition programs through four phases of the acquisition life cycle. These phases include a series of five Acquisition Decision Events (ADE) that provide the acquisition decision authority an opportunity to assess whether the program is ready to proceed through the acquisition life cycle phases. Figure 5 depicts the four phases of the acquisition life cycle and the associated ADEs. In addition, components and program offices have established program- level groups, such as Executive Steering Committees, to provide, among other things, assistance and support during the acquisition process. Border Patrol’s Requirements Management Process According to Border Patrol officials, in 2014, Border Patrol began implementing a new process to identify future technology needs. The Requirements Management Process (RMP), according to Border Patrol officials, is a new process designed to facilitate planning in order to fund and deploy operational capabilities, such as surveillance technology and tactical infrastructure, for border security operations. According to Border Patrol officials, Border Patrol will use information resulting from the RMP to fulfill DHS acquisition policy requirements, including information required for Acquisition Decision Events, as appropriate. Border Patrol is working to develop guidance to align the RMP with the DHS acquisition life cycle. The RMP consists of six steps as shown in figure 6. In the first step of the process, Border Patrol reviews strategic guidance to identify mission priorities and goals and assesses the state of the threat to be addressed. The second step, mission analysis, begins with the Capability Gap Analysis Process, which is intended to identify each station’s capability gaps by determining the difference between a station’s existing capabilities and the capabilities required to perform its mission- essential tasks. The identified shortfall in required capability is a capability gap. Under the RMP’s third step—planning—Border Patrol officials examine capability gaps in detail and determine courses of action—that is, solutions, which may include surveillance technologies, to close the capability gaps. For example, potential solutions could include adjusting the technologies or personnel deployed in a specific area or improving maintenance and repair of access roads. The solutions are documented in sector-specific Initial Requirements Documents. The fourth step— execution—involves Border Patrol leadership executing courses of action. Border Patrol officials stated that courses of action are options for Border Patrol commanders and executives to select and implement. Certain courses of action, including acquiring and deploying land-based surveillance technology, may need to proceed through the DHS acquisition life cycle as appropriate. Once implemented, these options are expected to resolve identified capability gaps in operations, according to Border Patrol officials. The fifth and sixth steps of the process— assessment and life-cycle management—involve implementing and monitoring solutions to determine their ability to resolve capability gaps, and gathering sector feedback on how the solutions affect border security operations. CBP Has Made Progress Deploying Technology along the Southwest Border, and Is Still Developing Guidance for Future Planning Processes Border Patrol Has Completed Deployment of Select Surveillance Technology to Arizona, Texas, California, and New Mexico, and Has Further Deployment Activities Underway As of October 2017, Border Patrol had initiated or completed the planned deployment of select technologies to sectors across areas in Arizona, Texas, California, and New Mexico. In 2014, we reported that Border Patrol had made progress deploying technologies and had completed deployments for two technology programs in Arizona—the Agent Portable Surveillance System (APSS) and the Thermal Imaging Device (TID) technologies. Since our 2014 report, Border Patrol has completed deployments of several additional technology programs. Specifically, according to Border Patrol officials, it has completed deployments of all planned RVSS, MSC, and Unattended Ground Sensors (UGS), as well as 15 of 53 IFT systems to Arizona. Border Patrol has also completed deployments of select technologies to Texas and California, including deploying 32 MSC systems to Texas and California. Border Patrol also has efforts underway for completing deployments of other technology programs, but some of those programs have not yet begun deployment or are not yet under contract. For example, as of October 2017, Border Patrol had not yet initiated deployments of RVSS to Texas because, according to PMOD officials, the program had only recently completed contract negotiations for procuring those systems. According to PMOD officials responsible for the RVSS program, Border Patrol has begun planning the designs of the command and control centers and towers, as well as real estate needs for the Rio Grande Valley sector. Additionally, Border Patrol initially awarded the contract to procure and deploy MVSS units to Texas in 2014 but, because of bid and size protests, did not award the contract until 2015, and the vendor that was awarded the contract did not begin work until March 2016. The deployment status of surveillance technologies is shown in table 1. Border Patrol’s Technology Programs Have Experienced Delays, but Are on Track against Revised Schedules and Cost Baselines Border Patrol has revised schedules and cost estimates for its three highest-cost programs—IFT, RVSS, and MSC—and as of October 2017, is on track to meet those revised schedules and estimates; however, risks remain in Border Patrol’s deployment efforts. Border Patrol has rebaselined (i.e., revised original schedule and cost goals) its three highest-cost programs—IFT, RVSS, and MSC—due to schedule, quantity, and cost estimating variances, among other changes to the programs’ original plans. According to our cost and schedule assessment guides, while rebaselining can be beneficial for quickly identifying new variances, reporting a program’s performance based on a rebaselined cost or schedule may not reflect the program’s overall cost and schedule performance or timeline. Deployment Schedules In March 2014, we reported that CBP had a deployment schedule for each of the seven technology programs planned for deployment at the time––IFT, RVSS, MSC, APSS, MVSS, TID, and UGS––and that four of the programs would not meet their originally planned completion dates. Specifically, we found that the three highest-cost programs (IFT, RVSS, and MSC) had experienced delays relative to their baseline schedules as of March 2013, which were current at the time of our review. We recommended that CBP ensure that scheduling best practices are applied to the IFT, RVSS, and MSC program schedules. DHS concurred with the recommendation and stated that CBP planned to apply scheduling best practices when revising the three programs’ schedules. Based on our assessment of the IFT, RVSS, and MSC programs’ revised schedules that CBP had completed as of January 2017, CBP did not apply all scheduling best practices. However, the revised programs’ schedules for the IFT, RVSS, and MSC reflect substantial improvements in quality and are consistent with the intent of our recommendation. In particular, CBP has improved the quality of its products for analyzing and quantifying risk to the programs’ schedules. Continuing to apply scheduling best practices in future updates will help better position CBP to identify and address any potential delays in its programs’ commitment dates. DHS approved Border Patrol’s rebaseline of the IFT program in December 2015, which extended the program’s completion date to 2020—five years beyond what Border Patrol had estimated in its original baseline schedule. The RVSS and the MSC programs’ completion dates were also extended because the scopes of the programs had increased, among other reasons. While Border Patrol’s revisions to its schedules are positive steps in helping the agency oversee its management of these programs, the programs continue to be behind schedule relative to their original planned baseline documents dated March and September 2012 for the IFT and RVSS programs, respectively, as shown in figure 7. In addition to revising program schedules, Border Patrol has revised the life-cycle cost estimates for the three highest-cost programs to reflect actual costs and include cost estimates for additional and ongoing work. For example, the MSC cost estimate increased by $294.7 million—from $107.2 million to $401.9 million—due to, among other reasons, the program’s expanded scope to Texas, California, and New Mexico. In December 2015, estimated life-cycle costs for the IFT program decreased from its original March 2012 baseline estimate by $211.5 million, in part because of lower-than-expected contract costs. However, from March 2012 to December 2015, IFT’s acquisition cost threshold increased by more than $50 million—from $288 million to $341 million—when CBP included the costs of contractor personnel supporting the program office, the cost of replacing SBInet systems, and actual costs through fiscal year 2014, rather than estimates. According to Border Patrol officials, a CBP policy change required them to include the contractor personnel support costs in the rebaseline, which was previously not required in the original cost baseline. Figure 8 shows original and revised cost estimates for the IFT, RVSS, and MSC programs. In March 2014, we reported that the three highest-cost programs (IFT, RVSS, and MSC) accounted for 97 percent of the Arizona Technology Plan’s estimated cost and that the life-cycle cost estimates for the two highest-cost programs—IFT and RVSS—reflected some, but not all, best practices for cost estimating. Reliable life-cycle cost estimates reflect four characteristics—they are (1) well-documented, (2) comprehensive, (3) accurate, and (4) credible. Our analysis of CBP’s estimates for the two highest-cost programs at the time of our March 2014 review showed that these estimates at least partially met three of these characteristics: well- documented, comprehensive, and accurate. In terms of being credible, these estimates had not been verified with independent cost estimates in accordance with best practices. We concluded that verifying life-cycle cost estimates with independent estimates in accordance with cost- estimating best practices could help better ensure the reliability of the cost estimates. We recommended that CBP verify the life-cycle cost estimates for the IFT and RVSS programs with independent cost estimates and reconcile any differences. DHS concurred with this recommendation, but stated then that it did not believe there would be a benefit from expending funds to obtain independent cost estimates and that if the costs realized to date continued to hold, there may be no requirement or value added in conducting full program updates with independent cost estimates. As part of our updates on CBP’s efforts to implement our 2014 recommendations, CBP officials told us that in fiscal year 2016, DHS’s Cost Analysis Division (CAD) would begin piloting its own independent cost estimate capability with the RVSS program. According to CBP officials, this pilot was an opportunity to assist DHS in developing its independent cost estimate capability. CBP selected the RVSS program for the pilot because the program was at a point in its planning and execution process where it could benefit most from having an independent cost estimate performed, as these technologies were being deployed along the southwest border beyond Arizona. According to CBP officials, CAD completed its independent cost estimate for the RVSS program in August 2016. CBP officials also told us that the RVSS life- cycle cost estimate was finalized and reconciled in March 2017. CBP reported that the component acquisition executive approved the reconciliation estimate in September 2017. According to CBP officials, CBP does not have plans to conduct an independent cost estimate and verification for the IFT. We continue to believe that independently verifying the life-cycle cost estimate for the IFT program and reconciling any differences, consistent with best practices, could help CBP better ensure the reliability of the estimate. Risks Affecting Schedule and Cost Goals While selected technology programs are on track to meet schedule and cost goals, according to Border Patrol officials, some programs have identified risks that may lead to schedule slips or cost growth in the future. Specifically, Border Patrol has experienced delays in completing deployments for planned technologies due to (1) land use and access- related issues; (2) technical issues; and (3) contracting challenges, among other factors. For instance, the IFT program continues to experience delays deploying IFTs to tribal lands in the Tucson sector in Arizona. Border Patrol officials stated that the IFT program has not received authorization from tribal land leaders to build an access road and deploy IFT tower systems on the tribe’s land. They also stated that the historic preservation officer for the tribal lands would need to issue a finding that the IFT would not have any negative impact on cultural resources before Border Patrol could proceed with deployment. In addition, RVSS program officials we met with noted that access to privately owned land is an issue of concern in Texas that could potentially delay RVSS deployment for the Rio Grande Valley sector. Border Patrol has also encountered delays in the IFT program as a result of technical issues identified during delivery of the IFT. For example, we previously reported that testing completed in November 2015 on IFT systems in Nogales had been delayed by 2 months in order for the contractor to address issues related to IFT cameras and operator interfaces. Additionally, Border Patrol has encountered schedule delays due to contracting challenges, such as renegotiations with the contractor after the contract was awarded. For example, according to Program officials, the MVSS contractor proposed a technical change to the system to address safety and maintenance concerns. Border Patrol agreed to the change, which led to delays. We have previously reported that program delays can result in increased costs and force agents to rely on legacy surveillance technologies. According to Border Patrol officials, program managers and Border Patrol are working to mitigate the risk of delays through quarterly executive steering committee meetings of program managers and representatives from other component and headquarters offices, such as DHS’s PARM. During these meetings, program managers discuss cost and schedule risks and evaluate options for mitigating those risks. For example, according to PARM officials, at one such meeting, officials reviewed the RVSS program and determined that it met cost criteria to receive additional DHS oversight. According to Border Patrol officials, Border Patrol has also used quarterly executive steering committee meetings to involve stakeholders and address potential risks as it moves forward with full production in the IFT program. As Border Patrol proceeds with these programs, it will be important to continue to find ways to mitigate the risk of delays in order to meet its revised schedules. Border Patrol Is Developing Guidance to Help Sectors Identify and Plan for Future Technology Needs and Deployments Border Patrol’s RMP and other initiatives are intended to help inform future technology deployment decisions, but, as we reported in February 2017, additional actions are needed to ensure station officials understand the process and their respective roles and responsibilities. Border Patrol officials reported that the Southwest Border Technology Plan is the baseline for identifying technology needs and planning technology deployments, and that changes to the plan are needed as threats and priorities evolve. To help address these changes and remain adaptive, in 2014, Border Patrol began implementing the RMP that, among other things, is intended to identify capability gaps in border security operations and identify solutions to those capability gaps. In February 2017 we found that Border Patrol had documented the RMP, but had not developed written guidance on how officials were to use the information and analyses resulting from the process when requesting tactical infrastructure—that is, fencing, gates, roads, bridges, lighting, and drainage infrastructure—for deployment purposes. For example, we reported that sectors varied in their understanding of how to use results from the Capability Gap Analysis Process when engaging in planning processes or when making resource allocation decisions. We recommended that Border Patrol develop and implement written guidance for the steps within its requirements process for identifying, funding, and deploying tactical infrastructure for border security operations, including clarifying the roles and responsibilities of the parties involved in the RMP. In response to our recommendation, Border Patrol officials reported that they are currently updating the RMP documentation, training, and guidance to the field. Border Patrol officials expect to have an updated Internal Operating Procedure and Manual for the RMP by the second quarter of fiscal year 2018. According to CBP officials, actions taken in response to our recommendation would apply to surveillance technology as well—not solely tactical infrastructure. By developing this written guidance, Border Patrol intends to reduce the risk of relevant agency officials not having the information needed to perform their appropriate role in the process. We will continue to monitor the progress of Border Patrol efforts related to the RMP to determine whether these actions meet the intent of our recommendation to fully develop and implement written guidance for the steps within the RMP. Until then, Border Patrol is less likely to have reasonable assurance that it has the best available information to inform future investments in surveillance technologies and resource allocation decisions among surveillance technologies. In addition to the RMP, future surveillance technology deployments will be affected by other ongoing DHS and CBP initiatives. Specifically, Border Patrol officials in the Strategic Planning and Analysis Division reported that the Domain Awareness: Land Surveillance initiative requirements documents and Southwest Border Capability Roadmap will also be taken into consideration throughout the RMP and will influence future surveillance technology deployments. CBP’s Domain Awareness: Land Surveillance initiative is intended to depict current CBP land domain awareness capabilities and inform future capabilities, which could help Border Patrol identify solutions during various phases of the RMP. Border Patrol officials stated that Border Patrol, with CBP’s Air and Marine Operations and U.S. Immigration and Customs Enforcement, is developing mission needs statements, concepts of operations, and capability operational requirements documents for (1) mobile, (2) fixed and relocatable, and (3) agent-portable capabilities which will help inform future technology deployments. Border Patrol also proposed a Southwest Border Capabilities Roadmap in April 2017 to assist with identifying solutions, such as surveillance technology, mobility and access, and personnel. This roadmap is intended to inform a balanced, risk-based investment strategy driven by capability gaps, geographic priorities, terrain, and other environmental factors, and to consider the evolving cross-border threat. The roadmap identifies specific requirements for persistent surveillance assets, such as RVSS, and was used to support CBP’s fiscal year 2018 budget justification for RVSS deployments in the Rio Grande Valley sector. To create this roadmap, Border Patrol officials reported reaching out to stations within 40 miles of the border to discuss their current gaps and how they would close them (either through physical barriers, manpower, or technology). Because Border Patrol is still in the planning phases for future technology deployments, it is too soon to tell how these efforts will assist Border Patrol in structuring and planning those deployments. CBP Is Taking Action to Better Link Performance Data to Planning Efforts, but Needs to Improve Data Quality Border Patrol Has Made Progress Identifying Performance Metrics for Technologies Border Patrol has made progress identifying performance metrics for the technologies under the Southwest Border Technology Plan, but additional actions are needed to fully implement our prior recommendations in this area. In November 2011, we found that CBP did not have the information needed to fully support and implement the ATP and recommended that CBP (1) determine the mission benefits to be derived from implementation of the ATP and (2) develop and apply key attributes for metrics to assess program implementation. We reported in 2014 that, in response to our recommendations, CBP had identified mission benefits expected from the implementation of the surveillance technologies under the ATP, but had not fully developed key attributes for performance metrics for the technologies. We recommended, among other things, that CBP analyze available data on apprehensions and seizures and technological assists, in combination with other relevant performance metrics or indicators, to determine the contribution of surveillance technologies to CBP’s border security efforts. CBP officials stated that they planned to develop objectives for each performance measure, at which time the agency would begin using the data to evaluate the contributions of specific technology assets. CBP also intended to establish a tool by the end of fiscal year 2016 that explained the qualitative and quantitative impacts of technology and tactical infrastructure on situational awareness in specific areas of the border environment. In September 2016, Border Patrol provided us a case study that assessed technology assist data, along with other measures, to determine the contributions of surveillance technologies to its mission. In April 2017, we reported that this was a helpful step in developing and applying performance metrics; however, the case study was limited to one border location and the analysis was limited to select technologies. In May 2017, Border Patrol officials demonstrated the agency’s new Tracking, Sign Cutting, and Modeling (TSM) system, which they said is intended to connect between agents’ actions (such as identification of a subject with a camera) and results (such as an apprehension) and allow for more comprehensive analysis of the contributions of surveillance technologies to Border Patrol’s mission. One official said that data from the TSM will have the potential to provide decision makers with performance indicators, such as changes in apprehensions or traffic before and after technology deployments. However, the TSM is still early in its use and officials confirmed that it is not yet used to support such analytic efforts. The official stated that over time it would be used to analyze performance on a systematic basis and provide information to decision makers. We continue to believe that it is important for Border Patrol to assess technologies’ contributions to border security and will continue to monitor the progress of the TSM and other Border Patrol efforts to determine whether these actions sufficiently meet the intent of our November 2011 recommendation to fully develop and apply performance metrics for its border technologies. Until then, Border Patrol is not well positioned to fully assess its progress in implementing the Southwest Border Technology Plan and determine when mission benefits have been fully realized. Border Patrol Measures Technologies’ Assistance in Apprehensions and Seizures, but Data Are of Limited Quality and Usability Border Patrol agents collect and report data on asset assists, which are instances in which technologies or other assets (such as canine teams, bicycle patrols, or air support from CBP’s Air and Marine Operations) contributed to an apprehension or seizure; however, the agency does not have sufficient controls to ensure the accuracy and reliability of that data. In March 2014, we reported that CBP was not capturing complete asset assist data on the contributions of its surveillance technologies to apprehensions and seizures and that Border Patrol agents were not consistently recording these data across locations. We recommended that CBP require data on asset assists to be recorded and tracked within the DHS Enforcement Integrated Database (EID), which contains data on apprehensions and seizures. Since then, Border Patrol has taken actions to better record asset assists and to expand the types of technologies that can be tracked, consistent with our prior recommendation. Specifically, in June 2014, Border Patrol issued guidance informing agents that the asset assist data field within the e3 Portal to the EID had become a mandatory data field. Additionally, when recording asset assists, agents initially could only choose from “camera,” “mobile surveillance system,” “scope truck,” “unattended ground sensor,” or “other” when selecting technologies. In May 2016, Border Patrol expanded the types of assets available for agents to choose from to include MSC, IFT, and APSS, among others. Border Patrol requirements for entering asset assist data into the e3 Portal and expansion of the types of assets listed have been positive steps to help better position Border Patrol to assess the contributions of surveillance technologies to border security efforts, as we recommended in 2014. However, we have identified issues with the completeness and reliability of the asset assist data. In particular, we analyzed data on asset assists from October 1, 2014 through May 3, 2017 and our analysis showed that agents incorrectly attributed some apprehensions and seizures to certain technologies rather than others. For example, stations in the Rio Grande Valley sector recorded assists from IFTs in nearly 500 instances from June through September 2016, which cannot be accurate, since the sector does not have IFTs. When we brought this issue to the attention of Border Patrol headquarters officials, they told us in December 2016 that they would discuss the matter with Rio Grande Valley sector officials. However, data from December 2016 through May 3, 2017 indicated that agents in the Rio Grande Valley sector continued to record asset assists from IFTs. Additionally, we found that one station in the Tucson sector with SBInet towers was recording asset assists from the SBInet towers as “other,” when Border Patrol headquarters officials told us that SBInet towers should be recorded as “IFT.” Moreover, our analysis showed that “other” (including “other” listed alongside additional assets) made up nearly 16 and 23 percent, respectively, of asset assists recorded in the Tucson and Rio Grande Valley sectors from October 1, 2016 through May 3, 2017. Border Patrol officials told us that “other” should be any technology not otherwise listed, and could include technologies or support that officials were interested in tracking locally. Officials said the large number of “other” assets could also be a result of agents not understanding their responsibilities or agents working to complete the asset assist data entry as quickly as possible so they could move on to other duties. According to Border Patrol officials, data integrity and quality checks are the responsibility of the individual sectors, and each station has a designated point of contact for data integrity and a system administrator to oversee data quality. However, Border Patrol has not provided written guidance to the sectors on how to oversee data integrity or conduct quality checks of asset assist data, and Border Patrol’s guidance on how to enter asset assist data is limited. According to Border Patrol officials, Border Patrol’s asset assist guidance for sectors consists of two training presentations. We reviewed the training slides for these presentations and found they included photographs and general descriptions of some technologies, along with two case examples for recording an asset assist. However, the slides did not discuss how sectors should conduct data integrity or quality checks. Furthermore, the slides did not address how agents should record assists for SBInet towers. The slides also did not explain why asset assist data are collected (other than that the Chief of the Border Patrol requires it), what it could be used for, or why it was important to ensure data were accurately recorded. Officials told us in June 2017 that the asset assist data were only used to respond to data requests from external agencies—the data were not being used for planning, budgeting, performance measurement, or other purposes. Standards for Internal Control in the Federal Government states that management should obtain relevant data from reliable internal and external sources in a timely manner based on the identified information requirements. Reliable internal and external sources provide data that are reasonably free from error and bias and faithfully represent what they purport to represent. Management should evaluate both internal and external sources of data for reliability. Additionally, management should periodically review policies, procedures, and related control activities for continued relevance and effectiveness in achieving the entity’s objectives or addressing related risks. If there is a significant change in the entity’s process (such as the addition of new asset assist fields), management should review this process in a timely manner to determine that the control activities are designed and implemented accordingly. Without sufficient guidance for sectors on how to enter and review asset assist information, Border Patrol does not have reliable data on asset assists that could help monitor the contribution of surveillance technologies to Border Patrol apprehensions and seizures and inform resource allocation decisions. Border Patrol Reviews Agent Feedback Data on Technology Performance to Improve Current and Future Deployments Border Patrol has a variety of mechanisms for collecting agent feedback on technology performance and for using that information to improve current and future deployments. For example, officials from the Border Patrol Program Management Office Directorate (PMOD) reported that they conduct required technology performance evaluations at specified intervals (e.g., a 6-month post-deployment review and an annual operational analysis) to regularly collect and evaluate agent feedback, and conduct monthly reviews of maintenance and repair requests. Officials said that this feedback is consolidated, prioritized based on cost effectiveness, and used to identify system upgrades (both for systems that have been deployed and for future deployments). The PMOD also collects feedback as part of the annual process for developing an operational analysis report. In the 2016 operational analysis for the MSC program, the PMOD assessed Border Patrol agents’ overall satisfaction with the MSC system, whether it enabled agents to perform their functions more easily and efficiently, and whether it met agents’ needs. Agents identified several MSC benefits, including performance improvements from the prior system (known as the Mobile Surveillance System), radars that exceeded the performance of other mobile systems’ radars, and targets being detected at farther ranges than the system specification. However, the analysis identified more opportunities for improvement, including the need for improvements to the MSC’s camera, video analytics, tracking, graphical user interface, engineering, and other changes. Border Patrol also reported using post-implementation reviews to collect agent feedback and identify improvements. For technologies under the ATP and Southwest Border Technology Plan, Border Patrol completed post-implementation reviews for the MSC in July and October of 2014, for the IFT in June 2016, and for the RVSS in October 2016. In August 2017, Border Patrol reported expecting to conduct reviews for the remaining technologies within 6 to 18 months of each technology reaching initial operating capability. A post-implementation review’s primary purpose is to determine the impact of the system on stakeholders, quantitative and qualitative performance of the system, and the ability of the system to meet identified goals. For example, the MSC review from October 2014 reported that the system was generally an improvement over the older Mobile Surveillance System; however, program risks included damage to trucks and sensors from Border Patrol agent operator error and the need to improve or widen access roads given the larger footprint of the MSC trucks (compared to the prior Mobile Surveillance System trucks). The review concluded with six recommendations to improve future assessments of the system and to plan for new sensor deployments. The recommendations to the PMOD and Border Patrol acquisition office included updating the life-cycle cost estimate to track manpower costs; using a skills-based qualification standard for MSC operators; and ensuring future Border Patrol surveillance systems include the ability to extract actual performance, operational, and environmental data. In August 2017, Border Patrol reported a range of actions underway to address these recommendations, including (1) developing updates to the MSC’s support system to better capture all service requests and maintenance work orders, (2) providing standardized training to each MSC operator with refresher training available upon request, and (3) updating key acquisition documents to reflect the increase in the number of MSCs to a full operating capability of 90 units. In addition to required reports, PMOD officials reported gathering agent feedback directly. For example, a PMOD official with responsibility for the IFT program reported conducting feedback meetings with agents. The meetings included both contractors and government personnel in order to ensure a shared understanding of agent-identified issues. The PMOD also reported conducting weekly, monthly, and real-time monitoring of trouble-tickets—that is, agent-generated reports of maintenance or other technical issues. PMOD officials reviewed the issues identified and prioritized them based on cost and the potential increases in capability. For issues beyond contractual requirements, the PMOD vets the requests and forwards them to senior Border Patrol leadership for approval and funding. Conclusions Since 2005, Border Patrol has spent more than one billion dollars deploying technologies to the southwest border, but is not yet positioned to fully quantify the impact these technologies have on its mission. We continue to believe that developing and applying performance metrics for its border technologies, in accordance with our prior recommendation, would help Border Patrol more fully assess its progress in implementing the Southwest Border Technology Plan and determine when mission benefits have been realized. Border Patrol has taken some steps toward tracking the performance of its surveillance technologies, including requiring agents to record when technologies assist in an apprehension or seizure. However, additional guidance to better ensure the quality of these data (including agent training and managerial review), would help Border Patrol determine the mission benefits of its surveillance technologies, which in turn could be used to inform Border Patrol’s resource allocation decisions. Recommendations for Executive Action The Chief of the Border Patrol should issue guidance for sectors to improve the quality and usability of its surveillance technology asset assist information to help ensure it has reliable data so that Border Patrol can be better positioned to measure the impact of these technologies on its border security efforts and inform future investments. (Recommendation 1) Agency Comments We provided a draft of this report to DHS for review and comment. DHS provided written comments, which are summarized below and reproduced in full in appendix I. DHS also provided technical comments, which we incorporated as appropriate. DHS concurred with our recommendation and described actions planned to address it. Specifically, DHS stated that Border Patrol will revise its training presentation concerning asset assists to include additional information on how sectors should conduct asset assist data integrity checks, why the data are collected, how the data can be used, and why Border Patrol needs to ensure asset assist data are accurately recorded. Border Patrol also plans to prepare and release a video concerning asset assists for all field office personnel. Border Patrol plans to complete these actions by February 28, 2018. If you or your staff have any questions about this report, please contact me at (202) 512-8777 or gamblerr@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix II. Appendix I: Comments from the Department of Homeland Security Appendix II: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Jeanette Henriquez (Assistant Director), Ashley Davis, Charlotte Gamble, Yvette Gutierrez, Eric Hauswirth, Nancy Kawahara, Marycella Mierez, Sasan J. “Jon” Najmi, and Claire Peachey made key contributions to this report.

Why GAO Did This Study The southwest border has long been vulnerable to cross-border illegal activity. In fiscal year 2016, Border Patrol apprehended over 409,000 illegal entrants. Border Patrol has employed a variety of land-based surveillance technologies to assist in securing the border. GAO has reported regularly on CBP's progress and challenges deploying surveillance technologies. GAO was asked to review CBP's use of surveillance technology. This report examines (1) the deployment status of surveillance technology programs and the extent to which CBP has developed plans for future technology deployments and (2) what data are available on the contributions of deployed technologies to CBP's border security efforts and the extent to which CBP has assessed technology performance. GAO analyzed technology program documents; interviewed CBP and Border Patrol officials; and conducted site visits to Arizona and south Texas to observe the operation of various land-based technologies. We selected these locations because CBP has deployed or has plans to deploy a mix of technologies there, among other factors. What GAO Found The U.S. Border Patrol, within the Department of Homeland Security's (DHS) U.S. Customs and Border Protection (CBP), has made progress deploying surveillance technology along the southwest U.S. border under its 2011 Arizona Technology Plan (ATP) and 2014 Southwest Border Technology Plan. The ATP called for deployment of a mix of radars, sensors, and cameras in Arizona; the 2014 plan expanded these deployments to the rest of the southwest border. As of October 2017, Border Patrol had completed the planned deployment of select technologies to Arizona, Texas, California, and New Mexico. For example, in Arizona, Border Patrol deployed all planned Remote Video Surveillance Systems (RVSS) and Mobile Surveillance Capability (MSC) systems, and 15 of 53 planned Integrated Fixed Tower (IFT) systems. Border Patrol also deployed all planned MSC systems to Texas, California, and New Mexico and completed contract negotiations to deploy RVSS to Texas. These technology programs have experienced delays, but are currently on track against revised program schedules and cost baselines. To plan for future technology deployments, Border Patrol reports it will use its Requirements Management Process (RMP)––a process designed to facilitate planning by, among other things, identifying capability gaps and collecting agents' feedback––and other initiatives. Border Patrol is currently developing written guidance for the RMP to ensure station officials understand their roles and responsibilities in the process. Border Patrol agents collect and report data on asset assists, which are instances in which technologies or other assets (such as canine teams) contributed to an apprehension or seizure; however, Border Patrol has not provided sufficient guidance to ensure the accuracy and reliability of that data. For example, agents incorrectly attributed some apprehensions or seizures to certain technologies rather than others. Stations in the Rio Grande Valley sector recorded assists from IFTs in about 500 instances from June through December 2016; however, this sector does not have IFTs. Data integrity and quality checks are the responsibility of individual sectors, but Border Patrol has provided limited guidance on how to ensure data quality. Without sufficient guidance to ensure the quality of asset assist data, Border Patrol is limited in its ability to determine the mission benefits of its surveillance technologies and use information on benefits to inform resource allocation decisions. What GAO Recommends GAO recommends that Border Patrol issue guidance to improve the quality and usability of its asset assist information. DHS concurred with GAO's recommendation.

gao_GAO-18-369

gao_GAO-18-369_0

Background Created in 1968, the SFSP is authorized under the Richard B. Russell National School Lunch Act and generally provides free meals to children age 18 and under in low-income areas during certain periods when school is not in session. Specifically, the SFSP operates during school summer vacation periods between May through September, vacation periods in any month for programs operating on a continuous school calendar, and certain other times for areas affected by an unanticipated school closure, such as for a natural disaster. However, the majority of SFSP meals are served to children during the summer months. In fiscal years 2007 through 2016, federal expenditures on SFSP increased, according to FNS data, though there was a slight decrease between fiscal years 2015 and 2016 (see fig. 1). SFSP Program Administration The SFSP is administered at the federal level by FNS through its national and regional offices. FNS is responsible for issuing regulations, instructions, and guidance; reviewing states’ program management and administration plans; overseeing program administration; and reimbursing states for meals served that meet program requirements. At the state level, the program is administered by state agencies and locally operated by state-approved sponsors, such as school districts, local government entities, or private nonprofit organizations. State agencies are responsible for approving, providing training to, and inspecting and monitoring sponsors and meal sites. Sponsors, in turn, are responsible for monitoring their SFSP meal sites, managing the meal service, and providing training to administrative staff and site operators. A sponsor may operate one site or multiple sites. Sites are physical locations in the community where children receive and consume meals in a supervised setting. According to FNS guidance, sites may be located in a variety of settings, including schools, parks, community centers, health clinics, hospitals, apartment complexes, churches, and migrant camps. SFSP Program Requirements States may approve different types of SFSP meal sites, including open sites, closed enrolled sites, and camps. Open sites operate in an area where at least half of the children are eligible for free or reduced-price school meals (referred to as “area eligible”), according to data from entities such as schools or the U.S. Census Bureau. Children are generally eligible for free or reduced-price school meals if their households have incomes at or below 185 percent of federal poverty guidelines. At open sites, meals are made available to all children in the area, and all meals served that meet program requirements are reimbursable. Closed enrolled sites, on the other hand, are open only to enrolled children, as opposed to the community at large. At closed enrolled sites, meals served to all children in attendance are reimbursable as long as at least half of the enrolled children are eligible for free or reduced-price school lunch. Unlike other types of sites, camps are reimbursed only for meals served to children who have been individually determined to be eligible for free or reduced-price school meals. SFSP meals must meet certain requirements in order to be eligible for federal reimbursement; for example, the meals must be served and consumed on-site at an approved site. Federal reimbursements for summer meals are provided for each breakfast, lunch, supper, or snack served to an eligible child at an eligible site that also meets federal requirements for menu components, scheduled meal times, and nutrition. For example, to meet nutritional requirements, a lunch or a supper must, at a minimum, include four components: 2 ounces of meat or a comparable serving of a meat alternate, 3/4 cup of fruits and/or vegetables (at least two kinds), a slice of bread or a comparable serving of another grain, and a cup of milk. In 2017, the federal reimbursement rate was $3.83 or $3.77 for each eligible SFSP lunch or supper served, depending on the type of meal site. Each site may serve up to two meals or one meal and one snack per day. Some flexibilities are available to FNS in implementing the SFSP program, under its waiver and demonstration authorities. Specifically, the National School Lunch Act authorizes the Secretary of Agriculture to waive, upon request of a state or eligible service provider, certain program requirements established under the National School Lunch Act or the Child Nutrition Act of 1966, as amended, including some for the SFSP. In order to grant a waiver request, the Secretary must determine that the waiver would facilitate the state or service provider’s ability to carry out the purpose of the program, and that the waiver will not increase the overall cost of the program to the federal government, among other things. In the event a waiver request is submitted, the Secretary is required to act promptly and state in writing whether the waiver request is granted or denied, and why. The Secretary is also required to periodically review the performance of waiver recipients, and submit an annual report to Congress summarizing the use of waivers and their effectiveness, among other details. In addition to this waiver authority, the Secretary is also authorized to carry out demonstration projects to develop and test methods of providing access to summer meals for low-income children in urban and rural areas, to reduce or eliminate the food insecurity and hunger of low-income children and improve their nutritional status. The Secretary is required to provide for an independent evaluation of the demonstration projects carried out under this authority, and submit an annual report to Congress on the status of each project and the results of the evaluations. Number of SFSP Meals Served Has Generally Increased since 2007, but Estimates of Children Participating in the SFSP Are Unreliable Number of SFSP Meals Served Has Increased by 32 Percent since Fiscal Year 2007 The total number of SFSP meals served nationwide during the summer— one indicator of program participation—increased from 113 million meals in fiscal year 2007 to 149 million meals in fiscal year 2016, or by 32 percent, according to our analysis of FNS data. The number of SFSP meals served has generally increased from year to year over this 10-year period. Most recently, meals decreased by 6 percent from 156 million meals in summer 2015 to 149 million meals in summer 2016, according to our analysis of FNS data (see fig. 2). Factors that may have affected year-to-year fluctuations include changes in funding for summer programs, sponsor participation, weather, and the number of weekdays available for sites to serve meals within a given summer, according to FNS and state agency officials we interviewed. For example, state agency officials in one of the three selected states we visited said they believe that reductions in state and local funding for summer programs that also provide meals, and turnover of sponsors, including losing one of the state’s largest sponsors in a recent summer, affected the total number of SFSP meals served in their state in 2016. According to our analysis of FNS data, SFSP lunches served in the summer months increased by over 17 million from fiscal year 2007 through fiscal year 2016, accounting for almost half of the total increase in the number of SFSP meals served in that period. However, when comparing across each of the meal types, supper and breakfast had the largest percentage increases over the 10-year period, 50 and 48 percent, respectively (see table 1). In comparison, the number of SFSP lunches served increased by 26 percent from fiscal years 2007 through fiscal year 2016. From fiscal year 2007 through fiscal year 2016, there were increases in the numbers of meals served in both SFSP and NSLP, the largest child nutrition assistance program. Specifically, SFSP lunches served in July increased from 32 million to 40 million, or 24 percent, from fiscal year 2007 to 2016, and NSLP lunches served in March increased from 328 million to 376 million meals, or 15 percent, according to our analysis of FNS data. Although the programs generally serve similar populations, different factors likely affected the number of meals served by each program, in part because NSLP serves children in schools during the school year and SFSP serves children in a variety of settings during the summer months. Estimates of Children Participating in SFSP Have Been Calculated Inconsistently and Are Unreliable Although states report the actual number of SFSP meals served to FNS, they estimate the number of children participating in SFSP, and information obtained from our state survey and FNS indicate that these participation estimates have been calculated inconsistently. FNS instructs state agencies on how to calculate a statewide estimate of children’s participation in the SFSP, referred to as average daily attendance (ADA), using sponsor-reported information on the number of meals served and days of operation in July of each year. However, states’ methods for calculating ADA have differed from state to state and from year to year, according to our review of states’ survey responses and FNS documents. For example, although FNS directed states to include the number of meals served in each site’s primary meal service— which may or may not be lunch—some states, according to our survey and FNS data, were calculating ADA using only meals served at lunch. FNS officials told us that these states were therefore not following the agency’s instructions. Further, some states have changed their methods for calculating ADA over time—five states reported in our survey that the method they used to calculate ADA in fiscal year 2016 differed from the one they used previously. While FNS clarified its instructions in May 2017 to help improve the consistency of states’ ADA calculations moving forward, ADA remains an unreliable estimate of children’s daily participation in SFSP for at least two reasons, according to our analysis. (See sidebar for the revised ADA calculation instructions.) First, ADA is based on summary data that does not account for existing variation in site days of operation, and second, it is based on July data, which does not reflect the month with the greatest number of meals served in every state. According to our analysis, ADA is an unreliable estimate of children’s participation in SFSP because it currently does not account for existing variation in the number of days that each site serves meals to children. Specifically, because FNS’s instructions indicate that sites’ ADAs are to be combined to provide a statewide ADA estimate, differences in the number of days of meal service are disregarded. As a result, ADA does not reflect the average number of children served SFSP meals daily throughout the month. Our analysis of site-level data from one of the selected states illustrates this limitation. In this state, multiple sites reported an ADA of 60 for July, yet two of those sites served meals to children on only 1 day of the month and another site served meals to children on 20 days. Although 120 children were served SFSP meals only 1 day in July across two of these sites, the combined ADA across all three sites, which we calculated following FNS’s instructions, inaccurately suggests an average of 180 children were participating in SFSP at these sites on a daily basis in July. According to our analysis, ADA is also an unreliable estimate of children’s participation in SFSP because it currently does not account for state variation in the month with the greatest number of SFSP meals served, potentially leading to an underestimate. According to FNS officials, the agency instructs states to calculate ADA for July because officials identified this as the month with the largest number of meals served nationwide. However, because of reasons such as state variations in school calendars, July is not the month with the largest number of meals served in every state. In one of the selected states, Arizona, using July to calculate ADA cuts the estimate almost in half. Specifically, we followed FNS’s instructions and calculated that Arizona’s ADA was 14,987 in July 2016 compared to 26,772 in June 2016. Nationwide, in summer 2016, 26 states served more SFSP meals in June or August than in July, according to our analysis of FNS data. However, without site level data on meals served and operating days, the extent to which these states had higher ADAs in June or August as compared to July is unknown. In its May 2017 memo to states revising the ADA calculation instructions, FNS said that it is critical that the agency’s means of estimating children’s participation in the SFSP is as accurate as possible because it helps inform program implementation at the national level and facilitates strategic planning and outreach to areas with low participation. In addition, Standards for Internal Control in the Federal Government state that agencies should maintain quality data and process it into quality information that is shared with stakeholders to help achieve agency goals. Although FNS has also collected information on other data that states collect on the SFSP, the agency has not yet used this information to help improve its estimate of children’s participation in the program. In 2015, FNS published a Request for Information, asking whether states or sponsors collect any SFSP data that are not reported to FNS. While FNS received responses from only 15 states, these responses suggest that some states collect additional data, such as site-level data that may allow for an improved estimate of children’s SFSP participation, potentially addressing the issues we found in our analysis. In response to the information FNS received, they followed up with up to 9 of the 15 states in 2016 and 2017 to explore the feasibility of collecting additional data and improving estimates of children’s participation. Although they took these steps, FNS officials told us they are cognizant of the burden on states and site operators that would be associated with additional reporting requirements. At this time, the agency has not taken further action to improve the estimate, such as addressing the reliability issues caused by variation in the number of operating days of meal sites and in the months with the greatest number of meals served by state. As a result, FNS’s understanding of children’s participation in the SFSP remains limited, which impairs its ability to both inform program implementation and facilitate strategic planning and outreach to areas with low participation. Other Federal and Nonfederal Programs Help Feed Low- Income Children over the Summer to Some Extent Other Federal Programs Provide Meals and Nutrition Assistance Benefits over the Summer Other federal programs that operate solely in the summer, as well as those operating year-round, help feed low-income children in the summer months. These programs include the NSLP Seamless Summer Option, which provides nutrition assistance benefits solely in the summer, and several federal programs that operate year-round. In July 2016, in addition to the 70 million meals provided through the SFSP, 26 million meals were provided to low-income children through school food authorities participating in the NSLP’s Seamless Summer Option, according to FNS data. The Seamless Summer Option was established in 2004, and according to FNS, streamlines administrative requirements to encourage school food authorities providing free or reduced-price meals during the school year under the NSLP and SBP to continue providing meals to low-income children when school is not in session. For example, officials from a national organization involved in summer meals told us the Seamless Summer Option makes it easier for school food authorities to provide summer meals because they continue working with the same state agency, reporting the same information to the state, and operating without having to transition to a separate program. Nonetheless, school food authorities can choose to provide free summer meals to children through either the SFSP or Seamless Summer Option, and the majority of states (34) reported in our survey that a greater proportion of school food authorities participated in the SFSP than the Seamless Summer Option in summer 2016. According to FNS and selected state officials, this may be related to the generally lower meal reimbursement rates school food authorities participating in the Seamless Summer Option receive compared to the rates received by those participating in the SFSP. In summer 2016, the Seamless Summer Option added to the geographic availability of summer meal sites in two of the three states we visited as part of our review. School food authorities provided summer meals through the Seamless Summer Option in Arizona and Illinois, but not in Massachusetts, based on our analysis of data provided by these states. In Arizona and Illinois, school food authorities participating in the Seamless Summer Option added 643 and 298 summer meal sites, respectively, in the month with the largest number of SFSP meals served in each state (see fig. 3). In addition, some of the Seamless Summer Option sites in these two states provided meals to children in areas where there were no SFSP sites. For example, Seamless Summer Option sites provided meals in areas near the northeastern and southwestern corners of Arizona that lacked nearby SFSP sites. In addition to the SFSP and the Seamless Summer Option, the Summer Electronic Benefit Transfer for Children (Summer EBT) demonstration provided nutrition assistance benefits to 209,000 low-income children in summer 2016 in select areas across 6 states and 2 Indian Tribal Organizations, according to FNS officials. Since the summer of 2011, Summer EBT benefits have been provided to eligible households on an electronic benefits transfer card, which households use to purchase eligible foods at authorized retailers. Specifically, the demonstration has provided monthly benefits of $30 or $60 per eligible child to households with children in areas with a perceived high level of need, based on the demonstration grantees’ assessments of the percentage of children eligible for free or reduced-price school meals and the availability of the SFSP. Consistent with this, three of the states that participated in Summer EBT in 2016 reported through our survey that these benefits helped children who were unable to access summer meals through the SFSP or the Seamless Summer Option. Further, according to an FNS- funded evaluation, Summer EBT improved food security among low- income children who participated in the demonstration. Specifically, the evaluation found the receipt of these benefits reduced the number of children in the demonstration experiencing very low food security between 2011 and 2013 by one-third. Some low-income children also receive nutrition assistance in the summer through federal programs that operate year-round. According to FNS data, in June 2016, 5.8 million infants and children participated in the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) and 3 million children participated in the Child and Adult Care Food Program (CACFP). In addition, an average of 19.2 million children participated each month in the Supplemental Nutrition Assistance Program (SNAP) in fiscal year 2016, according to FNS data. These benefits are provided year-round, including when school is in session and children may also be eligible to receive school meals. In our previous work on federal domestic food assistance programs, we reported that no one program alone is intended to meet a household’s full nutritional needs. At that time, several officials and providers told us that the variety of food assistance programs offers eligible individuals and households different types of assistance and can help households fill the gaps and address the specific needs of individual members. For example, a mother with two children may rely on SNAP for her household’s basic groceries, the NSLP to feed a school-age child during the school year, and WIC to obtain supplemental foods for herself and an infant. Nonfederal Programs Also Help Feed Low-Income Children in the Summer, but States and Local Organizations Reported That These Have Limited Reach Some low-income children also receive summer meals through nonfederal programs, according to our state survey and interviews with organizations involved in summer meals. Twenty-seven states reported in our survey that they were aware of other state- or non-state-funded programs that provided children of low-income households with meals in their states during the summer months. According to our analysis of state survey responses, local faith-based organizations and foodbanks were the most common types of entities operating these types of programs. Similarly, officials from FNS and two regional organizations we interviewed said they were aware of children receiving summer meals through nonfederal programs operated by faith- based and other community organizations. In addition, SFSP site operators at 6 of the 30 meal sites we visited in the selected states told us nearby foodbanks and faith-based organizations may also be providing children with free meals to some extent. For example, one of the meal sites we visited was operated by a foodbank that, in addition to the SFSP, provided food boxes to those in need and distributed food to other local community organizations to provide to persons in need of immediate assistance, including families with children. Although FNS and the majority of states do not collect data on nonfederal programs, results from our state survey and interviews with SFSP providers and organizations involved in summer meals indicate the reach of nonfederal programs is limited. In our survey, states reported that the geographic coverage of these nonfederal programs varied by state, with 11 states indicating that they operated in some portions of the state—the most common state response. In addition, 16 states reported that they were not aware of any nonfederal programs providing summer meals to children in their state (see fig. 4). Similarly, SFSP site operators at 24 of the 30 meal sites we visited were unaware of nonfederal programs providing meals to children in the areas in which they operated. In addition, officials from several national organizations involved in summer meals told us children have very few options for receiving summer meals beyond the federal summer meals programs. Specifically, officials from one national organization explained that food is often a significant part of the cost of a summer activity program for children and suggested that is one reason why organizations choose to participate in the SFSP. Although the SFSP provides for federal reimbursement of eligible meals and certain administrative and operating costs, nonfederal programs that provide children with summer meals may choose not to participate in the SFSP for several reasons, according to officials we interviewed from several organizations involved in summer meals. For example, some nonfederal program providers may not participate in the SFSP because they are unaware the program exists. Additionally, some nonfederal program providers may be aware of the SFSP, but choose not to participate because they do not want to follow certain program requirements, such as the nutrition or meal pattern standards. In addition, some providers may not participate in the program because they do not think they can handle certain aspects of the administrative workload associated with the SFSP. For example, a state official we interviewed told us the administrative workload associated with the SFSP can be particularly challenging, especially for smaller sponsors. Similarly, officials from a regional organization involved in summer meals told us one of the providers they work with who operated 10 meal sites chose to leave the SFSP because the paperwork required to operate the sites was too administratively burdensome for their volunteer site operators. States and SFSP Providers Face Challenges with Meal Sites, Participation, and Program Administration, and FNS Actions Have Addressed Some, but Not All Areas States and SFSP providers reported challenges with meal sites, participation, and administration, though federal, state, and local entities have taken steps to improve these areas. Half or more of states reported in our survey that SFSP issues related to meal site availability, such as in rural areas, increasing children’s participation, and program administration were moderately to extremely challenging (see fig. 5). Overall, 41 states reported facing at least one challenge with the SFSP, while 9 reported facing none. Challenges with the Availability of Meal Sites Rural Areas Availability of transportation, low population density, and limited meal sites pose challenges for SFSP in rural areas, according to states we surveyed, selected national organizations, and state and local officials in the three selected states we visited. More than two-thirds of states in our survey reported they faced a moderate to extreme challenge with limited options in rural areas to transport children to summer meal sites (37), as well as with the distance to summer meal sites in rural areas resulting in low child turnout that affects the financial viability of site sponsorship (36). As officials from one national organization explained, it may not be cost- effective for sponsors to operate in remote or rural areas if there are not enough meal sites or children participating in the program. Similarly, a sponsor in one of the selected states indicated that there are large parts of the state where the distances between meal sites are substantial, and travel between them takes several hours. An official from one of the selected states said transportation challenges can lead to underserved rural areas, including Indian reservations. Of the three states we reviewed, each had rural areas with few or no federally funded meal sites in summer 2016. However, a majority of the children in some of those areas were eligible for free or reduced price school meals, according to Census data provided by FNS, and would therefore be “area eligible” for the purposes of SFSP. For example, as shown in figure 6, “area eligible” locations in rural western parts of Arizona did not have any SFSP or Seamless Summer Option meals sites in June 2016, the month with the greatest number of summer meals served in that state. States and SFSP providers have responded to challenges with meal sites in rural areas by using other meal delivery approaches—efforts that FNS has supported through information sharing and grants. For example, according to one national organization involved in summer meals, some SFSP providers offer vans or buses to transport children to meal sites or partner with local bus authorities to give children free rides to meal sites. Instead of transporting children to sites, other sponsors transport meals to children through mobile meal delivery, an alternative summer meal model used in 48 states according to our survey. In this model, sponsors deliver meals by bus, using a route with state-approved stops in a community, and children consume the meal at the stop under a supervised setting. According to FNS officials and representatives from national organizations, this approach can be particularly helpful for providing summer meals to children in rural areas. State officials in two selected states told us they use mobile meal delivery to help fill gaps in meal service and help children overcome the lack of transportation or resources in their community. To serve children in very remote areas with limited resources, a sponsor in one of the selected states reported piloting a model involving delivering frozen meals every other week to such areas and supplying equipment, such as freezers and microwaves, to support meal service. To help sponsors address challenges related to meal sites in rural areas, FNS has shared information on alternative delivery models through its SFSP toolkit and webinars and has also provided related grant funding. For example, in summer 2011 and 2012, FNS funded the Meal Delivery demonstration project to provide meals to children in rural areas where low population density, long distances, and transportation issues made it difficult for children to get to SFSP sites, making site and sponsor operation financially unsustainable. The demonstration project funded meals to children in rural areas of Delaware, Massachusetts, and New York, providing food delivery to homes or drop-off sites near homes of eligible children. Area Eligibility More than half the states (30) in our survey reported they faced a moderate to extreme challenge reaching low-income children in communities that are not area eligible. Areas in which fewer than 50 percent of children qualify for free or reduced-price meals during the school year are not eligible to have open summer meal sites at which all children who come to the site can receive a free meal. As a result, some children who are eligible for free and reduced-price meals during the school year do not have open summer meal sites located in close proximity to their residences, according to several national organization officials and SFSP providers. Eligible children in these areas may instead be limited to other types of SFSP sites, such as closed enrolled summer meal sites, or nonfederal programs providing meals, if available. For example, in one of the selected states, a sponsor of SFSP sites funded meals without federal support at one site that they operated as an open site in order to serve low-income children residing in low-income housing. These children did not otherwise have access to a federally funded summer meals site, according to these officials, because the broader area was part of a school district that had a greater than 50 percent proportion of children from higher-income families. Recognizing that some children may reside in an area that is not area eligible but is immediately adjacent to such an area, FNS has allowed additional flexibility in establishing area eligibility for open meal sites. Specifically, in 2014 and 2016 policy memos, FNS expanded the ways in which states and sponsors can use Census data to establish area eligibility. For example, FNS has allowed states and sponsors to average Census data across adjacent geographic areas to determine area eligibility. FNS noted that these additional flexibilities help ensure meal sites can be located in more areas in which poor economic conditions exist. Limited Days of Operation Nearly all states (50) reported in our survey that the availability of meal sites throughout the summer months was a factor critical to the success of the SFSP, yet more than half the states (27) also reported they faced a moderate to extreme challenge with limited meal site days of operation. Nineteen of the 40 states that provided information about site days of operation reported 1 day as the shortest length of operation for SFSP sites in their state in fiscal year 2016. Limited meal site days of operation was a significant challenge in one of the three selected states we visited, as almost one-quarter of sites operated for only 1 to 2 weeks across a 2-month period in summer 2016, and an additional half of sites operated for 3 to 4 weeks across that same period, according to our analysis of state data. In contrast, in the other two selected states, the majority of sites (64 and 76 percent, respectively) operated for 5 or more weeks during a 2-month period. SFSP sites may have limited days of operation for various reasons, such as constraints with program administration and costs, according to interviews with a national organization official and a sponsor in one of the selected states. Some SFSP providers and national organizations involved in summer meals have responded to these challenges by working to extend the days of operation of meal sites—efforts that FNS has supported through related grant funding. Officials from one meal site located at a school in one of the selected states told us that 2017 was the first year the site stayed open an additional 4 weeks after summer school classes ended in an effort to expand participation, an extension made possible through support from an experienced sponsor. In addition, officials from a national organization involved in sponsoring summer meals told us they encourage their local sites to operate in August—a month where there are generally fewer summer meal service offerings—to meet children’s needs. At the federal level, under its demonstration authority, FNS funded the Extending Length of Operation Incentive project, a grant which provided an additional 50-cent reimbursement for all lunch meals served at sites in Arkansas in 2010 that offered meals for 40 or more days. Challenges with Children’s Participation Awareness of the SFSP Program and Meal Sites Two-thirds of states (34) reported through our survey that they also faced a moderate to extreme challenge with a lack of awareness of summer meal sites among children and families, a challenge also mentioned by SFSP providers in the selected states. Meal site operators in one selected state noted that making families aware that all children may receive a meal for free at open sites can be a challenge. For example, one sponsor operating a meal site in a school said the perception among some is that the meal program is only for children attending summer school, and not for others in the community. Although that site had outside banners and advertising to help address that misperception, another SFSP provider explained that having sufficient funds to market the SFSP and increase awareness among families is also a challenge. To address these challenges, state agencies, some SFSP providers, and FNS have taken steps to help promote awareness of the SFSP. For example, nearly all states (47) reported in our survey that they have increased their outreach efforts for the SFSP in the last 5 years. More than half of states (36) also reported increases in overall SFSP participation during that time, which they believe were related to their outreach efforts. The majority of states in our survey reported conducting outreach on the SFSP to groups including children, parents and guardians, and schools, among others, using methods such as flyers, email, newspapers, and social media (see fig. 7). Further, state agency officials and sponsors in the selected states reported that they have developed partnerships with state and local advocacy groups and community leaders, among others, to promote the SFSP. For example, one state agency official said they partner with local advocacy organizations to field calls from parents seeking information about summer meal sites through their hunger hotline. FNS has promoted the use of such partnerships, as well as traditional and social media, to raise awareness of the SFSP. In addition, FNS developed the Summer Meals Site Finder, an online mapping tool that provides information on summer meal sites nationwide. Attracting children of all ages to SFSP meal sites can also be a challenge, according to states and SFSP providers. More than half of the states (31) reported in our survey that they faced a moderate to extreme challenge with limited youth and teen participation at summer meal sites, and an official from a national organization involved in the SFSP explained that it is difficult to attract children to a meal site when the site is focused solely on food. Similarly, 46 states in our survey reported that providing age- appropriate programming and enrichment activities for children at summer meal sites is a factor critical to the success of the SFSP. However, some meal sites may lack the resources to add activities, according to some SFSP providers in the selected states as well as FNS and national organization officials. Attracting teens can be particularly challenging, in part because of meal service time periods, a lack of age- appropriate activities, and stigma, according to national organizations and providers we interviewed. For example, early morning meal sites generally attract younger kids as teens may be apt to sleep later in the summer, and teens may also perceive a stigma in participating in a free meal program and may face peer pressure not to eat. In addition, meal offerings at SFSP sites may also present challenges to teen participation. Specifically, because FNS bases minimum portion size requirements for meals on the needs of younger children, meals are not always adequate to meet the nutritional needs of teens, according to one sponsor we interviewed. Across the 30 meal sites in the 3 states we visited in summer 2016, we observed variety in the meals served during different meal services. (see fig. 8.) States and SFSP providers have collaborated with others and sought specific types of sites to help provide enrichment activities and attract certain age groups—efforts that FNS has supported through information sharing and related grant funding. Sponsors in the selected states said they have focused on partnerships with groups such as those focused on youth development, churches, libraries, and police or fire departments, to offer age-appropriate activities for children (see fig. 9). For example, programs with local police departments, such as Cops N Kids in one selected state, or libraries in two selected states, provided meal services in combination with youth development or other enrichment activities. (See sidebar for highlights on the Cops N Kids program.) One national organization official said activities at SFSP sites can help take away the stigma around the program because children are not just there for the meal. Efforts to rebrand the SFSP as a community event where entire families can participate at the meal site also can have this effect, which is why some sponsors in the selected states said they partnered with foodbanks to donate meals for adults. In addition, a sponsor in one selected state told us they adjusted their meal offerings to match the needs of children of different age groups, for example, by serving meals to younger children earlier in the day and meals to teens later in the day. To support participation from children of all ages, FNS has shared information on age-appropriate activities through its SFSP toolkit and provided related grant funding. For example, in 2010, FNS funded the Activity Incentive demonstration project, in which sponsors in Mississippi were provided with mini-grants to increase enrichment and recreational activities, such as education, tutoring, sports and games, arts and other activities, to draw children to meal sites. More than half the states reported in our survey that they faced a moderate to extreme challenge with limited state agency staffing (27), a limited amount of federal funding for SFSP administration (27), as well as ensuring sponsor participation to meet needs (28). In addition, 28 states reported in our survey that they faced a moderate to extreme challenge with sponsors not following program requirements. Limited staffing can affect a state agency’s ability to conduct efforts aimed at increasing participation, identifying potential sponsors, and reviewing and monitoring sponsors, according to national organization and state officials we interviewed. For example, increases in sponsors and sites requires additional staff and time to conduct pre-approval visits, sponsor and site reviews, vendor reviews, and technical assistance visits, which directly affects the amount of funding needed to support staff salaries and travel reimbursement, according to one state in our survey. However, because the SFSP administrative funds FNS provides to states are based on the number of meals served in the previous year, increasing the number of staff to help increase SFSP participation is difficult, according to a national organization official we interviewed. States reported a moderate to extreme challenge with the following issues related to ensuring sponsor participation: a lack of sponsors to meet summer meal needs, a lack of awareness of the summer meal program among potential sponsors or sites, completing federal requirements for monitoring of SFSP sponsors, and identifying potential sponsors. State agencies responsible for administering the SFSP reported relying on other resources and partners to help with program administration— strategies that FNS has supported through information sharing and its online tools. As discussed earlier, all three selected state agencies we interviewed told us they partner with advocacy groups to help expand and conduct outreach on the SFSP. Additionally, more than half the states in our survey reported several factors—which may ease the administrative burden on states—as critical to the success of the SFSP, including partnerships with SFSP sponsors (49) and retaining sponsors and sites over multiple summers (51). To support states’ use of alternative funding sources to help administer the SFSP, FNS has shared information on federal, state, and private funding and grant opportunities. FNS also developed the online Capacity Builder tool, which 35 states reported in our survey was moderately to extremely useful in identifying or confirming meal site eligibility in fiscal year 2017. States and SFSP Providers Also Reported Challenges with Meal Site Safety and Duplicative Paperwork, and FNS’s Efforts to Address These Areas Are Limited Seventeen states reported in our survey that ensuring summer meal sites are in safe locations was moderately to very challenging, a challenge that some states and SFSP providers have taken steps to help address. State officials and SFSP providers in the selected states reported that when crime has occurred near a site, there are concerns about ensuring children’s safety while they are consuming meals at the site, as well as the safety of site staff delivering meals. Some sponsors noted, in particular, parents’ concerns for the safety of their children at meal sites in light of criminal activities in the surrounding area. To ensure children continue to have access to meals, some sponsors noted that in the event of an immediate threat at an outdoor meal site, site staff are sometimes able to bring children to a nearby indoor space instead. States and SFSP provider officials in two selected states told us they have also used other strategies, including partnerships with local law enforcement agencies, to help address safety concerns during the meal service and ensure children have access to meals. For example, national organizations involved in summer meals and sponsor officials in the selected states said they encourage partnerships with local police departments to use police escorts at meal sites or to follow mobile meal routes in situations where safety at the meal site is a concern. When violence or crime has occurred near a site, some states and SFSP sponsors have also sought flexibility from FNS with respect to the federal requirement that children consume summer meals on site, according to state and local officials. FNS has used its available authorities to grant some states and sponsors flexibility with respect to the requirement that children consume summer meals on site, such as when safety at the site is a concern; however, FNS has not clearly communicated to all states and sponsors the circumstances it considers when deciding whether to grant this flexibility. According to our review of letters FNS sent to multiple states approving their requests for this type of flexibility, the agency identified a consistent set of circumstances that needed to be met for it to grant this flexibility. These circumstances were described in the letters the agency sent to states and generally included verification that violent crime activities occurred within both a 6-block radius of the meal site and 72 hours prior to the meal service. FNS’s letters to states indicate that when documentation was provided to the agency showing that these circumstances existed at a summer meals site on a particular day or days, meals consumed by children off site on those days were eligible for federal reimbursement. Although FNS has issued guidance on the general processes for requesting flexibility from program requirements under its waiver and demonstration authorities, these guidance documents do not detail the specific circumstances that the agency considers when deciding whether to grant flexibility from the on-site requirement due to safety concerns. FNS has communicated this information only in its responses to specific state and sponsor requests, and it has not communicated these circumstances more broadly to all states and sponsors. FNS officials explained that they review state and sponsor requests for flexibility due to safety concerns on a case-by-case basis. However, they also acknowledged that the set of circumstances used for approval of state and sponsor requests for flexibility, which we identified in their letters to states, has been used repeatedly. Further, states and sponsors reported challenges obtaining the specific data needed for approval of a site for this type of flexibility, hampering some providers’ efforts to ensure safe delivery of meals. For example, state agency and sponsor officials in one selected state said obtaining the crime data needed to qualify for the flexibility can be an administrative burden on sponsors, and these data are not consistently available in a timely manner. According to state agency and sponsor officials in one of the selected states, daily crime statistics are not available in all areas, and while a sponsor can sometimes access current data on crime in a city, the most recent available data on crime in suburban areas are sometimes one year old. FNS is aware of state and local challenges obtaining the necessary crime data, according to our discussions with FNS officials. FNS officials acknowledged that while they have granted some state and sponsor requests to allow children to consume meals off site in certain areas where violence or crime has occurred, some sponsors were unable to implement the flexibility because they could not obtain the necessary crime data. To help achieve agency objectives and address related risks, the Standards for Internal Control in the Federal Government state that agencies should communicate key information to their internal and external stakeholders. Although FNS officials told us they do not have one set of circumstances under which they approve these requests, our review found only one set of circumstances under which this type of flexibility has been approved. However, FNS has not broadly communicated the circumstances it considers in deciding whether to approve requests for flexibility with respect to the requirement that children consume summer meals on site in areas with violence or crime. Unless FNS shares this information with all states and sponsors, states and sponsors will likely continue to be challenged to use this flexibility, hindering its usefulness in ensuring safe summer meal delivery to children. In addition, FNS has issued reports to Congress evaluating some of its demonstration projects, as required under its statutory authorities, but the agency has not issued any such reports to Congress specifically on the use of flexibilities with respect to the on-site requirement in areas where safety is a concern. As previously discussed, the agency is required to annually submit certain reports to Congress regarding the use of waivers and evaluations of projects carried out under its demonstration authority. Furthermore, Standards for Internal Control in the Federal Government state that management should use quality information to make informed decisions and evaluate the entity’s performance in achieving key objectives and addressing risks. Yet, FNS has not evaluated nor reported on the use of waivers and demonstration projects in cases where safety was a concern. Although FNS requests reports from state agencies or sponsors that have received flexibility with summer meals delivery under FNS’s demonstration and waiver authorities, FNS officials told us they have not assessed whether their use of these flexibilities to address safety issues has been effective in ensuring safe meal delivery. FNS officials told us that they have not evaluated or reported on these flexibilities, in part, because they have limited information on their outcomes. Without understanding the impact of its use of these flexibilities, neither FNS nor Congress knows whether these flexibilities are helping provide meals to children. In addition to the challenges with safety at meal sites, sponsors also sometimes face administrative challenges when participating in multiple child nutrition programs that are operated by different state agencies or divisions within the same agency, according to officials from national and regional organizations and sponsors we interviewed. For example, officials from national organizations involved in summer meals told us the management of each child nutrition program and processes related to applications, funding, and oversight are fragmented in many states. For example, a sponsor in one of the selected states told us aspects of the SFSP and CACFP sponsor applications are highly duplicative and estimated it took 42 hours last year to complete duplicative paperwork. Another sponsor that provides school meals during the school year told us they had to fill out 60 additional pages of paperwork to provide summer meals, which coupled with having a state contact for the SFSP that was different from the one they worked with for the NSLP, was a significant burden for them. Officials from one national organization told us a lack of interoperability of some state agencies’ data systems has caused challenges and administrative burden for some sponsors. For example, in some states, different agencies oversee child nutrition programs, yet are unable to share data on sponsor approval, and therefore, sponsors are required to submit similar information to both, according to these officials. Duplicative paperwork can be particularly burdensome for some SFSP providers, as national organization officials and SFSP providers in the selected states said completing SFSP application paperwork can be especially challenging when a sponsor has staff shortages or no dedicated SFSP staff. Some selected states have worked with SFSP sponsors to help minimize the administrative burden. For example, state agency officials from one of the selected states said they have connected less-experienced sponsors to more-experienced sponsors in the community to help them with program administration. In one case, an experienced SFSP sponsor partnered with a small sponsor new to the program to help with SFSP administration, including helping them understand program rules and paperwork requirements. One SFSP sponsor also noted that their state agency took additional steps to ease administrative burden, such as making the forms for the CACFP more consistent with those for the SFSP and streamlining certain requirements for large and experienced sponsors, which the sponsor found helpful. At the federal level, FNS has established program and policy simplifications to help lessen the administrative burden on sponsors participating in multiple child nutrition programs, though the persistence of these challenges indicate that information about these simplifications has not reached all relevant state agencies. While FNS officials told us that some of the duplicative requirements may be a function of differences in statute, FNS provided guidance to states in 2011 and 2014 on simplified application procedures for institutions participating in CACFP that also wish to apply for SFSP. FNS noted in its guidance that in states where CACFP and SFSP are administered by different state agencies, state agencies are encouraged to work together to share information and streamline the application and agreement process as much as possible. FNS also addressed these simplifications in a state agency meeting in November 2017. Additionally, FNS provided guidance to states in 2012 on simplified application and review procedures for school food authorities participating in the NSLP that wish to also participate in the SFSP. Although FNS has shared this information with states in an attempt to make them aware of streamlining options, FNS officials noted that some states may choose not to implement them. Standards for Internal Control in the Federal Government state that management should externally communicate the necessary quality information to achieve the entity’s objectives, as well as periodically evaluate the methods of communication to ensure communication is effective and appropriate. FNS’s existing guidance addresses options for streamlining administrative requirements for sponsors participating in multiple child nutrition programs. However, information on program and policy simplifications available for sponsors participating in both NSLP and SFSP has not been shared with states recently, and challenges in this area persist, indicating this information has not reached all relevant state agencies. Without further efforts from FNS to disseminate information on current options for streamlining administrative requirements across child nutrition programs, overlapping and duplicative administrative requirements may limit children’s access to meals by discouraging sponsor participation in child nutrition programs. Conclusions The purpose of the SFSP is to continue to provide children in low-income areas with nutritious meals over the summer when school is no longer in session, and to that end, the program provided 149 million SFSP meals to children in fiscal year 2016. Although meals served are one indicator of participation, FNS’s current estimates of children participating in SFSP are unreliable. Without additional understanding of children’s participation in the SFSP, FNS lacks information critical for informing program implementation, strategic planning, and outreach. The majority of states nationwide and SFSP providers in the three states we visited reported experiencing a number of challenges with the SFSP, and FNS has taken important steps to address these challenges. Two key challenges identified by officials in the selected states and national organizations we interviewed are ensuring summer meal sites are in safe locations, and meeting administrative requirements when participating in multiple child nutrition programs. FNS has taken steps to address these challenges by providing flexibilities in how meals are delivered to children and streamlining options for those providers participating in more than one child nutrition program. However, a lack of clarity concerning the circumstances under which FNS grants flexibilities in areas of violence and crime, and a lack of information on its use of these flexibilities and their impact on program administration, hinder efforts to ensure program goals are met. Furthermore, absent a reminder to states regarding existing options for streamlining administration across multiple nutrition programs, some providers may continue to be discouraged from participating in these programs due to duplicative and burdensome administrative requirements, which may ultimately limit the provision of nutritious meals to children. Recommendations for Executive Action We are making the following four recommendations to FNS: The Administrator of FNS should improve its estimate of children’s participation in the SFSP by focusing on addressing, at a minimum, data reliability issues caused by variations in the number of operating days of meal sites and in the months in which states see the greatest number of meals served. (Recommendation 1) The Administrator of FNS should communicate to all SFSP stakeholders the circumstances it considers in approving requests for flexibility with respect to the requirement that children consume SFSP meals on-site in areas that have experienced crime and violence, taking into account the feasibility of accessing data needed for approval, to ensure safe delivery of meals to children. (Recommendation 2) The Administrator of FNS should evaluate and annually report to Congress, as required by statute, on its use of waivers and demonstration projects to grant states and sponsors flexibility with respect to the requirement that children consume SFSP meals on-site in areas experiencing crime or violence, to improve its understanding of the use and impact of granting these flexibilities on meeting program goals. (Recommendation 3) The Administrator of FNS should disseminate information about existing flexibilities available to state agencies to streamline administrative requirements for sponsors participating in the SFSP and other child nutrition programs to help lessen the administrative burden. For example, FNS could re-distribute existing guidance to state agencies that explains available flexibilities and encourage information sharing. (Recommendation 4) Agency Comments and Our Evaluation We provided a draft of this report to the Secretary of the USDA for review and comment. FNS officials provided technical comments, which we incorporated as appropriate. In addition, in oral comments, FNS officials, including the Deputy Administrator for Child Nutrition Programs, generally agreed with the recommendations in the report. As agreed with your offices, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies to the Secretary of the USDA and interested congressional committees. The report will also be available at no charge on the GAO website at www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-7215 or larink@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix IV. Appendix I: Objectives, Scope and Methodology This appendix discusses in detail our methodology for addressing three research objectives: (1) What is known about participation in the Summer Food Service Program (SFSP) and how has it changed in the last 10 years? (2) What other programs help feed low-income children over the summer? and (3) What challenges exist, if any, in providing summer meals to children, and to what extent does the U.S. Department of Agriculture’s (USDA) Food and Nutrition Service (FNS) provide assistance to states and sponsors to address these challenges? In addition to the methods we discuss below, to address all three research objectives, we reviewed relevant federal laws, regulations, and guidance; interviewed FNS officials in its headquarters and seven regional offices; and interviewed a broad range of regional and nationwide organizations involved in the SFSP. In addition, we coordinated with officials in USDA’s Office of Inspector General on their ongoing work in this area. Summer and School Meals Data To address our first objective about participation in the SFSP, we analyzed FNS data on meals served for fiscal years 2007 through 2016. Specifically, we analyzed the total number of meals served nationwide through the SFSP from fiscal year 2007 through fiscal year 2016. Each month, states report to FNS the number of meals served by meal type (breakfast, lunch, snack, and supper) and the number of meals served by meal and sponsor type (e.g., government, nonprofit, etc.) using the FNS- 418 form. To add context on these trends, we also analyzed and compared the number of SFSP lunches served in July with the number of free and reduced-price lunches served to children in March through the National School Lunch Program (NSLP), the largest child nutrition assistance program, from fiscal year 2007 through fiscal year 2016. Each month, states report to FNS the number of meals served through the NSLP using the FNS-10 form. To assess the reliability of SFSP and NSLP data, we (1) performed electronic testing of relevant data elements, (2) reviewed existing information about the data and the system that produced them, and (3) interviewed agency officials knowledgeable about the data. Electronic testing included, but was not limited to, checks for missing data elements, duplicative records, and values outside a designated range or valid time period. We determined that these data were sufficiently reliable to identify the number of SFSP meals served and assess change over time. To further examine what is known about participation in the SFSP, we also reviewed FNS’s data on estimates of children’s participation in the program and determined that these estimates have been calculated inconsistently and are unreliable. To assess the reliability of these data, we reviewed documentation about the estimates, interviewed FNS officials, and asked states about the estimate calculation in our survey. As described in our findings, FNS does not collect data on the number of children participating in the SFSP. Instead, FNS relies on states’ estimates of children’s participation, which are based on other data reported by sponsors, such as the number of meals served and meal service days in July. To address our second objective about other programs that help feed children in the summer, we reviewed FNS’s estimate of the number of meals served through the NSLP’s Seamless Summer Option in fiscal year 2016. FNS does not collect data on the number of meals served through the Seamless Summer Option. Instead, FNS annually estimates the number of Seamless Summer Option meals served nationally by aggregating the number of free and reduced-price breakfasts, lunches, and snacks served through the School Breakfast Program (SBP) and NSLP in July. As previously noted, states report these data monthly to FNS. Although FNS does not know the actual number of meals served through the Seamless Summer Option, agency officials told us they believe the number of summer meals provided through the NSLP is small relative to the number of meals served through the Seamless Summer Option during the summer months. They noted that their use of July NSLP data to estimate the Seamless Summer Option meals likely overestimates the number of these meals for July and underestimates the number of these meals for the entire summer. To assess the reliability of the July NSLP data, we (1) performed electronic testing of relevant data elements, (2) reviewed existing information about the data and the system that produced them, and (3) interviewed agency officials knowledgeable about the data. Electronic testing included, but was not limited to, checking for missing data and data that fell outside of a reasonable range or date for the specific time period (July). We determined that these data were sufficiently reliable to describe the number of meals served. In addition to the data FNS requires states to report, some states collect summer meals data at the meal site level and we used such data from the three selected states to address all three objectives. For objective one, to examine the number of meals served and days of operation at each summer meals site, we analyzed site-level data for 2 months from summer 2016, including the month with the largest number of SFSP meals served in each selected state: Arizona (June and July 2016), Illinois (July and August 2016), and Massachusetts (July and August 2016). Each state also provided us with data on the number and types of meals served at each SFSP site, the site location, and the duration of time each site operated over the summer. Using the data provided by the states, we calculated the average daily attendance (ADA) for each meal site based on FNS’s instructions and examined the variation in ADA across sites and months. For our second objective on other programs, these selected states provided similar site level data for the state’s Seamless Summer Option sites, if applicable. We assessed the reliability of these data by (1) performing electronic testing of relevant data elements, (2) reviewing existing information about the data and the system that produced them, and (3) interviewing agency officials knowledgeable about the data. We determined that the data were sufficiently reliable for the purposes of this report. For both our second objective on other programs and third objective about challenges in providing summer meals to children, we also examined meal site availability in the three selected states by mapping the locations of meal sites. On the maps, we included fiscal year 2016 area eligibility data from FNS’s Capacity Builder mapping tool, as provided by FNS. The site area eligibility data from FNS’s Capacity Builder is based on the U.S. Census Bureau’s 5-Year American Community Survey (ACS) estimates of children ages 0-12 and 0-18 eligible for free and reduced-price meals by Census block group and tract. According to FNS officials, FNS obtains 5-Year ACS estimates annually from the U.S. Census Bureau and updates its site area eligibility in the Capacity Builder accordingly. For fiscal year 2016, FNS used 2009- 2013 ACS data to identify and include site area eligibility in its Capacity Builder. Survey To help inform all of our research objectives, we conducted a survey of the state agencies that oversee the SFSP in the 50 states and the District of Columbia. We administered our web-based survey between August and October 2017 and received 100 percent response rate. The survey included questions about participation in the SFSP, factors critical to the overall success of the SFSP, outreach efforts, federal technical assistance, barriers and challenges in providing summer meals, alternative summer feeding models, the NSLP’s Seamless Summer Option and the federal Summer Electronic Benefit Transfer for Children demonstration, and nonfederal programs that provide children of low- income households with meals during the summer months. The survey also requested data on SFSP sites participating in the program in fiscal year 2016 and the method state agencies used to calculate ADA in SFSP on the FNS-418 form in fiscal year 2016. Because this was not a sample survey, there are no sampling errors. However, the practical difficulties of conducting any survey may introduce nonsampling errors, such as variations in how respondents interpret questions and their willingness to offer accurate responses. We took steps to minimize nonsampling errors, including pretesting draft instruments and using a web-based administration system. Specifically, during survey development, we pretested draft instruments with SFSP staff from four states (Michigan, New Mexico, North Carolina, and South Dakota) in May 2017. We selected the pretest states based on information provided by officials from FNS’s regional offices and national organizations involved in summer meals about state administration of summer meals programs, with the goal of selecting a group of states with varied experiences. In the pretests, we were generally interested in the clarity, precision, and objectivity of the questions, as well as the flow and layout of the survey. For example, we wanted to ensure definitions used in the surveys were clear and known to the respondents, categories provided in close-ended questions were complete and exclusive, and the ordering of survey sections and the questions within each section were appropriate. We revised the final survey based on pretest results. Another step we took to minimize nonsampling errors was using a web-based survey. Allowing respondents to enter their responses directly into an electronic instrument created a record for each respondent in a data file and eliminated the need for and the errors associated with a manual data entry process. We did not fully validate specific information that states reported through our survey. Site Visits To help inform all of our objectives and gather information about the SFSP directly at the local-level, we conducted 30 site visits in three states: Arizona (12 sites), Illinois (8 sites), and Massachusetts (10 sites) between June and July 2017, and interviewed organizations involved with the SFSP in each site visit state. We used U.S. Census Bureau data to select states and local areas within those states based on a high proportion of children in poverty, a mix of urban and rural locations, as well as a mix of sponsor and site type and diverse locations. We visited a wide variety of site locations including, but not limited to, schools, parks, community recreation areas, and libraries. At each SFSP site, we gathered information on local level factors related to SFSP participation and administration by interviewing the organization sponsoring the site, the site operators and staff, and those participating at the site using semi-structured questions. While interviewing SFSP sponsor organizations, we collected information on the sponsors’ roles in the SFSP, characteristics of the sites the organizations sponsored, outreach efforts, any challenges or barriers to SFSP administration and any efforts to address such challenges, relationships with the state agencies that administer the SFSP, relationships with FNS (national and regional offices), and the availability of nonfederally funded programs that provide meals to low-income children over the summer. During the interviews with site operators and staff, we collected information about site operation (e.g., site operating days, meals offered, etc.), any challenges to providing SFSP meals to children and any efforts to address such challenges, outreach efforts, and the proximity of the next closest meal site. The information we collected from those participating at the sites included their perspectives on the SFSP food, site food consumption habits, ease of travel to the site, and access to other SFSP sites. At each site, we made observations as to how the food was provided to the children, food consumption and waste, the approximate age range of the children being served, and availability of programs or activities (e.g., recreational sports). Using semi-structured questions, we also interviewed the state agencies responsible for administering the SFSP in the site visit states to gather further information on how the SFSP is administered in each state, statewide participation in the program, related data collection activities, any challenges to administering the program and any efforts to address such challenges, related outreach efforts, alternative meal delivery models being employed by SFSP sponsors, FNS guidance or technical assistance, and the availability of nonfederally funded programs that provide meals to low-income children over the summer. Appendix II: Select Questions and Responses from GAO’s Summer Food Service Program (SFSP) Survey Prompt For states that indicated there were other challenge(s), we provided an open-ended question that requested a description of the challenge(s) and 14 states provided descriptions of other challenges, not shown here. Prompt Prompt For states that indicated there were other challenge(s), we provided an open-ended question that requested a description of the challenge(s) and 8 states provided descriptions of other challenges, not shown here. Appendix III: Area Eligibility and Summer Meal Sites in Selected States Appendix IV: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Rachel Frisk (Assistant Director), Claudine Pauselli (Analyst-in-Charge), Melissa Jaynes, and Matthew Nattinger made key contributions to this report. Also contributing to this report were Susan Aschoff, Sarah Cornetto, Ying Long, Jean McSween, Mimi Nguyen, Almeta Spencer, and Ashanta Williams.

Why GAO Did This Study The SFSP, a federal nutrition assistance program, is intended to provide food to children in low-income areas during periods when area schools are closed for vacation. In the last decade, federal expenditures for SFSP have increased as the program has expanded, according to USDA data. GAO was asked to review the SFSP. This report examines (1) what is known about SFSP participation, (2) other programs that help feed low-income children over the summer, and (3) challenges, if any, in providing summer meals to children and the extent to which USDA provides assistance to address these challenges. GAO reviewed relevant federal laws, regulations, and guidance; analyzed USDA's SFSP data for fiscal years 2007 through 2016; surveyed state agencies responsible for administering the SFSP in 50 states and the District of Columbia; visited a nongeneralizable group of 3 states and 30 meal sites, selected based on Census data on child poverty rates and urban and rural locations; analyzed meal site data from the 3 states; and interviewed USDA, state and national organization officials, and SFSP providers, including sponsors and site operators. What GAO Found Nationwide, the total number of meals served to children in low-income areas through the Summer Food Service Program (SFSP) increased from 113 to 149 million (about 32 percent) from fiscal year 2007 through 2016. The U.S. Department of Agriculture (USDA) directs states to use the number of meals served, along with other data, to estimate the number of children participating in the SFSP. However, participation estimates have been calculated inconsistently from state to state and year to year. In 2017, USDA took steps to improve the consistency of participation estimates, noting they are critical for informing program implementation and strategic planning. However, GAO determined that the method USDA directs states to use will continue to provide unreliable estimates of participation, hindering USDA's ability to use them for these purposes. Other federal and nonfederal programs help feed low-income children over the summer to some extent, according to states GAO surveyed and SFSP providers and others GAO interviewed. For example, in July 2016, USDA data indicate about 26 million meals were served through a separate federal program that allows school meal providers to serve summer meals. Some children also received summer meals through nonfederal programs operated by faith-based organizations and foodbanks, though GAO's state survey and interviews with providers and national organizations indicate the reach of such efforts is limited. States and SFSP providers reported challenges with meal sites, participation, and program administration; USDA has taken steps to address these areas. Specifically, in GAO's survey, a majority of states reported challenges with availability and awareness of meal sites, as well as limited program participation and administrative capacity. National, state, and local officials have taken steps to address these issues, such as increasing outreach and offering activities to attract participation. In addition, 17 states in GAO's survey and providers in the states GAO visited reported a challenge with ensuring meal sites are in safe locations. To address this safety issue, USDA has granted some states and sponsors flexibility from the requirement that children consume meals on-site. However, USDA has not broadly communicated the circumstances it considers when granting this flexibility. Further, some states and sponsors that have requested this flexibility reported difficulty obtaining data to show these circumstances exist, hampering their ability to ensure safe meal delivery. What GAO Recommends GAO is making four recommendations, including that USDA improve estimates of children's participation in SFSP and communicate the circumstances it considers when granting flexibilities to ensure safe meal delivery. USDA generally agreed with GAO's recommendations.

gao_GAO-18-533

gao_GAO-18-533_0

Background NSF relies on two programs for bringing rotators into the agency: (1) the IPA program and (2) the VSEE program. The Office of Personnel Management develops policies on agencies’ use of the IPA program and promulgates program regulations. Rotators in NSF’s IPA and VSEE programs differ in key respects, including their employment status and compensation. IPA rotators. NSF enters into written agreements with rotators’ home institutions for all IPA assignments. The agreements detail rotators’ salaries and health, retirement, and other fringe benefits at their home institutions, as well as the cost-sharing amounts NSF and home institutions are to pay during rotators’ assignments. NSF reimburses its cost-sharing amounts to home institutions, which continue to pay rotators’ full salaries and benefits. NSF does not cap the salaries of IPA rotators; as a result, IPA rotators may receive salaries that exceed the maximum federal salary for the position they hold at NSF. In contrast, if an IPA rotator’s salary is less than the minimum federal salary for the position, NSF will supplement the salary to the minimum rate. VSEE rotators. NSF appoints VSEE rotators as federal employees on a nonpaid leave of absence from their home institutions. VSEE rotators receive their salaries directly from NSF but are not eligible for certain federal benefits, such as retirement; instead, NSF reimburses home institutions for the employer’s share of retirement, life insurance, and health benefits that would otherwise be discontinued. NSF’s policy is to set salaries for VSEE rotators that are generally comparable to the salaries the rotators would receive at their home institutions. In setting salaries, NSF also takes into account other sources of income, such as consulting, and allows for locality pay adjustments applicable to employees in the Washington, D.C., metropolitan area. However, because VSEE rotators are federal employees, NSF caps their salaries at the federal maximum for the position they hold at NSF. Both IPA and VSEE rotators are eligible for certain other types of reimbursement. In particular, rotators have the option of having NSF pay their moving expenses to and from Washington, D.C., or receiving per diem allowances in accordance with federal travel regulations for up to 2 years. In addition, NSF may reimburse rotators for travel-related expenses related to their participation in NSF’s Independent Research and Development program, which enables NSF staff to maintain their involvement with their professional research and research-related activities at their home institutions. Table 1 shows additional information on IPA and VSEE rotator expenses. Rotators are generally assigned to one of NSF’s seven directorates that support science and engineering research and education (see table 2). Each directorate is headed by an assistant director and deputy assistant director. Directorates are further subdivided into divisions, offices, or sections. Each division is headed by a division director and typically a deputy division director, and each office is headed by an office director and typically a deputy office director. All these positions are executive positions at NSF. At the staff level, NSF uses program directors—subject matter experts in the scientific areas they manage—to conduct reviews of proposals and recommend which projects the agency should fund. With an annual budget of about $7.5 billion, NSF funds approximately 24 percent of all federally supported basic research conducted by colleges and universities in the United States. In 2016, NSF established the Steering Committee for Policy and Oversight of the IPA Program. The steering committee serves as the primary body for considering policy on NSF’s use of IPA rotators and overseeing common approaches to budgeting and implementation of the IPA program. The committee’s membership includes NSF’s chief human capital officer, who serves as the chair, and several other NSF officials. The steering committee has established strategic principles for management of the IPA program. These principles include maintaining a balance between IPA rotators and federal staff and a commitment to ongoing improvement of the program. NSF officials told us that there is no similar steering committee for overseeing VSEE rotators. Instead, each VSEE rotator is individually overseen by his or her respective supervisor. For the agency as a whole, NSF’s Office of Information and Resource Management and its Division of Human Resource Management conduct human capital management. NSF officials stated that the head of the Office of Information and Resource Management serves as the Chief Human Capital Officer and develops and oversees NSF’s human capital approaches and strategies. These officials also told us that the Deputy Chief Human Capital Officer serves as the division director of Human Resource Management and is responsible for administering the division’s day-to-day operations. The Division of Human Resource Management administers the agency’s human capital policies as set forth in NSF’s personnel manual. NSF Maintained a Relatively Stable Number and Cost of Rotators and Used Them in Executive and Program Director Positions The numbers of rotators and their costs to NSF in proportion to other staff have remained relatively stable. Most rotators were IPA rotators, and were used in both executive and program director (staff-level) positions. NSF generally used VSEE rotators in program director positions. Most Rotators Were IPA Rotators, Comprising About 12 Percent of NSF’s Workforce and 17 Percent of Staff Costs on Average in Fiscal Years 2008-2017 Most rotators at NSF were IPA rotators, and the proportion of rotators relative to other staff has remained relatively stable over time (see fig. 1). During the 10-year period we reviewed, from fiscal year 2008 through fiscal year 2017, IPA and VSEE rotators comprised about 12 percent and about 3 percent, respectively, of NSF’s total workforce; and the number of IPA rotators ranged from 162 to 190 (about 11 to 12 percent of total staff), and the number of VSEE rotators ranged from 22 to 52 (about 1 to 3 percent of total staff). NSF Primarily Used IPA Rotators in Executive and Program Director Positions and VSEE Rotators in Program Director Positions NSF primarily used rotators across its seven scientific directorates, using IPA rotators in executive and program director positions and VSEE rotators in program director positions. The agency used rotators in these positions alongside NSF’s permanent staff to perform day-to-day agency operations, including managing the agency’s merit review process for determining which projects to fund. Use of IPA Rotators in Executive Positions NSF used IPA rotators in executive positions such as assistant director. According to agency officials, individuals in executive positions at NSF are responsible for setting the direction for the scientific area they are assigned, leading scientific and technical matters, establishing an organizational culture, overseeing outreach and collaboration with NSF stakeholders, and contributing to NSF and national policy development and implementation. For example, an executive IPA rotator that we interviewed told us that he emphasized forming partnerships with industry when setting the direction for his directorate, including issuing joint solicitations for research proposals with industry partners. In addition, according to NSF officials, individuals in executive positions provide guidance and team management for staff. The proportion of IPA rotators to federal employees in executive positions within NSF’s seven scientific directorates and other staff offices has generally increased since fiscal year 2012. As shown in figure 4, from fiscal year 2008 through fiscal year 2017, the number and proportion of executive positions filled by IPA rotators ranged from 18 of 98 (about 18 percent) in 2008 to 30 of 108 (about 28 percent) in fiscal year 2016. In November 2017, IPA rotators filled 29 of 88 (about 33 percent) executive positions within NSF’s seven scientific directorates. At that time, the proportion of executive positions filled by IPA rotators varied among directorates, as shown in table 3. For example, IPA rotators filled 4 of 8 (50 percent) of the executive positions in the Directorate for Social, Behavioral, and Economic Sciences and 2 of 14 (about 14 percent) of the executive positions in the Directorate for Mathematical and Physical Sciences. According to NSF officials, NSF often pairs IPA rotators and federal employees at the executive level so that each can benefit from the other’s experience and perspective. For example, in all but one directorate, an IPA rotator filled the assistant director position and a federal employee filled the corresponding deputy assistant director position. Two NSF executives we interviewed, including an IPA rotator and a federal employee, commented positively on the pairing of IPA rotators and federal employees at the executive level. For example, they said that rotators maintain close ties to the research community and federal employees may have more experience with NSF’s institutional history. One NSF executive told us that IPA rotators help keep the agency at the forefront of science because they have deep ties with the research community and regularly publish their own research. Additionally, a federal program director we interviewed told us that in one previous instance in which an IPA rotator filled an executive position without being paired with a federal employee, the rotator’s lack of institutional knowledge of NSF and the steep learning curve for the position caused inefficiencies during the rotator’s first year at NSF. The agency, however, does not require pairing IPA rotators and federal employees at the executive level, according to NSF officials. For example, in November 2017, IPA rotators filled both the division director and deputy division director positions in the Division of Behavioral and Cognitive Science and Division of Undergraduate Education. In our interviews with a nongeneralizable sample of NSF employees and rotators, we found mixed perceptions about the effect of NSF’s use of IPA rotators on opportunities for advancement for permanent employees. For example, in response to a question about this effect, one permanent NSF employee told us that she advanced to an executive position and that opportunities exist for advancement within the agency. In contrast, another NSF employee we interviewed told us that she did not feel there were opportunities for advancement because, in her view, executive vacancies created by the departure of rotators were exclusively filled with other rotators. NSF officials said that the agency has no policy that restricts repeatedly filling certain executive positions with rotators and that such a situation is a common practice. Nevertheless, NSF officials told us 32 of the 88 executives (about 36 percent) in NSF’s seven scientific directorates in November 2017 had held staff-level positions within the agency before becoming executives. Use of IPA and VSEE Rotators as Program Directors NSF uses both IPA and VSEE rotators in program director positions, which are staff-level positions. In fiscal year 2016, NSF had a total of 506 program directors, including 139 IPA rotators (about 27 percent) and 39 VSEE rotators (about 8 percent). According to NSF officials, program directors are responsible for conducting long-range planning and developing budgets for the areas of science represented by their program and for administrating the merit review process. In particular, IPA and VSEE rotators who serve as program directors help determine the projects that NSF funds. To do so, they review proposals, identify experts in their field to serve as external reviewers, and make funding recommendations to their respective division directors. NSF officials told us that, similar to the pairing of IPA rotators and federal employees at the executive level, permanent and rotating program directors frequently work together on a shared program so that each can benefit from the other’s experience and perspective. For example, a rotating program director we interviewed told us that she worked under the guidance of a program lead, who is typically a permanent employee. Another rotating program director told us that NSF’s permanent federal employees are good at training incoming rotators. NSF Adopted Rotator Program Cost- Management Strategies to Achieve the Greatest Savings with the Least Harm to Recruitment, but Results Are Unknown Beginning in fiscal year 2017, NSF adopted rotator program cost- management strategies expected to achieve the greatest savings with the least harm to recruitment, but NSF officials said it is too soon to determine the full results because these new strategies are being phased in for new IPA agreements only. NSF considered other strategies to manage rotator costs, but it did not adopt them, generally because NSF anticipated negative effects on rotator recruitment or because it estimated the resulting cost savings would be small. NSF Adopted Three Strategies to Manage IPA Rotator Costs in Fiscal Year 2017 and Has Not Yet Determined Their Full Results NSF has adopted three strategies to manage rotators’ costs in fiscal year 2017, but, NSF officials said it is too soon to determine the full results because these new strategies are being phased in for new IPA agreements only. All three of these strategies relate to IPA rotators; NSF officials told us that they have not considered or adopted any cost- management strategies related to VSEE rotators. The officials explained that any such strategies could affect NSF’s entire federal workforce because VSEE rotators are federal employees. The three strategies are: (1) obtaining a minimum 10 percent cost-share from each IPA rotator’s home institution, (2) limiting IPA rotators’ paid trips to their home institutions to 12 per year, and (3) no longer reimbursing IPA rotators for consulting income that they forgo while at NSF. NSF officials told us they expect to issue a report with the results of evaluations of all three strategies in December 2018. Cost-Sharing Pilot Program In October 2016, NSF implemented a cost-sharing pilot program that requires institutions covered by the program—those who entered into negotiations for new IPA agreements in fiscal year 2017—to pay for at least 10 percent of the IPA rotators’ salaries and fringe benefits. Implementing this cost-management strategy, and the other strategies that NSF adopted, was consistent with recommendations from NSF’s steering committee for oversight of IPA rotators. This cost-management strategy targeted NSF’s costs for IPA rotators’ salary and fringe benefits, which constitute the largest component of IPA rotators’ costs. For example, these costs were about $34.7 million, or about 89 percent of IPA rotator costs in fiscal year 2017. Previously, according to NSF officials, the agency requested an optional cost-share amount of 15 percent from rotators’ home institutions, but it typically received less because of variations in the amounts that home institutions provided. According to an October 2016 report from the task force on fiscal oversight, NSF decided on 10 percent for the cost-sharing pilot program because, historically, few home institutions provided the full 15 percent and NSF believed a requirement of 10 percent would not significantly affect its ability to recruit and hire IPA rotators. If a home institution is unable to provide the full 10 percent, the institution may request that NSF waive the cost-sharing requirement. According to NSF officials, such requests must be signed by a senior administrator at the rotator’s home institution and include the rationale for not being able to provide the required amount, the financial impact on the institution if it were to provide the full 10 percent, and associated documentation, among other things. Changes made in implementing this strategy, and the other strategies that NSF adopted, applied to new IPA agreements made in fiscal year 2017. These changes did not apply to IPA rotators with agreements made prior to 2017—even if those agreements are subsequently extended or renewed—or that were being negotiated at the time of the policy change, provided that the rotators’ appointment memoranda were already being reviewed by NSF’s Division of Human Resource Management. NSF officials told us that as of March 2018, the agency had not conducted full evaluations of this strategy or the other strategies because it was too soon to determine their full effects and NSF had not yet collected enough data to do so. Instead, NSF issued reports in January and March 2018 containing its preliminary analyses. In general, these preliminary reports found that the cost-management strategies resulted in savings to NSF. Similarly, our analysis of data from NSF found that cost sharing as a percentage of IPA rotators’ salary and fringe benefits increased from about 7 percent in fiscal year 2016 to about 8 percent in fiscal year 2017. NSF officials told us that of the 55 IPA rotators who were subject to the cost-sharing requirement in fiscal year 2017: the home institutions for 54 rotators met or exceeded the 10 percent cost-share requirement, and of those, 16 exceeded the cost-share requirement; and the home institution for 1 rotator did not cost-share because the rotator was from a Federally Funded Research and Development Center and NSF waived the cost-share requirement because cost- sharing would not decrease the overall federal cost. In November 2017, NSF decided to extend the cost-sharing pilot through at least the end of fiscal year 2018, to ensure a full evaluation could be conducted. In particular, NSF officials told us that they need more data and experience with this pilot program to better understand its effects, such as the ability to recruit potential IPA rotators. For example, one IPA rotator that we interviewed expressed concern with the cost-sharing requirement’s potential effect on small or publicly funded universities, which may lack funds to contribute to the cost of an IPA assignment. According to NSF officials, their evaluation will include an analysis of the cost of IPA rotators under the cost-sharing requirement and its effect on the IPA program, including recruitment. Limitation on NSF-Funded Trips to IPA Rotators’ Home Institutions Beginning in fiscal year 2017, for IPA rotators who entered into negotiations for new agreements in that fiscal year, NSF placed a limit of 12 agency-funded trips per year that rotators may take to their home institutions under the Independent Research and Development program. In our analysis of data from NSF, we found that NSF’s costs for IPA rotators under this program decreased from about $1.5 million (about 3 percent of IPA rotator costs) in fiscal year 2016 to $1.1 million (about 3 percent of IPA rotator costs) in fiscal year 2017. NSF officials told us that the new limit applies only to an IPA rotator’s trips to their home institution and does not limit travel to other locations for fieldwork or scientific conferences, among other things. These officials explained that NSF chose not to limit trips to these other locations because they are considered fundamental to IPA rotators’ research and are infrequent—occurring one to three times per year, on average, per IPA rotator. Additionally, rotators are permitted to use annual leave, leave without pay, or flexitime to take trips using non-NSF funds for activities performed on a rotator’s own time. In adopting this cost-management strategy, NSF sought to balance the benefits of IPA rotators’ travel with the travel costs. According to the Task Force on Fiscal Oversight’s October 2016 report, NSF’s support for travel benefits the agency by providing a way for program directors and executives to stay current in their scientific fields, conduct outreach with scientific communities, and provide oversight and stewardship of NSF’s programs and awards. NSF officials told us that the agency sought to control travel costs under the Independent Research and Development program by setting a reasonable limit to NSF-funded trips that would cause the least harm to rotators’ research so as not to discourage them from coming to NSF. As a result, NSF decided on a maximum of 12 trips per year under this program because, historically, more than 80 percent of the IPA rotator participants traveled to their home institution less than once per month. Elimination of Reimbursement for Lost Consulting Income In fiscal year 2017, for IPA rotators who entered into new agreements in that fiscal year, NSF ended reimbursements for consulting income that the rotators forgo as a result of their assignment to NSF. Previously, when an IPA rotator discontinued consulting activities during an IPA assignment, NSF would reimburse the rotator up to $10,000 a year. IPA rotators who entered into negotiations or agreements with NSF prior to this change may still receive this reimbursement. In fiscal year 2017, NSF’s cost for lost consulting reimbursements to IPA rotators was $150,000. This amount represented a decrease of about $160,000, or about 52 percent, from fiscal year 2016. NSF made this change because it determined that doing so would not negatively affect the IPA program. In particular, NSF found that other federal science agencies typically did not reimburse IPA rotators for lost consulting income and it concluded that IPA rotators typically do not expect NSF to offer reimbursement. NSF Did Not Adopt Certain Cost-Management Strategies It Considered Because of Small Cost Savings or Potential Negative Effects In addition to the three adopted strategies, NSF’s Task Force on Fiscal Oversight identified other potential cost management strategies for the IPA program. The task force reviewed various data on the costs that make up the IPA program, such as the number of IPA rotators who received a particular form of compensation or who would be affected by the potential strategies. In addition, the task force took into account anecdotal and other evidence on how IPA rotators might react to the strategies. Using input from the task force, NSF opted against the other potential strategies because it either (1) expected the resulting cost savings to be small or (2) anticipated potential negative effects from implementing them, such as increased difficulty in hiring IPA rotators. These potential cost-management strategies primarily related to IPA rotator compensation, as described below. Capping IPA rotators’ salaries. NSF decided against establishing a salary cap for IPA rotators at various levels between about $185,000 and $240,000 annually. The task force found that salary caps at lower levels would have greater cost savings because of the higher number of individuals covered by the cap, but that the caps would also pose a significant risk to NSF’s ability to recruit IPA rotators. In particular, the task force found that salary caps at lower levels would disproportionately affect IPA rotators in two of its directorates—the Directorate for Computer and Information Science and Engineering and the Directorate for Engineering—because of the higher salaries of individuals in positions associated with those fields. As a result, the task force recommended that NSF first assess the effects of its cost-sharing pilot program before proceeding with any cap on IPA rotators’ salaries. Reducing or eliminating IPA rotators’ supplemental pay. NSF decided against reducing or eliminating the supplemental pay that IPA rotators receive when their salary at their home institution is below the minimum for their NSF position. In fiscal year 2017, NSF’s cost for IPA rotators’ supplemental pay was $1.0 million (about 3 percent of IPA rotator costs). The task force recommended against this potential cost-management strategy because it would disproportionately affect IPA rotators in two of its directorates—the Directorate for Biological Sciences and the Directorate for Geosciences. In addition, the task force expected that any cost savings associated with this strategy would be small. Reducing IPA rotators’ per diem payments. NSF decided against reducing or eliminating per diem payments for lodging (excluding taxes), meals, and incidental expenses incurred during the length of rotators’ assignments. In fiscal year 2017, NSF’s cost for per diem payments was $3.1 million (about 8 percent of IPA costs). The task force concluded, based on its analysis of per diem costs and anecdotal evidence, that many IPA rotators would opt to depart NSF if NSF did not provide per diem payments. As a result, the task force recommended against this strategy. NSF Has Not Developed a Workforce Strategy for Using Rotators or Fully Evaluated Rotator Program Results NSF Has Not Developed an Agency-Wide Workforce Strategy for Balancing Rotators and Federal Staff As of June 2018, NSF had not developed an agency-wide workforce strategy for using rotators, as its IPA program steering committee recommended. In addition, NSF has not fully evaluated or developed plans to evaluate both IPA and VSEE rotator program results in terms of progress toward NSF’s human capital goals or programmatic results. As of June 2018, NSF had not developed an agency-wide workforce strategy that includes use of rotators, as NSF’s IPA program steering committee had recommended. In an August 2016 report on the IPA program, the steering committee stated that NSF did not have an agency- wide workforce strategy; instead, each directorate made decisions on its own about when and how to use IPA rotators in executive and program director positions. According to the report, an agency-wide framework would enable NSF to ensure an optimal balance of federal and rotator executives and program directors, which is a strategic principle that the steering committee developed for the IPA program. In February 2017, the committee issued an internal report to agency leadership that recommended expanding what was originally envisioned as a workforce strategy for the IPA program into a comprehensive agency-wide workforce strategy. The report stated that expanding the scope of the workforce strategy would have the greatest impact across the agency and would help NSF leadership in making strategic human capital decisions. The report outlined a process for developing a workforce strategy with various steps, including the following: Job analyses. The report recommended job analyses to review the roles and responsibilities of executive and staff-level positions and to identify the skills and capabilities required for successful performance of the work. According to the report, the steering committee’s working group for developing a workforce strategy found, based on its initial efforts to review position descriptions and roles and responsibilities, that some functions may be better served if performed by permanent federal employees and other functions by rotators. However, the working group concluded that NSF should obtain additional input and evidence before initiating large-scale changes in its workforce. Analysis of workforce gaps and surpluses. The report stated that identifying gaps and surpluses in the demand and supply for federal and rotator scientific staff would inform opportunities to optimize recruitment and retention efforts. The report recommended separate analyses for executive and scientific staff- level positions. Development of strategies to close workforce gaps and address surpluses. According to the steering committee’s report, examples of strategies include succession planning and rebalancing the mix of permanent federal staff and rotators to ensure an optimal workforce with the skills, experience, and capabilities to accomplish NSF’s science-related work. According to NSF officials, the agency’s Division of Human Resource Management was responsible for implementing the steering committee’s recommendation. In particular, it undertook an effort to work with senior leadership to develop a broad strategic workforce plan for the agency. However, in June 2018, NSF officials told us that they shifted their focus from developing a separate workforce strategy in order to focus instead on (1) development of a human capital operating plan, which agencies are required to develop and approve annually, and update as needed, under OPM regulations that went into effect on April 11, 2017; and (2) an Office of Management and Budget (OMB) memorandum issued in April 2017 directing agency heads to develop reform plans that identify ways to improve the efficiency, effectiveness, and accountability of their respective agencies. The NSF officials explained that they recognized the value in having a workforce strategy, but they did not consider it appropriate for the Division of Human Resource Management to develop a workforce strategy at the same time that the agency was completing the OPM and OMB plans. NSF did not specify how its efforts to complete the OPM and OMB plans would address the need the steering committee identified for an agency- wide framework that would enable NSF to ensure an optimal balance of federal and rotator executives and program directors. In particular, NSF’s human capital operating plan, which it approved in April 2018, does not discuss NSF’s use of rotators or include information on balancing the agency’s use of rotators with permanent staff. Furthermore, NSF has not yet determined how it will address its use of rotators as part of its agency reform plan. In particular, NSF officials told us in June 2018 that they may address the agency’s use of rotators under the workforce focus area of its reform plan, but that they were only just beginning to identify and select initiatives under this focus area and that these initiatives have not yet been finalized. The process the NSF steering committee laid out in its internal report, when implemented, would align with two key principles GAO has identified for effective strategic workforce planning. Specifically, it would align with the principles of (1) determining the skills and competencies that are critical to successfully achieving missions and goals, and (2) developing human capital strategies to address gaps and enable the contribution of critical skills and competencies needed for mission success. By incorporating the NSF’s steering committee’s recommendation for a workforce strategy—and the process outlined by the steering committee for developing this strategy—into its human capital operating plan or agency reform plan, NSF could better manage its use of rotators and balance them with its permanent staff. NSF Has Not Fully Evaluated the Rotator Program’s Progress toward Human Capital Goals or Their Programmatic Results We have previously found that high-performing organizations recognize the fundamental importance of measuring both the outcomes of human capital strategies and how these outcomes have helped the organizations accomplish their missions and programmatic goals. However, as of May 2018, NSF had not fully evaluated and did not have plans to evaluate the results of its IPA and VSEE rotator programs in terms of progress toward human capital goals and the contributions the programs made toward achieving programmatic results. One of GAO’s key principles for effective strategic workforce planning states that agencies should monitor and evaluate progress toward the agencies’ human capital goals and the contribution that human capital results have made toward achieving programmatic results. In particular, we previously found that evaluation activities can improve the effectiveness of workforce strategies by identifying shortfalls in performance and other improvement opportunities. OPM also requires agencies to develop a human capital operating plan that will support the evaluation of the agency’s human capital strategies. In March 2014, NSF published a summary of the results of focus groups with IPA rotators and their supervisors. This summary outlined benefits and challenges of the program from the perspectives of both groups, such as the benefit of bringing fresh perspective and new ideas to NSF and the challenge of recruiting and retaining qualified IPA rotators. However, the summary did not provide the agency’s assessment of progress towards programmatic results and human capital goals. For example, it summarized the benefits of the program from the standpoint of rotators and did not provide NSF’s assessment of how individual IPA rotators or the program as a whole contributed to NSF’s scientific mission. In addition, the summary did not provide an assessment of the extent to which the current workforce balance of federal and rotator executives and program directors is aligned with NSF’s work. In our semistructured interviews with federal staff and rotators in executive and staff-level positions at NSF, most were comfortable with the current balance, but three individuals raised concerns about the use of rotators in executive positions, suggesting that NSF could benefit from further analysis of its balance of rotators and federal staff. In April 2018, NSF adopted its human capital operating plan which identifies specific, short-term actions that the agency will take to achieve its human capital goals. In its plan, NSF identified strategies derived from NSF’s commitment to ongoing improvement, such as reviewing and realigning its workforce to meet future needs. Also, NSF’s process for developing a workforce strategy, outlined in the steering committee’s February 2017 internal report, included recommendations to conduct an assessment of the outcomes of workforce strategies and the impact of these outcomes on helping NSF accomplish its scientific mission and related programmatic goals. However, plans for this assessment did not include an evaluation of the agency’s rotator programs. Moreover, neither the steering committee’s February 2017 internal report nor NSF’s April 2018 report committed to conducting such an evaluation or specified how assessments described in its reports would address NSF’s rotator programs. For example, neither report specified how NSF would evaluate the extent to which the rotator programs have achieved NSF’s objectives, which we identified through our review of NSF documentation and interviews with NSF officials. These objectives include: bringing fresh perspectives from across the country and across all fields of science and engineering supported by NSF; helping influence new directions for research in science, engineering, and education, including emerging interdisciplinary fields; providing scientific leadership and management of NSF’s research and education programs; and providing opportunities for researchers to gain first-hand knowledge of the philosophy and mechanisms of federal support for research and bring this knowledge back to their home institutions. According to NSF officials, the agency has not separately evaluated the results of its rotator programs in part because rotators are blended into its permanent federal workforce, making it difficult to evaluate the results of its rotator programs separately from those of its overall workforce. In our December 2003 report on key principles for effective strategic workforce planning, we found that federal agencies in general have experienced difficulties in defining practical and meaningful measures that assess the effects human capital strategies have on programmatic results. However, without an evaluation of the extent of the rotator programs’ contributions toward NSF’s human capital goals or programmatic results, NSF is limited in its ability to demonstrate the programs’ benefits to external stakeholders, such as the Congress, and to adjust the programs, if warranted. Such adjustments could include increasing or decreasing the use of rotators overall or in certain types of positions, such as executive or staff-level positions. Conclusions In recent years, NSF has recognized the need to think more strategically about its use of rotators and has taken positive steps to manage its rotator programs. For example, beginning in fiscal year 2017, NSF adopted several strategies to manage the cost of rotators. However, as of June 2018, NSF had decided against developing a separate agency-wide strategy for balancing its use of IPA rotators and federal staff, as NSF’s steering committee for the IPA program recommended in February 2017. NSF officials said that they recognized the value in having a workforce strategy but wanted to focus instead on addressing OPM and OMB requirements related to workforce planning. By following through on the steering committee’s recommendation for a workforce strategy, NSF could better manage its use of rotators and balance them with its permanent staff. Moreover, as of June 2018, NSF had not fully evaluated the results of the rotator programs, as called for by key principles for effective strategic workforce planning. NSF officials told us they have not done so, in part, because rotators are blended into NSF’s permanent federal workforce, making it difficult to evaluate the results of its rotator program separately from those of its overall workforce. However, without an evaluation of the extent of the rotator programs’ contributions toward NSF’s human capital goals or programmatic results, NSF is limited in its ability to demonstrate the programs’ benefits to external stakeholders, such as the Congress, and to adjust the programs, if warranted. Recommendations for Executive Action We are making the following two recommendations to NSF: The NSF Director of Human Resource Management should complete the development of an agency-wide workforce strategy for balancing the agency’s use of IPA and VSEE rotators with permanent staff as part of NSF’s current agency reform planning efforts or updates to its human capital operating plan. (Recommendation 1) The NSF Director of Human Resource Management should evaluate the contributions of the IPA and VSEE rotator programs toward NSF’s human capital goals and the contributions the programs have made toward achieving programmatic results. (Recommendation 2) Agency Comments We provided a draft of this report to NSF for comment. In its written comments, which are reproduced in appendix I, NSF concurred with our recommendations and stated that implementation of the recommendations will enhance efforts to fulfill the agency’s mission and strengthen its workforce. NSF also provided technical comments, which we incorporated as appropriate. We are sending copies to the appropriate Congressional Committees, the Director of the National Science Foundation, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-3841 or neumannj@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix II. Appendix I: Comments from the National Science Foundation Appendix II: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the individual named above, Joseph Cook (Assistant Director), Nkenge Gibson, Kathryn Smith, and Douglas Hunker made key contributions to this report. Also contributing to this report were Antoinette Capaccio, Serena Lo, Timothy Guinane, Cynthia Norris, and Sara Sullivan.

Why GAO Did This Study NSF has identified potential benefits and challenges associated with its use of rotators. Benefits include fresh perspectives and close connections to the scientific community, while challenges include staffing turnover and higher costs for some rotators compared with permanent employees. GAO was asked to review NSF's use and management of the IPA and VSEE rotator programs, among other things. This report examines (1) the number, costs, and uses of NSF rotators for fiscal year 2008 through fiscal year 2017; (2) the strategies NSF has used to manage rotator costs and the results of these efforts; and (3) the extent to which NSF has a workforce strategy for using rotators and has evaluated the results of its rotator programs. GAO analyzed summary-level data on NSF's rotators; reviewed key documents; interviewed NSF officials; conducted semistructured interviews with a nongeneralizable sample of rotators and permanent federal employees selected from different scientific directorates within NSF; and compared NSF's management of the program to key principles for effective strategic workforce planning. What GAO Found The numbers of rotators—outside scientists, engineers, and educators on temporary assignment—at the National Science Foundation (NSF) and their costs in proportion to other staff remained relatively stable in fiscal years 2008 through 2017. Most rotators joined NSF under its Intergovernmental Personnel Act (IPA) mobility program. IPA rotators comprised about 12 percent of NSF's workforce and 17 percent of staff costs on average and were not subject to a federal salary cap. They remain employees of their home institutions, with NSF reimbursing the institutions for most of their salaries and benefits. The remaining rotators are considered temporary federal employees under the Visiting Scientist, Engineer, and Educator (VSEE) program; their salaries could not exceed the federal maximum for their positions. Beginning in fiscal year 2017, NSF adopted IPA rotator program cost management strategies expected to achieve the greatest savings with the least harm to recruitment, but NSF officials said it is too soon to determine the full results. For example, for new IPA rotators who had not yet begun negotiating their assignments, NSF began requiring their home institutions to pay for 10 percent of the rotators' salary and benefits. NSF officials told GAO they expect to issue a report evaluating the strategies in December 2018. NSF's IPA program steering committee recommended developing a workforce strategy for balancing the agency's use of rotators with federal staff, but as of June 2018, NSF had not developed a strategy or fully evaluated the IPA and VSEE rotator programs' results, as called for by GAO's key principles for effective strategic workforce planning. NSF officials said they recognized the value of a workforce strategy but were focusing instead on other workforce planning efforts, and they had not fully evaluated program results in part because rotators are blended into the agency's permanent workforce, making a separate evaluation difficult. Without a workforce strategy and evaluation of results, NSF is limited in its ability to manage and, if warranted, adjust its use of rotators. What GAO Recommends GAO recommends that NSF develop an agency-wide strategy for balancing the agency's use of rotators with permanent staff and evaluate the contributions of its rotator programs toward NSF's human capital goals and programmatic results. NSF agreed with GAO's recommendations.

gao_GAO-18-252

gao_GAO-18-252_0

Background As we have previously reported, 911 services have evolved from basic 911—which provided Americans with a universally recognized emergency number—to Enhanced 911 which also routes calls to the appropriate call center and provides information about the caller’s location and a call back number. NG911 represents the next evolution in 911 services by using IP-based technology to deliver and process 911 traffic. Under NG911, call centers will continue to receive voice calls and location information, but will also be able to accommodate emergency communications from the range of technologies in use today. In addition, NG911 systems provide call centers with enhanced capabilities to route and transfer calls and data, which could improve call centers’ abilities to handle overflow calls and increase information sharing with first responders. 911 Communications Process Generally speaking, 911 communications begin when a caller dials 911 using a landline, wireless, or Voice over Internet Protocol (VoIP) system. Once a 911 caller places an emergency call, a communications provider receives and routes the call to the appropriate call center, along with the caller’s phone number and location (i.e., street address for a landline caller, approximate geographic location for a wireless caller, and the subscriber’s address for VoIP). Calls and data may be routed to 911 call centers using legacy methods (i.e., routing calls across traditional telephone networks) or NG911 methods (i.e., routing calls and other data through IP-networks). Once the call reaches a call center, trained call takers and dispatchers determine the nature of the emergency and dispatch first responders, typically using a variety of equipment and systems, including call handling systems, mapping programs, and computer aided dispatch. Figure 1 illustrates the 911 communications and dispatch process. As illustrated in figure 1, NG911 systems use IP-networks capable of carrying voice plus large amounts of data. These emergency-services networks are typically deployed at the state or regional level with multiple call centers connecting to the network. However, the existence of an IP- network alone does not constitute an NG911 system. As defined by standards developed by the emergency communications community, an NG911 system should have the capability to, among other things: provide a secure environment for emergency communications; acquire and integrate additional data for routing and answering calls; process all types of emergency calls, including multimedia messages; transfer calls with added data to other call centers or first responders. While NG911 systems must possess certain capabilities, it is important to note that states and localities may make decisions about which capabilities they intend to use to best meet their needs. In addition, states and localities have the authority to make decisions about what NG911 equipment, systems, and vendors to use; thus, the configurations of these systems vary. NG911 Implementation According to a panel of experts convened by the National 911 Program, the transition to NG911 may require a variety of technical and operational changes to current 911 systems and processes. For example, technical changes can include upgrades to networks or installing new hardware or software in 911 call centers. Operational changes can include the need for additional training or the development of new policies and procedures (e.g., new procedures for processing or storing multimedia). These technical and operational changes may also have effects on 911 funding and state and local governance structures, which we will discuss in more detail later in this report. According to an FCC advisory body that examined NG911 systems architecture in 2016, while NG911 systems are implemented in a variety of ways at the state or local level, NG911 implementation can occur gradually and in phases. According to this model, NG911 implementation occurs on a continuum that begins with legacy 911 systems and ends with a fully deployed NG911 national end-state where all individual 911 call centers nationwide would be connected. The NG911 implementation model identifies activities that take place as part of the NG911 transition, many of which occur concurrently, such as: planning (e.g., conducting feasibility studies, preparing databases, establishing governance models); acquiring, testing, and implementing NG911 system elements (e.g., establishing an emergency-services IP-network, location-based call routing, processing multimedia); connecting call centers within a jurisdiction (i.e., jurisdictional end- state in which all call centers are fully NG911 operational, supported by agreements, policies, and procedures); and connecting NG911 systems nationwide (i.e., national end-state in which all call centers in the nation are fully NG911 operational, supported by agreements, policies, and procedures). In addition, because 911 services provide an essential function, the implementation of NG911 generally involves using both the legacy system and the NG911 system simultaneously for a period of time, according to the FCC advisory body, to ensure 911 services are not disrupted as new system elements are tested and implemented. State and Local Roles Deploying and operating 911 is the responsibility of 911 authorities at the state and local level. As we have previously reported, all 50 states and the District of Columbia collect—or have authorized local entities to collect—funding for 911 from telephone service subscribers, and methods within each state for collecting funds vary. FCC, as required by statute, reports to Congress annually on the states’ collection and distribution of 911 fees and charges. There are approximately 6,000 call centers nationwide that process 911 calls, often at the county or city level, and these centers can vary greatly in size and technical sophistication. The state and local governance structures that oversee 911 operations also vary by location. For example, we previously reported that some states collect fees or charges for 911 and administer a statewide 911 program. Other states authorize local entities to collect fees or charges for 911 and administer 911 programs at the local level. Still other states use a combination of these approaches. According to a panel of experts convened by the National 911 Program, historically, 911 authority has been coordinated and maintained locally with no requirement to coordinate with other jurisdictions. However, the transition to NG911 enables connection of 911 systems. Thus, as previously mentioned, the NG911 transition may require technological and operational changes, as well as changes to 911 policies and governance responsibilities for states and localities. Federal Roles While deploying and operating 911 is the responsibility of entities at the state and local level, federal agencies—including NHTSA, NTIA, FCC, and DHS—have responsibilities to support state and local implementation, including through facilitating coordination of activities among 911 stakeholders and administering federal grants, for example: NHTSA houses the National 911 Program as part of its Office of Emergency Medical Services (Office of EMS) to provide national leadership and coordination for the NG911 transition throughout the United States, as previously mentioned. According to NHTSA, the fiscal year 2017 budget for the National 911 Program was $2.74 million. Among other activities, which we will discuss later in this report, the National 911 Program surveys states on progress implementing NG911 and reports this survey data annually. FCC issues orders and regulations for 911 service providers on topics relevant to NG911, such as 911 reliability, location accuracy, and text- to-911. FCC also sponsors advisory bodies comprised of government and industry experts that study relevant topics and provide recommendations related to NG911, such as the Task Force on Optimal Public Safety Answering Point Architecture and the Communications, Security, Reliability, and Interoperability Council. While there are no federally mandated time frames for implementing NG911, the Next Generation 911 Advancement Act of 2012 requires specific actions of some federal agencies as outlined in table 1, below. In addition, according to the National 911 Program, as states and localities continue to implement NG911, and begin to explore interconnection with other states’ 911 systems, federal agencies may need to take steps to help ensure state NG911 networks are interoperable and connected. We will discuss actions taken by federal agencies to assist states and localities to implement NG911 later in this report. States and Localities Have Reported Varied Progress in Implementing NG911 and Identified Funding, Technology, and Governance Challenges Reported Progress among States Varies in Implementing Next Generation 911 According to NHTSA’s most recent national survey, state and local progress implementing NG911 varies, and about half of all states reported being in some phase of transition to NG911 in 2015. While a few states are well into statewide implementation, NHTSA officials told us that no state had completely implemented all NG911 functions. Additionally, as of the fall of 2017, none of the selected states we spoke with were processing multimedia—such as images or audio/video recordings—through their 911 systems due to concerns related to privacy, liability, and the ability to store and manage these types of data, among other things. The national survey data, based on responses from 45 states, measured the extent to which NG911 planning and acquisition of NG911 equipment and services were occurring, and the extent to which basic NG911 functions were operational at the state and local levels in 2015. Planning: This measure includes state and local NG911 plans for governance, funding, system components, and operations. In this context, system components refer to an emergency services IP-based network, NG911 software, system and information security, and databases, among other things, according to NHTSA’s survey. In total, 25 of 45 states reported having a state or at least one local NG911 plan in place; conversely, 18 states reported having no NG911 plan in place at either the state or local level—which may indicate they are in the early stages of planning for the NG911 transition or have not yet begun the transition to NG911. Acquisition: These measures identify states or local entities that have defined their NG911 needs and awarded contracts, and then installed and tested acquired NG911 components and services. Twenty-four states reported awarding at least one contract at the state or local level for NG911 components and services. Twenty-three states reported having installed and tested NG911 components and services at either the state or local level. NG911 services: This is a measure of 911 authorities that have some basic, functioning NG911 infrastructure in place. In total, 21 states reported having some level of basic NG911 services in place at the state or local level. Of these 21 states, 10 reported that all 911 authorities within the state were using NG911 technology to process emergency calls. Another 7 of these states reported that 25 percent or less of their state’s 911 authorities were using NG911 technology to process emergency calls. Officials Identified State and Local Funding, Technology, and Governance Challenges to Implementing NG911 Federal officials, industry stakeholders, and state and local 911 officials we interviewed from nine states identified a number of challenges to implementing NG911, including challenges related to funding, evolving technology and operations, and governance. Funding: State and local officials in four of nine selected states identified insufficient funding as one of the challenges they face in implementing NG911. Additionally, FCC, NHTSA, and industry reports noted that state and local financing strategies are generally insufficient to fully implement NG911. Specifically, these reports note that the need to provide new capital for NG911 implementation while simultaneously funding legacy operational costs during the transition can strain state and local funding. Limited funding: Officials in three states told us that their current funding may not be able to support the upfront costs of infrastructure and equipment acquisitions associated with the transition to NG911. Further, officials said they will need to simultaneously fund both the new NG911 and legacy 911 systems currently in operation until the NG911 systems are fully operational. To address these challenges, a Minnesota official told us about how the state leveraged economies of scale to reduce overall costs through cost sharing between multiple call centers and of call centers consolidating operations from 114 to 104 call centers. Additionally, a Virginia official told us that to cover the upfront costs of transitioning to NG911, the state plans to borrow from the state treasury and then repay the treasury with future-year fee collections. Fee diversion: Diversion of fees intended for 911 costs to non-911 activities may affect a state’s or locality’s ability to cover NG911 transition costs and necessitate identifying alternative funding sources. The FCC’s 2016 annual report on 911 fees indicates that for calendar year 2015, all but two of the states that responded to FCC’s 911 fee survey affirmed that their state or jurisdiction collects fees from phone users to support or implement 911 services. State and local authorities also determine how these 911 fees can be used. FCC’s report also indicated that eight states and Puerto Rico reported diverting a total of more than $220 million (or approximately 8.4 percent) of 911 fees collected to non-911 purposes. Some of these diverted funds were directed to other public safety programs, and others were diverted to either non-public safety or unspecified purposes. According to one state official, had it not been for 911 fees being diverted to non-911 purposes, funding would have been sufficient to cover the NG911 transition without having to go to the state legislature for additional funding. However, officials in the other eight selected states told us that either fee diversion was not an issue in their state or that the diversion of funds had not affected their state’s ability to implement NG911. Evolving technology and operations: Officials in eight states told us that the retirement of legacy infrastructure and the transition to IP-based systems introduces new technical and operational challenges for call centers and states, as well as for equipment and service providers. Interoperability: Officials in three selected states mentioned that connecting to neighboring networks—whether within or between states—could pose challenges. For example, officials mentioned that states and localities may have obtained different equipment, software applications, or service providers – all of which can make interconnections difficult. Officials in Maine and New Hampshire told us that differences in service providers can also be a challenge to seamlessly connecting to neighboring systems. In an instance where two states (Minnesota and North Dakota) have worked to connect their 911 systems, both states used the same service provider, which officials said allowed for fewer barriers to connection. Cyber risks: Officials in three states told us that the transition from a traditional system that only transmits voice traffic to an IP-based system that transmits voice and data traffic has significantly increased the risk of a cyber-attack. This can be a challenge because managing cyber risks is a new and evolving role for state and local 911 authorities. Approaching the transition to NG911 without managing these risks could result in disrupted or disabled call center operations and ultimately a delayed response to an emergency situation. Multimedia: Officials in three states mentioned potential implementation challenges related to accepting and processing multimedia such as audio recordings, images, and videos. More specifically, one official said they did not have procedures to manage or store these multimedia files once received. In addition, another official raised privacy and liability concerns. Call routing: One of the core services of an NG911 system is the ability to have calls routed to the appropriate call center based on a wireless caller’s physical location, instead of the location of the cellular tower that receives and transmits the call. An FCC-sponsored working group reported that there are several options for achieving this and each option has unique positive and negative aspects. One challenge officials in two states noted was that rather than a single, nationwide approach to routing these calls, state and local 911 authorities would need to work individually with the wireless carriers to determine how to best implement location-based call routing. Governance: FCC has noted that transitioning to NG911 will likely result in new roles and levels of coordination between state 911 authorities, local 911 authorities, 911 call centers, and 911 service providers. Further, relationships among authorities at the state and local level may change as states work to interconnect NG911 systems. State and local officials noted that these types of governance challenges can apply in a variety of situations, including within or between states. Evolving roles: As previously mentioned, 911 governance structures vary among states. These varying governance structures may pose different challenges. For example, some states have a centralized structure in which a single government agency is responsible for statewide 911 system’s administration and policy. Officials in two states told us that although they faced challenges transitioning to NG911, their states’ centralized 911 structure eased the transition in their states because there was uniformity in policy and technology, among other things, coming from a single statewide authority. In other states, 911 systems are primarily a local responsibility and organized with decentralized authorities and resources. In these instances, there may be specific challenges related to transitioning to an interconnected NG911 system. Such challenges may include the need for increased levels of coordination among numerous jurisdictions with potentially disparate organizational structures, levels of funding, and priorities. An official also noted that there are governance challenges related to connecting states and evolving relationships between 911 authorities and service providers. Informing decision makers: One of the challenges identified by officials in two states is differing levels of experience and understanding by state and local officials as to what NG911 priorities should be for timely implementation. To help with this understanding, the federal government is making efforts to educate state and local authorities on how to facilitate policymaker understanding as well as provide regular updates to stakeholders on recent NG911 developments. We discuss some of these efforts later in this report. Federal Agencies Are Addressing NG911 Implementation Challenges, but the National 911 Program Lacks Measurable Goals and Next Steps for the NG911 Roadmap Initiative Federal Agencies Are Taking Actions to Address NG911 Implementation Challenges While state and local entities have the primary responsibility for implementing NG911 technology and services, federal agencies are taking actions to assist state and local 911 entities to address NG911 implementation challenges. Actions taken include developing resources, offering technical assistance, and convening stakeholders. More specifically, we identified selected activities that were taken by NHTSA, NTIA, FCC, and DHS that address some of the funding, technology, and governance challenges raised by state and local 911 stakeholders, for example: Cost study: NHTSA’s National 911 Program and NTIA, in consultation with FCC and DHS, plan to issue a study of the range of costs for 911 call centers and service providers to implement NG911 systems. According to NHTSA officials, the cost study will present a nationwide view, rather than a state-by-state view, on the progress of NG911 implementation and its associated costs. Grant program: NHTSA and NTIA are preparing to jointly administer a $115 million grant program to improve 911 services, including the adoption and operation of NG911 services. In September 2017, NHTSA and NTIA issued a notice of proposed rulemaking outlining implementing regulations for the grant program. NHTSA and NTIA expect to award the grants in 2018. Technology standards: The National 911 Program issued an annual guide in 2017 that stressed the importance of using open technology standards for NG911 services. The guide provides a list of standards that have been recently updated and an analysis that identifies whether existing standards fully address NG911 processes and protocols. Cybersecurity guides: DHS issued a guide in 2016 that identified cybersecurity risks for NG911 and risk mitigation strategies. According to DHS officials, the National 911 Program provided input on this guide. In addition, an advisory body tasked by FCC to examine 911 call-centers’ architecture issued a report in 2016 that provided a cybersecurity self-assessment tool for call centers and guidance on cybersecurity strategies. Governance plans: To address challenges related to the evolving roles for state and local 911 authorities, the National 911 Program issued a guide in 2016 that provided practices for states to consider when interconnecting NG911 networks, and DHS issued a guide in 2015 for emergency communications officials for establishing, assessing, and updating their governance structures. In addition, an FCC advisory body issued a report in 2016 that identified NG911 governance approaches, issues, and recommendations for states, localities, and call centers to consider when planning for the deployment of NG911. In addition to federal agency efforts to assist the state and local 911 community, the National 911 Program is in the early stages of establishing an interagency initiative to create a National NG911 Roadmap. As part of this initiative, the National 911 Program plans to convene the 911 stakeholder community to identify tasks that need to be completed at the national level by the federal government and other public and private-sector organizations to support the creation of a national, interconnected NG911 system. Additional details regarding this planned activity are described in further detail later in this report. For additional information on federal actions to address state and local NG911 challenges, see appendix II. National 911 Program Lacks Goals and Performance Measures As the lead entity for coordinating federal NG911 activities, the National 911 Program has taken a variety of actions to assist the state and local 911 community, in collaboration with other federal agencies. However, the program lacks goals and performance measures to assess whether these activities are achieving desired results. National 911 Program officials stated that they initiate program activities based on feedback received from the 911 community. In addition, officials said the program’s activities fall within the tasks established in the Next Generation 911 Advancement Act of 2012. However, the National 911 Program does not have a means to assess its progress toward meeting its responsibilities established in the 2012 Act. National 911 Program officials said the Office of EMS—the office within NHTSA in which the program is housed—has a strategic plan, but it is outdated and does not contain specific goals or performance measures related to 911 or NG911 implementation. Officials said the Office of EMS has held preliminary discussions to begin updating its strategic plan by January 2019 and plans to include goals and performance measures related to 911 and NG911 services. Office of EMS officials told us the Office of EMS strategic plan will be jointly developed with the National 911 Program. However the Office of EMS had not yet developed a draft strategic plan during the time of our review. Federal internal control standards call for management to clearly define objectives in order to achieve desired results. According to these standards, an entity determines its mission, establishes specific measurable objectives, and formulates plans to achieve its objectives. These standards state that management sets objectives in order to meet the entity’s mission, strategic plan, and goals and requirements of applicable laws and regulations. In addition, our work on leading practices for managing for results indicated that an agency’s strategic goals should also explain what results are expected from the agency and when to expect those results. Further, these goals form a basis for an entity to identify strategies to fulfill its mission and improve its operations to support the achievement of that mission. As the lead entity for coordinating federal NG911 efforts, the National 911 Program faces a complex and challenging task of assisting the 911 community while the nation’s 911 systems undergo a major transformation. However, without specific goals and related performance measures, the National 911 Program is unable to assess how well its activities are achieving results in relation to its responsibilities identified in the 2012 Act. As the National 911 Program and the Office of EMS consider creating a strategic plan, ensuring that the plan includes specific goals and related measures for the National 911 Program would help officials better understand whether the program’s activities are effectively assisting states and localities in transitioning to a fully integrated national NG911 system, and help identify any programmatic changes that might be needed. National 911 Program Lacks Plans to Identify Roles and Responsibilities for the NG911 Roadmap Initiative and an Implementation Plan to Achieve Objectives As previously mentioned, the National 911 Program is in the early stages of establishing an interagency initiative to create a National NG911 Roadmap. This initiative will convene the 911 stakeholder community to identify national-level tasks that need to be completed by federal agencies and other organizations to realize a national, interconnected NG911 system. According to the National 911 Program, a list of the national-level tasks needed to advance NG911 implementation nationwide has not been created to date. In addition, state officials we spoke with said there are certain issues related to interoperability and cybersecurity that federal agencies need to address before states can connect their respective state NG911 systems. To address these issues, NHTSA’s National 911 Program issued a request for proposal (RFP) in August 2017 for managing the roadmap development process and awarded a contract in September 2017. While the National 911 Program is taking steps to develop a National NG911 Roadmap, the program does not have a plan to identify: (1) roles or responsibilities for federal entities to carry out national-level tasks or (2) how the program plans to achieve the roadmap’s objectives. Clarifying Roles and Responsibilities NHTSA’s NG911 roadmap RFP specifies that by identifying a list of national-level tasks that are developed and adopted by the 911 stakeholder community, the roadmap could serve as a blueprint to carry out these tasks and thereby ensure the interoperability of the nation’s NG911 system. However, the National 911 Program does not have plans for the entities participating in the development of the roadmap to be assigned roles and responsibilities for executing the roadmap’s national- level tasks. National 911 Program officials told us the National 911 Program does not plan to assign roles and responsibilities because NHTSA does not have the authority to require or assign tasks for other entities. Additionally, program officials view the simultaneous identification of tasks and assignments of responsibility for those tasks as a risk to facilitating a candid and productive discussion with entities participating in the roadmap initiative. However, officials stated it may be appropriate for agencies participating in the roadmap initiative to perform specific tasks after the roadmap is finalized. We have previously examined interagency collaborative mechanisms and identified certain key issues for federal agencies to consider when using these mechanisms to achieve results. Our prior work has found that following leading collaboration practices, such as clarifying roles and responsibilities of agencies engaged in collaboration, can enhance and sustain collaboration among agencies and provide an understanding of who will do what in support of meeting the aims of the collaborative group. As stated above, the RFP specifies that a roadmap developed by and adopted by 911 stakeholders could serve as a blueprint to carry out the roadmap’s tasks. Securing the commitment of agencies to assigned roles could help organize the collaborative group’s joint and individual efforts and thereby better facilitate decision making. As we have previously found, a lack of clarity on the roles and responsibilities of agencies participating in an interagency effort—such as the execution of the roadmap’s tasks—may limit agencies’ abilities to effectively achieve shared objectives. Given the complexity of the task and the number of agencies that could be involved, following selected leading collaboration practices for the roadmap initiative—particularly with regard to collaborating with roadmap stakeholders to clarify their roles and responsibilities (whether during the creation of the task list or afterwards)—could reduce barriers to agencies effectively working together to achieve the national-level tasks. Developing an Implementation Plan to Achieve Objectives While clarifying the roles and responsibilities of roadmap stakeholders for the execution of the roadmap’s tasks is an important collaborative step, the National 911 Program has additional responsibilities as the lead entity for the initiative. However, National 911 Program officials are unable to clearly articulate how the program will proceed following the completion of the roadmap. National 911 Program officials said without knowing the contents of the roadmap, it would be premature to specify how the roadmap’s national-level tasks would be completed. Officials stated that once the roadmap is completed, possible next steps may include identification of timelines, deadlines, and a mechanism for tracking progress, among other things, but officials stated that these steps are not required in the roadmap RFP. As stated above, federal internal control standards call for management to clearly define objectives in specific terms. According to these standards, management defines what is to be achieved, who is to achieve it, how it will be achieved, and the time frames for achievement. Without a clear plan for how the National 911 Program would take next steps to support the implementation of the roadmap’s objectives and tasks, the National 911 Program may not be prepared to take effective action once the roadmap is completed. We have previously found that having an implementation plan can assist agencies to better focus and prioritize goals and objectives, and align planned activities. Once the roadmap is completed, developing an implementation plan that details what is to be achieved and how it will be accomplished will place the National 911 Program in a better position moving forward to support the completion of the national-level tasks. Conclusions The current 911 system is undergoing a historic transition. With no federal requirement that states transition to NG911 services, federal leadership is critical to addressing interoperability challenges and promoting the goal of an interconnected national system. As the lead federal entity for fostering coordination and collaboration among federal, state, and local 911 authorities, the National 911 Program plays a critical role in coordinating NG911 implementation efforts to improve the nation’s 911 services. However, this program—in collaboration with other federal agencies— faces a complex and challenging task to help move approximately 6,000 independent 911 call centers toward an interconnected national NG911 system. In addition, given that the NG911 transition is still in its early stages and is an ongoing effort, it is difficult to assess the effectiveness of various federal actions to assist states and localities in the transition. In light of these challenges, without specific goals and related measures to assess effectiveness, the National 911 Program may be hindered in determining whether it is making progress towards its stated mission. Through the roadmap initiative, the National 911 Program has taken important first steps in identifying the need for actions at the national level, in order to fully realize the desired end-state of a national, interconnected NG911 system. However, while identifying needed next steps is essential, equally important to the collaborative effort’s success is (1) defining and agreeing on the roles and responsibilities of the entities best suited to undertake these actions, and (2) developing plans for how the National 911 Program will support implementation to achieve the roadmap’s objectives. If taken, these actions could help further NG911 implementation nationwide and help the National 911 Program and federal agencies in assisting states and localities to improve these lifesaving services. Recommendations for Executive Action We are making the following three recommendations to the Administrator of NHTSA regarding the National 911 Program: develop specific program goals and performance measures related to NG911 implementation. (Recommendation 1) in collaboration with the appropriate federal agencies, determine roles and responsibilities of federal agencies participating in the National NG911 Roadmap initiative in order to carry out the national-level tasks over which each agency has jurisdiction. (Recommendation 2) develop an implementation plan to support the completion of the National NG911 Roadmap’s national-level tasks. (Recommendation 3) Agency Comments We provided a draft of this report to the Departments of Transportation, Commerce, and Homeland Security and FCC for their review and comment. In its comments, reproduced in appendix III, the Department of Transportation agreed with the recommendations. The Departments of Transportation and Homeland Security also provided technical comments, which we incorporated as appropriate. The Department of Commerce and FCC had no comments. As agreed with your office, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies of this report to the appropriate congressional committees, the Secretary of the Department of Transportation, the Secretary of the Department of Commerce, the Secretary of the Department of Homeland Security, the Managing Director of the FCC, and other interested parties. In addition, the report will be available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-2834 or goldsteinm@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. Staff who made key contributions to this report are listed in appendix IV. Appendix I: Objectives, Scope, and Methodology Our objectives were to examine (1) progress states and localities are making to implement Next Generation 911 (NG911) and the challenges they have faced and (2) how federal agencies have addressed state and local implementation challenges and planned next steps. To describe state and local progress in implementing NG911 and background information on fee collection and costs, we analyzed select survey data elements from the 2016 National 911 Progress Report and the Eighth Annual Report to Congress on State Collection and Distribution of 911 and Enhanced 911 Fees and Charges, maintained by the National Highway Traffic Safety Administration (NHTSA) and the Federal Communications Commission (FCC) respectively. More specifically, we analyzed the most recent state-provided data (from calendar year 2015) related to the planning and implementation of NG911 at the state and local levels, as well as NG911 cost and 911-related revenue data. We assessed the reliability of these data by reviewing relevant documents and discussing data elements with staff responsible for collecting and analyzing the data. We also conducted our own testing to check the consistency of the data. We found the data from both sources to be sufficiently reliable for our purposes to describe states’ progress in implementing NG911 and provide background on 911 fee collection and costs. While these data provide the best nationwide picture of NG911 implementation and fee collection, and are reliable for our purposes, there are some limitations on how the data can be used. Since we did not validate the state-reported responses, our findings based on these data are limited to what states reported. Additionally, regarding the 2016 National 911 Progress Report data, there are limitations to (1) making comparisons between states because states have different approaches to implementing NG911 and (2) ascertaining year-over-year progress because reporting is voluntary and states’ response rates can vary year to year. To describe implementation challenges that states and local authorities may be encountering, we selected a non-generalizable sample of 10 states as case studies, based upon a variety of factors, including reported progress in implementing NG911, statewide planning and coordination, reported number of annual 911 calls, whether states diverted 911 fees to other uses, and variation in geographic location. We selected these states, in part, based on their responses to the two aforementioned surveys. Based on the aforementioned criteria, we selected the following states to include as case studies: California, Maine, Maryland, Minnesota, Nevada, New Hampshire, North Dakota, South Dakota, Vermont, and Virginia. We reviewed documents and interviewed state officials from all of these states except Nevada about NG911 implementation progress, challenges, federal actions, and any additional assistance needed. We contacted 911 officials in Nevada but did not receive responses. We also interviewed local officials in four of the selected states. While not generalizable to all states, the information obtained from our case studies provides examples of broader issues faced by states and localities in managing the NG911 transition. To determine how federal agencies have addressed state and local implementation challenges and planned next steps, we reviewed relevant statutes, regulations, and documentation of federal agency actions and plans, and our prior reports. We also interviewed officials from federal agencies, including NHTSA, the National Telecommunications and Information Administration (NTIA), FCC, and the U.S. Department of Homeland Security (DHS), about federal actions taken and plans for next steps. To understand planning activities undertaken by NHTSA’s National 911 Program, and its planned project to develop a National NG911 Roadmap, we reviewed the National 911 Program’s internal planning documents, the program’s request for proposal to develop a national roadmap, the program’s written responses to our questions, and interviewed National 911 Program officials. In addition, we interviewed officials from national associations representing emergency-response- technology companies, wireless and wireline phone carriers, emergency- communications entities, and groups representing deaf and hard-of- hearing consumers to gain their perspectives on federal actions taken and next steps. We assessed the National 911 Program’s strategic- planning activities against leading practices for performance management found in our prior work on strategic planning and goal setting and federal internal control standards. We assessed the National 911 Program’s planned activities for the national roadmap project against federal internal control standards and selected key practices to enhance interagency collaboration identified in our prior work. We conducted our work from January 2017 to January 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Selected Federal Actions since 2013 to Address State and Local Challenges in Implementing Next Generation 911 (NG911), as of October 2017 Challenges Funding Description of challenge State and local funding may not be sufficient to support costs associated with transitioning to NG911 equipment and infrastructure. Transitioning from legacy infrastructure to Internet Protocol-based systems presents technical and operational challenges such as interoperability and cybersecurity risks. Federal actions Grant resources: The National Highway Traffic Safety Administration’s (NHTSA) National 911 Program issued on its website a list clarifying which of the fiscal year 2016 emergency-communications grants may be used for NG911 services. Program officials said they developed this list in collaboration with the Department of Homeland Security (DHS). Cost study: NHTSA’s National 911 Program and the National Telecommunications and Information Administration (NTIA), in consultation with the Federal Communications Commission (FCC) and DHS, plan to issue a study of the range of costs for 911 call centers and service providers to implement NG911 systems and on the nationwide progress of implementing NG911 services. Grant program: NHTSA and NTIA are preparing to jointly administer a $115 million grant program to improve 911 services, including the adoption and operation of NG911 services. NHTSA and NTIA expect to award the grants in 2018. Funding mechanisms: An advisory body tasked by FCC issued a report in 2016 that identified common costs and funding mechanisms for 911 officials to consider. The report also introduced a 911 funding sustainment model designed for use by 911 officials to calculate their financial needs to support a transition to NG911 implementation. Guides on technology standards and procurement practices: In 2017, NHTSA’s National 911 Program issued an annual guide on emergency- communications technology standards that stressed the importance of using open technology standards for NG911 services. The National 911 Program issued another guide in 2016 that provides information on procuring goods and services related to NG911 such as practices for call centers to consider when developing their request for proposals and contracts. Examining emerging technology issues: In 2017, FCC tasked a public- private advisory council to recommend how FCC can promote the NG911 transition, enhance the reliability of NG911, and mitigate the threat of 911 outages. Prior to that tasking, the FCC advisory council issued a report in 2016 that explored location-based routing issues and discussed transition considerations from legacy 911 to NG911. NG911 cybersecurity guide and technical assistance: DHS, with input from NHTSA’s National 911 Program according to DHS officials, issued a guide in 2016 that identifies cybersecurity risks for NG911 and risk mitigation strategies. In addition, DHS provides NG911 technical assistance for states seeking assistance with strategic planning and technology integration. In a separate effort, an advisory body tasked by FCC to examine 911 call center architecture issued a report in 2016 that provides a cybersecurity self- assessment tool for call centers and guidance on cybersecurity strategies. Challenges Governance Description of challenge States may face a range of challenges related to evolving roles for state and local 911 authorities that could hinder NG911 implementation. Federal actions Guides on state and legislative planning: NHTSA’s National 911 Program issued guides on state 911 planning and legislative issues to consider for NG911 and awarded a contract in September 2017 to update those guides. In 2016, the National 911 Program issued a guidebased on the experiences of Iowa, Minnesota, North Dakota, and South Dakota that identifies practices to consider for states interconnecting NG911 networks across state lines. Exploring NG911 governance implementation issues: In 2016, an advisory body tasked by FCC issued a report that identifies NG911 governance approaches, issues, and recommendations for states, localities, and call centers to consider when planning for the deployment of NG911. In 2013, FCC also issued a report that details recommendations to Congress for transitioning from legacy 911 to NG911 networks. Guide on emergency communications governance structures: In 2015, DHS and the National Council of Statewide Interoperability Coordinators issued a guide that provides characteristics of effective governance approaches and best practices for officials to establish, assess, and update their governance structures. Appendix III: Comments from the U.S. Department of Transportation Appendix IV: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Andrew Huddleston (Assistant Director), Jean Cook (Analyst in Charge), Camilo Flores, Steven Rabinowitz, Malika Rice, Kelly L. Rubin, Michael Sweet, Hai Tran, Marika Van Laan, and Michelle Weathers made key contributions to this report.

Why GAO Did This Study Each year, millions of Americans call 911 for help during emergencies. However, the nation's legacy 911 system relies on aging infrastructure that is not designed to accommodate modern communications technologies. As a result, states and localities are upgrading to NG911, which offers improved capabilities, such as the ability to process images, audio files, and video. While deploying NG911 is the responsibility of state and local entities, federal agencies also support implementation, led by NHTSA's National 911 Program, which facilitates collaboration among federal, state, and local 911 stakeholders. GAO was asked to review NG911 implementation nationwide. This report examines: (1) state and local progress and challenges in implementing NG911 and (2) federal actions to address challenges and planned next steps. GAO reviewed relevant statutes, regulations, and federal agency reports and plans. GAO also analyzed NHTSA's survey data on state 911 implementation for calendar year 2015, the most recent year for which data were available, and interviewed federal officials, state and local officials from nine states (selected to represent different regions and various phases of NG911 implementation), and officials from industry and advocacy groups. What GAO Found The National Highway Traffic Safety Administration's (NHTSA) National 911 Program's most recent national survey on Next Generation 911 (NG911) implementation indicated that about half of states were in some phase of transition to NG911 in 2015, but that state and local progress varied. Specifically, 10 states reported that all 911 authorities in their state processed calls using NG911 systems; however, 18 states reported having no state or local NG911 transition plans in place—which may indicate these states were in the early phases of planning for the transition to NG911 or had not yet begun. GAO spoke with state and local 911 officials in 9 states, which were in various phases of implementing NG911, and found that none of the 9 selected states were accepting images, audio files, or video. State and local 911 officials identified a number of challenges to implementing NG911. Such challenges are related to funding, evolving technology and operations, and governance. For example, officials in 3 states said that the current funding they collect from telephone service subscribers may not be sufficient to support NG911's transition costs while simultaneously funding the operation of existing 911 systems. Federal agencies—including NHTSA, the National Telecommunications and Information Administration, the Federal Communications Commission, and the U.S. Department of Homeland Security—have responsibilities to support NG911 implementation, such as through coordinating activities and administering grants, and are taking actions to assist state and local entities in addressing challenges to NG911's implementation. Such actions include developing resources, offering technical assistance, and convening stakeholders to explore emerging NG911 issues. For example, as the lead entity for coordinating federal NG911 efforts, NHTSA's National 911 Program is developing resources on NG911 topics, such as federal funding and governance structures. While the National 911 Program is taking steps to facilitate the state and local transition to NG911, the program lacks specific performance goals and measures to assess its progress. Without such goals and measures, it is not clear to what extent the program is effectively achieving its mission. In 2018, the National 911 Program plans to establish an interagency initiative tasked with creating a National NG911 Roadmap. This roadmap is intended to identify next steps for the federal government in supporting the creation of a national, interconnected NG911 system. While the National 911 Program is taking steps to develop a list of national-level tasks as part of its roadmap initiative, the program does not have a plan to identify: (1) roles or responsibilities for federal entities to carry out these tasks or (2) how the program plans to achieve the roadmap's objectives. Collaborating with the appropriate federal agencies to determine federal roles and responsibilities to carry out the roadmap's national-level tasks could reduce barriers to agencies effectively working together to achieve those tasks. Furthermore, developing an implementation plan that details how the roadmap's tasks will be achieved would place the National 911 Program in a better position to effectively lead interagency efforts to implement NG911 nationwide. What GAO Recommends GAO recommends that NHTSA's National 911 Program develop performance goals and measures and, for the National NG911 Roadmap, determine agencies' roles and responsibilities and develop an implementation plan. NHTSA agreed with GAO's recommendations.

gao_GAO-19-56

gao_GAO-19-56_0

Background SNAP Work Requirements All SNAP recipients ages 16 through 59, unless exempted by law or regulation, must comply with general work requirements. (See fig. 1.) These requirements generally include registering for work, reporting to an employer if referred by a state agency, accepting an offer of a suitable job, not voluntarily quitting a job or reducing work hours below 30 hours a week, or participating in a SNAP E&T program or a workfare program—in which recipients perform work on behalf of the state—if assigned by the state agency. SNAP recipients are exempt from complying with these work requirements if they meet certain criteria, such as being responsible for caring for a dependent child under age 6 or an incapacitated person. In addition, per federal law, those who are employed for 30 or more hours per week are exempt from the work requirements. SNAP recipients who are subject to the work requirements—known as work registrants—may lose their eligibility for benefits if they fail to comply with the requirements without good cause. In addition to meeting the general work requirements, able-bodied adults without dependents (ABAWDs) must work or participate in a work program 20 hours or more per week, or participate in workfare, which is performing work to earn the value of their SNAP benefits. Participation in SNAP E&T, which is a type of work program, is one way for ABAWDs to meet the 20-hour-per-week ABAWD work requirement, but other work programs are acceptable as well. Unless ABAWDs meet these work requirements or are determined to be exempt, they are limited to 3 months of SNAP benefits in a 36-month period. (See fig. 2.) At the request of states, FNS may waive the ABAWD time limit for ABAWDs located in certain areas of a state or an entire state when certain circumstances are met. For example, a waiver may be granted if the area has an unemployment rate of over 10 percent or the number of jobs available is insufficient to provide employment for these individuals. If the time limit is waived, ABAWDs are not required to meet the ABAWD work requirement in order to receive SNAP for more than 3 months in a 36-month period, but they must still comply with the general work requirements. SNAP Employment and Training Programs Federal requirements for state SNAP E&T programs were first enacted in 1985 and provide state SNAP agencies with flexibility in designing their SNAP E&T programs, including whom to serve and what services to offer. The state can require some or all SNAP work registrants to participate in the SNAP E&T program as a condition of eligibility, an approach commonly referred to as a mandatory program. In mandatory programs, individuals can be sanctioned if they fail to participate in an assigned SNAP E&T activity. State SNAP agencies also may elect to exempt categories and individuals from participating in SNAP E&T, such as those living in rural areas or experiencing homelessness. In addition, states may exempt all work registrants from participation in SNAP E&T and only serve volunteers, an approach commonly referred to as a voluntary program. States also determine which types of services to provide participants through their SNAP E&T programs, although they must provide at least one from a federally determined list. This list includes job search programs, job search training programs, workfare, programs designed to improve employability through work experience or training, education programs to improve basic skills and employability, job retention services, and programs to improve self-sufficiency through self- employment. There are three types of federal funding streams for state SNAP E&T programs: 100 percent funds—formula grants for program administration, including planning, implementing, and operating a SNAP E&T program; 50 percent federal reimbursement funds; and ABAWD pledge funds— grants to states that pledge to serve all of their at-risk ABAWDs. While the federal allocation for 100 percent funds has generally been capped at $90 million over the last decade, some states do not obligate or expend their full allocation, and as a result, the following year FNS reallocates these funds to other states that request additional funds, according to FNS officials. Total federal expenditures on SNAP E&T programs increased from about $282 million in fiscal year 2007 to about $337 million in fiscal year 2016, according to FNS data (see fig. 3). Federal 50 percent reimbursement funds, which are generally not capped, constitute the largest portion of federal expenditures on SNAP E&T and were responsible for the majority of the increase in total federal SNAP E&T expenditures over the last decade. These funds are used for program administrative costs for operating SNAP E&T programs, as well as SNAP E&T participant expenses, such as transportation and dependent care costs. Program Support and Oversight In 2014, FNS created the Office of Employment and Training to provide support and oversight for the SNAP E&T program. Specifically, FNS expanded its headquarters staff dedicated to SNAP E&T from one to six full-time employees, and added a dedicated SNAP E&T official in each FNS regional office to provide technical assistance to states. FNS has also developed resources, such as a SNAP E&T Operations Handbook, intended to help states implement and expand their SNAP E&T programs. To inform its program support and oversight, FNS collects information on SNAP recipients and work registrants, as well as SNAP E&T program participants, services, and expenditures. More specifically, FNS periodically collects data from states on the total number of work registrants, ABAWDs, SNAP E&T participants, and participants in each type of SNAP E&T service. FNS also collects data from states on a sample of all households participating in SNAP each month as part of the Quality Control process. The Quality Control data include characteristics of SNAP recipients, including whether they are work registrants, for example. In addition, as a result of requirements in the Agricultural Act of 2014, FNS began collecting annual SNAP E&T outcome and participant characteristics data from states in January 2018. Furthermore, FNS collects quarterly information from states on SNAP E&T expenditures. States are also required to submit an annual SNAP E&T plan to FNS, including information on the services they plan to offer during the year and their projected budget and program participation numbers. Guidance on plan requirements is provided in an FNS handbook for states. FNS national and regional officials review the plans to ensure compliance with requirements, and plans must be approved by regional officials before SNAP E&T funding is allocated to a state. State SNAP E&T Programs Have Served a Small Percentage of Recipients over Time and Little Is Known about Participants and Outcomes State SNAP E&T programs have served a small and decreasing percentage of overall SNAP recipients over time, and although these data are generally reliable, FNS data on SNAP E&T program participant characteristics and outcomes are not reliable. State SNAP E&T programs have served a small percentage of SNAP recipients over the last decade potentially due in part to certain policy changes during that time, such as the increasing number of states moving from mandatory to voluntary SNAP E&T programs. The number of SNAP recipients served by SNAP E&T programs has also potentially been low because a limited number of those referred to state programs go on to participate in services. FNS’s lack of reliable SNAP E&T data, as well as the agency’s lack of a plan for using newly reported participant characteristics and outcome data to assess program performance, constrain FNS’s ability to understand the extent to which agency goals are being met. State SNAP E&T Programs Have Served a Small and Decreasing Percentage of SNAP Recipients over the Last Decade, Due to Various Factors According to FNS data, among the approximately 43.5 million total SNAP recipients, only a small percentage—0.5 percent, or about 200,000—were served by state SNAP E&T programs in an average month of fiscal year 2016, due to several factors. (See fig. 4.) First, according to FNS data, most SNAP recipients are exempt from work requirements for various reasons, under federal law and regulation. For example, according to FNS data, almost two-thirds of SNAP recipients were children, elderly, or adults with a disability in an average month of fiscal year 2016, and these groups generally are exempt from work requirements. As a result of federal exemptions, in an average month of fiscal year 2016, about 14 percent of SNAP recipients, or about 6.1 million individuals, were work registrants who were subject to work requirements, according to FNS data. Further, state SNAP agencies may elect to exempt individuals for whom participation is judged to be impractical or not cost effective. Moreover, SNAP work registrants may participate in other activities to comply with work requirements, such as other federal- and state-funded E&T programs. In recent years, the number and percentage of SNAP recipients and work registrants participating in SNAP E&T programs has decreased, according to FNS data. From fiscal years 2008 through 2016, the average monthly number of SNAP E&T participants decreased from about 256,000 to about 207,000, or by 19 percent, according to state data on SNAP E&T participants that were reported to FNS. (See fig. 5.) However, the data also show that over the same time period, the average monthly number of SNAP recipients increased from about 27.8 million to about 43.5 million, and work registrants increased from about 3.2 million to about 6.1 million. As a result, the percentage of total SNAP recipients participating in SNAP E&T programs decreased from about 0.9 percent to about 0.5 percent, and the percentage of work registrants participating in these programs decreased from approximately 8.1 percent to approximately 3.4 percent. The decline in SNAP E&T participation in recent years may have been influenced by certain policy changes, including states’ widespread use of waivers for ABAWDs. According to FNS data, from fiscal years 2008 through 2012, the number of states with statewide waivers due to economic conditions increased from 7 to 46 states, potentially enabling ABAWDS in these states to continue receiving SNAP benefits without meeting ABAWD work requirements. As a result, these waivers potentially reduced the number of ABAWDs nationwide who may otherwise have participated in SNAP E&T programs in order to continue receiving SNAP benefits. Further, according to FNS data, from fiscal years 2011 through 2015, the majority of states continued to operate under statewide waivers of the ABAWD time limit. According to FNS data, states have also increasingly moved from mandatory to voluntary SNAP E&T programs in recent years, another policy change that may have influenced SNAP E&T participation. In fiscal year 2010, 36 states operated mandatory programs; however, by fiscal year 2017, 19 states operated mandatory programs. (See fig. 6.) When states move to a voluntary program, they generally experience a decline in SNAP E&T participation, according to FNS officials and our analysis of FNS data. Specifically, of the 21 states that changed from a mandatory to a voluntary program from fiscal year 2010 through fiscal year 2016, 13 experienced a decrease in SNAP E&T participation in the year following the change—ranging from a 21 percent decrease to a 93 percent decrease. This trend was generally inconsistent with the trend in work registrants, as 9 of the 13 states that changed from a mandatory to a voluntary program and experienced a decrease in SNAP E&T participation also experienced an increase in their total number of SNAP work registrants during the same time period. Furthermore, voluntary programs are generally smaller overall than mandatory programs, according to our analysis of FNS data. In fiscal year 2016, for example, the 31 states operating voluntary programs together served less than half of the total number of SNAP E&T participants served by the 22 states operating mandatory programs, although these two groups of states had similar numbers of new work registrants. FNS officials told us that there are various reasons states may move to voluntary programs. For example, FNS officials said that many states have reported to them that offering employer-driven, skills-based, intensive employment and training services, such as vocational training or work experience, through voluntary programs yields more engaged participants with stronger outcomes. FNS officials stated that they have been actively encouraging states to offer these types of services because they believe these types of services are more effective in moving SNAP recipients, who may be more likely to have barriers to employment, toward self-sufficiency. However, they noted that SNAP E&T funding may not be sufficient to provide these types of services in mandatory programs that require participation by SNAP recipients and thus have higher participation. In addition, FNS officials told us that voluntary programs are less administratively burdensome than mandatory programs, as they allow states to focus on serving motivated participants rather than sanctioning non-compliant individuals. In addition, participation rates are low for SNAP recipients referred to the SNAP E&T program, according to FNS officials, state program officials, and available data, regardless of whether the state operates a mandatory or voluntary program. FNS officials said that engaging SNAP recipients who are referred to the program is a challenge across all states—a point confirmed by the states we selected and available data. Among the 11 states that reported data to FNS on SNAP E&T participation by those referred to the program, which included states operating mandatory and voluntary programs, the percentage of SNAP recipients who were sent a referral letter but did not participate in any activity ranged from 35 to 98 percent in fiscal year 2017. For 8 of these states, about 70 percent or more of SNAP recipients who were sent a referral letter did not participate in any activity. FNS officials, state officials, and SNAP E&T service providers in our selected states indicated that participation by SNAP recipients referred to SNAP E&T may be low for various reasons. For example, FNS officials told us that some recipients face barriers to participation, such as a lack of transportation, childcare, or treatment for mental health issues, yet they have not been exempted by the state. For example, SNAP E&T providers and state officials in our selected states noted that SNAP recipients in rural areas, in particular, experience challenges participating in the E&T program due to a lack of transportation to E&T services, as well as the limited range of available services and employment opportunities. State officials and providers in all five of the states we selected also noted that SNAP recipients with mental health needs or substance abuse issues usually require additional services to participate in the SNAP E&T program, such as intensive case management or treatment. Lack of awareness of E&T services may also affect participation, as three SNAP E&T providers we spoke with said that SNAP recipients can be transient, and as a result, may not receive referral letters provided by mail. Further, some SNAP recipients may decide not to participate, despite the potential loss of SNAP benefits, or others face certain barriers to employment that may deter them from participating. For example, formerly incarcerated SNAP recipients may be discouraged from participating in SNAP E&T due to past struggles finding employers willing to hire those with a criminal background. Low participation rates are common across other employment and training programs serving similar populations, and although FNS has not researched how to address this issue in SNAP E&T, other agencies have assessed ways to improve participation in these programs. For example, in our past work, we found that states faced challenges with low participation in employment and training activities by Temporary Assistance for Needy Families cash assistance recipients. Recognizing that states would benefit from strategies on how to increase engagement in such activities, the U.S. Department of Health and Human Services contracted for research on behavioral interventions that affect attendance rates for employment and training services. Researchers found that strategies such as sending text messages to participants—in addition to letters in the mail—could increase the likelihood that they would attend program activities, particularly when communications encouraged recipients to make a detailed plan to participate. FNS officials stated that they are aware of research on strategies to address low participation in E&T programs; however, they noted that they have not researched causes of low participation in the SNAP E&T program. FNS officials added that they believe states could take steps to make enrolling and participating in SNAP E&T activities less burdensome for SNAP recipients. Further, FNS officials acknowledged that states could potentially benefit from technical assistance on increasing the rates at which referred SNAP recipients participate in SNAP E&T activities, but the agency’s SNAP E&T technical assistance resources have generally not focused on this issue. In a recent policy brief, FNS indicated that collecting data on SNAP E&T participation can help state agencies and providers determine where attrition is occurring and point towards processes or services that need improvement. However, the brief did not provide strategies for improving processes or services to reduce attrition, and FNS officials acknowledged that they generally have not focused their resources on getting recipients to initially engage with service providers. Rather, FNS has focused its technical assistance resources on an approach intended to improve participation among those recipients who engage with the SNAP E&T program. Specifically, according to FNS officials, the agency’s resources have focused on encouraging SNAP E&T providers to offer more intensive services, including skills-based training, as these services may be better able to address SNAP recipients’ barriers to employment. Officials noted that they believe these types of services may be more responsive to SNAP recipients’ needs, which could increase participation in E&T. Assisting state efforts to increase the level of participation for SNAP recipients who are referred to the E&T program could help FNS achieve agency goals, as well as help SNAP recipients move toward self- sufficiency. Specifically, USDA’s fiscal year 2018 strategic plan includes increased participation in SNAP E&T as a strategy for supporting SNAP recipients in achieving self-sufficiency. Similarly, in a 2016 letter to states, FNS noted that expanding SNAP recipients’ access to employment and training services is critical to helping them transition off the SNAP program by becoming economically self-sufficient. If states continue to struggle with low participation in SNAP E&T, and FNS does not expand its technical assistance to include a broader array of strategies to increase participation, both FNS’s ability to meet its strategic goals, and the program’s ability to help recipients achieve self-sufficiency, will be hindered. Information on SNAP E&T Participant Characteristics and Outcomes Is Not Reliable Although data on the number of overall participants in SNAP E&T programs in an average month from one FNS dataset are generally reliable, data on SNAP E&T program participant characteristics and outcomes are not reliable, according to our analysis of state data on SNAP E&T programs reported to FNS and our discussions with FNS and state officials. Specifically, in our review of the three FNS datasets that include state-reported information on SNAP E&T, we found several issues that affect the reliability of these data. According to our analysis, these data reliability issues include widely varying counts of SNAP E&T participants, ABAWDs, and work registrants across the datasets; missing or incomplete data on work registrants, ABAWDs, SNAP E&T participant characteristics and outcomes, and SNAP E&T services within the datasets; and inconsistencies within and between quarterly and annual reports of SNAP E&T participants in one of the datasets. For example, according to FNS officials, some states inaccurately reported participation in a single SNAP E&T service that exceeded the state’s total number of SNAP E&T participants. FNS has taken steps to address some of the SNAP E&T data limitations, including providing additional training and guidance to states. For example, FNS provided training to states in July 2014 and September 2018 on how to accurately report SNAP E&T participant information through one of the state-reported datasets on SNAP E&T. In addition, in response to state questions regarding how to collect new outcome measures on SNAP E&T required by the Agricultural Act of 2014, FNS issued two memoranda in 2016 and 2017 providing additional policy clarifications. Recently, in 2018, FNS issued two memoranda providing clarifications on work requirements for ABAWDs and on SNAP E&T, in part to improve the reliability of data collected. Even with these efforts, our analysis suggests that FNS continues to lack reliable data on SNAP E&T programs for at least two reasons: imprecise instructions on data collection forms and staff confusion at the state level. Imprecise instructions on data collection forms. According to our analysis, state-reported data on SNAP E&T participants and characteristics are not reliable due to imprecise instructions on the respective data collection forms. For example, the form used by states to collect information on SNAP recipients nationwide asks states to indicate if recipients are work registrants, and if so, participate in employment and training programs. Although FNS officials told us that this was intended to capture SNAP E&T participants alone, the form does not specify this. As a result, FNS officials explained that they believe states are incorrectly reporting SNAP recipients participating in any E&T program. Without a reliable link to SNAP E&T participation, FNS is unable to use this source, which provides detailed information on SNAP recipients’ demographic, educational, and economic characteristics, to analyze SNAP E&T participant characteristics. Similarly, in the case of another state-reported data source, we found that the form used to collect data on the types of SNAP E&T services participants receive does not list or define required services. According to FNS officials, states report widely varying SNAP E&T services within the same categories. Staff confusion at the state level. According to FNS officials, there has been widespread confusion among states regarding the need to track ABAWDs when waivers are in place. Consequently, some states were not tracking ABAWD participation or properly documenting SNAP recipients’ ABAWD status in recent years, according to FNS officials and some of the state SNAP agency officials we spoke with. FNS noted the importance of accurately tracking ABAWDs following the expiration of the waivers and reinstatement of the time limit in a March 2015 memorandum to regional directors. Further, FNS officials told us that states should have continued to track ABAWDs even if the state was under a statewide ABAWD waiver. FNS noted in its 2015 memorandum that states that failed to accurately track ABAWDs risked potential overpayments, as ABAWDs who fail to meet work requirements are ineligible for benefits. Further, although we found generally reliable SNAP E&T participation data in one FNS dataset, staff confusion has also likely affected these participation data. FNS officials told us that some states may mistakenly include those referred into SNAP E&T programs who did not participate in a program activity in their count of SNAP E&T participants. Finally, in the case of SNAP E&T data on outcomes, FNS regional officials told us that state-level staff were confused by the two different definitions for completion of a SNAP E&T activity used by FNS—an issue which may have affected the reliability of the outcome data. FNS has acknowledged that it is important to have reliable data on the SNAP E&T program for program oversight. Recently, in August and September 2018, FNS presented information to states at a national conference and in a webinar regarding the interactions of the different state-reported SNAP E&T data sources, and the importance of these data for funding and planning purposes. In a July 17, 2009 memorandum, FNS also stated that it is important that the agency collect reliable data on SNAP E&T to satisfy the increasing demands of Congress, advocacy groups, and the public for an accurate picture of the types of activities provided and participation patterns in those activities. This is generally consistent with federal internal control standards and our previous work on SNAP E&T. Federal internal control standards state that agencies should maintain quality data in order to produce and share quality information with stakeholders to help achieve agency goals. Further, in our 2003 report on SNAP E&T, we found that no nationwide data existed on whether SNAP E&T programs helped participants obtain employment, and we recommended that FNS collect nationwide data on program participants and require states to collect outcome measures. However, at present, the lack of reliable state-reported data on SNAP E&T participant characteristics and outcomes hinders FNS’ ability to effectively oversee and monitor the SNAP E&T program. Without such information, states, FNS, and the Congress are unable to fully assess whether agency goals are being met through the SNAP E&T program. Further, the lack of reliable state-reported data on work registrants and ABAWDs affects FNS’s ability to monitor states’ implementation of program rules, including work requirements, and ensure program integrity. In addition, as data on work registrants and ABAWDs are used to allocate federal funds to states for SNAP E&T, unreliable estimates of these groups have funding implications. FNS’s ability to understand the extent to which agency goals are being met is further hampered because FNS has not yet determined how it will use newly reported data to assess the performance of state SNAP E&T programs. As a result of provisions in the Agricultural Act of 2014, FNS required states to report new data on SNAP E&T participants’ outcomes, such as the median quarterly earnings of certain program participants, and participant characteristics, such as the percentage of participants who have received a high school diploma. In addition, the Act requires that FNS assess the effectiveness of states’ performance. In the preamble to the relevant interim final rule, FNS described at a high level how it intends to use the data, including identifying which program activities are most successful at moving individuals into employment. However, FNS officials told us that they were not yet certain how they will use the data to make such determinations. In addition, regional officials we spoke with stated that the current data might not allow FNS to answer questions about whether the program is achieving its goals. Similarly, state SNAP E&T officials we spoke with during our review did not know how the recently collected data related to program performance. Specifically, state officials in all five states we selected indicated that they were not certain how FNS will use these data to assess states’ performance. Officials in three states also said that a lack of clarity about how these data relate to program goals has led to confusion. FNS officials told us that they have not determined how they will use the newly reported data or whether the current data are sufficient, in part, because the agency has instead focused its resources on assisting the states in submitting the data to meet the new reporting requirements. According to FNS regional and national officials, states required extensive technical assistance to obtain the requisite data and calculate the reporting measures. For example, one regional official said that his office had been providing the states technical assistance for a year and a half to prepare them for the new reporting requirements. States we spoke with also indicated that the data were time-consuming and challenging to obtain. For example, many states struggled to obtain data sharing agreements with workforce agencies for the required employment data. According to FNS officials, after receiving the first round of reports in January 2018, FNS officials continued to provide technical assistance to states to improve the quality of the data, and FNS required states to submit revised reports in May. However, as of August 2018, one state and one territory had not submitted the required reports to FNS, according to FNS officials. In the absence of FNS taking steps to determine how it will use the newly reported data to assess state effectiveness, questions about whether SNAP E&T programs meet their goals will remain unanswered. Further, states may continue to be challenged to report these data, and without information from FNS on how state performance will be assessed, states may lack clarity on how collecting these data will help contribute to program goals. As of October 2018, FNS officials said that they are exploring ways to improve their ability to collect and analyze all of the program data necessary to do a comprehensive assessment of state SNAP E&T. Our prior work has emphasized the importance of establishing how performance data relates to program goals. In addition, federal internal control standards state that management should determine whether performance measures for the defined objectives are appropriate for evaluating the agency’s performance in achieving those objectives. Federal internal control standards also state that management should communicate necessary quality information to relevant internal and external parties to help the agency achieve its objectives. States Have Increasingly Partnered with Various E&T Providers, but Some States Have Not Leveraged Available Workforce Development System Resources States Have Increasingly Partnered with Other Entities to Provide SNAP E&T Services In recent years, state SNAP agencies have increasingly partnered with other state and local organizations, such as nonprofit community-based social service providers, community colleges, and workforce agencies, to provide services to SNAP E&T participants, according to FNS officials and states we selected for our review. In fiscal year 2018, 50 state SNAP agencies partnered with at least one other organization to deliver SNAP E&T services, with the majority partnering with more than one, according to an analysis by FNS (see fig. 7). In that year, 36 states partnered with community-based social service providers, 33 states had partnerships with workforce agencies, and 24 states partnered with community colleges. FNS officials in all seven regions said that states have increasingly used an approach FNS refers to as third party partnerships in recent years to leverage outside funding to serve SNAP E&T participants. In this model, according to FNS officials, third party organizations use non-federal funding to provide allowable E&T services and supports to SNAP recipients, and state SNAP agencies are then eligible for a federal reimbursement of 50 percent of these expenditures. FNS has promoted this third party partnership model through various technical assistance resources provided to states, including an operations handbook and webinars, and has added a dedicated position for a SNAP E&T official in each regional office, in part, to help develop these partnerships. Federal 50 percent reimbursement funds expended increased from nearly $182 million to more than $223 million, or by 23 percent, from fiscal year 2007 to fiscal year 2016. According to FNS national officials as well as officials in some FNS regions and states, partnerships play a critical role in SNAP E&T programs because state SNAP agencies may lack the capacity, resources, and expertise to provide the type of intensive employment and training services FNS considers most likely to lead to self-sufficiency for SNAP recipients. For example, two of our selected states reported that they have partnered with community colleges to train participants for local in-demand occupations, including information technology, healthcare, and welding. According to officials in one FNS regional office, community- based social service providers and community colleges may have staff with expertise in workforce development, which SNAP agencies may not have, and this enables SNAP agencies to expand their programs and services without the expense of growing their own staff. According to officials in some FNS regions and some of our selected states, partnering with workforce agencies has enabled some states to provide training to participants using Workforce Innovation and Opportunity Act (WIOA) funds and supportive services using SNAP E&T funds, maximizing their ability to address participants’ needs. Officials in one of the states we visited also said that partnering with the workforce agency allows them to ensure basic E&T services, such as job search assistance, are available to SNAP recipients across all counties in their state. (See fig. 8.) FNS officials also said that these partnerships better position states to improve their program outcomes by tapping into providers currently serving communities that include SNAP recipients. For example, one of our selected states partnered with nonprofit community-based social service providers experienced in working with homeless and previously incarcerated populations. Officials in this state said that the providers tailor E&T services based on their knowledge of these populations’ unique barriers to employment. Further, officials in three of our five selected states said that some of the community-based social service organizations they partner with provide SNAP E&T participants with additional supportive services, including transitional housing, clothing, financial advising, and mental and physical health services, to address a broader set of barriers to employment. Some States Do Not Leverage Workforce Development System Resources for SNAP E&T, and FNS Has Not Assessed State Efforts to Utilize these Resources Although states are increasingly partnering with external entities to provide SNAP E&T services, according to FNS data for fiscal year 2018, 20 state SNAP agencies have not partnered with workforce agencies for SNAP E&T. According to FNS officials, the nationwide network of more than 2,500 American Job Centers, which are operated by state and local workforce agencies, can help to fill service gaps in areas lacking community based organizations or community colleges. However, despite the broad availability of E&T services such as job search assistance through American Job Centers, 12 state SNAP agencies directly provided job search or job search training for their SNAP E&T programs, according to their fiscal year 2017 state SNAP E&T plans. In addition, some states have not yet fully leveraged resources from the broader workforce development system, which includes workforce agencies, community-based organizations, and community colleges, to provide SNAP E&T services. For example, FNS data for fiscal year 2018 show that three states’ SNAP agencies operated their own SNAP E&T programs in fiscal year 2018 and did not involve existing workforce development system entities in the provision of these services. According to their fiscal year 2017 state plans, these states each offered one or two types of SNAP E&T services, and the services they offered—primarily job search and job search training—are considered less intensive by FNS officials. In contrast, states with workforce development system partnerships offered a broader range of services, as well as more intensive services, such as vocational education. For example, all 36 state SNAP agencies that offered vocational education did so through workforce development system partnerships. As previously noted, FNS officials have said that intensive services are likely more effective in moving SNAP E&T participants, who may be more likely to have barriers to employment, toward self-sufficiency. Overlap and a lack of coordination in federally-funded E&T programs is a long-standing concern, and relatedly, state SNAP agencies are required to make use of workforce development system resources for SNAP E&T, when possible. In our prior work, we found that SNAP E&T was 1 of 47 federally funded E&T programs, nearly all of which overlapped with at least one other program by providing similar services to similar populations. We noted that overlap among federal E&T programs raises questions about the efficient use of resources, and we highlighted the value of coordination between these programs to ensure efficient and effective use of resources. Consistent with our findings, federal regulations require that each component of a state agency’s SNAP E&T program be delivered through its statewide workforce development system, unless the component is not available locally through such a system. FNS national and regional officials, as well as state officials, described challenges states face in forming effective workforce development system partnerships. FNS officials said that challenges are often caused by differences in workforce agency and SNAP E&T program target populations and service delivery approaches. According to FNS, SNAP E&T participants often have more barriers to employment, such as low literacy and limited work experience, than the broader population served by workforce agencies. Because those with employment barriers could adversely impact the workforce agencies’ employment and earnings performance, which could jeopardize agencies’ workforce program funding, workforce agency staff are sometimes reluctant to serve SNAP E&T participants, according to FNS national and regional officials in three of the seven regions, as well as officials in one of our selected states. For example, officials in one region said that workforce agency staff had stopped serving SNAP E&T participants in the past when they realized the participants needed more supportive services or time in workforce programs to meet employment goals. Recognizing these challenges, in recent years, USDA has urged state SNAP agencies to collaborate with workforce agencies and others to improve coordination of E&T services. For example, in March 2016, USDA and the Department of Labor issued a joint letter encouraging state SNAP agencies and state and local workforce agencies to work together to develop shared strategies to better connect SNAP recipients with E&T opportunities through American Job Centers. FNS has also provided states with technical assistance materials on SNAP E&T and WIOA partnerships, which describe respective program requirements and how SNAP E&T and WIOA-funded workforce programs can complement one another. However, FNS has not ensured that all states take steps to identify potential workforce development system partners. Federal internal control standards state that agencies should collect complete and reliable information to ensure effective monitoring. FNS officials told us that they do not independently assess the availability of states’ workforce development system partners but instead rely on states to document this information in their state SNAP E&T plans, a key tool used by FNS for program monitoring. However, we found that 24 states did not provide information in their fiscal year 2017 SNAP E&T plans that would allow FNS to verify whether these states had assessed available workforce development system providers. For example, the states’ plans did not describe existing workforce development services in the state, despite FNS guidance that directs states to describe the statewide workforce development system and identify the E&T services that will be delivered through this system in their plans. States that are not fully leveraging resources available through the workforce development system may miss opportunities to provide a wider variety of services to SNAP E&T participants and serve a greater number of SNAP recipients through SNAP E&T programs. If state SNAP agencies do not assess workforce development system resources available in their state, they may lack awareness of potential partners and the resources they offer, potentially leading to an inefficient use of resources. In addition, without complete and reliable information on states’ available workforce development system resources, FNS is not able to ensure that states are complying with the requirement to deliver SNAP E&T services through their state workforce development systems. Conclusions FNS has made strides in recent years to provide additional support and oversight of states’ SNAP E&T programs, yet the agency lacks complete and accurate information on these programs, which may limit the effectiveness of its efforts. For example, SNAP E&T programs have served a small percentage of SNAP recipients over time, and while FNS recognizes that states lack information on strategies for increasing participation among those referred to the SNAP E&T program, it has not provided technical assistance in this area. As a result, FNS may miss opportunities to help more SNAP recipients receive program services intended to increase their self-sufficiency, a USDA strategic goal. FNS’s ability to assess whether the program is assisting the department in meeting this goal is also hindered because FNS lacks reliable data on SNAP E&T participant characteristics and outcomes. Without reliable data on SNAP recipients subject to work requirements and participation in SNAP E&T, the agency’s ability to monitor states’ implementation of program rules to ensure recipients are not receiving benefits for which they are ineligible is also limited. Further, because FNS has not yet determined how it will use the newly required outcome and participant characteristics data to assess state SNAP E&T programs, questions about program performance remain unanswered. In addition, without information from FNS on how state performance will be assessed, states will continue to lack clarity on how reporting these data will help contribute to program goals. Finally, because partnerships can be a crucial source of additional capacity, resources, and expertise for SNAP E&T programs, states that are not fully leveraging available workforce development system resources may miss opportunities to serve a greater number of SNAP recipients through SNAP E&T and provide a wider variety of services to SNAP E&T participants. In addition, states may provide overlapping or duplicative services and use resources inefficiently, because FNS has not ensured that all states take steps to identify potential workforce development system partners. Recommendations for Executive Action We are making the following four recommendations to FNS: The Administrator of FNS should identify and disseminate strategies to states and service providers for increasing the participation of SNAP recipients referred to the SNAP E&T program. (Recommendation 1) The Administrator of FNS should take additional steps to address data reliability issues in the state-reported data on SNAP E&T participant characteristics and outcomes, including steps to address imprecise instructions on data collection forms and staff confusion at the state level. (Recommendation 2) The Administrator of FNS should determine and communicate to states how the agency will use newly reported outcome and participant characteristics data to assess the effectiveness of state SNAP E&T programs. (Recommendation 3) The Administrator of FNS should take additional steps to assist states in leveraging available workforce development system resources. Such steps should include ensuring that state SNAP E&T plans provide the agency with sufficient information to verify that states have assessed available workforce development system providers. (Recommendation 4) Agency Comments We provided a draft of this report to USDA for review and comment. On November 5, 2018, the Deputy Associate Administrator for SNAP and FNS officials from SNAP’s Office of Employment and Training provided us with the agency’s oral comments. FNS officials told us that they generally agreed with the recommendations in the report. They noted that they have been implementing strategies to help states improve their SNAP E&T programs, including expanding the reach of the programs and improving the reliability of state reported data. FNS officials stated that the agency plans to build on these current efforts to address the recommendations. We acknowledge the agency’s ongoing efforts in our report but continue to believe that additional action is necessary to address our recommendations. FNS also provided technical comments, which we incorporated into the report as appropriate. As agreed with your offices, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies to the Secretary of the USDA, congressional committees, and other interested parties. In addition, this report will be available at no charge on the GAO website at https://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-7215 or larink@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix II. Appendix I: Objectives, Scope, and Methodology This appendix discusses in detail our methodology for addressing our two research objectives examining (1) what is known about Supplemental Nutrition Assistance Program (SNAP) employment and training (E&T) program participants and outcomes over time and (2) the extent to which state SNAP E&T programs have partnered with other programs offering similar services. We scoped our review of state SNAP E&T programs to include the 50 states, the District of Columbia, Guam, and the Virgin Islands. In addition to the methods we discuss below, to address both our research objectives, we reviewed relevant federal laws, regulations, and guidance; interviewed United States Department of Agriculture (USDA) Food and Nutrition Service (FNS) officials in its headquarters and seven regional offices; and reviewed relevant research from FNS and the USDA Office of Inspector General, as well as our prior work on SNAP E&T programs. Further, we interviewed representatives of a range of nationwide organizations knowledgeable about SNAP E&T and state officials from seven state SNAP E&T programs: Idaho, Louisiana, New York, Pennsylvania, Tennessee, Washington, and the District of Columbia. We also analyzed SNAP E&T expenditures using form FNS- 778 data for fiscal years 2007 through 2016, the most recent data available. The form FNS-778—Federal Financial Report—is a form used by FNS to collect quarterly expenditure data for state SNAP E&T programs. To assess the reliability of these data, we interviewed FNS and state officials, performed data testing, and reviewed relevant documentation. We determined these data to be sufficiently reliable for the purposes of our report. We excluded from our review the SNAP E&T pilot programs that were authorized by the Agricultural Act of 2014 because these are being evaluated separately by FNS. We conducted this performance audit from September 2017 to November 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. SNAP E&T Program Data To address our first objective, we analyzed data on SNAP E&T participation from three FNS data sources. First, we analyzed aggregate data on SNAP E&T participants collected from state SNAP agencies for fiscal years 2008 through 2016, the most recent data available. Second, for the same time period, we analyzed Quality Control data on individual SNAP recipients, work registrants, and SNAP E&T participants. Finally, we reviewed and analyzed aggregate participation data from state SNAP agencies’ fiscal year 2017 outcome and participant characteristics reports. Form FNS-583 Data We analyzed the average monthly number of SNAP recipients participating in SNAP E&T using the form FNS-583 data. The form FNS- 583—U.S. Department of Agriculture Food and Nutrition Service SNAP Employment and Training (E&T) Program Activity Report—is used by FNS to collect quarterly and annual participation data for state SNAP E&T programs. To assess the reliability of these data, we interviewed FNS and state officials, performed data testing, and reviewed relevant documentation. Data testing included checks for missing data elements, duplicative data, and values outside a designated range. We determined the data were sufficiently reliable to identify the number of average monthly SNAP E&T participants and to assess change over time. To further examine what is known about participation in SNAP E&T, we also assessed form FNS-583 data on work registrants and able-bodied adults without dependents (ABAWDs) participating in SNAP E&T for fiscal years 2008 through 2016. To assess the reliability of these data, we interviewed FNS and state officials, performed data testing, and reviewed relevant documentation. We determined these data to be unreliable for the purposes of our report. As described above, for example, FNS officials learned in recent years that there was widespread confusion among states regarding the need to track ABAWDs when waivers were in place. Consequently, some states were not tracking ABAWD participation or properly documenting SNAP recipients’ ABAWD status. FNS Quality Control Data We analyzed SNAP Quality Control data on individual SNAP recipients, work registrants, and SNAP E&T participants. The SNAP Quality Control database contains detailed demographic, economic, and SNAP eligibility information for a nationally representative sample of SNAP households. We estimated the number of SNAP recipients and work registrants for fiscal years 2008 and 2016 using the public use Quality Control dataset and calculated confidence intervals to determine if the change over time was statistically significant (see table 1). To assess the reliability of these data, we interviewed officials from FNS and the contractor responsible for maintaining the Quality Control dataset, as well as state officials; reviewed relevant technical documentation; and conducted data testing. For example, we compared the estimates we produced for fiscal years 2008 and 2016 to the publicly reported estimates in the annual Characteristics of Supplemental Nutrition Assistance Program Households reports for those years. We determined that the data, and the corresponding estimates in these reports, were sufficiently reliable for our purposes. As a result, for fiscal years 2009 through 2015, we relied on the estimates of SNAP recipients and work registrants published in the reports. We also analyzed SNAP Quality Control data on SNAP E&T participants for fiscal year 2016. To assess the reliability of these data, we interviewed officials from FNS and the contractor responsible for maintaining the Quality Control dataset, as well as state officials; reviewed relevant technical documentation; and conducted data testing. For example, we compared the estimate of SNAP E&T participants from the SNAP Quality Control dataset to the number of SNAP E&T participants reported by states on the FNS-583, which we had determined was reliable. From our review, we determined the Quality Control SNAP E&T participation data to be unreliable for the purposes of our report. As described above, for example, the form used by states to collect information on SNAP recipients nationwide asks states to indicate if recipients participate in employment and training programs. Although FNS officials told us that this was intended to capture SNAP E&T participants alone, the form does not specify this, and FNS officials said that some states are incorrectly reporting SNAP recipients participating in any E&T program. To determine the percentage of SNAP recipients and work registrants that participate in SNAP E&T, we used the data that we had determined were reliable. Specifically, we used the Quality Control data on SNAP recipients and work registrants, as well as the form FNS-583 data on SNAP E&T participants, for fiscal years 2008 through 2016. SNAP E&T Outcome and Participant Characteristics Data We also reviewed and analyzed fiscal year 2017 outcome and participant characteristics data reported by state SNAP agencies in the SNAP E&T Annual Report Federal Fiscal Year 2017. These data include information on SNAP E&T participants’ outcomes, such as the median quarterly earnings of program participants, and participant characteristics, such as the percentage of participants who have received a high school diploma. Certain outcome data were only collected by FNS for two quarters of fiscal year 2017, whereas participant characteristics data were collected for the entire year. We received copies of these data reports from FNS as states submitted their initial reports to FNS in early 2018. Subsequent to FNS’ review of these initial reports and their efforts to help states improve the accuracy and consistency of their reporting, FNS provided us with updated versions of the reports for many of the states. We used the reports to describe rates at which SNAP recipients referred to the SNAP E&T program participated in services—data that were reported by 11 states. We did not validate the accuracy of these data. State SNAP E&T Plans and FNS Program Characteristics Data To address our second objective on the extent to which state SNAP E&T programs have partnered with other programs offering similar services, we reviewed fiscal year 2017 SNAP E&T state plans for all 53 state SNAP agencies. Specifically, we reviewed the plans to determine which services states planned to offer through partnerships with other programs in that year and the extent to which states documented their use of available workforce development system resources. To supplement our review of the plans, we also analyzed fiscal year 2018 summary data from FNS on the number of state SNAP agencies that partnered with community-based organizations, workforce agencies, and community colleges, as well as the number with state SNAP agency-operated SNAP E&T programs. We also analyzed fiscal year 2010 and 2017 summary data from FNS on mandatory and voluntary programs to determine how the number of state SNAP agencies with each program type changed over time. To assess the reliability of the FNS summary data, we interviewed FNS and state officials and reviewed relevant documentation. We determined these data to be sufficiently reliable for the purposes of this report. State Interviews and Site Visits To help inform both of our objectives and gather additional information about state SNAP E&T programs, we selected five states: Delaware, Oregon, Kansas, Texas, and Virginia. We selected these states based on several criteria to ensure our sample included state SNAP E&T programs with different service delivery approaches and other program characteristics, as well as geographic diversity. Specifically, we considered state SNAP E&T participation and expenditures, including utilization of federal 50 percent reimbursement funds. In addition, we considered whether the state operated a mandatory or voluntary SNAP E&T program, a county- or state-administered program, and opted to be an ABAWD pledge state. We also considered whether the state submitted its SNAP E&T plan as part of a Workforce Innovation and Opportunity Act Combined State Plan. Using semi-structured questions, we interviewed officials from the state agencies responsible for administering SNAP in the five selected states. We gathered information on SNAP E&T administration at the state level, including information on partnerships; program participation and expenditures; data collection efforts, including those related to assessing program outcomes; and any challenges to administering the program, as well as efforts to address such challenges. We conducted site visits to our selected states in which services are provided through partnerships with local providers—Delaware, Oregon, Texas, and Virginia—and interviewed selected local program staff with knowledge of SNAP E&T program operations, participant characteristics, and coordination with the state SNAP agency who provide SNAP E&T services in both urban and rural areas. We conducted these visits in February and March 2018. During each site visit, we used semi- structured questions to gather information on the goals and mission of the providers’ organizations, types of services provided to SNAP E&T participants, needs and characteristics of SNAP E&T participants and how these might differ from those of other clientele, sources of funding used to provide services to SNAP E&T participants, and efforts to coordinate with the state SNAP agency. The local program staff we interviewed included representatives of workforce agencies, non-profit community-based organizations, a for-profit company, and community colleges. Information collected from state and local SNAP E&T officials during our site visits cannot be generalized to all SNAP E&T officials nationwide. Appendix II: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Rachel Frisk (Assistant Director), Kristen Jones (Analyst-in-Charge), Morgan Jones, and Kelly Snow made key contributions to this report. Also contributing to this report were Alex Galuten, Mimi Nguyen, Sam Portnow, Julia Robertson, Monica Savoy, Almeta Spencer, Jeff Tessin, Kathleen van Gelder, Nicholas Weeks, and Jessica L. Yutzy.

Why GAO Did This Study SNAP is the nation's largest federally funded nutrition assistance program. In fiscal year 2017, it provided about $64 billion in benefits. To maintain eligibility for benefits, certain SNAP recipients must comply with the program's work requirements, which may include participating in a state's SNAP E&T program if required by the state. This report examines (1) what is known about SNAP E&T program participants and outcomes over time and (2) the extent to which state SNAP E&T programs have partnered with other programs offering similar services. GAO reviewed relevant federal laws, regulations, and guidance; analyzed USDA data on SNAP recipients, work registrants, and SNAP E&T participants from fiscal years 2008 through 2016, the most recent data available; reviewed states' fiscal year 2017 SNAP E&T plans and outcome reports; and interviewed USDA officials and state officials in five states selected, in part, to reflect a range of SNAP E&T program characteristics. What GAO Found The Supplemental Nutrition Assistance Program's (SNAP) Employment and Training (E&T) programs, which are overseen by the U.S. Department of Agriculture (USDA) and administered by states, have served a small percentage of SNAP recipients over time, and information on participant characteristics and outcomes is limited. In an average month of fiscal year 2016, SNAP E&T served about 0.5 percent of the 43.5 million SNAP recipients. Further, since 2008, the percentage of SNAP recipients served by SNAP E&T has declined. Participation in SNAP E&T may be low, in part, because most SNAP recipients were exempt from work requirements, according to USDA data. In addition, SNAP recipients may participate in other activities to comply with work requirements. Although data on the number of recipients served in SNAP E&T are generally reliable, USDA lacks reliable data on participant characteristics and outcomes because of imprecise instructions on data collection forms and staff confusion at the state level. USDA has taken some steps to address these issues, but data reliability issues persist. As a result, USDA's ability to assess whether agency goals are being met through the SNAP E&T program is limited, as is the department's ability to monitor states' implementation of work requirements and ensure program integrity. In fiscal year 2018, most state SNAP agencies partnered with workforce development system entities, such as community colleges and workforce agencies, to provide services to SNAP E&T participants, according to USDA data. Regional and state officials reported that state SNAP agencies often have used these partnerships to leverage non-federal funding sources and provide additional capacity and expertise to help expand SNAP E&T services. However, 3 states operated their own SNAP E&T programs without partnering with any other program, and a total of 20 states lacked partnerships with workforce agencies, according to USDA data for fiscal year 2018. Federal regulations require that SNAP E&T services be delivered through the state's workforce development system unless the services are not available locally through this system. USDA and state officials described challenges to forming effective partnerships with workforce agencies, including perceived disincentives to serving SNAP recipients. However, states that are not fully leveraging resources available through the workforce development system may miss opportunities to provide a wider variety of services to SNAP E&T participants and serve a greater number of SNAP recipients through SNAP E&T. What GAO Recommends GAO is making four recommendations, including that USDA take additional steps to address SNAP E&T data reliability issues and to help states leverage available workforce development system resources. USDA officials generally agreed with our recommendations.

gao_GAO-18-456

gao_GAO-18-456_0

Background This section describes (1) the purpose of LEPs and the process that NNSA and DOD use to manage them, known as the phase 6.X process; (2) the management of the ongoing LEP for the W76 warhead—an important historical reference for the B61-12 LEP—and the status of the two other ongoing LEPs; (3) future nuclear modernization plans and our past conclusions and recommendations on the affordability of these plans; and (4) the objectives of the B61-12 LEP and the roles and responsibilities of NNSA and the Air Force in conducting the program. LEPs’ Purpose and the Phase 6.X Process for Managing Them NNSA and DOD undertake LEPs to refurbish or replace nuclear weapons’ components to extend their lives, enhance their safety and security characteristics, and consolidate the stockpile into fewer weapon types to minimize maintenance and testing costs while preserving needed military capabilities. NNSA and DOD jointly manage LEPs under a multi-step process known as the phase 6.X process. The B61-12 LEP is currently in phase 6.4 (production engineering) of this process. Figure 1 illustrates the phase 6.X process. The phase 6.X process and the roles and functions of DOD, DOE, and NNSA in nuclear weapon refurbishment activities are described in a guidance document known as the Procedural Guideline for the Phase 6.X Process. The document also calls for NNSA to formally update its program cost estimate and reissue it as the baseline cost report prior to entering phase 6.4. In January 2017, NNSA issued a supplemental directive that also directs the Office of Cost Estimating and Program Evaluation to prepare an independent cost estimate for each nuclear weapon system undergoing life extension before an LEP enters phase 6.4. The Procedural Guideline for the Phase 6.X Process also describes the roles and functions of two joint bodies that provide oversight and approval functions to LEPs and other nuclear weapons-related activities: the Nuclear Weapons Council and its Standing and Safety Committee. The Nuclear Weapons Council is the joint DOD and DOE activity that serves as the focal point for interagency activities to maintain the nuclear weapons stockpile. Its membership includes the Under Secretary of Defense for Acquisition, Technology and Logistics (generally the Chair); the Under Secretary of Defense for Policy; the Vice Chairman of the Joint Chiefs of Staff; the Commander of U.S. Strategic Command; and the Department of Energy’s Under Secretary for Nuclear Security, who also serves as the Administrator of the National Nuclear Security Administration. In addition, the Nuclear Weapons Council charters a Project Officers Group for each weapon system to provide a technical forum for weapon development and management activities. Each Project Officers Group is led by a project officer from either the Navy or Air Force, the two military services that maintain and operate nuclear weapons. Management of the W76-1 LEP and Other Ongoing LEPs According to B61-12 program officials, the W76-1 LEP—which NNSA expects to complete in fiscal year 2019—has served as an important historical reference as NNSA prepared its plans and cost estimates for the B61-12 LEP. In August 2017, NNSA issued a study documenting lessons learned from difficulties it encountered in managing the W76-1 LEP. According to the study, prior to the W76-1 LEP, NNSA had not undertaken full-scale weapon system design activities since the 1982 design of the W88 warhead. Among other findings, the lessons learned study stressed the importance of using modern tools to validate and manage an LEP’s system and technical requirements to maintain cost, schedule, and performance during all phases of the program. This finding is consistent with our March 2009 findings that NNSA and DOD established an unrealistic schedule for the W76-1 LEP, did not establish a consistent cost baseline, and did not effectively manage technical risks in the program. These problems resulted in delays, additional expenditures, and difficulties tracking the cost of the program. Notably, the program had to delay first production of the W76-1 from September 2007 to September 2008 when it encountered problems with the final test batch of a key material, known as Fogbank. We recommended that NNSA develop realistic schedules for the W76-1 and future LEPs that build in additional time for unexpected technical challenges that may delay the programs. NNSA agreed with our recommendation and has taken steps toward improvement in this area, which we continue to monitor. In addition to the B61-12 and W76-1 LEPs, NNSA and DOD are managing two other LEPs: the W88 Alteration 370 program and the W80- 4 LEP. Table 1 provides basic information on all four ongoing LEPs. Future Nuclear Modernization Plans and Their Affordability In addition to the four ongoing LEPs, NNSA and DOD have outlined plans for several future nuclear weapon modernization programs: Under NNSA’s current program of record, which does not yet reflect new requirements that will be generated based on the 2018 Nuclear Posture Review, NNSA has plans for life extension efforts to transition the nuclear stockpile to three interoperable ballistic missile warheads and two air-delivered weapons. NNSA has described this plan as the 3+2 strategy. To undertake this strategy, NNSA has proposed initiating a series of interoperable warhead programs between about 2020 and 2060. NNSA’s plans for the first ballistic missile warhead in the 3+2 strategy—the Interoperable Warhead 1—indicate that, if authorized by Congress, the warhead would cost an estimated $12.4 billion from 2020 to 2041. As we reported in August 2015, NNSA paused the Interoperable Warhead 1 program in fiscal year 2014 to provide more time to study the concept of interoperability and to reduce uncertainty about the agency’s ability to achieve necessary plutonium and uranium capabilities to support the LEP. Under its current program of record, NNSA plans to resume the Interoperable Warhead 1 program in fiscal year 2019. Under its current program of record, NNSA has also begun preliminary planning for Interoperable Warhead 2, Interoperable Warhead 3, and B61-12 follow-on programs that, if authorized, would start in the 2020s and 2030s. In the 2018 Nuclear Posture Review, DOD stated a near-term intention to modify a small number of existing submarine-launched ballistic missile warheads to provide a low-yield option, and a long- term intention to pursue a modern nuclear-armed sea-launched cruise missile. The NNSA Administrator stated in March 2018 that NNSA would continue to work with DOD to determine the resources, time, and funding required to address these and other policies specified in the Nuclear Posture Review. As we concluded in an April 2017 report, these plans come during a particularly challenging decade for NNSA’s nuclear modernization efforts, as the agency plans to simultaneously execute at least four LEPs along with major construction projects, such as efforts to modernize NNSA’s uranium and plutonium capabilities. We further concluded that NNSA’s modernization budget estimates for fiscal years 2022 through 2026 may exceed the funding levels programmed for modernization in future budgets, raising affordability concerns. Moreover, we concluded that NNSA had not addressed a projected “bow wave” of future funding needs—that is, an impending and significant increase in requirements for additional funds—or the mismatch between potential funding needs and potential funding available. We recommended that NNSA include an assessment of the affordability of NNSA’s portfolio of modernization programs in future versions of the Stockpile Stewardship and Management Plan—for example, by presenting options NNSA could consider to bring its estimates of modernization funding needs into alignment with potential future budgets. NNSA did not explicitly agree or disagree with our recommendation, but we will continue to monitor any actions NNSA takes in response to the recommendation. Objectives of the B61-12 LEP and the Roles and Responsibilities of NNSA and the Air Force The B61-12 LEP has several objectives: consolidating the nuclear bomb stockpile, improving the accuracy of the resulting weapon through a new guidance assembly, and addressing other age-related issues. Consolidating the stockpile. Under the B61-12 LEP, NNSA and the Air Force plan to consolidate and replace four of the five variants of the B61 that were in the active stockpile at the time the B61 LEP began. Improving accuracy. The B61-12 is to be equipped with a new tail kit guidance assembly that enables it to be delivered with greater accuracy than the B61 bombs it replaces, which are equipped with parachutes. More specifically, according to Air Force officials and documents, the assembly will provide the B61-12 with a guided freefall capability while retaining a ballistic (unguided) delivery capability. The greater accuracy of the B61-12 is to enable the B61-12 to meet all the military requirements for which past versions of the B61 were designed. Addressing other age-related issues. The B61-12 LEP is to extend the service life of the B61 by at least 20 years, make field maintenance of the weapon easier for Air Force technicians, and provide modern security features. NNSA manages its B61-12 LEP activities through a federal program office on Kirtland Air Force Base in Albuquerque, New Mexico, under the direction of the federal program manager. It manages the work of six government-owned, contractor-operated NNSA laboratories and sites that serve as design and production agencies for the LEP. Sandia National Laboratories, also located on Kirtland Air Force Base, serves as the systems-level integrator for the overall weapon design. Figure 2 shows the six sites participating in the B61-12 LEP and their respective roles. The Air Force’s responsibilities, in addition to managing the acquisition of the tail kit guidance assembly, include integrating the B61-12 with its delivery aircraft and the operational flight program software of these aircraft. The Air Force Nuclear Weapons Center, also at Kirtland Air Force Base and under the direction of the Air Force lead project officer, manages technical integration and other LEP-related tasks required to qualify, certify, and field the weapon. The delivery aircraft that carry the B61-12 are being designed to deliver the weapon in two different modes with two different systems, the second of which provides the enhanced capabilities offered by the new tail kit guidance assembly. System 1 aircraft will have an analog interface with the B61-12 that is designed to deliver the weapon in a ballistic mode, with the tail kit in a fixed position. System 2 aircraft will have a digital interface with the B61-12, enabling the guided delivery capability afforded by the tail kit assembly. Figure 3 illustrates the delivery aircraft for the B61-12. NNSA Substantially Incorporated Best Practices in Developing the B61- 12 Program Cost Estimate NNSA substantially incorporated most of the cost estimating best practices identified by our past work when it developed the $7.6 billion program cost estimate for the B61-12 LEP. Our cost estimating guide identifies best practices for developing a high-quality, reliable cost estimate and identifies four characteristics of such an estimate: it is comprehensive, well-documented, accurate, and credible. These four characteristics and some of the best practices that underlie them are illustrated in figure 4. We assessed the B61-12 program cost estimate by comparing it with the best practices identified in our cost estimating guide and found that it substantially met the criteria for all four characteristics of a high-quality, reliable cost estimate (see fig. 5). A summary of our assessment is presented below, including reasons that the program cost estimate substantially met the criteria under each of the four characteristics as well as some examples of the best practices that the cost estimate could have more fully incorporated. Appendix I provides additional information on our assessment. Comprehensive: Substantially Met. The program established a consistent and clearly defined work breakdown structure—a hierarchical structure that subdivides the work necessary to accomplish the program’s objectives into smaller elements—to ensure that costs were not double-counted or omitted. The clearly defined work breakdown structure also helped the B61-12 program office manage the process of integrating each site contractor’s estimate for the cost of its activities into the overall program estimate. To more fully incorporate the criteria for a comprehensive cost estimate, the program would have had to take additional steps, such as including the full life-cycle costs associated with the B61-12 weapon in the estimate. Specifically, the estimate would have had to include costs such as program costs incurred prior to phase 6.3, the cost of NNSA federal program office personnel, components that are being shared by different nuclear weapon programs (such as the weapon’s radar), and costs associated with maintenance of the B61-12 after the LEP ends and the weapon enters the stockpile. In addition, the estimate would have had to include an assessment of how the program would be affected if key assumptions, such as the timing of the delivery of the tail kit guidance assembly, did not hold true. Nevertheless, the program incorporated practices that substantially met the criteria for a comprehensive cost estimate, which we believe contributed to the program’s estimate being reliable. Well-documented: Substantially met. In our visits to NNSA sites and our associated review of site contractors’ documents, we found that site contractors provided detailed documentation of their contributions to the cost estimate to the B61-12 program office. At all of the sites we visited, experienced cost estimating teams captured specific information on the data and data sources used to inform their estimates. To more fully incorporate the criteria for a well-documented cost estimate, the documentation that the site contractors provided to the NNSA program office would have had to capture the reliability of the underlying data and discuss how the data were normalized. Nevertheless, the program incorporated practices that substantially met the criteria for a well-documented cost estimate, which we believe contributed to the program’s estimate being reliable. Accurate: Substantially met. Technical personnel at both the NNSA sites and the Albuquerque federal program office discussed program risks to ensure that the program estimate represented a most likely, unbiased cost. Furthermore, all of the site-level cost estimates we examined—which the federal program office integrates into the overall program cost estimate—drew on historical data from primary sources, including internal financial systems from either past B61 costs or previous LEPs. Use of such sources is consistent with the best practice of grounding the estimates in a historical record of cost estimating and actual experiences on other comparable programs. In addition, the federal program office routinely reviews contract performance reports from each of the B61-12 sites to track variances between estimated and actual costs on a monthly basis. To more fully incorporate the criteria for an accurate cost estimate, the program would have had to use site estimates that were calculated in base- year dollars and then uniformly adjusted for inflation at the program level, and clearly defined the method it used to determine inflation indexes. Instead, all of the site contractors developed their cost estimates in then-year dollars and applied varied inflation indexes. Nevertheless, the program incorporated practices that substantially met the criteria for an accurate cost estimate, which we believe contributed to the program’s estimate being reliable. Credible: Substantially met. The B61-12 LEP became the first LEP to undergo a statutorily required independent cost estimate, conducted by the Office of Cost Estimating and Program Evaluation. Additionally, a different NNSA office developed a third cost estimate for the program to aid in the preparation of NNSA’s budget materials. Each of these three estimates used a different methodology. NNSA used this third estimate to cross-check overall program costs. Moreover, to assess risk and uncertainty in the program, most of the site estimates we reviewed included a detailed, quantifiable risk assessment for their portion of the overall program estimate. To more fully incorporate the criteria for a credible cost estimate, the program’s sensitivity analysis would have had to more fully examine and document cost impacts for the overall estimate and the individual site estimates. Instead, according to NNSA officials, it focused primarily on schedule and critical path analysis. Moreover, to more fully incorporate the criteria for a credible cost estimate, the program would have had to address risk correlation and the calculation of confidence levels differently. In the program’s analysis of risks and uncertainties in the program, we found the program inconsistently examined correlation among program risks. Specifically, according to NNSA officials, to arrive at the 70 percent confidence level for the overall program cost estimate, the program office added site-level cost estimates together at the 50 percent and 70 percent confidence levels. As noted in our cost guide, adding risk results for the underlying estimates in this way results in an incorrect confidence level for the overall estimate. Nevertheless, the program incorporated practices that substantially met the criteria for a credible cost estimate, which we believe contributed to the program’s estimate being reliable. We consider a cost estimate to be reliable if the overall assessment ratings for each of the four characteristics are substantially or fully met—as was the case with the B61-12 program cost estimate, which substantially met these criteria. For that reason, we are not making recommendations related to the program’s use of cost estimating best practices. However, by fully incorporating all of the best practices for the four characteristics, NNSA can better ensure that its future cost estimates are of high quality and reliable. Varying Methods and Assumptions about Future Program Performance Led to Differences between the Estimates, but NNSA Did Not Document the Rationale for Adopting the Program’s Estimate Unchanged The $7.6 billion program cost estimate for the B61-12 LEP differs from the $10 billion independent cost estimate primarily because the program office used different methods and assumptions than the Office of Cost Estimating and Program Evaluation, which prepared the independent cost estimate. The B61-12 program developed its estimate by compiling site- specific cost and schedule estimates for activities at each of the NNSA sites participating in the LEP; in contrast, the independent cost estimate projected a cost and completion date by evaluating program activities completed to date and applying a historical model to estimate costs and durations for remaining activities. As noted in our cost guide, both of these methods are commonly applied. To reconcile the differences between the two estimates, high-ranking NNSA officials met with officials from the B61-12 program office and the Office of Cost Estimating and Program Evaluation to discuss the estimates in 2016. However, NNSA did not document the rationale for its decision to use the program office’s lower estimate unchanged or a plan for how it would take the independent cost estimate into consideration. We previously recommended that NNSA should establish a requirement for its management to document and justify key decisions based on a reconciliation of LEP cost estimates with the Office of Cost Estimating and Program Evaluation’s independent cost estimates. NNSA agreed with this recommendation. The B61-12 Program Office and the Office That Developed the Independent Cost Estimate Used Different Estimating Methods and Assumptions Cost estimating best practices specify that programs should develop a point estimate—the best guess at the program’s cost estimate, given the underlying data—by collecting, analyzing, and validating program data and then using one of several commonly used methods for estimating the program’s cost. Once a program has developed a point estimate, the program should compare it to an independent cost estimate, which gives an objective measure of whether the program’s point estimate is reasonable. In January 2017, NNSA issued two directives implementing statutory requirements for the Office of Cost Estimating and Program Evaluation to develop independent cost estimates for NNSA programs, including LEPs. The differences between the respective cost estimating methods, both of which are valid, used by the B61-12 program office and the Office of Cost Estimating and Program Evaluation are the primary reason for the differences between the program estimate and the independent cost estimate. According to B61-12 program officials, the program generally developed its point estimate by using a “bottom-up” method formally known as the “engineering build-up” cost estimating method. In using this method, a program subdivides the work necessary to accomplish its objectives into a work breakdown structure. The program then develops estimates of costs at the lowest level of the work breakdown structure, one piece at a time, and uses the sum of the pieces to form the overall estimate. To develop its cost estimate, the B61-12 program office required all participating NNSA site contractors to prepare and submit their own cost estimates for the work to be performed on the LEP and provided instructions on what data to provide to the program office. For example, these instructions specified that all sites must apply a bottom-up estimating approach that includes detailed quantities and integrated resource-loaded schedules for all work breakdown structure elements under their management. The program office then compiled the site- provided information in a database to arrive at a total program cost. The program office also aggregated schedule information from the sites, which maintain detailed resource-loaded integrated site schedules, to develop an NNSA Integrated Master Schedule. As we previously noted, the program office estimated in October 2016, based on this process, that NNSA’s portion of work on the B61-12 LEP would cost $7.6 billion and that the LEP would be completed in fiscal year 2025, with a first production unit date of March 2020. In contrast, to develop the independent cost estimate, the Office of Cost Estimating and Program Evaluation used an estimating method that employed data on the B61-12 LEP’s actual performance, coupled with historical information from the W76-1 LEP for stages of the phase 6.X process that the B61-12 LEP had not yet reached. Specifically, the office gathered data on 1,600 activities in the NNSA Integrated Master Schedule for the LEP. The office tracked these 1,600 activities from August 2014 through March 2016 by evaluating data from successive versions of the NNSA Integrated Master Schedule, which the B61-12 program office updates monthly based on actual program performance to date. In a memo summarizing the office’s independent cost estimate, the office stated that the program’s task completion rate lagged the baseline plan. The office concluded that the LEP’s first production unit date would occur 2 years after the March 2020 target date unless the program took measures to reduce the LEP’s scope by removing tasks, delaying activities until after the first production unit date, or relaxing requirements to accommodate less mature components than originally planned. The office also concluded, based on the program’s spending rate of approximately $45 million per month, that pre-first production unit costs would increase by about $1 billion over the program’s estimate. To estimate the cost and schedule of the program after the first production unit date, the office used both B61-12 LEP actuals and historical information from the W76-1 LEP, comparing W76-1 funding levels to B61- 12 spending levels. On the basis of its analysis, the office concluded that full-scale production of the B61-12 would cost approximately $1 billion more than the program office estimated. All told, the independent cost estimate projected that the B61-12 LEP would cost approximately $10.0 billion and take about 2 years longer—with a projected completion date in fiscal year 2027—barring changes to the program’s scope. The B61-12 program office and the Office of Cost Estimating and Program Evaluation also have differences of opinion regarding the continued validity of the August 2014 schedule performance data and its relevance to the independent cost estimate. According to B61-12 program officials, the information in the NNSA Integrated Master Schedule improved and changed after the Office of Cost Estimating and Program Evaluation gathered initial schedule performance data in August 2014 and used this information as a starting point to evaluate the program’s performance. These issues include the following: The officials described the data available to the Office of Cost Estimating and Program Evaluation in August 2014 as tentative, saying that the program can now use the NNSA Integrated Master Schedule to track performance at a more detailed level. According to B61-12 program officials, the program made important decisions that affected components on the program schedule’s critical path at the time the August 2014 schedule performance data were gathered. Subsequent to establishing the baseline, for example, the program office restructured the path to first production unit for high- explosives components, correcting errors that had been captured in the August 2014 data and changing to a more streamlined approach to qualify high-explosives components from legacy material. This decision affected the program’s critical path to first production unit, moving the completion date earlier. The program undertook schedule recovery efforts that eased schedule constraints affecting other program elements that were on the critical path at the time of the August 2014 data. As a result of these factors, B61-12 program officials said that the entire baseline schedule that the Office of Cost Estimating and Program Evaluation analyzed appeared more problematic than the updated schedule and that the entirety of the independent cost estimate was thrown off by the obsolete August 2014 data. Officials from the Office of Cost Estimating and Program Evaluation told us they disagree with the B61-12 program office’s assessment of the independent cost estimate schedule analysis. These officials said that they understand that the schedule baseline is continuously changing but that the independent cost estimate schedule analysis is not dependent on a particular baseline. Rather, they said that the analysis is based on actual schedule performance for the 1,600 activities that represent the scope required to achieve the design maturity and that the program office specified in August 2014 as needed to reach the first production unit (phase 6.5) milestone. Officials from the Office of Cost Estimating and Program Evaluation said that although they will not formally assess the B61-12 LEP’s schedule again until the end of phase 6.4 of the program, their informal analysis of NNSA Integrated Master Schedule data as of February 2017 still showed the same rate of activity completion that underpinned the office’s independent cost estimate. At some point, according to these officials, the program will have to double or even triple its rate of activity completion to finish the LEP on schedule, which will increase cost. In contrast, B61-12 program officials stated that, given the improved quality of the program’s integrated master schedule data, they expect that the independent cost estimate that the Office of Cost Estimating and Program Evaluation prepares at the end of phase 6.4 of the LEP will be closer to the program’s estimate than to the October 2016 independent cost estimate. Program officials also said that the program’s performance to date supports their position that the program cost estimate is accurate. The positions of the two offices also differ regarding the B61-12 program’s ability to accelerate work in the production stages of the LEP to ensure that the LEP meets its completion date. B61-12 program officials stated that they have options other than to complete tasks sequentially and at a steady rate, so they do not expect the “straight-line” level of productivity assumed in the independent cost estimate analysis to occur. For example, some parts do not have to be built in a particular sequence. Instead, program officials said, the production agencies can build different lots of components when they are ready, so technologies that are ready earlier than others can be moved to production in the war reserve lot while other components remain in earlier stages. They also said that the program would not maintain an even spending rate of $45 million per month, as suggested in the independent cost estimate. Rather, they noted, the program’s spending rate is currently $55 million a month, and the program plans for it to rise to $65 million per month as the current production engineering phase of the LEP draws to a close and the production agencies accelerate their activities. These factors notwithstanding, one Office of Cost Estimating and Program Evaluation official observed that activities in the later stages of an LEP remain complex and carry risks. The official cited the history of the challenges that led to the delay of first production of the W76-1, cited earlier in this report, and said that the B61-12 program faces the added challenges of having to integrate with several delivery aircraft and of having more electronic components than the W76-1. As noted in our cost estimating guide, studies have shown limited opportunity for getting a delayed program back on track after it is more than 15 percent to 20 percent complete. NNSA Officials Met to Reconcile the Program’s Cost Estimate and the Independent Cost Estimate but Did Not Document the Rationale for Adopting the Program’s Estimate Unchanged Cost estimating best practices specify that a program cost estimate and an independent cost estimate should be reconciled and that differences between them should be examined and discussed to achieve understanding of overall program risk. Officials from NNSA, including from the Office of Cost Estimating and Program Evaluation and the B61- 12 program office, told us that they held several discussions in 2016 regarding the differences between the program estimate and the independent cost estimate. These included a meeting with the second- highest ranking official in NNSA—the principal deputy administrator— during which the respective offices presented their estimates and explained the methods used to produce them. After these meetings, the principal deputy administrator and the NNSA Administrator agreed to approve the program estimate unchanged. According to B61-12 program officials, the program adapted some of its practices as a result of their interactions with the Office of Cost Estimating and Program Evaluation. For example, officials said that they changed the program’s procedure for baseline changes to ensure consistency across the participating sites. The program also began to conduct baseline execution index analyses, as the Office of Cost Estimating and Program Evaluation recommended. Baseline execution index analyses track a program’s execution of tasks to date by monitoring the percentage of activities that a program has completed early or on time and that have a baseline for completion within the month the analysis is conducted. According to program officials, similar to a schedule performance index in an earned value management system, the baseline execution index gives an alternate cumulative measure that gives a program an opportunity to improve as it proceeds. However, B61-12 program officials said that they did not document the rationale for adopting the program cost estimate without making changes informed by the independent cost estimate. They told us that any attempt to combine the results of the two estimates would have been difficult, considering the significant differences between the program’s cost estimating model and the Office of Cost Estimating and Program Evaluation’s model. We recommended in a January 2018 report that NNSA should establish a requirement for its management to document and justify key decisions based on a reconciliation of LEP cost estimates with the Office of Cost Estimating and Program Evaluation’s independent cost estimates. We concluded in the report that without a requirement for its management to document and justify key decisions based on a reconciliation of program cost estimates with the Office of Cost Estimating and Program Evaluation’s independent cost estimates, NNSA may not have assurance that the independent cost estimates are being appropriately incorporated into the LEP decision-making process, potentially decreasing the reliability of program cost estimates. Our prior work has shown that, in general, because the independent cost estimate team is outside the acquisition chain, is not associated with the program, and has nothing at stake with regard to program outcome or funding decisions, its estimate is usually considered more accurate than the program’s internal estimate. In addition, our prior work has shown that independent cost estimates are historically higher than program office cost estimates because the team conducting the independent cost estimate is more objective and less prone to accept optimistic assumptions. However, we have also found that because independent cost estimates are typically higher than program office cost estimates, in some cases management may choose to ignore them because the estimates are too high. NNSA agreed with our January 2018 recommendation, stating that by March 2018, it would establish a protocol to document management decisions regarding significant variances between LEP cost estimates and the independent cost estimates produced by the Office of Cost Estimating and Program Evaluation. However, NNSA has not provided evidence that it has done so. We continue to believe that documenting key decisions regarding cost estimates is particularly important in the context of LEPs, where decisions could increase a program’s costs by billions of dollars. NNSA and DOD Have Identified Risks and Are Taking Steps to Manage Them NNSA and DOD have identified and are managing various risks that could complicate efforts to meet the fiscal year 2025 completion date for the B61-12 LEP. Some of these risks that the agencies are managing are within the program’s areas of responsibility, such as an aggressive flight test schedule, and additional risks could be identified within these areas. To manage risks, the program uses a formal risk management process and has taken steps such as consolidating flight tests and holding more regular meetings between NNSA’s design and production agencies. The program also faces risks that program officials told us lie outside the program’s direct control—such as risks related to the F-35 delivery aircraft, nuclear certification, and NATO coordination issues—and officials said they have provided information to the responsible DOD organizations to help address these risks. NNSA and DOD Have Taken Steps to Identify and Help Manage Risks within the Program’s Responsibility, and They May Identify Additional Risks NNSA and DOD have taken steps to identify and help manage risks within the B61-12 LEP’s responsibility, and program officials said that they may identify additional risks as the program progresses. More specifically, the program has a formal risk management process through which it has identified risks and could identify additional risks as the program proceeds, according to agency officials. Risks already identified and being managed include risks related to the program’s aggressive schedule of flight tests and to finalizing design and coordinating procurement and delivery of components. Risk Identification Process The B61-12 LEP has a formal risk management process that has identified joint NNSA and Air Force risks within the program’s areas of responsibility that could significantly impact the overall program’s schedule, its cost, or the technical performance of the weapon. According to program officials and the Program Joint Risk Management Plan, this process calls for each program element in NNSA or the Air Force to be responsible for identifying and managing risks at the lowest level possible. After the program element reviews and documents a risk, it then reviews the risk to determine its applicability to be considered a joint risk—that is, a risk that has the potential to affect any of the top-level program milestones or the program’s ability to successfully meet system performance requirements. Program officials told us that the Air Force lead project officer decides whether to accept the risk into the joint risk list. Senior management oversees those risks through a formal management plan. The process includes continual reviews to identify new risks that may emerge. The Joint Risk Review Board meets as new potential risks are identified to review their likelihood and consequence. Officials from both the Office of Cost Estimating and Program Evaluation and DOD’s Office of Nuclear Matters told us that during phase 6.4 and thereafter, the program may still discover new risks—”unknown unknowns”—during technical tests to qualify components and the development of production processes. The process also has steps to manage risks and remove them from the joint high-risk list, if the Joint Risk Management Board judges them to have been resolved to closure or a low-risk status, according to program officials. NNSA and DOD program officials said that the program’s risk management process has resulted in the resolution of about three- quarters of the identified high risks on the joint risk list. They also observed that the program’s Selected Acquisition Reports, through which NNSA and the Air Force report to the congressional defense committees on the program’s cost and schedule, have been unchanged since 2013 regarding major program milestones. Program officials said that to provide a 90-day schedule buffer and add flexibility to the program’s schedule in the event of unexpected difficulties, the program has planned to reach phase 6.5 in December 2019, ahead of the phase 6.5 date of March 2020 that is reported in the Selected Acquisition Reports. However, other officials told us that it is too soon to say whether the program can manage the identified risks, or other unidentified risks, to prevent delays in a program that has relatively little schedule margin. Problems can emerge even during the first production stage of an LEP, as happened in the W76-1 LEP due to the Fogbank production challenges we discuss earlier in this report and in our March 2009 review of W76 and B61 modernization efforts. Managing Risks Related to the Program’s Aggressive Schedule of Flight Tests DOD and NNSA officials we interviewed generally agreed that the program faces risks in completing an aggressive flight testing schedule to support the first production unit deadline. According to the officials, the B61-12 program needs to complete more than 60 flight tests over a 3- year period to meet this deadline. Completing the tests entails actively coordinating with the Air Force organizations that manage the various aircraft that will carry the B61-12 weapon: the B-2 bomber and the F-15, F-16, F-35, and PA-200 fighters. According to B61-12 program officials, aircraft may not be available when needed for the planned flight testing. This risk is of particular concern for B-2 bombers, they said, because only one B-2 test unit is available and it is in heavy demand for other Air Force purposes. Program officials characterized the flight test schedule as aggressive and ambitious, but feasible, and told us the program has managed the risks caused by the tight testing timeframes by coordinating with the responsible organizations and consolidating tests to minimize the amount of time required on each type of aircraft. Further, when aircraft are not available as planned, program officials said they can revise the sequence of tests. For instance, they accommodated the unavailability of a B-2 test asset on a planned test date by moving up a test date on the F- 16. This schedule adjustment avoided a ripple effect of delays on the overall testing schedule, according to Air Force officials. The video in figure 6 shows an F-16 dropping an inert B61-12 bomb during a flight test on March 14, 2017. NNSA and DOD have identified and taken steps to manage risks related to finalizing the weapon’s design and coordinating the procurement and delivery of components. These risks include: Technical risks associated with the design and production of various components. Officials told us some components of the bomb and tail kit assembly are on the program’s list of joint risks. They said that their use of the joint risk management process calls management attention to potentially serious risks and helps the program to manage these risks as early and as continually as possible. For example, NNSA officials said that when technical risks arose in designing one classified component on the program’s critical path—potentially affecting the design schedule—they augmented the design team with additional scientists in an effort to ensure that the component would be completed in time to support the production schedule. Similarly, to manage design risks related to the exacting specifications for certain components, Kansas City National Security Campus is working to develop sufficiently precise gages to measure the required specifications during production. Late design changes from design agencies provided to the production agencies. NNSA’s Fiscal Year 2018 Stockpile Stewardship and Management Plan identified late changes to component design as a risk facing the B61-12 program and other LEPs. Contractor officials we interviewed from the Kansas City National Security Campus and the Pantex Plant said that late changes to weapon design requirements from the Sandia and Los Alamos design agencies could create schedule problems for establishing production processes at the production sites. Kansas City National Security Campus officials expressed concerns that some component requirements continue to change—some arising from testing results—which creates a tension between improving the design and stabilizing production requirements and processes. Pantex officials also told us about a potentially significant production delay if late design changes require Pantex to get new production tools or testers. Late design changes could occur as scientists at the design agencies analyze test results. Flight tests, for example, produce a volume of information. Officials at the federal program office in Albuquerque said that 4 test flights on the B-2, conducted in July 2017, produced 4 to 6 hours’ worth of data per flight. Officials at both the Pantex and Kansas City sites said they have developed management strategies to provide some flexibility in their production schedules, such as speeding production by having staff work longer shifts. Moreover, because of lessons learned from prior LEPs, officials at both sites told us that coordination between production sites and design agencies has significantly improved over past practices— specifically, by having ongoing engagement that started earlier in the weapon development process. For instance, Pantex officials told us that they hold monthly meetings with design agencies to discuss design changes. A Pantex official told us that, as a result of addressing production concerns early, design requirements developed at the design agencies are less likely to result in unanticipated production problems. Vendor risks associated with procuring various bomb components. According to NNSA officials we interviewed, some bomb components are procured through single commercial vendors, in small lots, or are unusual. Kansas City National Security Campus officials told us that they had to replace one vendor that could no longer provide a certain material and that they generally risk losing potential or existing vendors because vendors prefer contracts for larger volumes of components than NNSA needs for the B61-12 bomb. In addition, unique materials for the bomb include certain components with specific compositions of rubber and plastics. Officials at the Kansas City National Security Campus said that they have encountered difficulties with getting rubber and plastic components from vendors that consistently meet composition specifications or with sustaining vendors’ interest in producing small batches of precision-manufactured components. In one such instance, they said they improved incentives and communication with a vendor to avoid losing a source for a key component. The officials said they also contract with smaller vendors when larger vendors may not be interested in the size of the contract NNSA offers. Delays in delivery of components from other production sites to Pantex for full bomb assembly. NNSA and Pantex officials told us that Pantex, which will assemble the full B61-12 bomb, depends on the other production sites delivering the components in a timely manner. NNSA production sites are scheduled to provide components to Pantex 120 days before the first production unit date. Pantex and NNSA officials have identified some schedule flexibility for assembling the first production unit at Pantex, depending on which components have delayed deliveries. Specifically, if the delayed components are those needed later in the assembly process, such as the bomb’s nose assembly, Pantex could stay on schedule by assembling other delivered components until the delayed components are needed. Delivery of other components, such as detonators, is more time- sensitive, and it is essential that these be delivered on time for assembly to proceed as planned, according to Pantex officials. NNSA and DOD Have Taken Steps to Identify and Help Manage Risks outside the Program’s Direct Control According to program officials, certain risks that may have a bearing on the B61-12 LEP or that may affect the fielding of the weapon lie in areas outside the program’s direct control. Nevertheless, program managers have taken steps to coordinate with other responsible parties to help address these risks. For example, two of the three delivery aircraft designated as system 2 aircraft—the F-35 and the B-2—have not yet completed development and procurement of operational flight program software that will enable the aircraft to deliver the B61-12 with the enhanced accuracy offered by the tail kit assembly, a key feature of the LEP. B61-12 program officials told us that the program offices responsible for each of these aircraft must manage the development and procurement of the operational flight program software. To help inform the software development process, the B61-12 program provided the F-35 and B-2 program offices with information about the weapon’s interface with the airplane, including information from flight tests performed on an earlier version of the F-35, according to program officials. NNSA and DOD officials characterized B-2 development related to the B61-12 as significantly more advanced than F-35 development. Specifically, Air Force officials said that a developmental version of the B-2 operational flight program software was fielded and certified in 2017 and would undergo final weapon system demonstration flight tests in October 2019 and nuclear design certification in June 2020. By contrast, they said that the F-35 software will not be ready for nuclear design certification until January 2023, after the B61-12 program’s first production unit date. The F-35 program office will be responsible for funding tests and aircraft- weapon integration activities, according to the Air Force officials. Because of the need to defer some flight tests until the software for the B-2 and F- 35 aircraft is ready, only one of the three system 2 delivery aircraft has undergone testing of the B61-12 bomb’s capabilities in its system 2 setting: the F-15E, on which NNSA and DOD conducted the first system 2 tests of the B61-12 in August 2017. Other risks outside the program’s direct control concern nuclear certification and the NATO mission. Nuclear certification—ensuring that people and objects that come into contact with the weapon will not adversely affect its performance characteristics—is a prerequisite to fielding the B61-12 and other nuclear weapons, but it is the responsibility of the Air Force organizations that manage the delivery aircraft. In a classified report issued in January 2018, we discuss risks related to nuclear certification of dual capable aircraft, which are able to deliver conventional munitions or nuclear bombs. B61-12 program officials told us that they are working to address these risks by providing information on the weapon to all of the organizations that manage the delivery aircraft. Similarly, in another classified report issued in February 2018, we discuss a risk related to the NATO mission that may affect the B61-12 LEP; program officials told us that they are working to address this risk, as well. We made recommendations in the two classified reports related to these risks; the responsible agencies agreed with our recommendations and stated their intention to take action in response to them. Agency Comments We provided a draft of this product to NNSA and DOD for comment. NNSA provided technical comments, which we incorporated as appropriate. DOD indicated that it did not have any comments. We are sending copies of this report to the appropriate congressional committees, the Secretaries of Defense and Energy, the Administrator of NNSA, and other interested parties. In addition, this report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff members have any questions about this report, please contact me at (202) 512-3841 or bawdena@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made significant contributions to the report are listed in appendix II. Appendix I: Results of GAO’s Assessment of the B61-12 Life Extension Program Cost Estimate Compared with Best Practices Characteristic Comprehensive Best practice The cost estimate includes all life cycle costs. The cost estimate completely defines the program, reflects the current schedule, and is technically reasonable. The cost estimate work breakdown structure—a hierarchical structure that subdivides the work necessary to accomplish the program’s objectives into smaller elements—is product-oriented, traceable to the statement of work/objective, and at an appropriate level of detail to ensure that cost elements are neither omitted nor double-counted. The estimate documents all cost-influencing ground rules and assumptions. The documentation should capture the source data used, the reliability of the data, and how the data were normalized. The documentation describes in sufficient detail the calculations performed and the estimating methodology used to derive each element’s cost. The documentation describes step by step how the estimate was developed so that a cost analyst unfamiliar with the program could understand what was done and replicate it. The documentation discusses the technical baseline description, and the data in the baseline is consistent with the estimate. The documentation provides evidence that the cost estimate was reviewed and accepted by management. The cost estimate results are unbiased, not overly conservative or optimistic, and based on an assessment of most likely costs. The estimate has been adjusted properly for inflation. The estimate contains few, if any, minor mistakes. The cost estimate is regularly updated to reflect significant changes in the program so that it always reflects current status. Variances between planned and actual costs are documented, explained, and reviewed. The estimate is based on a historical record of cost estimating and actual experiences from other comparable programs. Characteristic Credible Best practice The cost estimate includes a sensitivity analysis that identifies a range of possible costs based on varying major assumptions, parameters, and data inputs. A risk and uncertainty analysis was conducted that quantified the imperfectly understood risks and identified the effects of changing key cost driver assumptions and factors. Major cost elements were cross checked to see whether results were similar. An independent cost estimate was conducted by a group outside the acquiring organization to determine whether other estimating methods produce similar results. Appendix II: GAO Contact and Staff Acknowledgments GAO Contact: Staff Acknowledgments: In addition to the contact named above, Jonathan Gill (Assistant Director), Rob Grace (Analyst in Charge), Terry Hanford (Senior Analyst), and Jennifer Leotta (Senior Operations Research Analyst) made key contributions to this report. Also contributing to this report were Antoinette C. Capaccio, Scott Fletcher, Penney Harwell Caramia, Cynthia Norris, Karen Richey, and Sara Sullivan.

Why GAO Did This Study Weapons in the U.S. nuclear stockpile are aging. To refurbish or replace nuclear weapons' aging components, NNSA and DOD undertake LEPs. The B61-12 LEP is the most complex and expensive LEP to date. In October 2016, NNSA formalized a program cost estimate of about $7.6 billion, which is lower than an independent cost estimate of about $10 billion. Senate Report 113-44 included a provision for GAO to periodically assess the status of the B61-12 LEP. This report assesses (1) the extent to which NNSA followed best practices for cost estimation in producing the program cost estimate for the B61-12 LEP; (2) the reasons for differences between the program cost estimate and the independent cost estimate and how the differences were reconciled; and (3) the extent to which NNSA and DOD have identified and managed program risks. GAO assessed the program cost estimate against best practices, reviewed NNSA and DOD documents, conducted site visits to four NNSA and Air Force sites responsible for design, production, and management activities, and interviewed NNSA and DOD officials. What GAO Found The National Nuclear Security Administration (NNSA) incorporated most cost estimating best practices to develop the program cost estimate for the B61-12 Life Extension Program (LEP), which seeks to consolidate four versions of a nuclear weapon—the B61 bomb—into a bomb called the B61-12. As shown in the figure below, the program substantially met best practices for ensuring the estimate was comprehensive, well-documented, accurate, and credible. The B61-12 LEP's program cost estimate differs from an estimate prepared by another NNSA office independent of the program primarily because the program used different methods and assumptions than the independent office. The program developed its estimate by compiling cost and schedule estimates for activities at each of the NNSA contractor sites participating in the LEP. In contrast, the independent office evaluated program activities completed to date and applied a historical model to estimate costs and durations for remaining activities. NNSA management met with officials from both offices to reconcile the estimates but did not document the rationale for adopting the program estimate unchanged. GAO recommended in a January 2018 report that NNSA document and justify such decisions, in part because GAO's prior work has shown that independent cost estimates historically are higher than programs' cost estimates because the team conducting the independent estimate is more objective and less prone to accept optimistic assumptions. In response to the January 2018 report, NNSA agreed to establish a protocol to document management decisions on significant variances between program and independent cost estimates, but it has not yet provided evidence that it has done so. NNSA and the Department of Defense (DOD) have identified and are managing risks that could complicate efforts to meet the LEP's fiscal year 2025 completion date. Risks within the program's areas of responsibility include an aggressive flight test schedule for bomb delivery aircraft. The program is managing these and other risks with a formal risk management process. The program has also taken steps to address risks outside its direct control, such as risks related to the readiness and certification of the weapon's F-35 delivery aircraft, by providing information to the responsible DOD organizations. What GAO Recommends GAO is making no new recommendations but discusses a prior recommendation that NNSA document and justify decisions regarding independent cost estimates. NNSA provided technical comments, which GAO incorporated as appropriate. DOD did not have any comments.

gao_GAO-18-589

gao_GAO-18-589_0

Background A reliable schedule is critically important for a successful 2020 Census. In February 2017, we added the 2020 Census to our High-Risk List because operational and other issues including scheduling are threatening the Bureau’s ability to deliver a cost-effective enumeration. We reported on concerns about the quality of the Bureau’s schedule and cost assessment, the Bureau’s capacity to implement innovative census-taking methods, and uncertainties surrounding critical information technology systems. Underlying these issues are challenges in such essential management functions as the Bureau’s ability to collect and use real-time indicators of schedule, cost, and performance; follow leading practices for scheduling, cost estimation, risk management, and IT acquisition, development, testing, and security; and cost effectively deal with contingencies including, for example, fiscal constraints, potential changes in design, and natural disasters that could affect the enumeration. Reliable scheduling practices are essential for managing tradeoffs between cost, schedule, and scope. Among other things, scheduling allows program managers to decide between possible sequences of activities, determine the flexibility of the schedule according to available resources, predict the consequences of managerial action or inaction in events, and allocate contingency plans to mitigate risk. Following changes in a program, the schedule is used to forecast the effects of delayed, deleted, and added effort, as well as possible avenues for time and cost recovery. Scheduling is important because the cost of counting the nation’s population has been escalating with each decade. The 2010 Census was the most expensive in U.S. history at about $12.3 billion, and was about 31 percent more costly than the $9.4 billion 2000 Census (in 2020 constant dollars). According to the Bureau, the total cost of the 2020 Census is now estimated to be approximately $15.6 billion dollars, more than $3 billion higher than previously estimated by the Bureau. Moreover, as shown in figure 1, the average cost for counting a housing unit increased from about $16 in 1970 to around $92 in 2010 (in 2020 constant dollars). At the same time, the return of census questionnaires by mail (the primary mode of data collection) declined over this period from 78 percent in 1970 to 63 percent in 2010. Declining mail response rates have led to higher costs because the Bureau needs to send temporary workers to each nonresponding household to obtain census data. The schedules we reviewed for this report—2020 Census Geographic Programs, 2018 End-to-End Test Address Canvassing, and 2018 End-to- End Census Test Nonresponse Follow-up—relate to the key activities of developing an accurate address list and following up with households that did not mail back their census forms. The Bureau relies on a complete and accurate address list to maximize the more cost-efficient self- response rate. The three projects we selected contribute to two of the Census Bureau’s largest field operations. The Bureau’s Geographic Programs Operation maintains the Bureau’s master address file and mapping data used to conduct the 2020 Census. The Bureau’s Geographic Programs Operation provides the most current address list to the Bureau’s Address Canvassing Operation, where Bureau staff make updates to the address list via in-office and in-field procedures. These updates are processed on an ongoing basis throughout the decade. The Bureau conducts its Nonresponse Follow-up Operation after Census Day by having enumerators go door-to-door to determine the housing unit status for addresses that do not self-respond to the 2020 Census, and enumerate households that are determined to be occupied. Selected Census Schedules Better Reflect Characteristics of a Reliable Schedule Compared to Prior Assessment, though Weaknesses Remain We have previously reported in our Schedule Assessment Guide that a reliable schedule can provide a road map for systematic execution of a program, and the means by which to gauge progress, identify and address potential problems, and promote accountability. The guide identifies four characteristics of a reliable schedule: Comprehensive: The schedule should identify all activities and resources necessary to accomplish the project. The schedule should cover the scope of work to be performed so that the full picture is available to managers. Well-constructed: Activities should be logically sequenced and critical activities that would affect the timelines of the schedule should be identified. Credible: All schedules should be linked to a complete master schedule for managers to reference and analyzed for how risk impacts the outcome of the schedule. Controlled: There should be a documented process for changes to the schedule so that the integrity of the schedule is assured. For a schedule to be reliable, it must substantially or fully meet all criteria for these four characteristics. These characteristics, their related leading practices, and their criteria are described in more detail in appendix II. In 2013, we assessed the Bureau’s 2020 Research and Testing and Geographic Support System Initiative schedules using these criteria. While the results exhibited some of the characteristics of a reliable schedule, important weaknesses remained. Both schedules substantially met one of the four characteristics (controlled) and minimally or partially met the other three characteristics (comprehensive, well-constructed, and credible). For this review, we assessed the 2018 End-to-End Census Test Address Canvassing, 2018 End-to-End Census Test Nonresponse Follow-up, and 2020 Census Geographic Programs projects’ schedules. We found that overall the selected schedules better reflected two of the four characteristics of a reliable schedule compared to our 2013 assessment (see figure 2). Examples of the extent to which these characteristics were met are provided below. For a more detailed explanation of our assessment results, see appendix III. Comprehensive–Selected Schedules Partially Meet Characteristic but Do Not Identify Needed Resources As with our 2013 schedule assessment, our 2018 analysis found that the Bureau is partially meeting the characteristics of a comprehensive schedule. For example, the projects we assessed reflect the work to be accomplished for the project schedules, and each project schedule includes estimates of the duration of each activity. Additionally, the 2018 End-to-End Census Test Address Canvassing and the 2018 End-to-End Census Test Nonresponse Follow-up project schedules contain clear start and finish milestones, and map to the census program work breakdown structure—a detailed definition of the work necessary to accomplish a program’s objectives. This leading practice of capturing all activities was substantially met, not fully met (see appendix III for a more detailed explanation), because while for each project all activities and milestones are mapped to their work breakdown structures by codes, there are no corresponding dictionaries to define the work. The absence of such a dictionary could potentially lead to confusion among staff in different census offices about the scope of the work they are responsible for performing. Our schedule guide states that a work breakdown structure dictionary is a valuable communication tool between systems engineers, program management, and other stakeholders because it provides a clear picture of what efforts have to be accomplished. Bureau officials stated that although their 2020 Schedule Management Plan requires each project to have a schedule work breakdown structure dictionary, as project schedules are updated, they have not created these required dictionaries. As an alternative, they noted that the 2020 Census Operational Plan includes details and definitions of the projects. Additionally, none of the three schedules we assessed include information about what levels of resources, such as labor and equipment, are required to complete the planned work—including this information is called resource loading. The Bureau’s 2020 Schedule Management Plan states that it is the responsibility of a representative from a project team and the schedule staff to assign resources to an individual project schedule, and that defining and assigning resources should be done following the testing phase of the 2020 Census Lifecycle. The Bureau is now in its implementation phase (see figure 3 below), so according to its management plan, resource loading should have begun. But it has not. For example, the 2018 End-to-End Census Test Address Canvassing project schedule did not include any resource information on the recruiting and hiring goals for the address canvassing field work. Instead, Bureau officials stated that they are estimating the cost of activities using a software tool separate from the current schedule management tool. They further stated that this Bureau-wide solution includes all 2020 Decennial Census staff as Decennial funded resources. However, the information in this separate tool has no effect on the durations or forecasted start and finish dates of detailed activities within individual projects. Furthermore, the separate tool does not always track all activities at the lowest level in the schedule, so that Bureau managers do not have reliable visibility with it on the efforts of the lowest level of detailed activities. Resource loading is important for any agency, but is particularly important for the Census Bureau, given its statutorily mandated deadlines. Missed deadlines or schedule slippage can easily jeopardize the quality of the 2020 Census, and there is little room for error given that census data are used to apportion the seats of the House of Representatives, redraw congressional districts, and allocate billions of dollars each year in federal financial assistance. In our schedule guide, we reported that including resources such as labor, materials, and overhead costs can make a schedule a more useful management tool. A resource-loaded schedule can help management with things such as computing labor and equipment hours, calculating total project and per- period cost, resolving resource allocation conflicts, determining whether all required resources will be available when they are needed, and establishing the reasonableness of the plan. For example, information on the resource needs of field operations in the 2018 End-to-End Census Test would assist management in determining if the appropriate resource allocations have been made for any given test activity. It would also aide in forecasting the likelihood that those resources will be available to complete the 2018 End-to-End Census Test Address Canvassing and Nonresponse Follow-up activities as scheduled. If the schedule does not allow insight into the current or projected allocation of resources for these test activities, the Bureau’s risk of key end-to-end test milestones slipping increases significantly. In 2009, we reviewed the Bureau’s schedule and recommended that the Bureau include estimates of the resources in the 2020 integrated schedule for each activity as the schedule was built. The Department of Commerce did not respond to the recommendation at that time. In our 2013 assessment of the Bureau’s schedule, Bureau officials stated that they hoped to begin identifying the resources needed for each activity in their schedules by early 2014. However, as of May 2018, the Bureau has not yet implemented this recommendation. Senior Bureau officials have now stated that the Bureau would require additional staffing in the Schedule Management Branch in the Decennial Census Management Division in order to plan for and implement resource loading. When the Bureau has resource loaded its schedule, it will be able to use the schedule more effectively as a management tool. Well-Constructed–The Bureau Demonstrated Improvement in Selected Schedules Compared to Prior Assessment Our 2013 assessment of the Bureau’s schedule reported that the Bureau only minimally met the characteristics of a well-constructed schedule. Our 2018 assessment found that two of the selected project schedules now substantially met this characteristic and one partially met it. In this assessment, Bureau officials linked many of the activities clearly and in a straightforward sequence in the schedule, which was not always the case in prior assessments. This improvement is important because it helps staff identify next steps as they progress through such things as acquiring and mobilizing the staff needed to conduct the address canvassing and nonresponse follow-up test field work, and helps managers identify the impact of changes in one activity on subsequent activities. For example, the schedule lays out the sequence of activities needed, such as developing training materials, recruiting field staff, training staff and equipping them with the tools needed to complete the test. Our assessment also concluded that two of the three project schedules we assessed have valid critical paths, which is the sequence of activities in the schedule that, according to their current status, lead to the program’s earliest completion date. A valid critical path allows management to focus on activities that will lead to the project’s success. The 2020 Census Geographic Programs project schedule partially met the well-constructed characteristic due to problems existing within the schedule’s sequencing logic. In particular, we found a large number of unjustified date constraints and lags. In part because of these sequencing issues, total float calculations—that is, the amount of time a predecessor activity can slip before its delay affects the program’s estimated finish date—appear unreasonably high. Additionally, this project schedule has activities on the critical path with long durations. For example, the project schedule for Geographic Programs included several long-duration activities on its critical path that relate to the Bureau’s collection of community boundary data—information essential to delineating geographic boundaries used in the tabulation of census data. These critical long-duration activities make it difficult to measure time- critical progress on such activities in the near term. These issues with how the schedule is constructed can also cause schedule users to lack confidence in the forecasted dates. Bureau officials acknowledged that the 2020 Census Geographic Programs project schedule had logic issues at the time because it was in the middle of a revision. Bureau officials stated that the standard process is to update the project schedule in an offline version and then assess the quality and impacts of changes before acceptance. According to Bureau officials, the Geographic Programs project schedule did not follow this process and was instead updated in the live version of the schedule because of time constraints. Credible–Selected Schedules Partially Meet Characteristic, but the Bureau Has Not Carried Out a Schedule Risk Analysis Our 2013 assessment of the Bureau’s schedule found that the Bureau minimally met the characteristics of a credible schedule. Our 2018 assessment of the Bureau’s schedule found that the Bureau’s scheduling practices for a credible schedule have improved. We found that there is now a clear relationship between lower-level activities and higher-level activities and milestones, and there is generally better consistency of dates between the project schedule and higher-level management documents. However, the Bureau has not carried out a systematic quantitative risk analysis on its schedule. A schedule risk analysis is a statistical simulation of the possible effects of threats, opportunities, and general uncertainty to a program’s schedule that results in a quantifiable level of confidence in meeting the program’s key milestone dates. While the Bureau has identified and continues to track risks to its 2018 End-to-End Test address canvasing and nonresponse follow-up efforts in risk registers, a quantitative risk analysis would illustrate the impact of risks on the project schedule, and how those risks would affect the Bureau’s ability to meet milestones on time. Such an analysis would also provide a measure of how much time contingency should be built in the schedule to help manage prioritized risks, and, implicitly, provide indications of where additional resources might be needed to stay on schedule. In response to our 2013 schedule assessment, Bureau officials said they were waiting for decisions about scheduling software before making decisions about a schedule risk analysis. As of May 2018, the Bureau has conducted three risk analyses to prove the software’s capability. However, the Bureau still had not conducted a schedule risk assessment on the current integrated master schedule used to manage the 2020 Census program. Without a schedule risk analysis, the Bureau cannot determine the likelihood of each project’s completion date; how much schedule risk contingency is needed to provide an acceptable level of certainty for completion by a specific date; which risks are most likely to affect the schedule; how much contingency time each risk requires; and the sequence of activities that are most likely to delay the project. Senior Bureau officials stated that a schedule risk assessment plan and process was approved by Bureau management in late May 2018 and that they hope to implement this plan in summer 2018. They intend to conduct an internal review over the next couple months to determine how to best use the information this risk assessment would yield. Follow through on their plans is critical to ensuring our recommendation is implemented. Controlled–Selected Schedules Substantially Meet the Characteristic As with our 2013 schedule assessment, our 2018 analysis reported that the Bureau’s scheduling practices are substantially meeting the characteristics of a controlled schedule. Our analysis determined that there are no date anomalies in the project schedules, such as planned dates in the past or actual dates in the future. We found the schedule was current as of the date delivered to us, and according to Bureau documents, the schedule is updated weekly following an established schedule process. Additionally, the Bureau reported that it has a schedule management process in place and a method for logging changes to the schedule in accordance with leading practices. Bureau officials reported that they monitor schedule trends, including bi-weekly schedule reliability checks using the Defense Contract Management Agency 14-Point assessment, a commonly used set of schedule integrity and reliability measures. Bureau officials also provided the May 2014 Program Change Management Process Strategy which defines the process for initiating changes to the integrated performance measurement baseline configuration; analyzing the impact of changes to project cost, schedule and scope; approving or disapproving changes; and updating project or product specifications and baselines. However, our assessment found that the Bureau did not fully meet this characteristic for a controlled schedule. The Bureau lacked sound documentation of the schedule in the form of a schedule basis document, and changes to the current schedules in the form of a schedule narrative. The current schedule should be documented in a schedule narrative with each update, including changes made to the schedule during status updates and changes that are justified along with their likely effect on future activities. The Bureau had not prepared such narratives. Additionally, none of the three schedules were supported by a schedule baseline document—a single document that defines the organization of a schedule, describes the logic of the network, describes the basic approach to managing resources, and provides a basis for all parameters used to calculate dates. Sound documentation helps with analyzing changes in the program schedule and identifying the reasons why actual schedule results vary from their estimates, thereby contributing to the collection of data that can be useful to evaluations of schedule efforts, and that can be used to support future estimates. While the Bureau has made improvements to implement the recommendations regarding the comprehensiveness and construction characteristics of the Bureau’s scheduling practices, the Bureau’s lack of resource loading and a risk assessment of the schedule continue to affect the reliability of the Bureau’s schedule. The schedule would be a more useful management tool if the Bureau increased the schedule’s reliability by addressing these weaknesses. To address these remaining weaknesses, we continue to believe that these recommendations are valid in order to ensure the 2020 schedule can support key management decisions. Agency Comments and Our Evaluation We provided a draft of this report to the Department of Commerce. In its written comments, reproduced in appendix IV the Department of Commerce agreed with our findings. We are sending copies of this report to the Secretary of Commerce, the Under Secretary of Economic Affairs, the Acting Director of the U.S. Census Bureau, and interested congressional committees. The report also is available at no charge on GAO’s website at http://www.gao.gov. If you have any questions about this report, please contact me at (202) 512-2757 or goldenkoffr@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff that made major contributions to this report are listed in appendix V. Appendix I: Objective, Scope, and Methodology This report assesses the extent to which the Bureau is using leading practices for scheduling key projects. We did this by focusing on three 2020 projects. We selected the three projects from the December 17, 2017, version of the 2020 Census integrated master schedule. That schedule consists of 255 total projects, of which 134 were remaining to be completed. We made our selections based on the cost of the projects, their significance to the 2020 Census, and the fact that they were in progress. The 3 projects selected for review are 2018 End-to-End Census Test Address Canvassing, 2018 End-to-End Census Test Nonresponse Follow-up, and 2020 Census Geographic Programs. We reviewed the project schedules and underlying sub-schedules to assess them against the 10 scheduling leading practices by: Checking for specific problems that could hinder the schedule’s ability to respond to changes. For example, we: Examined if there are any open-ended activities (i.e., activities with no predecessor and/or successors), Searched for activities with poor logic: For example, Start to Start successor only or Finish to Finish predecessor only which represent dangling logic, or Logic on summary tasks rather than attached to detailed tasks (summary tasks are for organizing the schedule and should not drive the logic). Looked for activities with constraints which keep the schedule rigid (e.g., start no earlier than, finish no later than, etc.), Identified any lags or leads which should only be used to show how two tasks interact and not to represent work, Determined if activities were resource loaded—which helps to cost out the schedule—and examine whether resources are over allocated or not available when needed, Examined the length of activity durations and compared them to the program management review cycle, Checked for horizontal and vertical integration within the schedule, Examined the schedule critical path to determine whether or not it was reliable and logical, Examined schedule float and determine if it was reasonable, and Examined whether the schedule was baselined, its status cycle, and what deviations there were from the original plan. We also determined if there were any actual start or finish dates recorded in the future and whether there was any broken logic between planned tasks. We also interviewed Bureau officials responsible for the 2020 schedule. We scored each scheduling leading practice on a five-point scale ranging from “not met” to “met.” We determined the characteristic assessment rating by assigning each best practice rating a number and taking the average. The numerical ratings and ranges of the resulting averages are as follows. We then compared these results with our prior assessments of the Bureau’s schedule, particularly those where recommendations were made, and we updated the status of those recommendations. Assessing only three key projects limits possible statements about the Bureau’s entire schedule. For example, if the Bureau is not following best practices in creating and maintaining the three project schedules, we can conclude that the larger integrated schedule is unreliable. This is because an integrated master schedule consolidates lower-level project schedules; errors and reliability issues in lower levels will be transferred to higher- level schedules. However, if the selected lower-level projects are deemed reliable, we cannot definitively determine the reliability of the integrated master schedule because the other projects that were not assessed may be unreliable. Appendix II: Description of Scheduling Leading Practices Characteristic Comprehensive Capturing all activities Description A schedule should reflect all activities defined in the project’s work breakdown structure and include all activities to be performed by the government and contractor. The schedule should realistically reflect the resources (i.e., labor, material, and overhead) needed to do the work, whether all required resources will be available when needed, and whether any funding or time constraints exist. The schedule should reflect how long each activity will take to execute. The schedule should be planned so that all activities are logically sequenced in the order they are to be carried out. The schedule should identify the critical path, or those activities that, if delayed, will negatively impact the overall project completion date. The critical path enables analysis of the effect delays may have on the overall schedule. The schedule should identify float—the amount of time an activity can slip in the schedule before it affects other activities—so that flexibility in the schedule can be determined. As a general rule, activities along the critical path have the least amount of float. Credible The detailed schedule should be horizontally traceable, meaning that it should link products and outcomes associated with other sequenced activities. The integrated master schedule should also be vertically traceable—that is, varying levels of activities and supporting subactivities can be traced. Such mapping or alignment of levels enables different groups to work to the same master schedule. The schedule should include a schedule risk analysis that uses statistical techniques to predict the probability of meeting a completion date. A schedule risk analysis can help management identify high priority risks and opportunities. Controlled Progress updates and logic provide a realistic forecast of start and completion dates for program activities. Maintaining the integrity of the schedule logic at regular intervals is necessary to reflect the true status of the program. To ensure that the schedule is properly updated, people responsible for updating should be trained in critical path method scheduling. A baseline schedule represents the original configuration of the program plan and is the basis for managing the project scope, the time period for accomplishing it, and the required resources. Comparing the current status of the schedule to the baseline can help managers target areas for mitigation. Appendix III: Assessment of the Extent to Which the Bureau Followed Scheduling Leading Practices Appendix IV: Comments from the Department of Commerce Appendix V: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments Other key contributors to this report include Ty Mitchell, Assistant Director; Juaná Collymore; Rob Gebhart; Yvette Gutierrez; Jason Lee; Kayla Robinson; Cynthia Saunders; and Timothy Wexler.

Why GAO Did This Study The Bureau is required by law to count the population as of April 1, 2020; deliver state apportionment counts to the President by December 31, 2020; and provide redistricting data to the states within 1 year of Census Day, April 1, 2021. To meet these statutory deadlines, the Bureau carries out hundreds of projects, which it manages with an integrated master schedule. Because census operations need to proceed in concert with one another, significant delays could propagate to other activities resulting in increased costs, reduced operational quality, or changes to the design of the census in order to compensate for lost time. This report determines the extent to which the Bureau is using leading practices for scheduling key projects. GAO selected three projects for review based on their cost and in-progress status. GAO analyzed schedules and their supporting documents against GAO's Schedule Assessment Guide. GAO also spoke with relevant Bureau officials regarding the three selected projects. GAO provided a draft of this report to the Department of Commerce, which agreed with the findings. What GAO Found The three census project schedules GAO reviewed better reflect characteristics of a reliable schedule compared to a GAO schedule assessment performed in 2013, but weaknesses remain. GAO reviewed three projects that contribute to two of the Census Bureau's (Bureau) largest field operations—address canvassing and nonresponse follow-up. The schedules for all three projects are better constructed and more credible than previously reviewed project schedules. For example, the Bureau has improved the logic of the relationship between activities, and better ensured that all schedules are linked together in a master schedule so that their interactions can be better managed. However, the three selected schedules have some of the same weaknesses GAO identified in other Bureau schedules in 2009 and 2013. For example, none of the selected schedules contain information on resource needs and availability. GAO has reported that such information assists program offices in forecasting the likelihood that activities will be completed as scheduled. It can also help management compute total labor and equipment hours, calculate total project and per-period cost, resolve resource conflicts, and establish the reasonableness of the plan. If the schedule does not allow insight into current or projected allocation of resources, then the likelihood is significantly increased that the program may slip or need additional resources to complete on time. In GAO's 2009 review of the Bureau's schedule, GAO recommended that the Bureau include in the 2020 master schedule estimates of the resources, such as labor, materials, and overhead costs for each activity as the 2020 schedule was built. The Department of Commerce did not respond to the recommendation at that time. Then, regarding GAO's 2013 assessment of the Bureau's schedule, Bureau officials stated that they hoped to begin identifying the resources needed for each activity in their schedules by early 2014. However, as of May 2018, the Bureau had not taken these steps. Senior Bureau officials have now stated that it would require additional staffing in order to plan for and implement this recommendation. Additionally, the Bureau has not conducted risk assessments for the project schedules GAO assessed. Schedule risk analysis—the systematic analysis of “what if” scenarios—is an established leading practice. Risk assessments are needed to determine the likelihood of the project's completion date; how much schedule risk contingency is needed to provide an acceptable level of certainty for completion by a specific date; risks most likely to delay the project; how much contingency reserve each risk requires; and the paths or activities that are most likely to delay the project. In 2013, GAO recommended the Bureau conduct risk assessments for its schedules. The Bureau said it had no disagreement with this recommendation. However, while Senior Bureau officials stated that a schedule risk assessment plan and process were approved by Bureau management in late May 2018, it has not yet implemented this recommendation. GAO believes that these prior recommendations still apply and can help the Bureau improve the reliability of its 2020 schedule.

gao_GAO-18-526

gao_GAO-18-526_0

Background History of Civil Aviation between the United States and Cuba In January 1961, the United States severed diplomatic relations with Cuba, followed by a total economic embargo declared by President Kennedy in February 1962. The resulting restrictions, including prohibitions on civil aviation between the United States and Cuba, remained in place over the subsequent 37 years until the Clinton Administration announced the start of public and private charter air service operations between the United States and Cuba in 1999. Charter service was the exclusive means of air transport between the United States and Cuba from the time these flights were announced in 1999 until August 2016. In February 2016, United States and Cuban officials signed a memorandum of understanding reestablishing regularly scheduled commercial air service between the two countries. Specifically, this memorandum of understanding allowed U.S. air carriers to operate 20 daily scheduled round trip flights between the United States and Havana and 10 daily round trip flights between the United States and each of the 8 other Cuban airports, as shown in figure 1. The reestablishment of scheduled commercial flights between the United States and Cuba followed a March 2016 Obama Administration change to Office of Foreign Assets Control (OFAC) travel regulations with regards to educational travel. While travel to Cuba for tourist purposes is prohibited, U.S. persons may be authorized to travel to Cuba for certain activities including family visits and educational activities. Specifically, this change allowed individuals traveling under the educational category to create their own schedule of travel and interaction with the Cuban people rather than being required to travel under this category only through a licensed group. On August 31, 2016, U.S. airlines began offering regularly scheduled commercial flights to Cuba. In June 2017, President Trump directed the Department of Treasury through OFAC to revise various categories of travel, and OFAC revised the categories in November 2017 to again generally require U.S. persons to travel to Cuba as part of a licensed group. The revised categories do not change the ability of public charter and scheduled commercial flights between both countries to operate. However, since these changes to the travel categories were announced, four air carriers that had been awarded scheduled round-trip flights between the United States and Cuba returned all or some of their allotted flights, citing lack of market demand. See figure 2 for further detail on the history of civil aviation between the United States and Cuba. DHS Responsibilities for Ensuring the Security of U.S.-Bound Flights from Cuba Consistent with the Aviation Transportation Security Act and in accordance with existing statutory requirements, TSA assesses the effectiveness of security measures at foreign airports (1) served by a U.S. air carrier, (2) from which a foreign air carrier operates U.S.-bound flights, (3) that pose a high risk of introducing danger to international air travel, and (4) that are otherwise deemed appropriate by the Secretary of Homeland Security. The Secretary of DHS delegated to the TSA Administrator the responsibility for conducting foreign airport assessments, but retained responsibility for making the determination whether a foreign airport does not maintain and carry out effective security measures. In carrying out this function, the statute identifies measures that the Secretary must take in the event that he or she determines that an airport is not maintaining and carrying out effective security measures based on TSA assessments which can include, in some cases, revoking the authority of U.S. carriers to operate at the airport. In addition, TSA is to conduct inspections of U.S. air carriers and foreign air carriers operating U.S.-bound flights from foreign airports to ensure that they meet applicable security requirements. Currently, the Global Compliance Directorate, within TSA’s Office of Global Strategies, is responsible for conducting foreign airport assessments and air carrier inspections. TSA began performing foreign airport assessments and air carrier inspections in Cuba in 2007, when only charter carriers operated flights between the United States and Cuba. The first foreign airport assessment after regularly scheduled commercial air service resumed was conducted in Sierra Maestra Airport in Manzanillo, Cuba on October 19, 2016, and the first air carrier inspection after scheduled commercial service commenced was conducted for an American Airlines flight on October 24, 2016, at Juan Gualberto Gomez International Airport in Varadero. TSA’s Process for Conducting Foreign Airport Assessments TSA assesses the effectiveness of security measures at foreign airports including those airports in Cuba offering U.S.-bound public charter and scheduled commercial flights using select aviation security standards and recommended practices adopted by International Civil Aviation Organization (ICAO), a United Nations organization representing 192 countries. ICAO is a specialized agency of the United Nations with a primary objective to provide for the safe, orderly, and efficient development of international civil aviation. ICAO member nations (i.e., contracting states) agree to cooperate with other contracting states to meet standardized international aviation security measures. ICAO standards and recommended practices address operational issues at an airport, such as ensuring that passengers and baggage are properly screened and that unauthorized individuals do not have access to restricted areas of an airport. ICAO standards also address non- operational issues, such as whether a foreign government has implemented a national civil aviation security program for regulating security procedures at its airports and whether airport officials implementing security controls are subject to background investigations, are appropriately trained, and are certified according to a foreign government’s national civil aviation security program. TSA utilizes 44 ICAO standards and recommended practices it sees as most critical in conducting its foreign airport assessments, which cover broad categories, including: passenger and cabin baggage security; and quality control. TSA uses a risk-informed approach to schedule foreign airport assessments across all foreign locations, including Cuba. TSA defines risk as a function of threat, vulnerability, and consequence. The agency uses various data sources to assess the likelihood of a location being targeted by bad actors, the protective measures in place to prevent an attack, and the impact of the loss from a potential attack. TSA categorizes airports into three risk tiers, with high risk airports assessed more frequently than moderate and low risk airports. TSA’s assessments of foreign airports are conducted by a team of inspectors, which generally includes one team leader and one team member. According to TSA, it generally takes 3 to 7 days to complete a foreign airport assessment. However, the amount of time and number of team members required to conduct an assessment varies based on several factors, including the size of the airport and the threat level to civil aviation in the host country. At the close of an airport assessment, inspectors brief foreign airport and government officials on the results as well as any recommendations for corrective actions and prepare an internal report. As part of the report, and as shown in table 1, TSA assigns a vulnerability score to each ICAO standard and recommended practice assessed as well as an overall vulnerability score for each airport, which corresponds to the level of compliance for each ICAO standard and recommended practice that TSA assesses. If the Secretary of Homeland Security determines that an airport does not maintain and carry out effective security measures, he or she shall, after advising the Secretary of State, take secretarial action. This generally includes notification to the appropriate authorities of security deficiencies identified, notification to the general public that the airport does not maintain effective security measures, publication of the identity of the airport in the Federal Register, and, when appropriate, modification of air carriers operations at that airport. According to TSA officials, no secretarial actions have been issued for Cuban airports since the resumption of public charter flights between the United States and Cuba in 1999 and scheduled commercial flights in 2016. TSA’s Process for Conducting Air Carrier Inspections Along with conducting airport assessments, the same TSA inspection teams also conduct air carrier inspections in foreign locations. During these inspections, a TSA inspection team examines each air carrier’s implementation of applicable security requirements, including their TSA- approved security programs, any amendments or alternative procedures to these security programs, and applicable security directives or emergency amendments. The frequency of air carrier inspections at each airport depends on a risk-informed approach and is influenced, in part, by the airport’s vulnerability to security breaches, since the security posture of each airport varies. In general, TSA’s procedures require it to conduct air carrier inspections at each airport on an annual or semi- annual basis depending on the airport’s vulnerability level, with some exceptions. At the close of an air carrier inspection, results are recorded into TSA’s Performance and Results Information System (PARIS) database. If an inspector finds that an air carrier is not in compliance with any applicable security requirements, additional steps are taken to correct and record those specific violations ranging from on-the-spot counseling for minor violations to sending a warning notice and/or a letter of correction, to issuing notices of civil penalties for more egregious violations. In extreme cases, TSA may withdraw its approval of an air carrier’s security program or suspend the air carrier’s operations. TSA’s Assessments and Inspections in Cuba Generally Followed Standard Operating Procedures, but Improvements Can Help Ensure they Occur at the Established Frequency TSA’s Foreign Airport Assessments and Air Carrier Inspections in Cuba Followed Standard Operating Procedures During fiscal years 2012 through 2017, TSA inspectors generally followed standard operating procedures for documenting foreign airport assessment results as required by TSA’s 2010 and 2016 Foreign Airport Assessment Program Standard Operating Procedures and Global Risk Analysis and Decision Support (GRADS) Business Rules. Similarly, TSA inspectors generally followed standard operating procedures for documenting air carrier inspection results in fiscal years 2016 and 2017 as required by the PARIS Business Rules. TSA also resolved reported deficiencies in a timely manner, and conducted foreign airport assessments at established intervals as required by TSA’s 2010 and 2016 procedures. Documentation: We found that data in most of the assessment reports TSA created in fiscal years 2012 through 2017 were generally complete with some reports missing some required information. Specifically: One airport assessment report did not answer required questions about training for aircraft pre-flight security checks and whether or not passenger screening met the requirements of Cuba’s national civil aviation program. Another airport assessment report did not indicate which security measures were being used to screen checked baggage, which is typically included in TSA’s airport profile report. A third airport assessment report did not have complete information regarding unescorted access to restricted areas. TSA officials explained that although inspectors did not document this information in the appropriate data fields within the report, they did record this information elsewhere within assessment documentation. We also found that data in air carrier inspection reports were generally complete and error-free. However, TSA was unable to provide full documentation for some of the air carrier inspections it conducted in Cuba in fiscal years 2016 and 2017. TSA officials attributed these missing documents to human error. We also identified errors or missing data fields in most of the air carrier inspections reports with complete documentation. For example: In reviewing air carriers’ compliance with a TSA security requirement for air carriers to notify U.S.-bound passengers that loaded firearms are prohibited in checked baggage, some inspection reports indicated that air carriers were simultaneously in compliance and not in compliance. Inspectors failed to document air carriers’ compliance with a TSA security requirement to prohibit unauthorized access to checked baggage during some air carrier inspections. The errors and missing data we identified constituted a relatively small proportion of the data in each inspection report, which include information on air carriers’ implementation of various TSA security requirements. TSA attributed these to human error and has since issued guidance and updated its air carrier inspection report template designed to better ensure that air carrier inspections are fully documented and less likely to contain such errors or missing data fields. Recording, Tracking, and Resolving Findings: We found that TSA generally followed procedures to record and track deficiencies identified during assessments at foreign airports and whether they have been resolved by the host government during subsequent visits. Among the foreign airport assessments conducted in Cuba in fiscal years 2012 through 2017, TSA recorded findings in several of them. In nearly all of the reports with findings, TSA followed its SOPs by recording findings and their root causes in an internal document and tracking the status of host country action to resolve each finding. In one report, TSA failed to record the root cause of a deficiency. This issue has been identified in a prior GAO report, and TSA is taking steps to resolve the issue by better documenting the root cause of each deficiency. We also found that TSA followed procedures to record, track, and resolve findings from air carrier inspections. Among the air carrier inspections TSA performed in fiscal years 2016 and 2017, TSA recorded several violations. In each instance, TSA recorded the root cause of each violation in PARIS, resolved each violation with on-the-spot counseling or investigation, and closed all air carrier findings in fiscal years 2016 and 2017 after air carriers took corrective action. Timeliness: During fiscal years 2012 through 2017, TSA generally completed foreign airport assessments in Cuba within the scheduled time frames per TSA’s policy. However, TSA explained that lapses can occur and that such deferments often take place worldwide due to scheduling conflicts, logistical issues, and operational concerns. TSA Inspections of Air Carriers Did Not Always Occur at the Established Frequency Our analysis of TSA air carrier inspection data from fiscal years 2012 through 2016—a period in which public charter flights accounted for nearly all commercial air traffic between the United States and Cuba— revealed that TSA did not always inspect air carriers operating U.S.- bound flights from Cuba each fiscal year at frequencies established in TSA’s standard operating procedures. In general, public charter flights are operated by air carriers but arranged or sponsored by a charter operator. Consistent with scheduled service, TSA requires air carriers operating U.S.-bound public charters to adopt and implement a TSA- approved security program. For inspection purposes, TSA does not differentiate between scheduled service and public charter service and inspects these operations to the same TSA security program requirements. According to TSA’s Operational Implementation Plans for fiscal years 2012 through 2016, TSA’s stated objective was to inspect 100 percent of air carriers operating U.S.-bound flights from foreign locations at the frequency established in its standard operating procedures. Specifically, depending on an airport’s vulnerability rating, TSA’s standard operating procedures provide that air carriers are to be inspected on either an annual or semi-annual basis. However, our analysis of TSA inspection data during fiscal years 2012 through 2016 identified that among the air carriers we selected for our analysis, TSA conducted little over half of the required inspections in Cuba at the frequency established in its standard operating procedures. For example, our analysis revealed that TSA inspected an air carrier in September 2013 and April 2015, but did not do so in fiscal year 2014—a year in which this air carrier operated a total of 127 U.S.-bound flights. In response to our analysis, TSA officials explained that host government requests to reschedule inspections and the flight schedule data used to track public charter flights hinder TSA’s efforts to inspect 100 percent of air carriers operating U.S.-bound public charter flights in Cuba. Among the air carriers we selected for our analysis, TSA officials told us that 10 of the required air carrier inspections were not conducted at the established frequency due to external factors, including host government requests to reschedule TSA inspections. The officials told us that when planned air carrier inspections are deferred, TSA works with the host government to reschedule the inspection as close as possible to the original inspection date. In some instances, TSA has been unable to reschedule air carrier inspections within the specified time frame based on their risk level, and as a result, did not conduct the air carrier inspection at the established frequency. For example, TSA officials told us that the Cuban Government deferred air carrier inspections planned for June 2015 at one airport to November 2015 (in fiscal year 2016). Although TSA completed these inspections as rescheduled, the inspections were not conducted at this airport in fiscal year 2015, as required by its standard operating procedures. In another example, the officials told us that TSA did not conduct air carrier inspections at an airport in fiscal year 2014 because of deferrals and logistical challenges that hampered its attempt to reschedule. As a result, TSA did not conduct air carrier inspections at this airport—originally planned for July 2014—until 9 months later. Further, the flight schedule data TSA uses do not reliably identify and track public charter operations in Cuba. In an effort to conduct 100 percent of air carrier inspections due for completion each fiscal year, TSA develops an annual Master Work Plan which it uses to schedule air carrier inspections in Cuba and other foreign locations at the start of each fiscal year. According to TSA officials, TSA inspectors develop the Master Work Plan by collecting flight schedule data from a variety of sources, including past plans, past inspection data, Wikipedia, Secure Flight data, bi-annual flight schedules provided by air carriers, and airline and airport websites, among others, to identify the universe of air carriers requiring inspection in the upcoming fiscal year and track flight schedules. However, TSA officials told us that these flight schedule data are not always reliable and provide limited visibility into the universe of air carriers operating U.S-bound public charter flights from Cuba. For example, the flight schedule data TSA currently uses may fail to identify that an air carrier is operating U.S.-bound flights from a specific Cuban airport. In one such instance, TSA officials told us that during a planned air carrier inspection at one Cuban airport, TSA inspectors learned that the air carrier they intended to inspect had contracted with a different air carrier to operate the flight on its behalf. TSA was previously unaware that the air carrier contracted to operate the flight was operating U.S.-bound flights from that Cuban airport and proceeded to inspect it. Although external factors, including host government deferrals and flight schedule data, are outside of TSA’s control, TSA officials acknowledged that a tool that better corroborates and validates the flight schedule data it uses to track air carriers requiring inspection each fiscal year would improve the reliability of these data and help TSA ensure air carrier inspections in Cuba occur at the frequency established in its standard operating procedures. As of January 2018, TSA officials told us they were developing a new tool intended to more reliably track flight schedules worldwide. Specifically, TSA officials told us that this tool is intended to analyze the aggregate flight data it currently uses and corroborate and validate flight schedule information. According to TSA officials, the tool may help improve the reliability of the flight schedule data TSA uses to track air carriers requiring inspection each fiscal year. However, since this tool is still under development, TSA has yet to demonstrate whether it will ultimately improve the reliability of flight schedule data among public charters in Cuba. Further, since the tool relies on the data sources TSA already uses, the tool is unlikely to provide TSA with improved visibility into the universe of U.S.-bound public charters requiring inspection beyond those operations of which TSA is already aware. Without the ability to reliably identify and track U.S.-bound public charter operations in Cuba, TSA will be at risk of continuing to fall short of its stated goal of completing 100 percent of required air carrier inspections and, therefore, cannot ensure that all air carriers are implementing TSA security requirements for U.S.-bound flights departing Cuba. Developing and implementing a tool that corroborates and validates the data TSA currently uses can help TSA improve its ability to track flight schedules and schedule inspection visits to coincide with air carrier operations. Taking additional steps to better identify the universe of air carriers operating U.S.-bound flights from Cuba can provide TSA with greater assurance that it is accurately identifying all air carriers operating U.S.- bound flights from Cuba that require inspection. These steps can better position TSA to meet its goal of inspecting all air carriers operating U.S.- bound public charter flights from Cuba to the United States at least once per year—as established in its standard operating procedures—and help them ensure that these air carriers are implementing TSA security requirements. TSA Assessments of Cuban Airport Security Found Mixed Levels of Compliance TSA Found Mixed Levels of Compliance with ICAO Standards and Recommended Practices at Cuban Airports TSA found mixed levels of compliance with ICAO standards and recommended practices at Cuban airports during fiscal years 2012 through 2017. Specifically, of the Cuban airport assessments TSA conducted during this period, several resulted in no findings–meaning that TSA inspectors determined the airport was fully compliant with each ICAO standard and recommended practice the airport was assessed against. Of the remaining foreign airport assessments that did result in findings, TSA inspectors found that most of the airports were fully compliant with all but one or two of the ICAO standards and recommended practices. The instances of noncompliance fall within the following five categories: Access Control: During an assessment at one airport, TSA inspectors observed that a section of fencing along the perimeter had deteriorated and needed repair. TSA inspectors subsequently recommended that the fencing be repaired and, during a follow-up visit, TSA inspectors found that the perimeter fence had been repaired. During an assessment at another airport, TSA inspectors found that a checked baggage conveyor belt door was left open and unsecured. During subsequent visits, TSA inspectors observed that the baggage conveyor belt door was properly secured. Quality Control: During assessments at two airports, TSA inspectors observed that a comprehensive audit of these airports had not been conducted, in accordance with ICAO standards. TSA officials stated that if non-compliant findings such as these remain open, TSA will follow up on the finding until a TSA official is able to reassess the finding during a subsequent assessment. Aircraft and Inflight Security: During assessments at two airports, TSA inspectors found that airport officials did not have a formal oversight process in place to monitor air carriers to ensure that they performed an aircraft cabin search prior to departure. TSA officials stated they will follow up on such findings and look to ensure, for example, that corrective actions asserted by airport officials have been taken—in these cases, by ensuring trained security coordinators to conduct aircraft security searches have been assigned. Passenger and Baggage Security: During an assessment at one airport, TSA inspectors observed an issue with passenger screening. During a follow up visit, TSA inspectors observed passenger screening and determined the issue had been resolved. Fencing: During an assessment at one airport, inspectors found that the concrete perimeter wall was not topped with barbed wire, and during another assessment at a different airport, inspectors determined the perimeter fence needed to be augmented in height and manner of construction to increase its effectiveness. TSA officials stated that they plan to follow up on these findings during their next scheduled assessments. At another airport, TSA observed that excessive vegetation potentially compromised a section of airport perimeter fencing. TSA subsequently recommended that the issue be addressed and aviation authorities stated their intention to make necessary repairs. Most Inspections Showed Air Carriers Fully Implemented All TSA Security Requirements and Cuban Personnel Continue to Oversee Security Measures for Each U.S.-bound Flight Most Inspections Showed Air Carriers Fully Implemented All TSA Security Requirements TSA’s air carrier inspection results show that, among the air carriers operating U.S.-bound scheduled commercial and public charter flights from Cuba that TSA inspected in fiscal years 2016 and 2017, more than two-thirds of these inspections resulted in no findings. A result of no findings means that TSA inspectors determined that air carriers operating these flights fully implemented all requirements in their TSA-approved security program at the time of inspection. For example, air carriers fully implemented security requirements such as access controls, area security, and checked baggage screening. TSA also found that air carriers generally implemented requirements concerning signs and notifications, passenger screening, and aircraft search at the time of inspection. For the one-third of inspections where air carriers had not fully implemented requirements, issues ranged from failure to notify U.S.- bound passengers that carry-on items and checked baggage are subject to search to inadequate aircraft searches. TSA subsequently closed each finding after the respective air carriers took corrective actions. These findings include: Bilingual Signs/Notifications: TSA inspectors discovered that air carriers at several airports failed to properly notify U.S.-bound passengers that all carry-on items and checked baggage are subject to search. TSA inspectors resolved each violation with on-the-spot counseling and recommended that Cuba’s airport security agency, the Empresa Cubana de Aeropuerto y Servicios Aeronáuticos (ECASA), post signs at the ticket counters or verbally advise U.S.-bound passengers that their property is subject to search and subsequently closed each finding. Figure 3 shows an example of bilingual signage, posted by ECASA in response to a violation, listing prohibited items at a Cuban airport. TSA-Approved Amendments to Air Carriers’ Security Programs Allow Carriers to Use Cuban Personnel to Oversee Security Measures for Each U.S.- bound Flight To implement their TSA-approved security programs, air carriers operating U.S.-bound flights from Cuba requested, and TSA approved, an amendment regarding the fulfillment of Ground Security Coordinator (GSC) roles and responsibilities at Cuban airports that went into effect in December 2017. In general, air carriers are required to designate a trained GSC for each U.S.-bound scheduled and public charter flight. Each designated GSC serves as the air carrier’s authorized representative for all security-related matters and must be present at the airport from the time the air carrier opens the first ticket counter for the day until the air carrier’s last flight scheduled for that day departs. For each U.S.-bound flight, designated GSCs are responsible for reviewing the implementation of relevant security requirements, including those outlined in each air carrier’s TSA-approved security program, such as the screening of passengers and checked baggage, aircraft security, and the prevention of unauthorized access to secure areas of the airport. Air carrier officials we spoke with told us that they generally contract with locally based GSCs or directly employ GSCs at foreign locations to serve as their authorized representatives and oversee security matters for each U.S.-bound flight. However, air carriers operating at Cuban airports have been unable to designate their own GSCs to review security matters for U.S.-bound flights for two reasons. First, the Government of Cuba controls most sectors of the economy and employs the majority of the Cuban workforce. As a result, according to an airline official we spoke with, there are no private security firms or trained GSCs in Cuba that air carriers can contract with to serve as their authorized representatives and review security matters for each U.S.-bound flight at Cuban airports. Second, TSA officials told us that the Government of Cuba employs Aviation Security Technicians (AST) to review security matters for each U.S.- bound flight at Cuban airports. According to these officials, ASTs undergo a training regimen similar to that of a GSC and can execute GSC roles and responsibilities. As a result, the Government of Cuba has not allowed air carriers to permanently station air carrier-employed GSCs at Cuban airports because, according to TSA officials, it believes ASTs already provide for these roles and responsibilities. Prior to the resumption of regularly scheduled commercial service between the United States and Cuba in August 2016, TSA responded to this issue by approving amendments to each air carrier’s security program. These amendments allowed air carriers operating in Cuba to utilize Cuban ASTs instead of their own designated GSCs to oversee security matters for each U.S.-bound flight at Cuban airports, provided ASTs are trained to execute all GSC functions in accordance with TSA requirements. Under these amendments, according to TSA officials, Cuban ASTs were responsible for overseeing security measures including Secure Flight prescreening as well as passenger and checked baggage screening, among others, whereas the air carriers were responsible for performing security measures aboard the aircraft, including cabin searches and preventing unauthorized access to the aircraft, among others. An official from one air carrier we spoke with stated that they found AST performance to be at least equivalent in quality to the performance of GSCs they contract with at other foreign airports. TSA officials anticipated that once regularly scheduled commercial service between the United States and Cuba commenced in August, 2016, the Government of Cuba would permit air carriers to designate their own GSCs to review security matters for each U.S.-bound flight at Cuban airports. As a result, TSA determined that it would not renew the existing amendments, but would permit both U.S.-bound scheduled commercial and public charters to operate under the existing amendment until it expired in September 2017. However, TSA officials told us that during a meeting in Havana in October 2016, the Government of Cuba informed TSA and air carriers that ASTs would continue to perform GSC functions at Cuban airports and that air carrier personnel were not authorized to perform GSC functions within Cuba. In August 2017, the Government of Cuba reiterated that it would not permit air carriers to designate GSCs at Cuban airports and that Cuban ASTs would continue executing these functions. In light of the situation, TSA decided in September 2017 to renew the amendments to air carriers’ programs allowing them to continue utilizing ASTs instead of their own designated GSCs at Cuban airports. These new amendments will expire in September 2019, at which point TSA, air carriers, and the Government of Cuba may revisit the GSC issue. Conclusions Since 2007, TSA’s air carrier inspections have played a vital role in ensuring that air carriers operating U.S.-bound flights from Cuba meet security requirements designed to further ensure civil aviation security keep passengers out of harm’s way. These inspections allow TSA to identify security deficiencies and help air carriers address them through, for example, on-the-spot counseling. Exemplifying the importance of these inspections, TSA aims to inspect each air carrier operating flights from Cuba to the United States at each airport from which flights operate, in accordance with its standard operating procedures. However, for the air carriers selected for our analysis, many of the inspections in fiscal years 2012 through 2016 did not take place within the established time frames. While delays in inspections can occur due to deferments from host governments, our analysis revealed that many air carrier inspections that did not occur within the required time frames were because the flight schedule data TSA uses do not reliably identify or track public charter operations—which account for the majority of flights between the United States and Cuba in fiscal years 2012 through 2016. Without the ability to reliably identify and track U.S.-bound public charter operations in Cuba, TSA will be at risk of continuing to fall short of its stated goal of completing 100 percent of required air carrier inspections and, therefore, cannot ensure air carriers are implementing TSA security requirements for U.S.-bound flights departing Cuba. TSA has a tool under development that if successfully implemented, may help corroborate and validate the flight schedule data TSA uses and assist TSA in more reliably tracking U.S.-bound public charters from Cuba. Taking steps to better identify the universe of all public charters requiring inspection in Cuba would also help better position TSA to ensure that these air carriers are meeting essential security requirements. Recommendations for Executive Action We are making the following recommendation to TSA: The Administrator of TSA should instruct the Office of Global Strategies to improve TSA’s ability to identify all public charter operations requiring inspection in Cuba and develop and implement a tool that corroborates and validates flight schedule data to more reliably track air carriers’ public charter operations between the United States and Cuba. (Recommendation 1) Agency Comments and our Evaluation We provided a draft of our report to DHS for its review and comment. In June 2018, DHS provided written comments, which are noted below and reproduced in full in appendix II. DHS and the Department of Transportation provided technical comments in the prior sensitive report, which we also incorporated as appropriate in this report. DHS concurred with our recommendation in the report. The Department of State did not comment on the report. DHS concurred with our recommendation to develop and implement a tool that corroborates and validates flight schedule data to more reliably track air carriers’ public charter operations between the United States and Cuba. In its response letter, DHS described the challenges it faces in scheduling inspections for air carriers that have entered into lease or codeshare agreements with other carriers. We acknowledge the challenges TSA faces in identifying the correct flights and responsible regulated parties when scheduling inspections under the conditions described and are encouraged by TSA’s planned steps to better identify public charter flight operations and shared flights. DHS’s response letter describes steps that TSA is taking to develop a tool that aims to better analyze flight data to use in scheduling inspections and prompts manual confirmation of flight information when the automated system identifies lower confidence of flight operations. During the course of our review, TSA described this concept and explained how it plans to use it to better identify scheduled flights for air carrier inspections. However, as DHS indicates in its response letter, TSA is still exploring how to best integrate public charter flights into this tool. DHS also described planned improvement to TSA’s Master Work Plan (MWP) to corroborate and validate flight schedule data. While DHS does not specify what these improvements include and how they will lead to more reliable tracking of air carriers’ public charter operations between the United States and Cuba, we agree that improving the scheduling tool that is used to plan inspections is a good place to start. DHS also described planned updates to the rules that guide the management of data in its MWP. Specifically, TSA plans to record anomalies in operations identified before, during, and after visits, such as trip dates that were changed or air carriers that were scheduled to be inspected, but were not, as well as the reason why. Our analysis discovered some of these anomalies and explaining them required TSA to engage in a lengthy process of tracking down historical information that was not readily available. These improvements, if implemented, will be a helpful step in providing better historical information to track and validate carrier operations. Finally, DHS described TSA’s plans to work with aircraft operators, foreign air carriers, and U.S. Government agencies to directly obtain flight information. These efforts, if implemented as planned, represent a positive step for TSA in corroborating and validating flight schedule data to more reliably track air carriers’ public charter operations between the United States and Cuba. DHS acknowledges that these efforts are underway with an estimated completion date of March 2019. We will continue to monitor TSA’s progress in implementing these planned actions. We are sending copies of this report to the appropriate congressional committees, the Secretary of the Department of Homeland Security, the Secretary of the Department of State, and the Secretary of the Department of Transportation. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff members have any questions about this report, please contact William Russell at (202) 512-6360 or RussellW@gao.gov. Key contributors to this report are listed in appendix III. Appendix I: Objectives, Scope, and Methodology This report examines: (1) the extent to which the Transportation Security Administration (TSA) complied with its standard operating procedures (SOP) when assessing aviation security at Cuban airports in fiscal years 2012 through 2017, 2) the results of TSA’s Cuban airport assessments in fiscal years 2012 through 2017 and how these results compare to those for airports in the Caribbean region, and 3) the results of TSA’s air carrier inspections for Cuba in fiscal years 2016—when commercial scheduled air service between the U.S. and Cuba resumed—and 2017. This report is a public version of a prior sensitive report that we provided to you in May 2018. The sensitive report included part of an objective related to how the results of TSA’s foreign airport assessments for Cuba compared to others in the Caribbean region. TSA deemed some of the comparison results related to this objective to be sensitive, which must be protected from public disclosure. This public report also omits certain information that TSA deemed to be sensitive related to the specific number of airport assessments and air carrier inspections performed by TSA in Cuba, results of those assessments and inspections, and TSA’s risk-based approach in identifying U.S.-bound public charter operations from Cuba, among others. To provide context regarding the scale and magnitude of our findings, without disclosing sensitive information, we characterized specific numbers as some, many, or several. Although the information provided in this report is more limited in scope, as it excludes such sensitive information, it addresses the same objectives and uses the same overall methodology as the sensitive report. To collectively address all three objectives, we reviewed the relevant laws and regulations pursuant to which TSA conducts foreign airport assessments and air carrier inspections. We also reviewed various TSA documents on program management and strategic planning, including TSA’s master work plans for scheduling air foreign airport assessments and air carrier inspections. Specifically, we reviewed TSA’s 2016 standard operating procedures, which prescribes program and operational guidance for assessing security measures at foreign airports and inspecting air carriers and inform TSA personnel at all levels of what is expected of them in the implementation of the program. We also reviewed TSA’s Operational Implementation Plans, which establish program goals. In addition, we reviewed the job aids that TSA inspectors use during each assessment and inspection, which are intended to ensure that the TSA-specified International Civil Aviation Organization (ICAO) aviation security standards and recommended practices and air carrier implementation of TSA security requirements are fully evaluated during each assessment and inspection. To understand how TSA assesses and manages its Cuban airport and air carrier risk information, we obtained and reviewed documents on TSA’s methodology for assigning individual risk rankings (called tier rankings) to each Cuban airport it assesses. We also, interviewed TSA officials located at headquarters and in the field and interviewed other federal stakeholders, such as the Department of State and the Department of Transportation (DOT). Lastly, to obtain air carriers’ perspectives on aviation security in Cuba, we interviewed representatives from three air carriers that DOT licensed to operate scheduled commercial flights between the United States and Cuba. While the information obtained from these interviews cannot be generalized to all air carriers DOT licensed, these interviews provided insights into the carriers experiences. We outline the specific steps taken to answer each objective below. To determine the extent to which TSA followed its standard operating procedures when assessing aviation security in Cuba in fiscal years 2012 through 2017, we examined documentation for each of the foreign airport assessments conducted during the entire period and all air carrier inspections conducted in fiscal years 2016 and 2017 in Cuba for completeness and errors. For each finding resulting from Cuban airport assessments and air carrier inspections we reviewed, we examined the extent to which TSA followed its SOPs when following up and closing findings. We also analyzed Cuban airport assessment and air carrier inspection data to determine if TSA performed each assessment and inspection at the frequency established in its SOPs. Lastly, we met with TSA officials at headquarters and in the field to discuss how TSA inspectors apply their SOPs when assessing Cuban airports and inspecting air carriers in Cuba. To determine the completeness of TSA’s Cuban airport assessments in fiscal years 2012 through 2017, we analyzed and compared these assessment reports to TSA’s SOPs and the job aids which instruct inspectors on how to complete their assessments. In performing this analysis, we reviewed whether TSA inspectors followed their SOPs when assessing and documenting each Cuban airport’s compliance with applicable ICAO standard and recommended practices and the extent to which these documents contained missing data fields. Similarly, we reviewed documentation for each air carrier inspection TSA performed in fiscal years 2016 and 2017 for errors and completeness by analyzing and comparing these documents to TSA’s SOPs. In performing this analysis, we reviewed whether TSA inspectors followed their SOPs when inspecting and documenting each air carriers’ implementation of requirements in their TSA-approved security program and the extent to which these documents contained errors or missing data fields. When we identified discrepancies in the documentation for TSA’s Cuban airport assessments or air carrier inspections in Cuba, we met with TSA officials to discuss the cause of the discrepancies. To determine whether TSA inspectors followed their SOPs when recording, tracking, and resolving findings discovered during Cuban airport assessments and air carrier inspections in Cuba, we reviewed TSA’s SOPs governing finding follow up, closure and documentation of each finding, the status of each finding, and the actions TSA took to close findings. Specifically, we reviewed TSA findings discovered during Cuban airport assessments in fiscal years 2012 through 2017 by analyzing TSA’s Open Standards and Recommended Practices Finding Tool (OSFT), which TSA uses to monitor and track a foreign airport’s progress in resolving security deficiencies identified by TSA inspectors during previous assessments. To determine whether TSA inspectors followed their SOPs in response to a finding resulting from air carrier inspections in fiscal years 2016 through 2017, we reviewed TSA documentation of each finding and documentation on TSA’s findings response, follow-up, and closure, including air carrier inspection reports and enforcement investigative reports. To determine whether TSA performed Cuban airport assessments and air carrier inspections at the frequency established in TSA’s SOPs, we analyzed TSA data for all airport assessments from fiscal years 2012 through 2017. We also analyzed TSA air carrier inspection data from fiscal years 2012 through 2016 for a non-probability sample of 5 of the 18 air carriers operating U.S.-bound flights from Cuba that TSA inspected during this period along with flight traffic data for Cuba for these air carriers from the Department of Transportation’s Bureau of Transportation Statistics T-100 data bank, which contains data on all U.S.-bound departures from foreign airports, among other things. To assess the reliability of the T-100 data, we reviewed documentation on system controls and interviewed knowledgeable officials from the Bureau of Transportation Statistics. After determining that the T-100 data were sufficiently reliable for our intended use, we compared these data against inspection data for select air carriers. To assess the reliability of TSA’s assessment and inspection frequency data, we reviewed program documentation on system controls, interviewed knowledgeable officials from TSA and checked TSA’s frequency data for any potential gaps and errors. To select air carriers for our analysis, we identified air carriers (five in total) operating public charters flights—which accounted for the majority of flights from Cuba to the U.S. in fiscal years 2012 through 2016—that: 1) Operated at least 4 U.S.-bound flights in a single month or greater than 25 U.S.-bound flights within a fiscal year from one or more Cuban airports, and 2) DOT licensed to operate scheduled commercial flights following the policy change under the Obama Administration. Since we selected a non-probability sample of air carriers, the results of our analysis cannot be generalized to all air carriers that operated U.S.- bound flights from Cuban airports during this period, but did provide us with insights about TSA’s adherence to the frequency of air carrier inspections in accordance with its SOPs. To determine how TSA inspectors apply their SOPs when assessing Cuban airports and inspecting air carriers in Cuba, we interviewed officials at TSA headquarters and conducted site visits to TSA’s Miami Regional Operations Center (ROC) in Florida and in Cuba. During our site visit at the Miami ROC, which is responsible for conducting airport assessments and air carrier inspections in the Caribbean and South America, we met with the ROC manager and the TSA inspectors who conducted foreign airport assessments and air carrier inspections in Cuba. During these meetings, we discussed TSA’s assessments and inspections in Cuba, how they follow the SOPs when performing these assessments and inspections, and their perspectives on Cuban aviation security compared to other locations. On our visit to Cuba, we observed TSA inspectors from the Miami ROC conduct four air carrier inspections at Frank Pais Airport in Holguin and Antonio Maceo Airport in Santiago de Cuba. To describe the results of TSA’s Cuban airport assessments and air carrier inspections in Cuba, we obtained and analyzed relevant program documents and interviewed TSA officials on the results of its evaluations in Cuba. Specifically, we reviewed documentation for all Cuban airport assessments performed in fiscal years 2012 through 2017. We also analyzed TSA’s foreign airport assessment program vulnerability tracker, which TSA uses to record and track the vulnerability scores it assigns to each Cuban airport. Specifically, the tracking sheet contains vulnerability scores for each ICAO standard and recommended practice used in each assessment, as well as overall vulnerability scores of 1 through 5 assigned to each airport after each assessment. This overall airport vulnerability score is a representation of compliance or noncompliance with all ICAO standards and recommended practices against which TSA assesses Cuban airports. To describe air carrier inspection results in Cuba in fiscal years 2016—when scheduled commercial service between the U.S. and Cuba resumed—and 2017, we analyzed inspection data from all air carrier inspections TSA performed in Cuba during this period and reviewed each air carrier’s compliance with requirements in its TSA- approved security program, such as aircraft search and passenger screening. We also interviewed TSA managers and inspectors about their roles and responsibilities in determining and documenting inspection results in Cuba. We conducted this performance audit from February 2017 to May 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. We subsequently worked with TSA from May 2018 to July 2018 to prepare this nonsensitive version of the original sensitive report for public release. This public version was also prepared in accordance with these standards. Appendix II: Comment from the Department of Homeland Security Appendix III: GAO Contacts and Staff Acknowledgements GAO Contacts Staff Acknowledgments In addition to the contact above, Kevin Heinz (Assistant Director); Josh Diosomito (Analyst-in-Charge); David Alexander; Bruce Crise; Taylor Hadfield; Eric Hauswirth; Tom Lombardi; Heidi Nielson, and Kevin Reeves made key contributions to this report.

Why GAO Did This Study On August 31, 2016, as part of a shift in U.S. policy toward Cuba, air carriers resumed scheduled commercial flights between the United States and Cuba, a route previously only open to public and private charter carrier operations. In June 2017, travel restrictions were revised to require U.S. travelers going to Cuba to travel as part of a licensed group. TSA, the agency responsible for securing the nation's civil aviation system, assesses Cuban airports and inspects air carriers operating U.S-bound flights to ensure they have effective security measures in place. GAO was asked to review TSA's assessments of Cuban aviation security. This report examines (1) the extent to which TSA followed its standard operating procedures when assessing aviation security at Cuban airports in fiscal years 2012 through 2017; (2) the results of TSA's Cuban airport assessments in fiscal years 2012 through 2017; and (3) the results of TSA's air carrier inspections for Cuba in fiscal years 2016—when commercial scheduled air service between the United States and Cuba resumed—and 2017. GAO reviewed TSA policies and procedures, observed TSA air carrier inspections in Cuba, and compared TSA data on assessments and inspections to data from the Department of Transportation. What GAO Found The Transportation Security Administration (TSA) generally followed its standard operating procedures when documenting and resolving findings from its foreign airport assessments and air carrier inspections at Cuban airports in fiscal years 2012 through 2017. However, TSA did not perform all the required inspections of air carriers operating U.S.-bound public charter flights from Cuba. Specifically, GAO found that for the five air carriers selected for analysis, TSA performed approximately half of air carrier inspections in Cuba at the frequency established in its standard operating procedures in fiscal years 2012 through 2016. Of the inspections TSA did not perform, over half were not performed because TSA was not able to identify or reliably track U.S.-bound public charter operations from Cuba. Improving TSA's ability to identify public charters requiring inspection in Cuba and implementing a tool it is currently developing that more reliably tracks air carrier operations would better position TSA to meet its goal of inspecting all air carriers operating U.S.-bound public charter flights from Cuba at the frequency established in its standard operating procedures. Several of the Cuban airports TSA assessed in fiscal years 2012 through 2017 were fully compliant with International Civil Aviation Organization Standards at the time of assessment. The remaining airport assessments reported instances of noncompliance within the five categories: access control, quality control, aircraft and inflight security, passenger and baggage screening, and fencing. The majority of air carrier inspections TSA performed for Cuba in fiscal years 2016 and 2017 resulted in no findings, meaning that TSA determined air carriers operating these flights fully implemented all requirements in their TSA-approved security program at the time of inspection. The remaining inspections resulted in findings, which TSA closed after air carriers took corrective action. This is a public version of a sensitive report issued in May 2018. Information that TSA deemed to be sensitive is omitted from this report. What GAO Recommends GAO recommends that TSA improve its ability to identify all public charters requiring inspection in Cuba and develop and implement a tool that more reliably tracks public charter operations between the United States and Cuba. TSA concurred with our recommendation.

gao_GAO-18-214

gao_GAO-18-214_0

Background Radiological material is used throughout the world for medical and industrial purposes. Possession of this material within the United States requires a license from NRC or from one of the 37 Agreement States to which NRC has relinquished regulatory responsibility. NRC and Agreement States issue two types of licenses authorizing the possession and use of radiological materials: specific licenses and general licenses. Specific licenses are issued for devices that typically contain larger quantities of radiological material, such as medical equipment used to treat cancer, cameras used for industrial radiography, and moisture and density gauges used in construction. Devices approved for use under a general license, by contrast, such as luminous exit signs, normally contain relatively small quantities of radiological material. Such devices are designed with inherent safety features, are widely available commercially, and do not require NRC or Agreement State approval to possess. Not all radiological material requires an NRC license for possession. For example, there is naturally occurring radioactive material in ceramics, fertilizers, and granite tile that does not require a license. This report focuses on radiological material that requires specific licenses for possession and use. Beyond requiring specific licenses for possession of radiological material, NRC may also require a general or specific license to import such material. Generally, NRC will issue an import license when the recipient of the material is authorized to receive and possess the material being imported. When issuing licenses for the possession of radiological material, NRC and Agreement States take steps to ensure companies are legitimate. Specifically, NRC and Agreement State officials are to conduct pre- licensing visits with all unknown applicants, using detailed screening criteria. According to NRC, the purpose of the site visit is to have a face- to-face meeting with the applicant to determine whether there is a basis for confidence that the applicant will use the radiological materials sought as represented in the application when the applicant receives the license. NRC has established a 14-point checklist to guide pre-licensing site visits and has developed a list of questions and activities related to each applicant’s business operations, facility, radiation safety operations, and personnel qualifications, to scrutinize the applicant and provide a basis for confidence that the applicant will use the radiological material as specified in the license. In 2003, the International Atomic Energy Agency published a system— which NRC adopted in 2004—that ranks quantities of individual radionuclides into one of five categories on the basis of their potential to harm human health. Under this system, a given radionuclide is considered dangerous when gathered in sufficient quantity and in close enough proximity to people to cause direct human health effects. A category 1 quantity, if not safely managed or securely protected, is likely to cause permanent injury to a person who handles or is otherwise in contact with it for more than a few minutes. Being close to this amount of unshielded material for a period of a few minutes to an hour will probably be fatal. A category 2 quantity, if not safely managed or securely protected, can cause permanent injury to a person who handles or is otherwise in contact with it for a short time (minutes to hours). Being close to this amount of unshielded radioactive material for a period of hours to days can be fatal. A category 3 quantity, if not safely managed or securely protected, can cause permanent injury to a person who handles or is otherwise in contact with it for some hours. Being close to this amount of unshielded radioactive material for a period of days to weeks can be fatal. Category 4 and 5 quantities are unlikely to cause permanent injury. In addition to categorizing radionuclides on the basis of their potential to harm human health, NRC has identified 16 radionuclides that are sufficiently attractive for use in a dirty bomb or for other malicious purposes. These 16 radionuclides of concern, shown in table 1, warrant enhanced security and protection measures—such as cameras, alarms, and other physical security measures—under NRC regulations. Radiological material is imported into the United States by both express consignment couriers arriving by air and air cargo carriers. Express consignment couriers, such as FedEx, move cargo for the public under express commercial services and provide door-to-door delivery. Air cargo carriers transport radiological material in cargo containers on commercial airlines. We have previously reported on the disparity in portal monitor deployment between the express consignment and air cargo environments. There are dozens of portal monitors in U.S. airports servicing express couriers, but few servicing air cargo carriers. According to CBP officials, handheld monitors are used to scan radioactive material at airports where portal monitors are not available. The CBP mission includes the border enforcement of the customs, immigration, and agriculture laws and regulations of the United States and enforcement on behalf of numerous federal agencies. The mission includes enforcement of the laws relating to the importation and exportation of merchandise into and out of the United States. In addition, the agency’s mission includes denying entry to terrorists and their weapons and criminals and their contraband. CBP’s Office of Field Operations is responsible for passenger and cargo processing activities related to border security, trade, immigration, and agricultural inspection at the nation’s air, sea, and land ports of entry. Prior to importing goods into the United States, information is submitted to CBP declaring the contents of shipments. This information includes, among other things, a description of goods, the name of the recipient, the port of entry, and a tariff code that classifies goods. CBP uses various data systems to track shipments into the United States and identify shipments for license verification. According to CBP, the Automated Commercial Environment is the primary system for processing shipments entering the United States, and it enables the government to make determinations about whether to admit goods into the country. The data stored in the Automated Commercial Environment are also used to ensure proper duty is collected for imported goods. CBP officials also view Automated Commercial Environment data in CBP’s Automated Targeting System, which is a decision support tool that analyzes shipment data to assess risk and identify potential violations. The Automated Targeting System includes automated alerts, which notify CBP officials when they need to take additional actions before shipments can be released. Information about NRC and Agreement State licenses for radiological material is included in NRC’s Web-Based Licensing System, which includes information about NRC and Agreement State licenses for category 1 and 2 quantities of radiological material. In addition, the Web-Based Licensing System includes up-to-date information on all NRC and six Agreement States’ specific licenses, including licenses that authorize possession of radiological material below the category 2 threshold. However, licenses for quantities of material below the category 2 threshold issued by 31 Agreement States are not kept in the system. The CBP data systems used to identify shipments for license verification are listed in table 2. CBP’s Policies and Procedures Require License Verification for Imported Radiological Material CBP has implemented a policy and procedures requiring CBP officials at airports to contact experts within a centralized CBP office to verify licenses for radiological material being shipped into the United States. Specifically, CBP issued its “Radiation Detection Standard Operating Procedures Directive” policy in March 2014, outlining when CBP officials at ports of entry are required to contact internal experts at CBP’s Teleforensic Center who possess the technical expertise to verify that NRC and Agreement State licenses for radiological materials are legitimate. The function of the Teleforensic Center is to provide field CBP officials with assistance in resolving scientific and technological questions, including detection, isolation, and control of potential threats that may result from the presence of chemical, biological, radiological, or nuclear materials. The Teleforensic Center is staffed with scientists with expertise in a range of scientific disciplines, including chemistry, biology, explosives, radiological science, and nuclear science. The Teleforensic Center has established a hotline to receive requests for license verification, among other things, and the experts are available 24 hours a day, 7 days a week. CBP’s 2014 policy requires CBP personnel to verify the legitimacy of NRC and Agreement State licenses for all commercial imports of industrial and medical radionuclides that require a license from NRC or one of the 37 Agreement States. To implement this policy, CBP has established procedures for private-sector entities and CBP. These procedures, which apply equally to all quantities and types of licensable radiological material, can be broken down into three parts: submission of paperwork, identification of material, and verification of the license by experts in the Teleforensic Center, as outlined in figure 1. Once the shipment information is entered into the Automated Commercial Environment, CBP data systems identify which shipments of radiological material require license verification. Specifically, CBP’s Automated Targeting System uses certain information to identify shipments requiring license verification. Details about this information are omitted from this report because they were deemed to be sensitive by CBP. Once shipments are flagged as containing licensable radiological material, an alert is sent to CBP officials at the airport informing them that the material requires a license from NRC or an Agreement State. The alert outlines the steps the officials need to take to verify that the license is legitimate. Among other things, the alert explicitly states the phone number for contacting the Teleforensic Center and includes instructions for handling the material. According to CBP procedures, CBP officials are not allowed to release the shipment until they receive approval from the Teleforensic Center. Officials at the Teleforensic Center primarily use NRC’s Web-Based Licensing database to verify the legitimacy of licenses granted by NRC. However, as we’ve previously reported, licenses for some radiological material that are granted by Agreement States are not kept in that database, requiring the center’s experts to also call specific points of contacts at Agreement States to verify these licenses. CBP officials told us that Agreement State offices typically are not open 24 hours a day, 7 days a week, occasionally requiring CBP to hold shipments until an official can be reached. In addition to consulting the Web-Based Licensing database and contacting Agreement State officials to verify licenses, experts at the Teleforensic Center can also request additional information from CBP officials at the airports. After the experts verify that a license is legitimate, they give approval to the CBP officials at the airport to release the shipment. CBP officials at the airport then document the release. CBP officials at the four airports we visited and experts at the Teleforensic Center told us that it typically takes 30 to 90 minutes to verify a license, but it can take longer if the experts have to consult with an Agreement State. If a license cannot be verified, the shipment is returned to the sender or, in the case of an illegal shipment, seized and referred to proper law enforcement officials, as outlined in the policy. CBP Has Not Verified All Licenses as Required by Its Policy and Procedures CBP has not verified all licenses for radiological materials as required in its policy and procedures. During the 21-month period we reviewed, CBP personnel at airports across the country did not verify the legitimacy of a significant number of shipments CBP considered as containing potentially dangerous radiological material. CBP officials at two of the four airports we visited may not have verified the legitimacy of licenses for many of the shipments of radiological material imported during the 21-month period, which was not consistent with CBP policy. After we brought this issue to CBP’s attention, it issued additional guidance. However, this guidance was not clear and caused confusion at the two airports we visited where actions continued to be taken that were not consistent with CBP policy. CBP Is Not Consistently Implementing Its Procedures, Potentially Leaving Many Shipments of Radiological Material Unverified CBP officials did not verify the legitimacy of licenses for many of the shipments of radiological material imported from January 1, 2015, to September 30, 2016. We found that during this time frame, CBP officials stationed at airports nationwide did not make the required calls to verify licenses for a significant number of shipments of radiological material identified by CBP as requiring license verification—leaving many licenses unverified over this 21-month period. These shipments came through airports across the United States and, according to CBP officials we interviewed, arrived by both express courier and air cargo companies. At two of the four airports we visited, we observed that CBP officials were taking actions that were consistent with CBP policy. Specifically, we noted the following: At one airport, officials responsible for reviewing shipments of imported radiological material told us that they call the Teleforensic Center whenever they receive an alert from CBP’s data system, consistent with CBP policy. In addition, the officials said that they send any requested information to the Teleforensic Center and wait for approval from the center before releasing shipments. At another airport, officials responsible for reviewing shipments of imported radiological material told us that they also call the Teleforensic Center whenever they receive an alert and only release shipments upon receiving approval. However, at the remaining two airports we visited, officials responsible for reviewing shipments of imported radiological material took actions that were not consistent with CBP policy to verify the legitimacy of radiological shipments entering the country. As a result, officials at these airports had not verified hundreds of licenses as required under CBP’s policy because the officials misunderstood what they were required to do. In discussions with these officials, some described taking actions that were not consistent with the license verification requirements. Details about the extent of verification are omitted from this report because the information was deemed to be sensitive by CBP. At one airport, CBP officials told us they typically verify licenses on- site without calling the Teleforensic Center. This airport had more than 100 shipments of licensable radiological material during the 21-month period for which CBP provided data, but officials only made a few calls to the Teleforensic Center to verify licenses during this time, leaving many shipments of material unverified. Instead of calling the Teleforensic Center as required, the CBP officials said that they reviewed the shipment paperwork and looked for anomalies. CBP officials said that they undertake this paperwork review regardless of the risk of the radiological material in the shipment. For example, they said they would use this approach to verify licenses for category 1 materials, which NRC and the International Atomic Energy Agency classify as likely to cause permanent injury to a person who comes into contact with them. The officials told us that they call the Teleforensic Center only when there is something wrong with the shipment. Officials at CBP headquarters told us that this procedure does not comply with their verification policy and would not be effective. At the second airport, CBP officials we interviewed told us that license verification was conducted by private-sector express couriers overseas, negating the need for officials at the port of entry to call the Teleforensic Center. The CBP officials at the airport believed that a Memorandum of Understanding (MOU) between CBP and private companies delegates responsibility to express couriers to scan material with radiation detection equipment. These CBP airport officials said that express couriers also verify licenses as part of this process. However, the MOU between CBP and express couriers does not address the verification of licenses for radiological shipments. CBP headquarters officials we interviewed told us that the airport’s practice does not comply with the agency’s verification policy and confirmed that the Teleforensic Center is the only entity that can verify licenses. The headquarters officials also reiterated to us that license verification is not conducted by overseas private-sector companies. Officials we interviewed from an express courier that ships radiological material also told us that they do not verify licenses. This airport made few calls to the Teleforensic Center to verify licenses during the 21-month period, according to the data provided to us by CBP. CBP Issued Additional Guidance, but This Guidance Did Not Initially Improve Compliance In February 2017, we briefed CBP headquarters officials on our findings from the site visits to the four airports. We included in our briefing a summary of findings from our site visits and information on the number of calls made by CBP officials to verify licenses. At this meeting, CBP headquarters officials indicated that they would look into why calls were not made. Subsequently, in March 2017, in response to this briefing, CBP headquarters issued additional guidance to remind all field officials of CBP’s license verification policy. The guidance states that CBP officials must contact the Teleforensic Center to verify the license for all shipments of licensable radiological material. In addition, the guidance states that shipments may not be released from the airport until experts at the Teleforensic Center have completed verification of the license. The guidance was issued in the form of a “muster”—a type of memorandum addressed to all CBP field offices to emphasize CBP policy. Once such a memorandum is issued, CBP relies on local officials to interpret and pass along this information to those working directly with the shipments. According to CBP officials, the guidance was communicated to managers and then the managers communicated this information to front-line staff through weekly meetings and informal discussions. However, the muster was not successful in correcting previous misconceptions at the two airports we visited where officials’ actions were not consistent with CBP policy and the muster did not fully resolve their noncompliance with CBP policy. In part this was because, according to officials, they found the muster confusing. In April and May 2017, several weeks after CBP issued the muster, we contacted officials at the four airports we previously visited. Based on interviews with CBP officials at the two airports where actions were not consistent with CBP policy before the muster, we determined that they were continuing to take actions that were not consistent with CBP policy after they received the muster. For example, CBP officials at one of the two airports said they were continuing to conduct license verification without the assistance of the Teleforensic Center. Officials at this airport told us that they believed their actions were consistent with the policy, even though they had not altered their actions in response to the muster. Similarly, at the other airport where actions were not consistent with the CBP policy before the muster, officials told us again that license verification can be conducted overseas by express couriers, citing the MOU allowing express couriers to scan material with radiation detection equipment. When we discussed the content of the muster with CBP officials in headquarters in June 2017, they acknowledged the muster was confusing and stated it needed to be further clarified. Subsequent to our June 2017 meeting with CBP officials, they provided additional data that suggested an increase in calls to the Teleforensic Center. In July 2017, CBP officials told us they planned to issue additional clarifications. Subsequently, in November 2017, CBP issued an additional muster emphasizing its policy to call the Teleforensic Center for all shipments of licensable radiological material. CBP headquarters officials told us that they were unaware, until we informed them, that selected airports were not calling the Teleforensic Center to verify licenses and that licenses were not being verified for some imported radiological material. This is because CBP does not have a mechanism, such as a monitoring system, to ensure that all required license verifications are occurring. Such a system could also conduct checks to ensure CBP officials are following agency policy. The challenge to creating such a system is that CBP houses the data necessary to create it in separate systems that do not communicate with each other, and these systems are currently run by different offices with differing missions within CBP. Federal standards for internal control state that management should establish and operate monitoring activities to monitor the internal control system and evaluate the results. Until CBP develops a monitoring system to help ensure that CBP officials comply with the license verification policy, the agency will not have reasonable assurance that it can identify activities that are inconsistent with its policy and take corrective action as necessary. CBP Policies and Procedures Are Not Effective at Ensuring Only Properly Licensed Radiological Material Is Imported CBP policies and procedures are not effective at ensuring that only properly licensed radiological material is imported into the United States. Specifically, CBP’s procedures for identifying licensable radiological material do not ensure that all shipments of radiological material are identified and verified, resulting in the exclusion of a significant number of shipments that possibly contained radioactive material during the 21- month period we reviewed. Moreover, CBP’s current policy and procedures treat all radiological shipments with the same level of scrutiny and do not target resources based on the risk of the material. Details about these issues are omitted from this report because the information was deemed to be sensitive by CBP. CBP’s Procedures Do Not Ensure That All Shipments of Radiological Material Are Identified and Verified CBP procedures for identifying licensable radiological material do not effectively implement its policy to verify the license for all shipments of licensable radiological material. We identified the following issues that result in limitations in CBP’s procedures. Specifically, the data system that CBP uses to implement its procedures does not sufficiently identify all shipments of potentially dangerous radiological materials. To implement its procedures, the agency chose to use an existing data system designed to process all types of imports into the United States. This system uses general customs information to identify the contents of shipments. Consequently, of the 44,152 shipments that could contain licensable radiological material, the system alerted CBP officials that they were required to verify relatively few licenses from January 1, 2015, to September 30, 2016. In addition, CBP’s license verification procedures do not currently target the higher-risk radiological materials. CBP’s method for identifying the contents of shipments does not include information that describes the quantity of radiological material. Specifically, categories 1, 2, and 3 quantities of radiological material can cause permanent injury or death to a person in contact with them for some period of time. As a result, according to a senior CBP official, it is safer to assume all shipments of radiological material are dangerous until proven otherwise. Federal standards for internal control recommend that agencies design control activities to achieve objectives and respond to risks. Until CBP develops a robust system that can identify all shipments of radiological material that pose risk, it will not have reasonable assurance that it has the appropriate policies and procedures necessary to verify licenses for these shipments. Furthermore, as we reported in December 2016, an essential element of enterprise risk management is to examine risks considering both the likelihood of the risk and the impact of the risk on the mission, in order to help prioritize risk response. Although CBP officials recognize that their current system and procedures have limitations and do not allow them to fully implement the agency policy to verify all shipments of radiological material that enter the United States, we found that they have not developed a system nor revised their procedures to address the issues we identified. Of particular concern is that CBP has not conducted a comprehensive assessment of (1) the information not currently included in the automated alert to determine what additional information would indicate shipments that may contain dangerous material or (2) how to create a more risk-based approach that distinguishes between higher- and lower-risk categories of radiological materials. Until it conducts such an assessment, CBP will not know how to adjust its current procedures to ensure that it is identifying all shipments of potentially dangerous radiological material and targeting its limited resources to those that pose the greatest risk. Conclusions CBP has implemented a policy and procedures intended to ensure that the tens of thousands of shipments of potentially dangerous radiological material imported through U.S. airports each year are properly licensed. However, CBP’s procedures do not effectively implement CBP’s policy of ensuring that only properly licensed radiological material gains entry to the United States. This is because CBP does not have a monitoring system to help ensure that CBP officials at airports nationwide are complying with the license verification policy. Until CBP develops such a system, the agency will not have reasonable assurance that it can identify activities that are inconsistent with its policy and take corrective action as necessary. In addition, CBP’s procedures for identifying licensable radiological material do not ensure that all shipments of radiological material are identified and verified. This is the result of CBP’s automated alert, which currently does not include all relevant information needed to identify such shipments. Additionally, CBP procedures do not distinguish between high-risk categories of radiological materials and lower-risk categories; therefore CBP cannot target its limited resources to the shipments that pose the greatest risk. CBP has not conducted a comprehensive assessment of the information not currently included in the automated alert and does not know which shipments pose the greatest risk. A comprehensive assessment could help CBP gain a better understanding of information not currently included in the automated alert, and it could better position the agency to make appropriate changes to its existing system and procedures, as well as target its limited resources toward the quantities of material that pose the greatest risk. Recommendations for Executive Action We are making the following three recommendations to CBP: The Commissioner of CBP should develop a monitoring system to help ensure that CBP officials comply with license verification policies and procedures. (Recommendation 1) The Commissioner of CBP should conduct a comprehensive assessment of information not included in the automated alert to determine what information is needed to identify licensable radiological material. (Recommendation 2) The Commissioner of CBP should develop a system that better identifies shipments of radiological material that pose the greatest risk and revise CBP’s policies and procedures as necessary to verify licenses for these shipments. (Recommendation 3) Agency Comments We provided a draft of this product to the Department of Homeland Security (DHS) and NRC for review and comment. DHS provided written comments, reproduced in appendix I, in which it concurred with our three recommendations. DHS stated that it will take the following actions, among others, to address our recommendations: (1) include a monitoring process in an updated version of its policy addressing license verification, (2) conduct a comprehensive assessment of information not included in the automated alert to determine what information is associated with dangerous material, and (3) develop an intelligence-driven process that identifies shipments of radiological materials that pose the greatest threat. In addition, DHS and NRC provided technical comments, which we incorporated as appropriate. We are sending copies of this report to the appropriate congressional committees, the Secretary of Homeland Security, the Chairman of the U.S. Nuclear Regulatory Commission, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions concerning this report, please contact me at (202) 512-3841 or trimbled@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made significant contributions to this report are listed in appendix II. Appendix I: Comments from the Department of Homeland Security Appendix II: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the individual named above, Ned Woodward, Assistant Director; Jeffrey Barron; Richard Burkard; Kendall Childers; Cindy Gilbert; Cynthia Norris; Danny Royer; Jerry Sandau; Travis Schwartz; and Kiki Theodoropoulos made key contributions to this report.

Why GAO Did This Study Thousands of shipments containing radiological material enter the United States each year through airports across the country. Radiological material is used in various medical and industrial applications, and possession requires a license from the Nuclear Regulatory Commission (NRC) or one of the 37 states to which NRC has relinquished licensing authority. Failure to verify the licenses could allow terrorists to acquire radiological material for a dirty bomb, which uses explosives to disperse the material. GAO was asked to review CBP policies and procedures related to license verification. This report examines, among other things, (1) the extent to which CBP follows its policies and procedures, and (2) the effectiveness of these policies and procedures. GAO reviewed relevant policies and procedures, analyzed CBP data related to radiological material shipments and license verification, interviewed CBP and NRC officials, and selected four airports to visit based on expected traffic of radiological shipments. What GAO Found U.S. Customs and Border Protection (CBP) agency officials at U.S. airports have not verified the legitimacy of all licenses for imported radiological materials as required by CBP's policy. The policy requires CBP officials, when alerted, to verify licenses by calling experts in a centralized CBP office. CBP officials at two of four airports GAO visited said they were calling as required. However, CBP officials at the other two airports did not verify many licenses from January 1, 2015, through September 30, 2016, and headquarters officials were unaware of non-compliance with CBP policy. Also, GAO found that during this time frame nationwide, CBP officials were alerted to verify licenses for a significant number of shipments of licensable radiological material for all U.S. airports, but they did not make all the required calls—leaving numerous shipments potentially unverified over this 21-month period. This situation occurred because CBP does not have a monitoring system to ensure that officials make license verification calls as required. Until CBP develops a monitoring system for license verification, it will not have reasonable assurance that it can identify activities inconsistent with its policy and take corrective action. CBP procedures cannot effectively implement the agency's policy that its officials verify all radiological material shipments imported into the United States. The procedures are not effective for this policy in part because they rely on automated alerts that are based on some but not all relevant information that could indicate potentially dangerous radiological material. Consequently, CBP's current system and procedures cannot ensure that all such materials will be identified. Under federal internal control standards, agencies are to design control activities to achieve objectives and respond to risks. However, CBP does not have the information it needs to develop a robust system or revise its procedures because it has not conducted a comprehensive assessment of the information not included in its automated alert system. In particular, CBP has not assessed relevant information not currently included in the automated alert or how to create a more risk-based approach that distinguishes between higher- and lower-risk quantities of radiological materials. Without such an assessment, CBP may be unable to develop a system or procedures that best support its policy for verifying imported radiological materials. This is a public version of a sensitive report GAO issued in September 2017. Information CBP deemed sensitive has been omitted. What GAO Recommends GAO recommends that CBP develop a monitoring system to help ensure that CBP officials comply with the agency's license verification policy, conduct an assessment to determine relevant information that is not included in the automated alerts, and develop a system that allows it to identify shipments of greatest risk. CBP concurred with GAO's three recommendations and outlined actions to implement those recommendations.

gao_GAO-18-484T

gao_GAO-18-484T_0

Consolidation Plans Did Not Fully Conform with Leading Capital Decision- Making Practices and GAO Recommendation Has Not Been Implemented In our September 2014 report, we found that DHS and GSA planning for the DHS headquarters consolidation did not fully conform with leading capital decision-making practices intended to help agencies effectively plan and procure assets. Specifically, we found that DHS and GSA had not conducted a comprehensive assessment of current needs, identified capability gaps, or evaluated and prioritized alternatives that would help officials adapt consolidation plans to changing conditions and address funding issues as reflected in leading practices. At that time, DHS and GSA officials reported that they had taken some initial actions that may facilitate consolidation planning in a manner consistent with leading practices. For example, DHS had an overall goal of reducing the square footage allotted per employee across the department in accordance with workplace standards, such as standards for telework and hoteling. As we reported in 2014, DHS and GSA officials acknowledged that new workplace standards could create a number of new development options to consider, as the new standards would allow for more staff to occupy the space at St. Elizabeths than previously anticipated. DHS and GSA officials also reported at that time that analyzing different leasing options could affect consolidation efforts. However, we found that the consolidation plans, which were finalized between 2006 and 2009, had not been updated to reflect these actions. In addition, we found in September 2014 that funding for the St. Elizabeths project had not aligned with what DHS and GSA initially planned. We reported that according to DHS and GSA officials, the funding gap between what DHS and GSA requested and what was received from fiscal years 2009 through 2014, was over $1.6 billion. According to these officials, this gap created cost escalations of over $1 billion and schedule delays of over 10 years relative to original estimates. We found in 2014 that these delays posed challenges for DHS in terms of its leasing portfolio. Specifically, DHS’s long-term leasing portfolio was developed based on the original expected completion date for St. Elizabeths development in 2016. In 2014, DHS and GSA reported that they had begun to work together to consider changes to the DHS headquarters consolidation plans, but they had not announced when new plans would be issued. Furthermore, because final documentation of agency deliberations or analyses had not yet been developed, it was unclear if any new plans would be informed by an updated comprehensive needs assessment and capability gap analysis as called for by leading capital decision-making practices. Therefore, in our September 2014 report we recommended that DHS and GSA conduct various assessments and analyses and use the results to inform updated DHS headquarters consolidation plans. DHS and GSA concurred with this recommendation and stated that their forthcoming draft St. Elizabeths Enhanced Consolidation Plan would contain these analyses. As of April 2018, however, the agencies had not submitted updated plan information to Congress that would either meet the requirements of the DHS Headquarters Consolidation Accountability Act or address our recommendation. According to DHS officials, the agencies prepared a comprehensive response to the Act, including updated analyses, but the information is no longer current and now needs to be revised and revalidated before it is submitted to Congress. Officials told us that the updated consolidation plans and analyses assumed that the project would receive more funding in fiscal years 2017 and 2018 than was appropriated. Further, officials told us that the current Administration is expected to provide input on the planned DHS component occupancies at the St. Elizabeths campus. We continue to believe that DHS and GSA attention to following leading capital decision-making practices—including having a consolidation plan that justifies future actions—is critical given the project’s multi-billion dollar cost and impact on departmental operations. Cost and Schedule Estimates for the St. Elizabeths Project Did Not Reflect Leading Practices and GAO Recommendation Has Not Been Implemented In our September 2014 report, we found that DHS and GSA cost and schedule estimates for the headquarters consolidation project at St. Elizabeths did not conform or only minimally or partially conformed with leading estimating practices, and were therefore unreliable. Furthermore, we found that in some areas, the cost and schedule estimates did not fully conform with GSA guidance relevant to developing estimates. In 2014, we found that DHS and GSA cost estimates for the headquarters consolidation project at St. Elizabeths did not reflect leading practices, which rendered the estimates unreliable. For example, we found that the 2013 cost estimate—the most recent available at the time of our 2014 report—did not include (1) a life-cycle cost analysis of the project, including the cost of repair, operations, and maintenance; (2) was not regularly updated to reflect significant changes to the program including actual costs; and (3) did not include an independent estimate to determine whether other estimating methods produce similar results. In addition, a sensitivity and a risk and uncertainty analysis had not been performed to assess the reasonableness of the cost estimate. We have previously reported that a reliable cost estimate is critical to the success of any program. Specifically, we have found that such an estimate provides the basis for informed investment decision making, realistic budget formulation and program resourcing, meaningful progress measurement, proactive course correction when warranted, and accountability for results. Accordingly, in 2014, we concluded that DHS and GSA would benefit from maintaining current and well-documented estimates of project costs at St. Elizabeths—even if project funding is not fully secured. In 2014, we also found that the 2008 and 2013 schedule estimates (the estimates available at the time of our review) did not include all activities for both the government and its contractors necessary to accomplish the project’s objectives and did not include schedule baseline documents to help measure performance as reflected in leading practices and GSA guidance. For the 2008 schedule estimate, we found that resources (such as labor, materials, and equipment) were not accounted for and a risk assessment had not been conducted to predict a level of confidence in the project’s completion date. In addition, we found the 2013 schedule estimate was unreliable because, among other things, it was incomplete in that it did not provide details needed to understand the sequence of events, including work to be performed in fiscal years 2014 and 2015. In 2014, we concluded that developing cost and schedule estimates consistent with leading practices could promote greater transparency and provide decision makers needed information about the St. Elizabeths project and the larger DHS headquarters consolidation effort. However, in commenting on our analysis of St. Elizabeths cost and schedule estimates, DHS and GSA officials said that it would be difficult or impossible to create reliable estimates that encompass the scope of the entire St. Elizabeths project. In response to our findings, officials said that given the complex, multiphase nature of the overall development effort, specific estimates are created for smaller individual projects, but not for the campus project as a whole. Therefore, in their view, leading estimating practices and GSA guidance cannot reasonably be applied to the high-level projections developed for the total cost and completion date of the entire St. Elizabeths project. GSA stated that the higher-level, milestone schedule currently being used to manage the program was more flexible than the detailed schedule we proposed, and had proven effective even with the highly variable funding provided for the project. However, our September 2014 review found this high-level schedule was not sufficiently defined to effectively manage the program. For example, our review showed that the schedule did not contain detailed schedule activities that included all government, contractor, and applicable subcontractor efforts. In our 2014 report, we recognized the challenges of developing reliable cost and schedule estimates for a large-scale, multiphase project like St. Elizabeths, particularly given its unstable funding history and agreed that incorporating cost- and schedule- estimating leading practices could involve additional costs. However, we also concluded that unless DHS and GSA invest in these practices, Congress risked making funding decisions and DHS and GSA management risked making resource allocation decisions without the benefit that a robust analysis of levels of risk, uncertainty, and confidence provides. Therefore, in our September 2014 report we recommended that DHS and GSA develop revised cost and schedule estimates for the remaining portions of the consolidation project in accordance with leading practices. DHS and GSA concurred with the recommendation. As of April 2018, however, the agencies had not submitted revised cost and schedule information to Congress that would either meet the requirements of the DHS Headquarters Consolidation Accountability Act or address our recommendation. GSA is leading efforts to revise the project’s cost and schedule estimates, and according to GSA officials, the revised figures will take into account leading cost- and schedule- estimation practices, including a risk assessment. We continue to believe that creating up-to-date, reliable cost and schedule estimates for DHS headquarters consolidation should be an integral part of DHS and GSA efforts to reassess the project. Without this information, it will be more difficult for agency officials and Members of Congress to make informed decisions regarding resource allocations and compare competing funding priorities. DHS Did Not Consistently Apply Its Acquisitions Guidance When Overseeing the St. Elizabeths Project, but Has Taken Steps to Implement GAO’s Recommendation In our September 2014 report, we also found that DHS had not consistently applied its major acquisition guidance for reviewing and approving the headquarters consolidation project. Specifically, we found that DHS had guidelines in place to provide senior management the opportunity to review and approve its major projects, but DHS had not consistently applied these guidelines to its efforts to work with GSA to plan and implement headquarters consolidation. Part of the inconsistency was the result of DHS designating the headquarters consolidation project as a major acquisition in some years but not in others. For example, we found that in 2010 and 2011, DHS identified the headquarters consolidation project as a major acquisition and included the project on DHS’s Major Acquisitions Oversight List. Thus, the project was subject to the oversight and management policies and procedures established in DHS major acquisition guidance; however, the project did not comply with major acquisition requirements as outlined by DHS guidelines. For example, we found that the project had not produced any of the required key acquisition documents requiring department-level approval, such as life-cycle cost estimates and an acquisition program baseline, among others. As we reported in 2014, in 2012, the project as a whole was dropped from the list. Subsequently, in 2013 and 2014, DHS included the information technology (IT) acquisition portion of the project on the list, but not the entire project. DHS officials explained that they considered the St. Elizabeths project to be more of a GSA acquisition rather than a DHS acquisition because GSA owns the site and the majority of the building construction is funded through GSA appropriations. In our 2014 report, we recognized that GSA had responsibility for managing contracts associated with the headquarters consolidation project. However, we also noted that a variety of factors, including the overall cost, scope, and visibility of the project, as well as the overall importance of the project in the context of DHS’s mission, made the consolidation project a viable candidate for consideration as a major DHS acquisition. By not consistently applying this review process to headquarters consolidation, we concluded that DHS management risked losing insight into the progress of the St. Elizabeths project, as well as how the project fits in with its overall acquisitions portfolio. Thus, in our September 2014 report we recommended that the Secretary of Homeland Security designate the headquarters consolidation program a major acquisition and apply DHS acquisition policy requirements. DHS concurred with the recommendation. As of April 2018, DHS has made some progress implementing this recommendation. For example, on September 16, 2014, DHS issued an Acquisition Decision Memorandum designating the DHS-funded portions of the headquarters consolidation program as a Major Acquisition Program to be overseen by the departmental Acquisition Review Board (ARB). DHS also made progress implementing this recommendation by conducting and documenting an ARB of the program in November 2016. The ARB process provided DHS greater oversight of headquarters consolidation, and provided a forum for officials to consider a wide range of issues affecting consolidation efforts, such as funding and project scope. In addition, in January 2018, DHS officials reported that they were working to align headquarters consolidation program documentation to meet the spirit of DHS acquisition policy guidance. We will reassess the status of this recommendation after the consolidation plan and cost and schedule estimates are updated and submitted to Congress per the DHS Headquarters Consolidation Accountability Act. At that time, we believe there will be more certainty about the future direction of the project overall, and DHS’s funded portion in particular, and we will be better able to assess the level of DHS acquisitions oversight for the project. Chairman Perry, Ranking Member Correa, and Members of the Subcommittee, this concludes my prepared statement. I look forward to responding to any questions that you may have. GAO Contact and Staff Acknowledgments If you or your staff members have any questions about this testimony, please contact Chris Currie, Director, Homeland Security and Justice Issues, at (404) 679-1875 or curriec@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this statement. GAO staff who made key contributions to this statement are John Mortin (Assistant Director), Karen Richey (Assistant Director), Juaná Collymore, Jennifer Leotta, Thomas Lombardi, David Lutter, and Erin O’Brien. This is a work of the U.S. government and is not subject to copyright protection in the United States. The published product may be reproduced and distributed in its entirety without further permission from GAO. However, because this work may contain copyrighted images or other material, permission from the copyright holder may be necessary if you wish to reproduce this material separately.

Why GAO Did This Study DHS and GSA have been managing efforts to consolidate DHS executive leadership, operational management, and other personnel at one secure headquarters location rather than at multiple locations throughout the Washington, D.C., metropolitan area. The consolidation is to include the development of multi-billion dollar headquarters facilities at the St. Elizabeths campus in Washington, D.C. In September 2014, GAO issued a report entitled: Federal Real Property: DHS and GSA Need to Strengthen the Management of DHS Headquarters Consolidation (GAO-14-648). This statement summarizes the key findings and recommendations from this report, and provides a status update as of April 2018 on DHS and GSA implementation of GAO's recommendations. To complete the September 2014 report, GAO compared DHS and GSA capital planning efforts against applicable leading practices, interviewed officials, and reviewed cost and schedule estimates for the St. Elizabeths project. To assess subsequent DHS and GSA actions to implement GAO's September 2014 recommendations, GAO conducted periodic follow-up with agency officials. What GAO Found In its September 2014 report, GAO found that Department of Homeland Security (DHS) and General Services Administration (GSA) planning for the DHS headquarters consolidation at the St. Elizabeths campus in Washington, D.C. did not fully conform with leading capital decision-making practices intended to help agencies effectively plan and procure assets. Specifically, GAO found that DHS and GSA had not conducted a comprehensive assessment of current needs, identified capability gaps, or evaluated and prioritized alternatives that would help officials adapt consolidation plans to changing conditions and address funding issues as reflected in leading practices. GAO recommended that DHS and GSA conduct various assessments and analyses and use the results to inform updated DHS headquarters consolidation plans. The agencies concurred with this recommendation. In its September 2014 report, GAO also found that DHS and GSA cost and schedule estimates for the headquarters consolidation project at St. Elizabeths did not conform or, only minimally or partially conformed, with leading estimating practices, and were therefore unreliable. Thus, GAO recommended that DHS and GSA develop revised cost and schedule estimates for the remaining portions of the consolidation project in accordance with leading practices, and the agencies concurred with this recommendation. The DHS Headquarters Consolidation Accountability Act of 2015, enacted in April 2016 would, according to the accompanying Senate committee report, ensure that DHS and GSA fully address the recommendations from GAO's September 2014 report and provide Congress the information needed to make sound decisions regarding the project. Among other things, the Act required DHS, in coordination with GSA, to submit information to Congress, including various assessments and updated cost and schedule estimates related to the DHS headquarters consolidation. As of April 2018, however, DHS and GSA had not submitted the information to Congress that would either meet the requirements of the Act or address GAO's recommendations. DHS and GSA officials cited funding instability as one challenge to updating consolidation plans and cost and schedule estimates. What GAO Recommends Among other things, GAO recommended in its September 2014 report that DHS and GSA develop revised DHS headquarters plans that reflect leading practices for capital decision making and also reliable cost and schedule estimates. DHS and GSA concurred with our recommendations.

gao_GAO-19-17

gao_GAO-19-17_0

Background The Buy American Act of 1933 was enacted during the Great Depression when there was a need to create and preserve jobs for American workers, and it established a preference for the federal government to buy domestic end products. Many of the products the federal government buys—including aircraft engines and medical supplies—are end products that may be subject to the requirements of the Buy American Act. Further, the Buy American Act does not apply to products that are purchased for use outside the United States or obtained through contracts under the micro-purchase threshold, which was generally $3,500 in fiscal year 2017. end products manufactured in the United States provided that (a) the product is a commercially available off-the-shelf item; or (b) the cost of the components mined, produced, or manufactured in the United States exceeds 50 percent of the total cost of all components. End products that are not considered domestic under the Buy American Act are treated as foreign. This characterization is based on the origin of the end product—that is, where the product is manufactured or produced—and not the vendor’s location. For example, a vendor located in Finland may supply end products manufactured in the United States, in which case these products would be treated as domestic products. Conversely, a vendor located in the United States may supply end products manufactured in Finland. In this case, the end products would be considered foreign. Buy American Act Exceptions and Waivers Although the Buy American Act establishes a preference for domestic end products, there are situations in which agencies can procure foreign end products through established exceptions to the Buy American requirements. In addition, under the Trade Agreements Act of 1979, the United States has waived domestic purchasing requirements—including the Buy American Act—for certain acquisitions of foreign end products from countries that are party to international trade agreements or are considered designated countries by the U.S. Trade Representative. Exceptions to the Buy American Act In implementing the Buy American Act, the FAR sets forth several exceptions that permit federal agencies to buy foreign end products. These include situations when a domestic end product is not produced in sufficient quantities or cases where the cost would be unreasonable to buy a domestic end product. The steps that contracting officers must take to determine or document an exception will vary depending on the circumstances of the acquisition. For example, a written determination from the Head of Contracting Activity (HCA) or a delegate may be necessary to determine non-availability in some cases. However, a written determination may not be required when an acquisition is conducted through full and open competition, is synopsized, and no domestic offer is received. Other exceptions to the Buy American Act restrictions on the purchase of foreign products, such as the exception for commercial information technology, are blanket exceptions that do not require a written determination. In addition, some agencies have specified additional considerations that must precede a determination and what level of authority is appropriate for certain determinations. The five Buy American Act exceptions that apply government-wide and the corresponding determination standards in the FAR are listed in Table 1. Individual federal agencies may make blanket determinations of situations in which the Act’s restrictions should not apply to that agency’s procurements, when it is not in the public interest to restrict the purchase of foreign end products. For example, over the years, DOD has entered into reciprocal procurement agreements with 27 foreign counterparts. DOD determined that it would be inconsistent with the public interest to apply the Buy American Act restrictions on products from these 27 qualifying countries. Thus, if an offer includes end products from a qualifying country, those products are not restricted by the Buy American Act and the acquisition of qualifying country end products does not require higher approval. This public interest exception for qualifying countries applies only to contracts awarded by DOD. Trade Agreements Act Waivers Federal agencies can purchase eligible foreign end products from designated countries when the Buy American Act’s requirements are waived because of the terms of an international trade agreement or other criteria, such as a designation by the U.S. Trade Representative as a least developed country. In accordance with the Trade Agreements Act of 1979, the president has the authority to waive the Buy American Act. For eligible products that come from countries covered by the World Trade Organization’s Government Procurement Agreement, Free Trade Agreements, and the Israeli Trade Act, the Buy American Act has been waived so that these items receive nondiscriminatory consideration and are on equal footing with domestic end products. In total, these agreements cover approximately 60 countries—overlapping with all but two of the DOD qualifying countries. Appendix II highlights the overlap. Unlike DOD’s blanket public interest exception for qualifying countries, the Buy American Act requirements are only waived under a trade agreement if the acquisition is of a certain value set by the U.S. Trade Representative. Current trade agreement thresholds, at or above which the requirements are waived, range from $25,000 for contracts for eligible products from Canada to $180,000 for the 45 other parties to the World Trade Organization’s Government Procurement Agreement. Table 2 lists the parties eligible for trade agreements and the associated threshold for supply contracts. The FAR specifies certain conditions in which trade agreements do not apply, even if the acquisition is above the requisite threshold value set by the U.S. Trade Representative. In these cases, the Buy American Act would apply. These conditions include, but are not limited to: acquisitions that do not use full and open competition, when the limitation of competition would preclude the procedures applicable to acquisitions covered by trade agreements; certain sole-source acquisitions for commercial items using simplified acquisition procedures; acquisitions set aside for small businesses; acquisition of ammunition, arms, or war materials, or for purchases indispensable for national security or national defense purposes; and acquisitions from federal prison industries or nonprofit agencies employing people who are blind or severely disabled. If the contracting officer determines that a trade agreement applies to a particular acquisition, which waives the Buy American restrictions, that determination does not require additional review at a higher level. This is similar to other circumstances where Buy American Act restrictions do not apply, such as for the acquisition of products for use outside the United States or contracts valued below the micro-purchase threshold. Certifying Product Country of Origin The Buy American Act’s applicability is based on the country of origin of the product being supplied, rather than the country of the vendor offering the product to the government. Vendors who propose to do business with the U.S. government are required to certify as to where their products are manufactured or produced—whether in the United States or in a designated country covered by the Trade Agreements Act. Vendors can provide an annual certification applicable to all of their contracts through the federal government’s contractor registry, known as the System for Award Management (SAM). Through SAM, a vendor identifies the country of origin for foreign products associated with a broad category of products. For example, a vendor could state that it provides aircraft components that originate in France and Mexico. Vendors also have the option not to certify the origin of their products in SAM, but instead provide information about foreign end products in their individual offers for contracts. Contracting officials include the relevant clauses in solicitations and contracts in accordance with regulation to require vendor certification. For example, the clause at FAR 52.225-2, Buy American Certificate, requires the offeror to certify that each end product is a domestic end product, or list any foreign end products and their country of origin. Federal Procurement Data System-Next Generation Once a contract is awarded, the awarding agency must enter certain information into FPDS-NG, a government-wide database for contract awards and obligations. The Office of Federal Procurement Policy (OFPP) within the Office of Management and Budget provides the overall direction for FPDS-NG, which is managed by the General Services Administration. FPDS-NG data can be populated through the individual systems agencies use to develop contracts. Agencies are responsible for the quality of the information transmitted to FPDS-NG, including data captured on the contract value and whether the foreign product acquisition is authorized by one of the Buy American Act exceptions or a trade agreement. This information is recorded at the contract level, or at the delivery order level for orders from indefinite delivery contracts. For certain product categories—essentially those that represent end products—FPDS-NG requires that contracting staff enter information in the “Place of Manufacture” drop-down data field, as shown in Figure 1. This field must be populated for all reported manufactured end products, including those valued under the micro-purchase threshold, which at the time of our review was generally $3,500. Options in this field include indicating that the product is made in the United States, or that it is made outside the United States and qualifies under one of the Buy American Act exceptions, or that it is subject to the requirements of a trade agreement instead of the Buy American Act requirements. In 2018, FPDS-NG data on agencies’ historical reporting of the use of Buy American exceptions were added to the website on which agencies post contracting opportunities (www.fbo.gov). According to OFPP, this allows vendors selling domestic products to more easily see how agencies acquire foreign goods pursuant to Buy American Act exceptions. Agencies Report Applying Buy American Act Exceptions and Waivers but Data Quality Issues Exist In fiscal year 2017, the federal government obligated approximately $7.8 billion for the acquisition of foreign end products, which accounts for less than 5 percent of total federal contract obligations for end products in that year. We observed differences in how civilian agencies and DOD apply Buy American Act exceptions and waivers. In our review of 38 contracts and orders from four agencies—DOD, HHS, DHS, and VA—we found 6 instances where the place of manufacture information was misreported in FPDS-NG. We further identified system limitations in how FPDS-NG captures information. Foreign End Products Accounted for Less Than 5 Percent of Contract Obligations for Products Potentially Subject to the Buy American Act in Fiscal Year 2017 Based on our analysis of FPDS-NG data, almost 40 percent of federal contract obligations in fiscal year 2017—totaling approximately $196 billion—were for domestic and foreign end products, such as aircraft parts, that may be subject to the Buy American Act. Less than 5 percent of these obligations—approximately $7.8 billion—were reported as foreign end products. This is consistent with the information agencies reported in FPDS-NG in the previous 4 years, with foreign end products accounting for approximately 3 to 8 percent of goods subject to Buy American Act restrictions between fiscal years 2013 through 2016. The foreign end products in fiscal year 2017 primarily came from South Korea, the United Kingdom, Afghanistan, Canada, Mexico, and the United Arab Emirates, which together accounted for almost half of the total foreign end products reported. Appendix III shows the federal government’s obligations for foreign end products from various countries for fiscal year 2017. The procurement of foreign end products is permitted by the flexibilities available under the Buy American Act’s exceptions and waivers. Agencies also procured foreign end products through means separate from the exceptions allowed under the Buy American Act, primarily in cases where the Act would not apply. Agencies reported obligating more than $700 million to procure foreign end products by applying one of the five government-wide Buy American Act exceptions—such as domestic non-availability or unreasonable cost—in FPDS-NG for fiscal year 2017. Agencies reported obligating approximately $550 million to procure foreign end products as permitted by the Trade Agreements Act, which waives the Buy American Act’s domestic preference requirements for US trading partners when eligible products are covered by trade agreements and are above certain dollar thresholds. DOD also obligated nearly $2.9 billion to procure foreign products from countries with which it has reciprocal procurement agreements, using what is referred to as the DOD qualifying country exception. This is an exercise of the authority available to agencies under the Buy American Act’s public interest exception. DOD determined that it is not in the public interest to restrict the purchase of foreign end products from 27 countries. All but two of these qualifying countries are also US trading partners, so some of these awards for eligible products may be authorized by a trade agreement. However, the qualifying country exception allows DOD to procure foreign end products without regard to dollar thresholds or other trade agreement eligibility limitations. Agencies also procured foreign end products, such as fuel, to be used outside the United States, in which circumstance the Buy American Act’s requirements do not apply. For fiscal year 2017, these obligations accounted for almost $3.7 billion—about 47 percent of all dollars obligated for foreign end products, as reported in FPDS-NG. Figure 2 highlights fiscal year 2017 obligations, including agencies’ reported spending on foreign end products under the Buy American Act exceptions and other means. DOD Buys Most Foreign End Products for Use Outside the United States while Civilian Agencies Report a Wide Array of Buy American Act Exceptions and Waivers DOD accounted for more than 80 percent—roughly $6.4 billion—of the total obligations for foreign end products in fiscal year 2017. Almost all of DOD purchases were either for use outside of the United States, so were not subject to Buy American Act restrictions, or were reported under the public interest exception for DOD qualifying countries. In contrast, civilian agencies report a more varied mix of the exceptions and waivers of the Buy American Act. The civilian agencies—which are unable to apply DOD’s qualifying country exception—were more likely to report buying foreign end products based on trade agreements or another exception to the Buy American Act requirements. Figure 3 shows how DOD and the civilian agencies acquired foreign end products authorized by the various exceptions and waivers of the Buy American Act. From our review of FPDS-NG data, the civilian agencies are more likely to cite one of the five Buy American Act exceptions or a trade agreement waiver when buying foreign end products, and thus take corresponding actions to document or approve the authority cited. For example, in our review of contracts from four agencies, VA obligated $71,000 for medical imaging equipment from Canada, and had to consider whether a trade agreement waiver applied. The manufacturer was determined to be the only source available and the contracting officer determined the acquisition was authorized by a Buy American exception based on domestic non-availability, which can require additional review. In contrast, DOD may make a similar contract award for equipment from Canada based on the qualifying countries exception. DOD acquisitions, then, may be authorized by exceptions such as domestic non-availability when a required item does not come from a qualifying country. For example, we reviewed a $744,000 DOD award for vehicle equipment that was only available from South Africa—which is not one of the DOD qualifying countries and not covered by any of the trade agreements—so the acquisition was authorized by the domestic non-availability exception. In addition, the civilian agencies also reported buying foreign end products for use outside the United States but to a lesser extent than DOD. For example, this included one of the contracts we reviewed, an HHS award for Ebola vaccines manufactured in the Netherlands, with $44.7 million obligated in fiscal year 2017. This contract was reported as used outside the United States because it is primarily stored overseas. Coding Errors and System Limitations Highlight Data Reliability Issues FPDS-NG is the primary means for capturing procurement data regarding the Buy American Act, but we found that agencies may not always input reliable information on the extent to which exceptions or waivers authorized the acquisition of foreign end products. In addition, some aspects of how FPDS-NG is structured could lead to additional data reporting errors. In the non-generalizable sample of 38 contracts and orders we examined from DOD, HHS, DHS, and VA, we found 6 awards where information related to the Buy American Act was incorrectly reported in FPDS-NG. In three of the six contracts, agencies recorded the wrong exception or waiver, most often because of an error when reporting the place of manufacture in FPDS-NG. For example, DOD reported a $22,000 contract for vehicle equipment from South Africa as a Buy American Act exception due to unreasonable cost. But the contract file indicated that the exception that applied was domestic non-availability. DOD officials acknowledged the error and corrected it in FPDS-NG during the course of our review. In the three remaining contracts, agencies misreported whether an end product came from the United States or another country. For example, DHS incorrectly recorded that an $18 million contract was for aircraft accessories and other parts manufactured in the United States, even though file documentation showed the contract was for Italian-produced spare parts from the original equipment manufacturer. The Italian- produced spare parts were available from existing inventory maintained by the manufacturer and were needed immediately to meet a mandatory operational requirement. Officials from DHS acknowledged the recording oversight, attributed it to a mistake when entering information in FPDS- NG, and have since corrected the error in response to our observation. Additionally, FPDS-NG has system limitations that could hinder complete and accurate reporting of Buy American Act information: DOD Qualifying Country Exceptions and Trade Agreement Waivers. FPDS-NG requires that information on the type of Buy American Act exception or waiver applied be provided when end products are reported as foreign. But FPDS-NG does not identify errors associated with this process. For example, we reviewed an $8.3 million DHS contract for engines manufactured in Germany that was recorded as a DOD qualifying country exception in FPDS-NG, although this exception is not available to civilian agencies. Contracting officials corrected the data in FPDS-NG during the course of our review. Further, FPDS-NG does not prevent agencies from reporting trade agreement waivers when the contracts are valued below applicable thresholds or waivers do not apply, such as for small business set asides. For example, in the fiscal year 2017 data we reviewed, more than 5 percent of contract obligations reported for trade agreement waivers were for awards set-aside for small businesses, which would not be eligible under the Trade Agreements Act. OFPP officials noted that because of the various dollar thresholds applicable to different trade agreements, adding automatic thresholds in FPDS-NG to guide contracting staff in reporting an applicable trade agreement could lead to additional data errors in the procurement database. Awards under the Micro-purchase Threshold. Although the Buy American Act requirements do not apply to contract awards valued below the micro-purchase threshold—generally $3,500 in fiscal year 2017—the FPDS-NG ‘Place of Manufacture’ field does not have an option to indicate whether a contract is under the threshold. Instead, contracting officers entering information for awards under the micro- purchase amount must still state whether the product is domestic or foreign. If the product is foreign, the officials must select a Buy American Act exception authorizing the purchase, even though no exception is needed at these dollar levels. As a result, when agencies report in FPDS-NG that a Buy American Act exception or waiver applied for a procurement valued at less than $3,500, that information would not be accurate. Based on our review, this may have involved about $16 million in fiscal year 2017 obligations. Awards for both Foreign and Domestic Products. When reporting data for contracts that include multiple end products from both the United States and a foreign country, FPDS-NG only allows for one country of origin to be identified. Contracting officers told us that they typically will report a foreign end product in FPDS-NG when the foreign products account for the preponderance of the contract value. Thus, in cases where a contract includes foreign end products that do not account for the preponderance of the contract’s value, the value of these foreign end products would not be reported in FPDS-NG. We have previously reported that FPDS-NG has similar limitations in other fields, such the type of product or service provided, which prevent contracting officers from identifying more than one condition. According to OFPP, a recent change in the FAR requiring contract reporting at the line item level should provide greater transparency of all products included in a contract. Buy American Act Exceptions and Waivers under Indefinite Delivery Contracts. The way FPDS-NG captures data for Buy American Act exceptions and waivers for some indefinite-delivery contracts results in inaccurate data reporting. When an indefinite- delivery contract is initially awarded, FPDS-NG functionality does not give contracting staff the option to enter information for the ‘Place of Manufacture’ field. Instead, this information is typically captured once an order is placed on the contract. In our review of FPDS-NG data across the four agencies, however, we found that in some cases obligations are reported on the initial indefinite delivery contract so the Buy American Act exceptions or waivers are not recorded. This occurred with multiple agencies, but particularly at HHS, where information for almost 28 percent of HHS obligations for end products in fiscal year 2017 was not captured in FPDS-NG because the obligations were reported in the system through the initial contracts rather than orders. As a result, the applicability of the Buy American Act for HHS contracts totaling almost $1.9 billion in fiscal year 2017 was unreported in FPDS-NG. DOD, DHS, and VA officials told us they identified FPDS-NG reporting as an area of concern. GAO Standards for Internal Controls in the Federal Government state that management should use quality information to support objectives, and that such data should be complete and accurate. In response to the 2017 Executive Order calling for federal agencies to assess their implementation of the Buy American Act requirement, OFPP officials told us they are identifying potential strategies for improving the information agencies submit to FPDS-NG. As OFPP weighs potential options for FPDS-NG reporting, implementing enhancements to reduce data entry errors and ensure that the data collected are complete and accurate would help enable the system to provide the most useful information possible. Ensuring information is correctly reported in FPDS- NG is critical because the data are used to inform procurement policy decisions and facilitate congressional oversight. Agency Approaches to Training and Guidance Varied and Certain Contracting Practices Highlight Buy American Act Implementation Concerns The four agencies we reviewed—DOD, HHS, DHS, and VA—took different approaches to provide training and guidance for the Buy American Act requirements. Contracting officers faced challenges when procuring products subject to the Buy American Act. For example, we found instances in which contracting officers applied a waiver or exception to contracts where the waiver did not apply and did not have complete guidance for required determinations or reviews. There also were challenges in confirming product origin information when vendors did not provide consistent information. Agency Approaches to Guide Implementation of Buy American Act Requirements Varied The four agencies we reviewed varied in the mix of training and guidance provided to aid contracting officers in implementing the requirements of the Buy American Act. Three of the four agencies—DOD, DHS and VA— supplemented the federal acquisition regulation, which implements the requirements of the Buy American Act and Trade Agreements Act, with their own acquisition regulations. In addition, DHS and DOD have recently updated existing training or added new training and guidance. VA issued policy memoranda in 2017, emphasizing the importance of meeting Buy American Act requirements, but has not added training or provided specific guidance. HHS does not provide department-level training or guidance related to the Buy American Act. Most of the DHS and DOD contracting officers we spoke to reported that they had attended training and several found the guidance provided by the training to be helpful. HHS and VA contracting officials described confusion due to the lack of resources available at their respective agencies. DOD Updated Its Buy American Act Training in 2017 In 2017, in response to a series of recommendations from the DOD Inspector General to re-emphasize Buy American Act training and guidance, the Defense Acquisition University introduced an updated training course that specifically focuses on the requirements and implementation of the Buy American Act. While not mandatory, a June 2017 memo notified all DOD services and the defense agencies that members of their contracting workforce should complete this training as part of their professional development. At the current pace of enrollment, DOD officials anticipate approximately 18,000 people will have taken this course by the end of September 2018, which is a seven-fold increase over previous graduation rates. Incorporated into these trainings were supplemental on-the-job tools to assist contracting officers when awarding contracts for end products subject to the Buy American Act requirements. One such tool is a flowchart outlining applicable solicitation provisions or contract clauses based upon the awarded contract’s total dollar value. DOD contracting officials we interviewed from Defense Logistics Agency’s (DLA) Land and Maritime division had completed the agency-level Buy American Act training and said it served as a good refresher, with some noting that most of the training they had received on the subject came when they were first hired. DOD provides regulations and guidance on Buy American Act requirements through both the Defense Federal Acquisition Regulation Supplement (DFARS) and the accompanying Procedures, Guidance and Information. DOD contracting officers use the provisions and clauses in DFARS to address the public interest exception for DOD qualifying countries. In addition, as a part of the updated training, the Defense Pricing and Contracting Office developed two documents to provide additional Buy American Act guidance. One outlines a step-by-step approach contracting officers can follow to determine whether the Buy American Act applies to their particular procurement and, if so, whether the use of an exception or waiver is appropriate. The second assists contracting officers with evaluating all offers—foreign and domestic— when price is the determining factor. In addition, we found that DLA supplements the available Defense Acquisition University training and guidance with a robust level of support, including annual training and subject matter expertise. DLA contracting officers told us that while they found the updated training helpful, they also appreciated the training course internal to their agency, as it addresses the types of procurements they typically handle in their day-to- day work, such as buying spare parts. Further, DLA contracting officers noted that they use the job aid provided through the local training. DHS Added Buy American Act Training and Revised Policy DHS introduced training courses in 2017 that specifically focus on the requirements and implementation of the Buy American Act, including a mandatory training course for DHS contracting officers. DHS reported that 94 percent of contracting staff had taken the required course as of April 2018. DHS developed these courses in response to the 2017 Executive Order to ensure its staff was familiar with the Buy American Act requirements. Incorporated into these training courses are supplemental on-the-job tools to assist contracting officers when awarding contracts for end products subject to the Buy American Act requirements, such as a flowchart outlining applicable solicitation provisions or contract clauses based upon contract dollar value. Contracting officials generally view the training and tools they received as beneficial. For example, several DHS contracting officials we interviewed said that the agency’s new course provided a helpful review on the topic, while one contracting officer specifically noted that the course materials are useful to new staff, to help them understand the Act’s waivers and exceptions. DHS also revised its acquisition manual in December 2017 to add further detail regarding the Buy American Act requirements. Specifically, DHS updated its acquisition manual to provide contracting officers more explicit FPDS-NG reporting instructions for procurements subject to the Buy American Act, as well as discretion to purchase domestic end products at or below the micro-purchase threshold. Additional changes include increasing the documentation and level of managerial review required to use several of the exceptions to the Buy American Act. For example, prior to 2018—which includes the time period in which the DHS contracts and orders we reviewed were awarded—the head of individual contracting offices had the authority to approve domestic non-availability and unreasonable cost exceptions, with a notification made to the DHS Chief Procurement Officer. But under the new policy, the use of these exceptions must have the concurrence of the HCA—who is responsible for contracting activities within individual DHS components—and be approved by the department’s Chief Procurement Officer. Table 3 outlines these changes. In September 2017, VA issued guidance to reinforce existing Buy American Act requirements. The policy memorandum encourages the HCAs within VA to institute reviews of awarded contracts subject to the Buy American Act to ensure compliance. As of September 2018, policy officials did not know how many HCAs had taken this step. Further, the guidance emphasizes the importance of Buy American Act training for its acquisition workforce. VA policy officials explained that the Buy American Act is introduced in several VA training courses, but the agency does not have a specific course on implementing the Buy American Act requirements or provide additional instruction or tools. During the course of our review, VA officials said that some of the HCAs had added internal training on the Buy American Act. In addition, VA contracting staff has the option of taking training offered outside the agency, such as the updated Defense Acquisition University course on the Buy American Act. This training is not required. Contracting officials we spoke to at VA said they struggled with the details of awarding contracts subject to Buy American Act requirements because they are not provided sufficient agency-specific training and guidance on the topic. Moreover, several contracting staff noted an increased need for training due to recent changes in VA contracting practices. Specifically, in response to a 2016 Supreme Court decision, VA has increased contracting efforts with veteran-owned small businesses through the Veterans First Contracting Program. As a result, contracting officials explained they have reduced their use of schedule contracts, in which the determinations related to the Buy American Act requirements were made with the initial awards. As one contracting officer explained, prior to this change, more than 90 percent of her division’s procurements were through VA schedule contracts in which Buy American Act applicability had already been established. However, this shift in contracting practices means contracting officers will more frequently need to consider the applicability of the Buy American Act, but contracting officers have not received specific guidance or training to do so. Noting the significance of this change, one contracting officer stated she approached VA management to obtain Buy American Act training for her division, but such training was not available. Federal internal controls state that agencies should ensure that training is aimed at developing and retaining employee knowledge, skills, and abilities to meet changing organizational needs. In September 2018, we reported that VA was experiencing difficulties implementing multiple aspects of the Veterans First policy, and we recommended that VA provide more targeted implementation training. As VA moves forward to implement this training, incorporating the Buy American Act requirements will be important to provide greater assurance that staff has the knowledge and skills needed to navigate the changing procurement environment. HHS Does Not Provide Department-wide Buy American Act Training or Guidance HHS does not have agency-level Buy American training or guidance. The HHS Acquisition Regulation Supplement does not address foreign acquisitions. HHS officials told us that efforts to develop guidance that would address Buy American Act requirements are underway, but they do not know when they will be finalized and made available to contracting officers, and could not describe the extent to which they will address Buy American Act implementation. The HHS contracting officers we interviewed discussed informal Buy American Act training their divisions had developed because department-level training was not available. For example, at HHS’ National Institutes of Health, a contracting official told us about a training course she recently developed because her office was taking on the administration of additional contracts for which the Buy American Act requirements would apply. Contracting officers at HHS’ Office of Biomedical Advanced Research and Development Authority described informal training on the agency’s contract writing system— included as part of their weekly internal staff meetings—that provides additional guidance on how to appropriately complete certain data fields relevant to the Buy American Act. Agencies Face Challenges Applying Waivers and Exceptions Where Guidance and Training Is Incomplete or Does Not Exist In our analysis of 38 contracts from across the four agencies, we found that agencies faced various levels of challenge in applying the Trade Agreements Act waivers and Buy American Act exceptions to acquire foreign end products. This was particularly apparent in cases where contracting officers had to determine if a trade agreement applied or cases which required a determination that a domestic end product was not sufficiently available, in accordance with the domestic non-availability exception to the Buy American Act requirements. Contracting staff also had difficulty determining the origin of products in light of incomplete or conflicting information. Trade Agreements Act Incorrectly Applied Of the six contracts we reviewed reporting that a trade agreement applied to the foreign end products purchased, we found two cases in which this waiver did not apply to the contracts in question. The value of the contract is one determining factor for whether a trade agreement waives the Buy American Act requirements, although the FAR also states additional factors that would affect applicability under a trade agreement. The two contracts we found, both from VA, had total dollar values at award— $8,435 and $11,950, respectively—that were less than any of the thresholds at which trade agreement waivers of the Buy American Act are applicable. Both contracts were for products from countries that are party to the World Trade Organization Government Procurement Agreement, so the value of the acquisition would have to be equal to or exceed $191,000—the threshold that was in effect at the time of award—for waivers from Buy American requirements to apply. Contracting officials in both cases were generally unaware that the applicable threshold was not met, making the trade agreement waiver inapplicable. Although VA has added Buy American Act guidance since these contracts were awarded early in fiscal year 2017, the information currently available does not provide sufficient detail to assist contracting officers when awarding contracts in these situations. For example, the guidance VA provided contracting officers in September 2017 does not emphasize consideration of the applicable trade agreement thresholds or include information on how contracting officers should determine the applicable waiver or exception. When contracting officers procure foreign end products, the type of waiver or exception used to support the purchase matters—particularly when required additional steps and review are not completed because the wrong waiver or exception was applied. We found that the two VA contracts with foreign end products were incorrectly reported as the Trade Agreements Act waiver having applied. If one of the other Buy American Act exceptions permitting purchases of foreign end products had applied, contracting officers may have been required to obtain higher-level review or complete a written determination. In addition, we reviewed contracts that show some of the complexities contracting officers face beyond applying the dollar thresholds when determining if an award for foreign end products is eligible under the Trade Agreements Act waiver of the Buy American Act. Specifically, we found instances where DHS contracting officials took different approaches for non-competed awards for similar foreign-manufactured products. For example, we reviewed a non-competed $58 million DHS award for acquiring spare aircraft parts from an original equipment manufacturer located in a foreign country that is party to the World Trade Organization Government Procurement Agreement. DHS reported in FPDS-NG that the procurement was waived by the Trade Agreements Act. However, we also reviewed two other sole-source awards from DHS for similar products—spare aircraft parts from two separate manufacturers in foreign countries that are also party to the World Trade Organization Government Procurement Agreement—that were instead reported as subject to the Buy American Act, but excepted due to the non-availability of domestic products. Contracting officers may come to different conclusions for similar products, in part, because of the multiple factors that have to be considered when determining whether an acquisition is subject to the Buy American Act and whether any waivers or exceptions apply. However, available guidance does not always address these complexities. For example, agencies need to decide if other conditions, such as the procurement of products deemed indispensable for national security or national defense purposes apply to an acquisition that would make a trade agreement inapplicable. Further, if the product’s country of origin is considered a designated country under the World Trade Organization Government Procurement Agreement, officials need to determine that the product is eligible under that agreement. DHS updated its training and guidance for the Buy American Act, which includes a job aid outlining at what dollar values solicitation provisions and contract clauses under a trade agreement waiver are applicable. However, it does not address other situations in which contracts may not be eligible under the Trade Agreements Act, such as non-eligible products or products for national defense purposes. For the other agencies in our review, we found that DOD’s updated Buy American Act training and its acquisition supplement both address trade agreement eligibility, but HHS does not yet have Buy American guidance to address this topic. Federal internal control standards state that agencies should communicate quality information internally to achieve their objectives and that they should select the appropriate methods of communication. When written guidance is not available, agencies may miss opportunities to ensure appropriate steps are taken to meet Buy American Act requirements. Guidance for Determining Domestic Non-Availability Exceptions Incomplete Our review of 38 contracts also included 8 contracts for foreign end products pursuant to the domestic non-availability exception. In certain situations, such as when contracts are awarded without full and open competition, this exception requires an approved written determination. The FAR establishes requirements for domestic non-availability determinations, but agencies can delegate the level of review required or specify information to be included in the determination. Three of the agencies we reviewed—DOD, DHS, and VA—provide supplemental guidance on the process for making determinations, including who must make the determination when applying a domestic non-availability exception. However, DHS policy officials told us that when the agency uses the domestic non-availability exception for a sole-source acquisition, the written justification that the FAR requires for non- competed awards should suffice as the documentation to support the non-availability determination as well. The practice of using sole-source justifications to support Buy American determinations is not addressed in DHS guidance. According to DHS policy updated in 2018, determinations of domestic non-availability must be concurred with by the HCA and approved by the Chief Procurement Officer. Federal and DHS acquisition regulations, however, state that some justifications can be approved at a lower level. In the absence of further guidance, this difference in approval levels could result in inconsistent application within the department. In addition, as previously noted HHS does not yet have Buy American Act guidance so the department does not provide information on how to make determinations. According to federal internal control standards, agencies should communicate quality information internally to achieve their objectives and that they should select appropriate communication methods. When written guidance is not available, agencies may miss opportunities to ensure that contracting officers take the steps needed to meet requirements when applying a domestic non-availability exception. Inconsistent Information and Guidance Limits Efforts to Accurately Determine Product Origin Knowing the country of origin of the products the federal government buys is necessary to implement the Buy American Act, but contracting officers do not always have access to accurate information on originating countries. The FAR and the DFARS provide various clauses which, when incorporated into contracts, require vendors to certify that the end products they provide to the government are domestic and, if necessary, declare the foreign countries from which they provide products. Vendors frequently certify this information through the System for Award Management (SAM), the government-wide system used to collect vendors’ annual representations and certifications. Contracting officers may rely on the information vendors provide about their product origins, but they are generally expected to take actions to verify incomplete or conflicting information when they have reason to believe that a vendor will be providing a non-compliant product. We found that SAM certifications and offers did not always include accurate information on end products from foreign countries. In 6 of the 38 contracts that we reviewed—from DHS, HHS, and VA—the vendors certified that they only provided domestic end products although the end products provided were foreign. In all of these cases, the contracting officers knew that the acquisitions included foreign end products. For example, we reviewed two DHS awards for spare aircraft parts from an Italian-based company, one of which was reported in FPDS-NG under the domestic non-availability exception to the Buy American Act, and the other which was incorrectly reported as being manufactured in the United States but has since been corrected. Contracting officials said they knew the parts were made in Italy based on extensive experience contracting with the company and, in part, because they had visited the production location. Contracting officials—including some at HHS and VA—said they use SAM as their primary source to determine whether the vendor is offering domestic end products. Others reported some awareness of the limitations of SAM certifications. At all four agencies, contracting officials emphasized that it is important to ask questions when end product origin information is not readily available—or if there is conflicting information—but agency guidance that we reviewed does not address this need or provide information on how to do so. Only the local training offered by DOD’s DLA addresses other sources of information, which officials said was helpful because it is specific to the industries with which they work. Instead, some officials described how they rely on their experience to know how to verify products’ origins but this can be problematic, particularly with newer staff. For example, in one contract we reviewed VA contracting officials acknowledged that a new contracting specialist at VA did not follow-up when the product origin certification was not provided and assumed all of the items procured were domestic. During the course of our review, the contracting specialist’s supervisor said that she contacted the vendor and learned that some of the items provided were in fact foreign end products. The foreign products were not considered to account for the preponderance of the contract so were not reported in FPDS-NG, but the contracting officer was acting with incomplete information at the time of award. Further, in 4 of the 38 contracts that we reviewed, it is not clear how contracting staff took steps to obtain product origin information in situations where it was not provided in SAM. In these cases—which include contracts for both domestic and foreign end products—the vendors had opted not to certify their product origins in SAM, but instead said that they would provide the information with their individual proposals. However, based on the information in the contract files, the proposals did not include this information. For example: Three of the contracts we reviewed from HHS—all reported as purchasing end items manufactured in the United States—did not certify product origin in SAM. The supervising contracting officer for two of the awards explained that his contracting staff regularly check the vendor’s written representations and certifications provided in the offer, because the SAM certifications are general and do not always apply to the specific equipment they buy. However, the three contract files we reviewed did not include manufacturing or origin information. The vendor for a DHS contract that was reported as manufactured in the United States did not certify this information in SAM. The contracting officer said that he checks SAM for product origin information, but in the documents we reviewed there is no evidence of the information in the contract file. Federal internal control standards state that agencies should communicate the necessary quality information needed to achieve the agency’s objectives, thereby enabling personnel to address risks. Providing guidance regarding the situations in which contracting officers should verify product origin information with vendors may help agencies better meet the requirements of the Buy American Act. Conclusions Although purchases for foreign end products account for less than 5 percent of federal procurement spending in fiscal year 2017, it is important that these purchases be consistent with the domestic- purchasing restrictions in the Buy American Act. This requires that Buy American Act exceptions and trade agreement waivers be used only when applicable, and that agencies report accurate data on the extent to which they are used. However, data reporting errors by contracting staff and FPDS-NG limitations mean that data on the use of exceptions and waivers are not fully captured. The federal agencies all have responsibilities to ensure Buy American Act data are accurate and complete. The lack of good data can hinder congressional oversight of the extent to which foreign end products are procured as authorized by one of the exceptions or waivers of the Buy American Act. Agencies have taken varied approaches for providing information to contracting officers that navigate the complexities and nuances associated with applying the different Buy American Act exceptions or trade agreement waivers. DOD has added such detailed information through its revised training course and policy guidance. Adding these types of targeted information to address challenging areas would help contracting officers at other agencies implement the Buy American Act’s domestic preferences, as well as related exceptions and waivers. Further, to accurately determine how exceptions and waivers apply requires complete product origin information. Although the responsibility to certify the origins of products supplied to the federal government rests with the contractors, contracting officers would benefit from resources that help them identify information that may be inconsistent, to ensure that accurate information is available. Recommendations for Executive Action We are making four recommendations, one each to the Office of Management and Budget, DHS, VA, and HHS. The Director of the Office of Management and Budget should instruct the Office of Federal Procurement Policy: To facilitate additional training to improve the understanding of the contracting workforce regarding the Buy American Act requirements; and To facilitate clarifying revisions to FPDS-NG, where needed, and provide training and guidance for recording Buy American Act information in FPDS-NG to improve the accuracy of the Buy American data. (Recommendation 1) The Secretary of Homeland Security should clarify existing guidance in the Homeland Security Acquisition Manual or update training to help contracting officials: Identify the factors that should be considered in order to determine the applicability of the Trade Agreements Act and waiver of the Buy American Act; Document determinations of the use of Buy American exception for domestic non-availability and ensure the required approvals are obtained, particularly when such determinations are evidenced through justifications for other than full and open competition; and Identify sources of information available for determining product origin and the steps they should take to verify information that is inconsistent. (Recommendation 2) The Secretary of Veterans Affairs should clarify existing guidance, or provide training or other instruction, to help contracting officials: Address the applicability of the Buy American Act requirements and provide instruction on how to implement the requirements, including in any training developed to implement the Veterans First policy; Identify the factors that should be considered in order to determine the applicability of the Trade Agreements Act and waiver of the Buy American Act; and Identify sources of information available for determining products’ origins and the steps they should take to verify information that is inconsistent. (Recommendation 3) The Secretary of Health and Human Services should provide guidance, training, or other instruction to help contracting officials: Identify the factors that should be considered in order to determine the applicability of the Trade Agreements Act and waiver of the Buy American Act; Document determinations of the use of Buy American exceptions for domestic non-availability and ensure the required approvals are obtained; and Identify sources of information available for determining products’ origins and the steps they should take to verify information that is inconsistent. (Recommendation 4) Agency Comments and Our Evaluation We provided a draft of this report to DOD, HHS, DHS, VA, and the Office of Management and Budget for review and comment. DOD reviewed the report, but did not offer comments. HHS, DHS, and VA provided written responses, which are reproduced in Appendices IV, V, and VI of this report, respectively. A senior official within the Office of Federal Procurement Policy (OFPP) at the Office of Management and Budget provided a response via email. In addition, HHS, DHS, and OFPP provided technical comments, which we incorporated into the report where appropriate. In their responses, HHS, DHS, VA agreed, and OFPP generally agreed, with our findings and recommendations. The written response from HHS and DHS included information on the steps each agency plans to take to address the recommendations. Specifically, HHS stated that the agency will evaluate ways to provide additional training and guidance to contracting officials. DHS stated that it will provide guidance on the applicability of the Buy American Act and the Trade Agreements Act in certain situations and the documentation and approvals required when awarding non-competed contracts that require an exception. Additionally, DHS plans to update training regarding actions contracting officers should take when there are discrepancies in product origin information. VA concurred with our three-part recommendation and described some of the actions the agency plans to take in response. However, VA’s comments do not fully address our recommendation. Specifically, we recommended that VA clarify guidance or provide training to identify factors that could help contracting officers determine the applicability of Trade Agreements Act waivers of the Buy American Act. The comments from VA, however, only restate the existing Buy American Act exceptions and make no mention of Trade Agreements Act waivers. Further, we recommended that VA identify sources of information regarding product origin and the steps to be taken to verify inconsistent product origin information. VA’s response only noted that contracting officers are responsible for conducting market research and ensuring that all product origin requirements are met. VA did not outline any additional steps the agency would take to help contracting officers navigate the complexities inherent in this area. Going forward, VA will need to develop a more robust and responsive approach in order to fully implement our recommendation. We are sending copies of this report to the appropriate congressional committees; the Secretaries of the Departments of Defense, Health and Human Services, Homeland Security, and Veterans Affairs; the Director of the Office of Management and Budget; and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-4841 or woodsw@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix VII. Appendix I: Objectives, Scope and Methodology The objectives of this report are to assess the extent to which (1) the federal government procures foreign products through Buy American Act exceptions and waivers; and (2) selected agencies provide training and guidance to implement the Buy American Act requirements. To address both of these objectives, we reviewed relevant laws and policies, such as sections of the Federal Acquisition Regulation (FAR); the Buy American Act as amended; the Trade Agreements Act of 1979 as amended; federal acquisition regulation supplements from audited agencies such as the Department of Defense Federal Acquisition Regulation Supplement (DFARS); the Executive Order “Buy American, Hire American” of 2017; the World Trade Organization’s Agreement on Government Procurement; and memorandums, policy, guidance, and instructions related to the Buy American Act. To assess the federal government procurement of foreign products, including those procured through citing exceptions and waivers of the Buy American Act, we analyzed data from the Federal Procurement Data System-Next Generation (FPDS-NG) for fiscal year 2017, which was the most recent and complete data available at the time of our review. We analyzed procurement data in FPDS-NG across the federal government in fields relevant to the Buy American Act’s domestic preference requirements, including the product service code, country of product origin, and place of manufacture, in addition to fields such as the contract value and dollars obligated. We reviewed the place of manufacture field in particular as it contains information on how the Buy American Act applies to the contract, including whether the preponderance of the obligations is for manufactured end products and, if so, whether they are manufactured in or outside of the United States. When manufactured outside of the United States, this field also captures the reason the purchase was permissible, which we analyzed to assess the dollar obligations associated with the various Buy American exceptions or trade agreement waiver reported, as well as when products were used outside of the United States. We also analyzed data from FPDS-NG to identify the countries where foreign end products were reported to be manufactured and the associated dollars obligated in fiscal year 2017. In addition, we met with officials from the Office of Management and Budget, Office of Federal Procurement Policy to better understand ongoing reviews of the data in FPDS-NG that pertains to the Buy American Act. In our analysis of FPDS-NG data, we took steps to minimize issues that might affect data reliability. Specifically, we analyzed FPDS-NG data to identify potential errors and inconsistencies, such as non-eligible agencies reporting the use of exceptions for DOD qualifying countries, or reporting trade agreement waivers for contracts valued less than minimum thresholds for trade agreements. We made minor adjustments to minimize potential data reporting issues, including aggregating the exceptions reported, and where appropriate, limiting our analysis to one year of data, fiscal year 2017. Based on these steps, we determined that FPDS-NG data were sufficiently reliable to allow us to calculate the approximate extent of obligations for foreign end products and the use of the Buy American Act exceptions and the Trade Agreements Act waiver. However, we are unable to precisely determine the amount spent on foreign end products through the use of exceptions and waivers because of the reporting errors and data system limitations we identified in this report. Using FPDS-ND data, we identified four agencies—the Departments of Defense (DOD), Health and Human Services (HHS), Homeland Security (DHS), and Veterans Affairs (VA)—that had the highest fiscal year 2017 obligations in the product codes for manufactured products, which are potentially subject to the Buy American Act restrictions. In addition, to identify trends and determine if there were variations in reported obligations for foreign end products in the past, we reviewed FPDS-NG data on the Buy American exceptions and trade agreement waivers in fiscal years 2013 through 2017. To assess the extent to which selected agencies are providing training and guidance to implement the requirements of the Buy American Act, we reviewed training course materials and regulations, policies, and other guidance available at the four agencies in our review—DOD, HHS, DHS, and VA—to determine the extent to which they address the Buy American Act requirements. In addition, we reviewed training materials available to government employees through sources such as the Federal Acquisition Institute. We interviewed policy officials from the four agencies to understand how training and guidance had been implemented. We further reviewed relevant inspector general reports from the DOD Inspector General issued between 2015 and 2018, which made several recommendations to improve compliance with the Buy American Act, among other requirements. Within the four agencies, we selected contracting offices that reported obligating fiscal year 2017 dollars for awards with foreign end products and awards with US-manufactured end products. We specifically focused on offices that reported a sufficient amount of foreign end product obligations and a sufficient number of contract awards to allow us to select multiple contracts. We also considered offices with a variety of Buy American exceptions and waiver types reported, in order to select a mix of contracts. The contracting offices selected were as follows: DOD: Defense Logistics Agency, Land and Maritime HHS: National Institutes of Health and the HHS Office of the Assistant Secretary for Preparedness and Response DHS: United States Coast Guard VA: Veterans Health Administration From these offices, we selected a non-generalizable sample of 38 contracts and delivery orders awarded in fiscal year 2017. At each agency, we selected awards to include a mix of end items produced by domestic and foreign manufacturers and, when products were reported as foreign manufactured, a mix of the various exceptions and waivers cited. We also include awards across a range of value for dollars obligated above the micro purchase threshold—ranging from approximately $5,000 to more than $100 million—to ensure we reviewed awards both above and below the various thresholds at which the Trade Agreements Act waiver might apply. Additionally, our sample included awards for similar types of end products across agencies, including aircraft parts at DOD and DHS and medical supplies at HHS and VA, to compare practices in different agencies. We originally selected 40 awards for review—10 from each agency—but removed two awards from our sample. One was an HHS award that we determined was awarded using Other Transaction Authority and was not subject to the Buy American Act. The second excluded contract was from DHS, which was modified after award to reflect that it was an information technology service rather than a product. As a service, it would not be subject to the Buy American Act. We reviewed the contract files for each of the 38 awards in our sample, including documentation such as the contract and task order award, solicitations, vendors’ offers or response to proposals, determination and finding memos, and FPDS-NG output documents. In addition, we reviewed the certifications each vendor provided in the System for Award Management (SAM) at the time of contract award. We interviewed contracting officials responsible for each of the 38 contracts and task orders to understand how they addressed the Buy American Act requirements, including how they determined exception or waiver applicability and product origin. We also reviewed any agency-specific or local training and guidance, tools, or job aids available to assist contracting officers in implementing the Act’s requirements We conducted this performance audit from October 2017 to December 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on audit objectives. Appendix II: Countries with Free Trade Agreements or Reciprocal Defense Procurement Agreements with the United States The United States maintains trade relationships with other countries whose specific negotiated terms results in different levels and types of applicability for waivers and exceptions to the Buy American Act. Figure 4 depicts the range of relationships that the United States maintains with other nations that allow for less restrictive purchasing of foreign end products by the federal government. Appendix III: Federal Government Procurement of Foreign End Products, by Country The federal government purchases foreign end products from various countries. Figure 5 highlights the different amounts of contract obligations for foreign end products from these countries for fiscal year 2017. The highest category, over $500 million, includes 4 countries that account for almost 40 percent of all federal procurement of foreign end products. Countries where the federal government obligated less than $5 million for the procurement of foreign end products are not included. Appendix IV: Comments from the Department of Health and Human Services Appendix V: Comments from the Department of Homeland Security Appendix VI: Comments from the Department of Veterans Affairs Appendix VII: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Candice Wright, Assistant Director; and Jennifer Dougherty, Analyst-in-Charge, managed this review. Skip McClinton; Erin Stockdale; Adam Cowles; Stephanie Gustafson; Julia Kennon; Anne Louise Taylor; and Robin Wilson made key contributions to this report.

Why GAO Did This Study The Buy American Act of 1933, as amended, is the main U.S. law promoting domestic purchasing. The Act permits agencies to buy foreign end products only under certain exceptions, such as when domestic items are not available at a reasonable cost. Further, U.S. trade agreements waive the Buy American restrictions for certain products. GAO was asked to review implementation of the Buy American Act. This report assesses the extent to which (1) the federal government procures foreign products through Buy American Act exceptions and waivers; and (2) selected agencies provide training and guidance to implement the Act. GAO reviewed laws, regulations, and policies related to the Buy American Act and analyzed data for fiscal year 2017 from FPDS-NG. GAO also analyzed a non-generalizable sample of 38 contracts from DOD, HHS, DHS, and VA—the agencies with the most obligations for products in fiscal year 2017. The 38 awards selected include a mix of foreign and domestic products, as well as dollars obligated. Finally, GAO interviewed cognizant contracting and policy officials from the selected agencies. What GAO Found According to data reported in the Federal Procurement Data System-Next Generation (FPDS-NG) in fiscal year 2017, foreign end products accounted for less than 5 percent—about $7.8 billion—of federal obligations for products potentially subject to the Buy American Act. Federal agencies procured foreign products using exceptions to Buy American Act requirements, as well as through waivers or when the Buy American Act did not apply, as shown in the figure. The amount of foreign end products purchased could be greater than reported in FPDS-NG, however, due to reporting errors and system limitations. GAO found that 6 of the 38 contracts reviewed from the Departments of Defense (DOD), Health and Human Services (HHS), Homeland Security (DHS), and Veterans Affairs (VA) inaccurately recorded waiver or exception information. FPDS-NG system limitations compound these errors because it does not fully capture Buy American Act data. Among other things, the database does not always enable agencies to report the use of exceptions or waivers on contracts for both foreign and domestic products, reducing data accuracy. The Office of Management and Budget (OMB) is considering strategies to improve Buy American Act data. The four agencies GAO reviewed varied in their approaches to Buy American Act training and guidance. DOD reports that it will have trained more than 18,000 personnel by the end of 2018. DHS reports training almost 1,400 people—approximately 94 percent of its contracting staff—as of April 2018. Some VA courses mention the Act, but none is focused specifically on implementing its requirements. HHS does not have agency-level training or guidance on the Act. GAO found that contracting officers for the contracts it reviewed face challenges implementing Buy American Act requirements. Having specific and targeted Buy American Act guidance and training can better ensure that agencies meet the Act's requirements. What GAO Recommends GAO is recommending that OMB take steps to improve Buy American Act data and that HHS, DHS, and VA improve agency guidance and training on implementing the Act. All of the agencies either concurred or generally concurred with GAO's recommendations.

gao_GAO-18-51

gao_GAO-18-51_0

Background The federal government plans to invest about $96 billion in fiscal year 2018 for IT that is critical to the health, economy, and security of the nation. However, prior IT expenditures have often resulted in significant cost overruns, schedule delays, and questionable mission-related achievements. For example, The Department of Health and Human Services’ website, Healthcare.gov, and its supporting systems, which were to facilitate the establishment of a federal health insurance marketplace by January 2014, encountered significant cost increases, schedule slips, and delayed functionality. In a series of reports, we identified numerous planning, oversight, security, and system development challenges faced by this program. For almost two decades, the Department of Veterans Affairs (VA) has undertaken numerous initiatives with the Department of Defense (DOD) that were intended to advance the ability of the two departments to share electronic health records. In our report of the departments’ efforts in 2015, we reported that the departments had not identified outcome-oriented goals and metrics to clearly define what they aimed to achieve from their interoperability efforts, resulting in numerous failures. During most of the last 20 years, VA has also been planning to modernize its system separately from DOD. Recently, the Secretary of VA announced that the department plans to use the same electronic health record system that DOD is in the process of acquiring. However, the significant challenges that have confronted VA in its efforts contributed to our designation of VA health care as a high risk area. The Department of Homeland Security’s U.S. Citizenship and Immigration Services’ Transformation Program, which was initiated to address processing inefficiencies and transform the agency’s current paper-based system into an electronic account-based system, has faced continual management and development challenges, limiting its progress and ability to achieve its goals of enhanced national security and system integrity, better customer service, and operational efficiency. The U.S. Citizenship and Immigration Services estimates that the program’s cost increased by approximately $1 billion and its schedule was delayed by over 4 years from its initial approved baseline. The Office of Personnel Management’s Retirement Systems Modernization program, which was intended to improve the efficiency and effectiveness of its retirement claims processing, was canceled in February 2011 after the agency had spent approximately $231 million on a third attempt to automate the processing of federal employee retirement claims. As previously stated, due to the challenges associated with acquiring IT across the federal government, in 2015, we added improving the management of IT acquisitions and operations to our list of high-risk areas. We recently issued an update to our high-risk report and determined that, while progress has been made in addressing the high- risk area of IT acquisitions and operations, significant work remains to be completed. For example, as of May 2017, OMB and federal agencies had implemented 380 (or about 47 percent) of the 803 recommendations that we had made from fiscal years 2010 through 2015 related to IT acquisitions and operations. OMB’s Mission and Oversight of Federal IT Programs By law, OMB is to oversee federal agencies’ management of information and information technology. Within OMB, primary responsibility for oversight of federal IT has been given to the Administrator of the Office of Electronic Government and Information Technology, who is also called the Federal Chief Information Officer (Federal CIO). According to OMB, this oversight responsibility covers about 800 major and nearly 5,700 non-major IT investments across the federal government. As a part of the oversight, the E-Gov office develops policy and reviews federal agencies’ IT strategic plans. In addition, OMB establishes processes to analyze, track, and evaluate the risks and results of IT investments made by executive agencies, and issues guidance on processes for selecting and overseeing agency privacy and security protections for information and information systems. OMB has also implemented a series of initiatives to improve the oversight of underperforming investments and more effectively manage IT. These initiatives include the following: Federal IT Dashboard. In June 2009, to further improve the transparency into and oversight of federal agencies’ IT investments, OMB deployed the Federal IT Dashboard, a public website with information on the performance of these investments. OMB provided guidance to the agencies on the information they should maintain and update on the dashboard. Currently, the dashboard displays information on the cost, schedule, and performance of close to 800 major IT investments at 26 federal agencies. In addition, agencies are to submit ratings from their CIOs to the dashboard, which, according to OMB’s instructions, should reflect the level of risk facing an investment relative to that investment’s ability to accomplish its goals. The public display of these data is intended to allow OMB, other oversight bodies, and the general public to hold agencies accountable for mission-related outcomes. Over the past 7 years, we have issued a series of reports that have noted both significant steps OMB has taken to enhance the oversight, transparency, and accountability of federal IT investments by creating the dashboard, as well as issues with the accuracy and reliability of the data it contains. TechStat reviews. In January 2010, the Federal CIO began leading TechStat reviews—face-to-face meetings to discuss whether to terminate or turn around IT investments that are in danger of failing or are not producing results. These meetings involved OMB and agency leadership and were intended to increase accountability and improve performance. OMB reported that federal agencies achieved over $3 billion in cost savings or avoidances as a result of these reviews in 2010. Subsequently, it empowered agency CIOs to begin holding their own TechStat reviews by June 2012. OMB’s 2015 guidance specified that the TechStat reviews were to be held with agency leadership, not led by OMB as in the past, and that agencies were only required to notify OMB of the meetings’ occurrence and report the results. In November 2015, we testified that OMB had conducted only one TechStat review between March 2013 and October 2015, and had not listed any related savings in its quarterly reporting to Congress since June 2012. In April 2017, OMB reported that it had not led a TechStat review since 2015. A report issued by the Federal CIO Council in January 2017, entitled The State of Federal Information Technology, noted that shifting TechStat reviews from OMB to agencies had diminished the executive scrutiny and impact of the initiative. PortfolioStat sessions. To better manage existing IT systems, in 2012, OMB launched the PortfolioStat initiative, which required agencies to conduct an annual, agency-wide portfolio review to, among other things, reduce commodity IT spending and demonstrate how their investments aligned with the agency’s mission and business functions. These reviews were to be held between the Federal CIO and agency leadership. In 2014 and 2015, OMB’s PortfolioStat guidance also called for it and agencies to identify high impact IT programs that merited additional support and oversight by OMB and/or agency leadership, and for these programs to be discussed during a PortfolioStat session. The 2015 guidance also changed the frequency of the PortfolioStat sessions from annually to quarterly, and the level of participation to no longer require attendance by the Federal CIO or the agency’s Deputy Secretary. The Federal Information Technology Acquisition Reform provisions (commonly referred to as FITARA) enacted as a part of the Carl Levin and Howard P. ‘Buck’ McKeon National Defense Authorization Act for Fiscal Year 2015 aimed to improve federal IT acquisition and operations and recognized the importance of these OMB initiatives by incorporating certain requirements into the law. For example, among other things, the act requires OMB to publicly display investment performance information and review federal agencies’ IT investment portfolios. Further, as previously mentioned, the December 2014 explanatory statement for the Consolidated Appropriations Act, 2015, stated that OMB was to identify the top 10 high priority IT programs under development in the federal government and report on their status quarterly. Additionally, in December 2015, in the explanatory statement for the Consolidated Appropriations Act, 2016, Congress stated that USDS, an OMB component, was to provide a quarterly status report on its current projects, including the top 10 high priority programs. USDS’s Mission and Organization The current mission of USDS is to deliver better government services to the American people through technology and design. USDS is focused on, among other things, improving the nation’s most important digital services used by the public, and modernizing procurement processes and practices for the digital era. To execute its mission, USDS recruits private sector experts, such as IT engineers and designers and leading civil servants, and deploys small teams to federal agencies. It selects which projects it will apply resources to and generally initiates the effort with the federal agency that owns the IT projects. In August 2016, we reported that USDS had developed procedures and criteria for selecting and prioritizing projects to work on. Current Administration’s Efforts to Improve Federal IT The current administration has initiated additional efforts aimed at improving federal IT, including digital services. Specifically, in March 2017, the administration established the Office of American Innovation, which has a mission to, among other things, make recommendations to the President on policies and plans aimed at improving federal government operations and services and modernizing federal IT. In doing so, the office is to consult with both OMB and the Office of Science and Technology Policy. Further, in May 2017, the administration established the American Technology Council, which has a goal of helping to transform and modernize federal agency IT and how the federal government uses and delivers digital services. The Federal CIO and the USDS Administrator are members of this council. OMB Identified High Priority IT Programs, but Did Not Ensure Federal CIO Oversight or Continued and Timely Reporting OMB reported that it undertook a structured approach to identifying the top 10 high priority IT programs it reported to Congress in June 2015 and June 2016. Specifically, staff in the E-Gov office, including the Unit Chief for the Agency Oversight and Implementation team, stated that they chose the top 10 programs from a list of high impact programs that OMB separately maintains. They added that their approach was not guided by any documented procedures or scoring techniques to distinguish the programs. Analysts in the E-Gov office told us that, to identify the high impact programs from which the top 10 high priority programs were selected, they used program information from IT portfolio summaries, monthly IT dashboard updates, and quarterly integrated data collection submissions that OMB receives from agencies. They also considered several additional factors, such as risk exposure, public impact, public use, criticality to agency mission, size, and cost. In addition, they considered input from USDS leadership and OMB budget examiners, as well as our reports, inspectors general reports, and CIO risk ratings. Further, the analysts stated that OMB sought input from officials of the 24 Chief Financial Officers Act agencies to gain a better understanding of the importance of each program. In the end, based on all of the information considered, E-Gov staff made a judgment call regarding which of the agencies’ programs to identify as being high impact. According to these staff, on average, two high impact programs are identified for each of the 24 agencies, with at least one program being identified for each agency, and as many as four programs being identified for some larger agencies. Further, according to the staff, the Federal CIO approved the process for selecting the high impact programs. In addition, in determining the top 10 high priority programs, the E-Gov staff stated that they identified on the high impact list those programs that they believed were representative of the most important IT programs across federal agencies. They also considered other factors, such as whether a program had generated legislative interest or had performance issues. They added that the Federal CIO then approved the list of top 10 programs that they had selected. OMB subsequently issued two reports to Congress—in June 2015 and June 2016—that identified the top 10 high priority programs. Along with the status of each program, the reports identified the IT investments that were a part of each program as well as total IT spending, average CIO risk rating, major milestones, and the level of involvement USDS has with the program. Table 1 lists the programs that OMB identified in the two reports. As shown in the table, OMB made two changes to the top 10 list in 2016. Specifically, it replaced the Social Security Administration’s Service Modernization program with the agency’s Disability Case Processing System due to a change in focus within the agency. In addition, it replaced VA’s Medical Appointment Scheduling System with VA’s Medical 21st Century Development Core program, which is a larger effort that encompasses the scheduling system. OMB Oversight Efforts Did Not Ensure Federal CIO Involvement E-Gov staff reported that the high priority programs were already receiving greater oversight than what was provided to the other major programs due to their high impact designation. Specifically, the staff stated that OMB provides additional oversight to high impact IT programs by regularly communicating about them through quarterly, monthly, weekly, or daily meetings with agency CIOs and program staff, depending on the risk and profile of the program. In contrast, communication is much less frequent for other major IT systems. The staff added they also discuss high impact programs with agencies during quarterly PortfolioStat meetings. Further, they stated that they can request that agencies perform a TechStat review, if needed, for a troubled high impact program. Most agencies that owned the high priority programs identified in the June 2016 report confirmed that OMB had already provided increased oversight for the high priority programs when they were originally designated as high impact programs. Specifically, officials from six of the eight agencies stated that the programs were discussed during quarterly PortfolioStat sessions, and officials from five agencies stated that OMB had provided action items for them to address with regard to their programs. Further, officials from six agencies stated that OMB’s oversight included periodic meetings (e.g., daily, weekly, or bi-weekly). Its oversight also included participation in a TechStat review of a high impact program performed by the Social Security Administration. Nevertheless, while additional OMB oversight of the high impact programs (and, accordingly, the identified high priority programs) is a positive step, the Federal CIO was not directly involved in this oversight. According to the E-Gov staff, the Federal CIO does not typically get involved with overseeing individual IT programs due to the large number of programs. However, the results of past CIO-led TechStat reviews suggest that the Federal CIO’s involvement in overseeing such programs does have significantly positive results. Specifically, as previously mentioned, CIO-led TechStat reviews of IT investments performed in 2010 resulted in $3 billion in savings and cost avoidance. Further, during a September 2016 Comptroller General forum to explore challenges and opportunities for improving federal IT acquisitions and operations, current and former CIOs and other participants pointed to the importance of OMB’s oversight and guidance, and specifically, to the role of the Federal CIO in helping to ensure effective IT governance. The participants cited specific OMB initiatives undertaken by the Federal CIO, including the TechStat reviews that had resulted in greater accountability and positive results. Thus, without the involvement of the Federal CIO more directly and regularly in the oversight of high impact and high priority programs, including leading TechStat reviews for the programs, OMB is likely to miss significant opportunities to improve accountability for and achieve positive results from the federal government’s IT investments. OMB Has Stopped Reporting on High Priority IT Programs OMB issued two reports on the high priority programs—one in 2015 and another in 2016. While these reports provided the requested information they were not issued quarterly. According to the E-Gov staff, OMB was not able to report quarterly on the programs because of other competing reporting requirements, limited resources available to draft the report, and the amount of time it takes to get its reports fully reviewed. Moreover, the staff stated that they stopped issuing the high priority reports in 2016 because they believe the explanatory statement to the Consolidated Appropriations Act, 2016, no longer directed them to continue reporting on the programs. Specifically, the explanatory statement directed USDS to provide a quarterly report to the Committees on Appropriations of the House and Senate describing the status of current USDS teams and projects, including the top 10 high priority programs, a list of USDS accomplishments, and agency project proposals. Both E-Gov staff and USDS staff said they determined this to mean that OMB should report on the USDS projects considered to be high priority given USDS’s responsibilities. In addition, USDS staff stated that they did not receive any feedback from congressional stakeholders indicating otherwise. However, continued identification and reporting on the top ten high priority programs, and not just USDS projects, would further enhance congressional oversight by providing congressional stakeholders with information on high priority programs that is not readily available. Such information could also be useful for current administration IT governance entities such as the Office of American Innovation and the American Technology Council, to assist them with prioritizing their efforts to modernize federal agency IT. USDS Has a Defined Project Selection Process, but its Reporting Did Not Address High Priority Programs As Requested and Was Not Timely USDS has developed a process for identifying and prioritizing the IT projects to which it provides support. Moreover, as we have previously reported, its project selection process is consistent with best practices, which state that organizations should establish and implement procedures for prioritizing projects that include identifying selection criteria to help consistently select projects based on their contributions to the strategic goals of the organization. In explaining the process used to identify projects, USDS staff stated that they obtained input from various sources, including the Federal IT Dashboard, leadership of the relevant federal agency, E-Gov analysts, and GAO reports. In addition, the staff said they considered the high impact programs identified by the E-Gov office; they also coordinated with E-Gov analysts through monthly meetings to incorporate issues of significance to E-Gov into their selection process. To further facilitate the selection process, the USDS staff established three questions as criteria for prioritizing agencies’ IT projects, in the following order of importance: (1) What will do the greatest good for the greatest number of people in the greatest need? (2) How effective and cost efficient will the USDS investment be? (3) What potential exists to scale or reuse a technological solution across the government? The staff said they used the criteria to create a list of all potential projects, including their descriptions and information on resource needs; they updated the list when they identified additional projects that met the criteria. This list was subsequently used by USDS leadership to make decisions about which projects to pursue. According to the staff, an important consideration when selecting a project was whether there was executive sponsorship from the agency. They added that executive sponsorship ensures that USDS has the help it needs to make changes to the projects, and it affects the efficiency and cost-effectiveness of USDS’s investment. Projects that were not selected went into a backlog that is to be used to select a project when a team becomes available to work on a new engagement. USDS has issued two reports to Congress on the status of its projects— one in December 2016 and the other in July 2017. The report issued in December 2016 summarized the status of 11 projects that USDS is engaged in at federal agencies, and 3 broader initiatives that are intended to improve the performance and cost-effectiveness of government digital services. According to USDS staff, the 11 projects had broad impact and had made the most significant progress. For example, these included the VA’s Vets.gov project, which is intended to assist the department in developing a new digital application for healthcare, and DOD’s Defense Travel System, a system that facilitates travel for all DOD employees. Further, USDS reported on the status of their efforts for these three initiatives: modernizing procurement processes, development of federal shared services, and hiring top technical talent. The July 2017 report included summaries for 10 projects, including 5 new projects. Specifically, the new projects were the General Service Administration’s Login.gov project, the Small Business Administration’s effort to modernize small business certification for government contractors, DOD’s Advisor Network system, DOD’s Defense Personal Property System, and transforming federal IT procurement through digital acquisition training. In addition, 6 projects that were discussed in the December 2016 report were not included in the July 2017 report. Table 2 provides a complete list of the projects identified in the two reports. Our analysis determined that four of the projects identified in USDS’s reports were among the high priority programs that OMB’s E-Gov office had identified in its June 2015 and June 2016 reports to Congress. These projects were Healthcare.Gov, Disability Claim Processing, modernizing the immigration system at the Department of Homeland Security, and improving the Visa program at the Department of State. However, USDS’s report does not specify this or provide an update on the status of the other high priority programs. USDS staff stated that they did not address all of the top 10 high priority programs because, as stated earlier in this report, they interpreted Congress’s 2016 request as being focused on USDS’s priority projects and not on the programs previously identified by E-Gov. As mentioned earlier, however, continuing to identify and report on the top 10 high priority IT programs while also reporting on USDS’s projects would further enhance congressional oversight and current administration IT governance entities’ efforts by providing stakeholders with information on high priority programs and USDS projects that is not readily available. Further, although USDS was directed to report quarterly, it did not do so. Instead, it issued a report in December 2016, nearly a year after Congress’s direction, and in July 2017, nearly 7 months after its first report. In discussing this matter, USDS staff said they were not able to report quarterly due to the time and effort needed to prepare and review a report. As a result, USDS’s reporting did not provide congressional stakeholders with the timely information needed to support their oversight responsibilities. Conclusions While OMB’s 2015 and 2016 reports to Congress on the top 10 high priority programs included the status of the programs, their total IT spending, and other information to assist Congress in monitoring the progress of critical programs, OMB did not issue the reports quarterly as directed in the explanatory statement. In addition, OMB does not plan on continuing to issue the top 10 high priority reports because it believes that, in 2016, Congress directed the agency to instead focus on providing a status of USDS’s most important projects. Further, OMB’s December 2016 and July 2017 reports on USDS’s projects did not address the top 10 high priority programs across the government. However, continued reporting on the top 10 high priority programs would further enhance congressional oversight by providing congressional stakeholders with information that is not readily available on those programs in the greatest need of attention. Reporting on the top 10 high priority programs could also be useful for IT governance entities such as the Office of American Innovation and the American Technology Council, to assist them with prioritizing their efforts to modernize federal agency IT. Moreover, OMB did not issue the high priority programs and USDS reports on a quarterly basis as requested. Without OMB’s quarterly reporting on the progress of both the top 10 high priority programs and the status of the USDS projects, congressional stakeholders and others may lack the timely information they need to support their oversight and other responsibilities. Finally, while additional OMB oversight of the high impact programs (and, accordingly, the identified high priority programs) is a positive step, the Federal CIO was not directly involved in the oversight of these programs. Based on the positive impact of direct Federal CIO involvement in leading investment reviews in the past, such involvement could significantly improve program outcomes. Recommendations for Executive Action We are making the following three recommendations to OMB: The Director of OMB should continue to identify and report to Congress on the status of the top 10 high priority IT programs and the extent to which USDS is involved in the programs, as was done in June 2015 and June 2016. In doing so, the Director should ensure that these reports are issued quarterly. (Recommendation 1) The Director of OMB should ensure that the Federal CIO is directly involved in the oversight of high priority programs. (Recommendation 2) The Director of OMB should continue to report on the status of USDS projects. In doing so, the Director should ensure that the reports are issued quarterly. (Recommendation 3) Agency Comments and Our Evaluation We received comments on our draft report via e-mail from the OMB liaison to GAO. In the comments, OMB did not specifically state whether it agreed or disagreed with our recommendations. Rather, OMB stated that it has concerns with GAO’s alternative interpretations of law and that GAO’s findings and conclusions are rooted in an incorrect legal interpretation of OMB’s annual appropriation. Specifically, it stated that GAO considers reporting requirements specific to an annual appropriation to apply for all future annual appropriations. However, OMB’s characterization is incorrect, as we did not assert this legal conclusion. As stated in our report, in the explanatory statement accompanying the Consolidated and Further Continuing Appropriations Act, 2015, Congress directed OMB to identify the 10 highest priority IT investment programs (referred to in our report as the top 10 high priority programs) that are under development across federal agencies and report on their status each quarter. Subsequently, the explanatory statement accompanying the Consolidated Appropriations Act, 2016, directed USDS to provide a quarterly status report on, among other things, current USDS projects, including the top 10 high priority programs. Our report does not conclude that the language of either explanatory statement establishes a legally binding requirement, whether applicable only to the subject fiscal year or beyond. GAO’s conclusions are based on the view that continued identification and reporting on the top 10 high priority programs, and not just USDS projects, would help to enhance congressional oversight. Identifying and reporting on the top 10 high priority programs is important because such information is not readily available. We have revised relevant statements in the report to clarify our message in this regard. Further, while GAO did not assert a legal conclusion, we have, nonetheless, removed all references to our “interpretation” of the explanatory statement so as to avoid the inference that we are making legal conclusions. OMB also said our conclusion that it has stopped reporting altogether is incorrect. However, our report does not state that OMB has stopped reporting altogether. Rather, our report states that OMB stopped issuing reports on the top 10 high priority IT programs due to its interpretation of the 2016 explanatory statement and, instead, switched to reporting on the status of USDS’s projects. In addition, OMB stated that, while it agreed that the reports have not been submitted on a quarterly basis, it provided Congress with the requested information for the four quarters of the relevant fiscal years. In addition, OMB stated that USDS is currently discussing with congressional stakeholders whether providing quarterly briefings, instead of reports, would address the quarterly reporting requirement. As noted in our report, timely (i.e., quarterly) information would enhance congressional and other stakeholders’ oversight responsibilities. Therefore, we maintain that our recommendation for reporting quarterly is appropriate. Finally, OMB stated that it disagreed with our conclusion that the lack of personal Federal CIO involvement in high priority IT programs had resulted in inadequate oversight. While we state that the Federal CIO was not directly involved in overseeing the high priority programs, we did not conclude that this resulted in inadequate oversight. Rather, we stated that the results of past CIO-led TechStat reviews suggest that more direct and regular involvement of the Federal CIO would improve accountability and achieve positive results for the federal government’s investments. Thus, we continue to believe that our recommendation to ensure that the Federal CIO is directly involved in overseeing high priority programs is appropriate. OMB also provided technical comments, which we have incorporated into the report as appropriate. We are sending copies of this report to interested congressional committees, the Director of the Office of Management and Budget, and other interested parties. In addition, the report will be available at no charge on GAO’s website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-9286 or pownerd@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix I. Appendix I: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Sabine Paul (Assistant Director), Scott Borre (Analyst in Charge), Nancy Glover, Lori Martinez, Bradley Roach, and Marshall Williams, Jr. made key contributions to the report.

Why GAO Did This Study The federal government plans to spend almost $96 billion on IT investments in fiscal year 2018; however, as GAO has previously found, too often these investments have cost overruns and schedule delays. To enhance oversight of IT programs, for 2015, Congress directed OMB to identify the 10 highest priority IT programs that are under development across federal agencies and report on their status each quarter. Further, for 2016, Congress directed USDS to provide a quarterly report of current USDS projects, including the top 10 high priority programs. GAO was asked to review OMB's oversight of high priority programs. This review evaluated (1) OMB's process for identifying, overseeing, and reporting on the high priority IT investment programs and (2) USDS's process for identifying and prioritizing its projects, including its consideration of the high priority programs, and its reporting on the projects. GAO analyzed OMB memorandums and reports to Congress and interviewed OMB staff, including from USDS. In addition, GAO compared USDS's processes with IT management best practices. What GAO Found The Office of Management and Budget's (OMB) Office of E-Government and Information Technology (E-Gov) reported that it undertook a structured approach to identify the top 10 high priority IT programs in reports to Congress in June 2015 and June 2016. Specifically, OMB staff stated that they chose the top 10 programs from a longer list of agency programs requiring additional oversight (referred to as high impact programs). E-Gov staff reported that the high priority programs were already receiving greater oversight than what was provided to the other major programs due to their high impact designation. This additional oversight included frequent meetings with agency Chief Information Officer (CIO) leadership and quarterly meetings with OMB staff. However, the Federal CIO was not directly involved in this oversight. According to E-Gov staff, the Federal CIO does not typically get involved in individual programs due to the large number of programs. However, past experience has shown that Federal CIO involvement has had a significant impact. For example, Federal CIO-led reviews of troubled projects, known as TechStat reviews, resulted in $3 billion in savings in 2010. Until OMB ensures that the Federal CIO is more directly involved in the oversight of these high priority programs, it may be missing a key opportunity to improve accountability and achieve positive results. OMB's 2015 and 2016 reports to Congress on the top 10 high priority programs identified the status of the programs and major milestones. However, the reports were not issued on a quarterly basis, as directed. E-Gov staff stated that they were unable to do so because of other competing reporting requirements and the limited resources available to draft and fully review the report on a quarterly basis. In addition, OMB stopped issuing the reports on the top 10 high priority IT programs after June 2016. OMB stated that Congress' 2016 direction to the U.S. Digital Service (USDS)—an OMB component—to provide a quarterly report that described the status of USDS teams and projects, including the top 10 high priority programs, meant that OMB should only report on USDS projects considered to be high priority. However, continued identification and reporting on the top 10 high priority programs, and not just USDS projects, would further enhance congressional oversight by providing congressional stakeholders information that is not readily available on those programs in the greatest need of attention. USDS issued reports to Congress on the status of its key projects in December 2016 and July 2017; however, the reports did not address the top 10 high priority programs as directed by Congress, according to OMB staff, because of OMB's interpretation of Congress's direction. In addition, the reports were not issued quarterly, as directed. USDS staff attributed this to the time and effort needed to review and prepare the report. However, continuing to identify and report on the top 10 high priority programs while also reporting on USDS's projects would help to enhance congressional oversight and current administration IT governance entities' efforts by providing stakeholders with information that is not readily available. What GAO Recommends GAO is making three recommendations to OMB for enhancing the oversight of high priority programs and continuing to report on both these programs and USDS projects. OMB neither agreed nor disagreed with the recommendations but disagreed with several of GAO's conclusions, which GAO continues to believe are valid as discussed in the report.

gao_GAO-18-74

gao_GAO-18-74_0

Background Collectively, the ongoing GPS acquisition effort aims to (1) modernize and sustain the existing GPS capability and (2) enhance the current GPS system by adding an anti-jam, anti-spoof cybersecure M-code capability. Figure 1 below shows how GPS satellites, ground control, and user equipment—in the form of receiver cards embedded in systems—function together as an operational system. Modernizing and sustaining the current GPS broadcast capability requires launching new satellites to replace the existing satellites that are near the end of their intended operational life as well as developing a ground control system that can launch and control both existing and new satellites. Sustaining the current GPS broadcast capability is necessary to ensure the quality and availability of the existing broadcast signals for civilian and military GPS receivers. The ongoing modernization of GPS began with three programs: (1) GPS III satellites; (2) OCX to control the satellites; and (3) MGUE increment 1 (which develops initial receiver test cards for military ships, ground vehicles, or aircraft). Table 1 describes these programs. Delays to OCX of more than 5 years led the Air Force to create two additional programs in 2016 and 2017 to modify the current GPS ground system to control GPS III satellites and provide a limited M-code broadcast. As a result, there are currently five total GPS modernization programs. Table 2 provides a description of the two new programs. All of the original GPS modernization programs—GPS III, OCX, and MGUE—have experienced significant schedule growth during development. Table 3 outlines several schedule challenges in the modernized GPS programs. We found in 2015 that unrealistic cost and schedule estimates of the new ground control system and receiver card development delays could pose significant risks to sustaining the GPS constellation and delivering M- code. At that time, we also made five recommendations so that DOD would have the information necessary to make decisions on how best to improve GPS modernization and to mitigate risks to sustaining the GPS constellation. We made four OCX-specific recommendations targeted to identify underlying problems, establish a high confidence schedule and cost estimate, and improve management and oversight. For MGUE, we recommended the Air Force add a critical design review before committing resources to allow the military services to fully assess the maturity of the MGUE design before committing test and procurement resources. DOD concurred with the four recommendations on OCX and partially concurred on the MGUE recommendation. Since 2015, our annual assessment of DOD weapon systems has shown that some of the original GPS programs have continued to face cost or schedule challenges, increasing the collective cost to modernize GPS by billions of dollars. Appendix III outlines the cost increases that have resulted. Key GPS Modernization Points According to our analysis, over the next decade or more, DOD plans to achieve three key GPS modernization points: (1) constellation sustainment, (2) M-code broadcast, and (3) M-code receivers fielded. Figure 2 shows the current sequencing of the three points and the intervals when they are planned to be achieved, if known. Throughout this report, we will use figures based on this one to highlight the separately-managed programs DOD plans to synchronize to achieve each of the three identified modernization points. Some GPS capabilities require the delivery of more than one program, which must compete for limited resources, such as testing simulators. The Air Force coordinates the interdependent activities of the different programs and contractors in order to achieve each modernization point. GPS Satellite Constellation The satellites in the GPS constellation broadcast encrypted military signals and unencrypted civilian signals and move in six orbital planes approximately 12,500 miles above the earth. What is a Global Positioning System (GPS) satellite orbital plane and how many are there? The GPS constellation availability performance standards commit the U.S. government to at least a 95 percent probability of maintaining a constellation of 24 operational GPS satellites to sustain the positioning services provided to both civilian and military GPS users. Therefore, while the minimum constellation consists of satellites occupying 24 orbital slots—4 slots in each of the six orbital planes—the constellation actually has 31 total satellites, generally with more than four in each plane to meet the 95 percent probability standard. These additional satellites are needed to provide uninterrupted availability in case a satellite fails. The constellation includes three generations of satellites with varying capabilities and design lives. An orbital plane is an imaginary flat disc containing an Earth satellite’s orbit. One orbital plane, as is shown above, represents the trajectory a GPS satellite follows as it circles the Earth in space. The GPS constellation has six orbital planes. Each contains at least 4 satellites that allow the constellation to meet the minimum requirement of 24 satellites. We found in 2010 and 2015 that GPS satellites have proven more reliable than expected, greatly exceeding their initially predicted life expectancies. Nevertheless, the Air Force must regularly replace satellites to meet the availability standard, since operational satellites have a finite lifespan. Excluding random failures, the operational life of a GPS satellite tends to be limited by the amount of power that its solar arrays can produce. This power level declines over time as the solar arrays degrade in the space environment until eventually they cannot produce enough power to maintain all of the satellite’s subsystems. Consequently, the Air Force monitors the performance of operational satellites in order to calculate when new satellites need to be ready to join the constellation. The 10 GPS III satellites currently under contract and in production with Lockheed Martin will provide a range of performance enhancements over prior GPS satellite generations. The GPS III satellites were designed to provide a longer life than previous generations, greater signal accuracy, and improved signal integrity—meaning that the user has greater assurance that the broadcast signal is correct. When they are eventually controlled through the OCX ground control system, the satellites will also offer a stronger M-code signal strength than prior GPS satellite generations. They will also include an additional civilian signal known as L1C, which will permit interoperability with European, Japanese, and other global navigation satellite systems for civilian users. Figure 3 describes the evolution of GPS satellite generations, including capabilities and life-span estimates. Ground Control Segment The current GPS ground control segment, OCS, primarily consists of software deployed at a master control station at Schriever Air Force Base, Colorado, and at an alternate master control station at Vandenberg Air Force Base, California. The ground control software is supported by 6 Air Force and 11 National Geospatial-Intelligence Agency monitoring stations located around the globe along with four ground antennas that communicate with the moving satellites. Information from the monitoring stations is processed at the master control station to determine satellite clock and orbit status. As each of the three ground control segment programs—COps, MCEU, and OCX—is completed or partially completed, they will each introduce new capabilities, eventually culminating in the delivery of the full M-code broadcast planned for January 2022. Receiver Cards GPS receiver cards determine a user’s position and time by calculating the distance from four or more satellites using the navigation signals on the satellites to determine the card’s location. All warfighters currently acquire, train with, and use GPS receivers. Until MGUE receiver cards are developed and available for production, all DOD weapon systems that use GPS will continue to use the current GPS Selective Availability/Anti- Spoofing Module (SAASM) receiver card or an older version. The Ike Skelton National Defense Authorization Act for Fiscal Year 2011 generally prohibits DOD from obligating or expending funds to procure GPS user equipment after fiscal year 2017 unless that equipment is capable of receiving M-code. Under certain circumstances this requirement may be waived or certain exceptions may apply. The increment 1 receiver cards range in size from approximately 2 inches by 3 inches for the ground card up to 6 inches by 6 inches for the aviation/maritime card. Figure 4 below shows an illustration of a MGUE receiver card. DOD has previously transitioned its weapon systems gradually from one generation of GPS receivers to the next. For example, some weapon systems have either upgraded or are still in the process of upgrading to the current SAASM receivers that were introduced in 2003, while others are still equipped with older cards. DOD anticipates that the length of time necessary to transition to MGUE will require users to operate with a mix of receiver cards. Hundreds of different types of weapon systems require GPS receiver cards, including ships, aircraft, ground vehicles, missiles, munitions, and hand-held devices, among others, across all military services. The Air Force funds the MGUE program, providing funding to the military services so they can acquire, integrate, and operationally test the receiver cards on four service-specific lead platforms. These platforms are intended to test the card in the military services’ ground, aviation, and maritime environments: (1) Army—Stryker ground combat vehicle; (2) Air Force—B-2 Spirit bomber; (3) Marine Corps—Joint Light Tactical Vehicle (JLTV); and (4) Navy—DDG-51 Arleigh Burke destroyer. Figure 5 depicts selected weapon systems that will need to install M-code capable receiver cards. Acquisition Risks Persist on GPS III Satellites but Do Not Threaten Sustainment of the Constellation in the Short Term The Air Force has made some progress toward ensuring continued constellation sustainment since our September 2015 report and should be able to sustain the current service because of the length of life of the current satellites. The current GPS constellation is now projected to meet its availability performance standard (in the absence of operational GPS III satellites) into June 2021—an increase of nearly 2 years over previous projections. This increase will give the Air Force more schedule buffer in the event of any additional delays to the GPS III satellite program. However, the Air Force still faces technical risks and schedule pressures in both the short and long term. In the short term, schedule compression with the first GPS III satellite is placing the satellite’s launch and operation at risk of further delays. In the long term, most of the satellites under contract will have been launched before operational testing is completed, limiting Air Force corrective options if issues are discovered. Figure 6 shows the schedule for programs that need to be delivered to modernize and sustain the GPS satellite constellation. Progress: Programs Advancing to Support Constellation Sustainment Requirements The Air Force has made progress since our last report in September 2015 on the three programs (GPS III, OCX, and COps) needed to support GPS constellation sustainment, readying both ground control and the satellite for the first GPS III satellite’s launch, testing, and eventual operation. Raytheon delivered OCX block 0, the launch and checkout system for GPS III satellites, in September 2017. The Air Force took possession of OCX block 0 in October 2017 and will finally accept it at a later date after OCX block 1 is delivered. Lockheed Martin completed the assembly, integration, and testing for the first GPS III satellite and in February 2017 the Air Force accepted delivery in advance of its currently scheduled May 2018 launch. As noted earlier, because of delays to OCX block 1, the Air Force initiated the COps program to ensure an interim means to control GPS III satellites. Without COps, no GPS III satellites can join the constellation to sustain it until OCX block 1 is operational in fiscal year 2022. In September 2016, COps formally started development, establishing a cost baseline of approximately $162 million to meet an April 2019 delivery. The COps program began software coding in November 2016, after a design review established that the product design would meet the Air Force’s intended needs. Short-Term Challenges: Compressed and Concurrent Schedules, Component Issues with the First GPS III Satellite The Air Force continues to struggle with keeping multiple, highly compressed, interdependent, and concurrent program schedules synchronized in order to sustain and modernize the GPS constellation. Figure 7 shows some of the schedule challenges of the three programs needed for constellation sustainment and modernization. Launching and operating the new GPS III satellite is a highly complex effort, since it requires synchronizing the development and testing schedules of OCX block 0, the first GPS III satellite, and the COps programs. For the Air Force to achieve its objective of making the first GPS III satellite operational by September 2019, numerous challenges (discussed below) must be addressed in the next 18 months on all three programs. If any of the three programs cannot resolve their challenges, the operation of the first GPS III satellite—and constellation sustainment—may be delayed. OCX Block 0 and Pre-Launch Testing Schedules With the goal of launching the first GPS III satellite in March 2018, the Air Force restructured its pre-launch integrated satellite and ground system testing in the summer of 2016, compressing the overall testing timeframe from 52 weeks to 42 weeks. More OCX block 0 delays in early fiscal year 2017 complicated Air Force test plans, resulting in changes to the sequence and timing of events, the introduction of concurrency at various points throughout the testing, the use of incomplete software in early testing, and an increase in the likelihood of discovering issues later in pre- launch integrated testing. Air Force officials stated that some pre-launch testing revisions streamlined the overall test plan since the merging of certain test events allowed multiple objectives to be met by the same event. Raytheon delivered OCX block 0, the launch and checkout system for GPS III satellites, in September 2017. The Air Force took possession of OCX block 0 in October 2017 and will finally accept it at a later date after OCX block 1 is delivered. However, if issues requiring corrective work are discovered during subsequent integrated testing, the GPS III launch schedule may be delayed further since there is minimal schedule margin on OCX block 0 for correcting any additional problems that may be found. First GPS III Satellite Capacitors There are hundreds of capacitors—devices used to store energy and release it as electrical power—installed in each GPS III satellite. In 2016, while investigating capacitor failures, the Air Force discovered that the subcontractor, then known as Exelis (now Harris Corporation), had not conducted required qualification testing for the capacitor’s operational use in GPS III satellites. The Air Force conducted a review of the components over many months, delaying program progress while a subcontractor qualified the capacitor design as suitable for use on the GPS III satellite. However, the Air Force concluded that Harris Corporation failed to properly conduct a separate reliability test of the particular production lot from which the questionable capacitors originated. The Air Force directed the contractor to remove and replace the capacitors from that production lot from the second and third GPS III satellites. After weighing the technical data and cost and schedule considerations, the Air Force decided to accept the first satellite and launch it “as is” with the questionable capacitors installed. The COps program is also pursuing a compressed and concurrent development and testing schedule to be operational as planned in September 2019. The COps acquisition strategy document acknowledges that the program’s timeline is aggressive. DOT&E has highlighted the compressed COps schedule as a risk, since the limited time between the developmental and operational testing permits little time for the evaluation of test results and resolution of any deficiencies found. The COps program has already begun drawing from its 60-day schedule margin, with a quarter of this margin used within the first 5 months after development started. According to Air Force officials, this margin use was the result of unplanned delays certifying a software coding lab. Additionally, the program schedule has concurrent development and testing, which in our previous work we have noted is often a high risk approach but is sometimes appropriate for software development. COps faces further schedule risk from its need for shared test assets, particularly the GPS III satellite simulator, a hardware- and software- based ground system that simulates GPS III function, which is also required by the GPS III and OCX programs. According to a DOT&E official, the OCX program receives priority over COps for the use of the GPS III satellite simulator, since the testing asset is heavily needed in the development of the ground control system. Because of the competing demands for this resource, which Air Force and DOT&E officials maintain requires lengthy and complex software reconfigurations to repurpose the simulator from one test event to the next, the Air Force is using a less realistic and purely software-based simulator for the testing of COps, where possible. Short-Term Risk Mitigation: Nearly 2 Years of Schedule Buffer to When First GPS III Satellite Needed Recent data show that the current satellites in the GPS constellation are expected to remain operational longer than previously projected, creating an additional, nearly 2-year schedule buffer before the first GPS III satellite needs to be operational to sustain the current GPS constellation capability. The Air Force projected that the first GPS III satellite needed to be operational by September 2019 based on 2014 satellite performance data. However, our analysis of the Air Force’s more recent May 2016 GPS constellation performance data indicates that, in order to continue meeting the constellation availability performance standard without interruption, the operational need for the first GPS satellite is now June 2021. This projection incorporates updated Air Force data from the current satellites that take into account an increase in solar array longevity expected for IIR and IIR-M satellites, according to Air Force officials. The Air Force is likely to meet the constellation’s June 2021 operational requirement because there are seven GPS III satellites planned to be launched by June 2021. Figure 8 shows the events leading to the launch and operation of the first GPS III satellite, achieving constellation sustainment once the first GPS III is operational, and subsequent GPS III launches that continue to support sustainment. The nearly 2-year buffer between planned operation and actual need for the first GPS III satellite permits the Air Force additional time to resolve any development issues. Because of this additional 2-year schedule buffer, we are not making a recommendation at this time to address the short term challenges we have identified but will continue to assess the progress of each of the programs and risks to constellation sustainment in our future work. Long-Term Challenge: Most GPS III Satellites Under Contract Will Have Launched before Operational Testing Confirms Satellite Performance The Air Force risks additional cost increases, schedule delays, and performance shortfalls because operational testing to confirm that GPS III satellites work as intended with OCX blocks 1 and 2 will not be completed until after the planned launch of 8 of the 10 GPS III satellites currently under contract. Due to delays to the OCX final delivery, the new ground control system will not be completed in time to control the GPS III satellites for the first few years they are in orbit (approximately 3.5 years). Consequently, GPS III operational testing will now occur in three phases— 1. in late fiscal year 2019 to confirm the satellites can perform similarly to the existing GPS satellites with COps; 2. in fiscal year 2020 to confirm the GPS III satellites can perform some of the new M-code capabilities with MCEU; and 3. in fiscal year 2022 to confirm the GPS III satellites can perform all of the new M-code capabilities with OCX blocks 1 and 2. The first GPS III satellite is projected to complete operational testing of legacy signal capabilities in September 2019. By that point, the Air Force plans to have launched 3 of the 10 GPS III satellites, the fourth satellite is expected to be delivered, and major integration work will be underway on satellites 5 through 8. Therefore, if satellite shortcomings are discovered during any phase of the operational testing, the Air Force will be limited to addressing such issues through software corrections to satellites already on orbit. If any of the three phases of operational testing reveals issues, the Air Force may face the need for potentially costly contract modifications and delivery delays for satellites not yet launched. To offset this risk, the Air Force has obtained performance knowledge of GPS III satellites through ground testing of the first satellite, and findings from this testing have driven modifications to all ten satellites. Because of the rigor of the ground testing of the first satellite, Air Force officials maintain that the knowledge that might be obtained through on-orbit operational testing of the first GPS satellite would be minimal. However, a DOT&E official said that ground testing is limited to assessing system responses that are induced through the testing process and therefore may omit phenomena that might be experienced in actual system operation on orbit. We will continue to track the progress of operational testing in our future work. Modernizing GPS Military Broadcast Challenged by High- Risk Development Schedules DOD has established high-risk schedules for modernizing the GPS broadcast, or M-code signal, produced by GPS satellites. These risks are manifest in different ways. In the near term, the Air Force plans to provide a limited M-code broadcast—one that does not have all of the capabilities of OCX—in the MCEU program in fiscal year 2020. However, the MCEU schedule is high risk for its dependency on the timely completion of the COps program, for its aggressive schedule, and because of competition for limited test resources. Further, the full M-code broadcast capability, planned for fiscal year 2022, is at high risk of additional delays because (1) it is dependent on unproven efficiencies in software coding, (2) the program has not yet completed a baseline review, which may identify additional time needed to complete currently contracted work, and (3) there are known changes to the program that must be done that are not included in the proposed schedule. High-Risk Programs Underlie Strategy to Deliver M-code Broadcast Capability As noted above, the Air Force’s plans for delivering the M-code broadcast involve two separate high-risk programs—MCEU and OCX blocks 1 and 2—delivered at separate times to make an operational M-code signal available to the warfighter. Figure 9 highlights the current forecasted operational schedules to deliver limited M-code broadcast capabilities with MCEU and full M-code broadcast with OCX. The MCEU program, created because of multiple delays to OCX and to partially address that program’s remaining schedule risk, is itself a high- risk program that is dependent on the timely development of COps. Estimated to cost approximately $120 million, MCEU formally entered the acquisition process in January 2017 as a software-specific program to modify OCS. To develop MCEU, Lockheed Martin officials stated they will leverage personnel with expertise maintaining and upgrading OCS as well as utilize the staff working on COps. With a planned December 2019 delivery for testing and a September 2020 target to begin operations, the MCEU program faces several schedule risks. The Air Force’s proposed plan anticipates a compressed software development effort, which the Air Force describes as aggressive. The Air Force has also identified potential risks to the MCEU schedule from competing demands by GPS III, OCX, COps, and MCEU for shared test resources. Air Force officials specifically noted competing demands for the GPS III simulator test resource. If development or testing issues arise in these other programs, those issues could delay the availability of the satellite simulator and thereby disrupt the planned MCEU development effort. According to program officials, the Air Force is working to mitigate this threat to the MCEU program through the use of a software-based simulator, when possible. Additionally, MCEU software development work is dependent on the timely conclusion of the COps effort—which, as previously mentioned, itself has an aggressive schedule and faces competition for a limited test resource. Air Force program officials have said that some Lockheed Martin staff planned to support MCEU will need to transfer from the COps effort. However, after reviewing the staffing plans at the MCEU contractor kickoff, Air Force officials said this is no longer viewed as a significant risk. OCX blocks 1 and 2 Raytheon has made some progress starting coding for OCX block 1 and taken the first steps toward implementing and demonstrating initial software development efficiencies that may benefit development for OCX blocks 1 and 2. The software efficiencies are built up in seven phases and need to be completed before the development process reaches each of the phases to take full advantage of the efficiencies they will create. Once ready, the efficiencies are inserted at different points in the software development schedule. For example, as of August 2017, the first of seven phases implementing the software development improvements was nearly complete, while the second phase was approximately two-thirds complete. Both are needed in place for insertion when the next phase of coding begins. Further, the Air Force proposed a new rebaselined schedule in June 2017 as the final step to getting the program back on track after declaring a critical Nunn-McCurdy unit cost breach in 2016 when the program exceeded the original baseline by more than 50 percent. A Nunn- McCurdy unit cost breach classified as critical is the most serious type of breach and requires a program to be terminated unless the Secretary of Defense submits a written certification to Congress that, among other things, the new estimate of the program’s cost is reasonable and takes other actions, including restructuring the program. In October 2016, DOD recertified the program, with a 24-month schedule extension. Under this newer proposed schedule Raytheon forecasts delivering blocks 1 and 2 in December 2020 with 6 months of extra schedule—a 30-month schedule extension—to account for unknown technical issues before OCX blocks 1 and 2 are due to the Air Force in June 2021. The Air Force projects operating OCX in fiscal year 2022 after completing 7 months of operational testing post-delivery. Three factors place delivery of OCX blocks 1 and 2 in June 2021 at high risk for additional schedule delays and cost increases: First, the newly proposed June 2017 rebaselined schedule assumes significant improvements in the speed of software coding and testing that have not yet been proven, but will be introduced at various periods as software development proceeds. Whether Raytheon can achieve the majority of these efficiencies will not be known until the end of fiscal year 2018. However, the Defense Contract Management Agency, which independently oversees Raytheon’s work developing OCX, noted in July 2017 a number of risks to the schedule, including that some initial assumed efficiencies had not been demonstrated. Specifically, they noted for initial coding on block 1 that Raytheon had achieved only 60 percent of the software integration maturity planned to that point in time in conjunction with greater numbers of software deficiencies that will require more time than planned to resolve. Second, the proposed rebaseline schedule has not yet undergone an integrated baseline review (IBR) to verify all of the work that needs to be done is incorporated into that schedule. The IBR is a best practice required by the Office of Management and Budget on programs with earned value management. An IBR ensures a mutual understanding between the government and the contractor of the technical scope, schedule, and resources needed to complete the work. We have found that too often, programs overrun costs and schedule because estimates fail to account for the full technical definition, unexpected changes, and risks. According to prior plans, the IBR would have taken place in early 2017, but it has been delayed multiple times for a number of reasons. A significant and recurring root cause of delays on the OCX program has been a lack of mutual understanding of the work between the Air Force and Raytheon. The IBR start was scheduled for November 2017 with completion in February 2018. Once conducted, the review may identify additional work not in the proposed schedule that needs to be completed before delivery. For example, Raytheon is conducting a review of hardware and software obsolescence. If significant additional obsolescence issues are found that need to be resolved before OCX blocks 1 and 2 are delivered, the projected delivery date may need to be delayed further at additional cost. Third, the OCX contract will likely be modified because the Air Force needs to incorporate into its contract with Raytheon a number of changes that are not currently a part of the proposed schedule. According to Air Force and contractor officials, negotiations are under way to determine which of these changes will be incorporated before OCX blocks 1 and 2 are delivered and which may be added after delivery. Air Force officials said that the incorporation of changes should be completed by February 2018. Schedule risk assessments for OCX blocks 1 and 2 delivery vary, making it unclear when the full M-code broadcast will finally be operational. Government assessments of Raytheon’s performance continue to indicate more schedule delays are likely. Table 4 shows the varying assessments of potential schedule delays by the Defense Contract Management Agency and the Air Force to the proposed June 2021 delivery date and the subsequent operational date that occurs 7 months later. In 2015, we made four recommendations to the Secretary of Defense, one of which was to use outside experts to help identify all underlying problems on OCX and develop high confidence cost and schedule estimates, among others, in order to provide information necessary to make decisions and improve the likelihood of success. To date, none of these recommendations have been fully implemented but DOD has taken steps to address some of them. Further, because the Air Force has undertaken the COps and MCEU programs to provide interim capabilities to mitigate OCX delays for the full broadcast capability, we are not making additional recommendations at this time but will continue to monitor progress and risks to the acquisition of OCX. Greater Coordination Needed to Prevent Duplication of Effort Developing and Fielding M-code Receivers While technology development for the M-code receiver cards is underway, DOD has developed preliminary—but incomplete—plans to fully develop and field M-code receiver cards across the more than 700 weapon systems that will need to make the transition from the current technology. DOD has prepared initial cost and schedule estimates for department-wide fielding for a fraction of these weapon systems. While the full cost remains unknown, it is likely to be many billions of dollars greater than the $2.5 billion identified through fiscal year 2021 because there is significant work remaining to verify the initial cards work as planned and to develop them further after the MGUE increment 1 program ends. Without greater coordination of integration test results, lessons learned, and design solutions DOD is at risk of duplicated development work as multiple weapon system programs separately mature and field similar technologies on their own. Further, with the full M-code broadcast available in fiscal year 2022, a gap—the extent of which is unknown—between operationally broadcasting and receiving M- code exists. Figure 10 highlights the gap between the time the M-code signal will be operational and the undefined time M-code can be used by the military services. DOD Has Made Some Progress in Developing Technology for New M- code Receiver Cards The Air Force program to develop initial M-code receiver test cards has made progress by establishing an acquisition strategy for this effort and maturing receiver test cards. In January 2017, DOD approved the MGUE increment 1 program to formally begin development, and it defined the criteria to end the program as (1) verifying technical requirements on all types of final receiver test cards; (2) certifying readiness for operational testing by the Air Force Program Executive Officer; (3) completing operational testing for the four lead platforms for, at a minimum, at least the first card available; and (4) completing manufacturing readiness assessments for all three contractors. Within the MGUE increment 1 program, contractors are making progress toward delivering final hardware test cards and incremental software capabilities. For example, one contractor has achieved its initial security certification from the Air Force, which is a key step toward making the MGUE increment 1 receiver test card available for continued development and eventual procurement. Further, the MGUE increment 1 program is also conducting risk reduction testing in preparation for formal developmental verification testing, an important step that ensures the receiver cards meet technical requirements. Programs throughout DOD can make risk-based decisions to develop and test the receiver test cards after technical verification of the card’s hardware and software. According to MGUE program officials, this is significant because it allows non-lead platforms to obtain and work with the cards sooner than the end date of operational testing on lead platforms. Significant Development Work Remains to Eventually Field M-code Receiver Cards Although the Air Force has made progress in maturing receiver test cards, significant development work remains to reach the point where the cards can ultimately be fielded on over 700 different weapon systems. For example, for MGUE increment 1, the Air Force must define additional technical requirements in order for the M-code receiver cards to be compatible and communicate with existing weapon systems. The Air Force will also need to conduct operational tests for each of the lead platforms—the Stryker ground combat vehicle; B-2 Spirit bomber; JLTV; and DDG-51 Arleigh Burke destroyer—before the full M-code signal is available with OCX. Because these tests will instead be conducted with the limited signal provided by MCEU, DOD risks discovering issues several years later once full operational testing is conducted. Further, according to military service officials and assessments by DOT&E, this operational testing will only be minimally applicable to other weapon systems because those other weapon systems have different operational requirements and integration challenges than the four lead platforms. As a result, additional development and testing will be necessary on an undetermined number of the remaining weapon systems to ensure the receiver cards address each system’s unique interfaces and requirements. In 2018, DOD will also formally begin development for MGUE increment 2. Increment 2 will provide more compact receiver cards to be used when size, weight, and power must be minimized, such as on handheld receivers, space receivers, and munitions where increment 1 receiver cards are too large to work. The military services are working to mitigate some of these development challenges. For example, Army officials told us they do not plan to field MGUE receiver cards on its lead platform, the Stryker, due to ongoing gaps in technical requirements. In addition, there is not a lead platform to demonstrate increment 1 on munitions since munition requirements were planned to be addressed in increment 2. However, to address its needs, the Army has initiated efforts to modify the MGUE increment 1 receiver card for some munitions that would otherwise need to wait for MGUE increment 2 technologies. Individual munition program offices within other military services have begun to do so as well. According to military service officials from the Army, Navy, and Marine Corps, it is essential that user needs are met by increment 2, or they will have to conduct additional development and testing. The Army previously identified gaps in increment 1 that the Air Force has either addressed in increment 1, has deferred to increment 2, or will need to be addressed outside of the MGUE increment 1 and 2 programs. Army and Navy officials also stated that they were concerned that any disagreements in requirements for increment 2 could lead to further fielding delays. Finally, the transition from existing GPS receiver cards to M-code receiver cards is likely to take many years. We recently reported that transitioning all DOD platforms to the next generation of receiver cards will likely take more than a decade. A lengthy transition has happened before, as previous efforts to modernize GPS to the current receiver cards, begun in 2003, are still underway and the older receiver cards are still being used. As a result, DOD anticipates that warfighters will have to operate with a mix of older and newer receiver cards. DOD Has Begun Cost and Schedule Planning; Full Cost Is Unknown but Likely to be Many Billions of Dollars DOD has begun collecting preliminary information on M-code requirements for individual weapon systems. In December 2016, the USD AT&L directed the military services, the Missile Defense Agency (MDA), and Special Operations Command (SOCOM) to submit implementation plans with M-code investment priorities across weapon systems and munitions, including projected costs and schedules. According to DOD, these M-code implementation plans are intended to provide DOD with a management and oversight tool for the fielding effort. In February 2017, each organization submitted its own implementation plan to USD AT&L. These plans were then briefed to the PNT Executive Management Board and PNT Oversight Council in February and March, respectively. However, these implementation plans are preliminary and based on assumptions about the Air Force’s ability to achieve MGUE increment 1 and 2 technical requirements, the timeline required to do so, and the amount of development and test work that will remain for the receiver cards to be ready for production and fielding after the programs end. Since the MGUE increment 2 program has not started development, it has not yet finalized requirements. Once approved, the increment 2 program office will produce an acquisition strategy, schedule, and cost estimate. However, after the MGUE increment 2 program ends there is no detailed plan for completing development, testing, and fielding of M-code receiver cards for weapon systems across the department. DOD has preliminary cost and schedule estimates for some weapon programs, but lacks a total cost at this point because the department does not include all efforts initiated by programs to meet specific needs, including those outside the MGUE increment 1 and 2 programs. The initial M-code implementation plans responded to what was requested but do not individually identify what the total cost will be for each organization to develop and field M-code receiver cards, so a total cost can be determined across DOD. Because USD AT&L required that the implementation plans include funding and schedule estimates for 2 to 3 years while directing that plans be resubmitted, at a minimum, every 2 years, weapon systems that will need M-code but were not considered an immediate priority were not included in the initial submissions. In addition, the military services, MDA, and SOCOM provided only initial cost estimates. According to military service officials, these estimates were based on the current MGUE increment 1 program schedule and technical development and include risk-based decisions to partially fund specific programs until the MGUE increment 1 program matures. According to a USD AT&L official, the plans would both facilitate M-code implementation planning for the department and inform the issuance of waivers. The official stated that as the acquisition programs critical to providing M-code capability mature, future implementations plans should provide more comprehensive estimates of cost and schedule to achieve M-code implementation for the department. Our analysis of the M-code receiver card implementation plans found that initial funding estimates indicate a cost of over $2.5 billion to integrate and procure M-code receiver cards on only a small number of weapon systems out of the hundreds of types that need M-code receiver cards. The full cost will be much larger—likely many billions of dollars because the majority of the weapon systems that need M-code receiver cards are not funded yet or are only partially funded, according to the M-code implementation plans. Specifically, the military services, MDA, and SOCOM identified 716 types of weapon systems in their February 2017 implementation plans that require almost a million M-code receiver cards. For example, the JLTV fleet—which provides protection for passengers against current and future battlefield threats for multiple military services—is one type of weapon system that will eventually need almost 25,000 receiver cards. Of the 716 types of weapon systems that will need M-code receiver cards, only 28—or less than 4 percent—are fully funded through fiscal year 2021. The remainder have either partially funded M- code development and integration efforts (72 weapon systems), or do not yet have funding planned (616 weapon systems). Additionally, the preliminary estimates to develop and procure M-code receivers on selected weapon systems do not all include funding beyond fiscal year 2021 that will be needed for further development, integration, and procurement. This means that DOD and Congress do not have visibility into how much additional funding could be needed to fully fund the remaining 96 percent of all weapon systems that need M-code receivers. Figure 11 shows the M-code development and integration efforts that are funded, partially funded, or unfunded through fiscal year 2021 across DOD weapon systems that will need M-code receiver cards. Because the implementation plans are a first step toward providing DOD leadership insight on this large set of acquisitions and they will be updated at least every 2 years by the different organizations within DOD, we are not making a recommendation at this time. However, we will continue to monitor DOD’s cost and schedule planning. DOD Risks Duplication of Effort Integrating and Testing M-code Receiver Cards The level of development and procurement effort beyond MGUE increments 1 and 2 is significant and will require close coordination among the military services, MDA, and SOCOM. While Joint Staff officials stated that the DOD Chief Information Officer is working with the military services and Joint Staff to produce a user equipment roadmap to help guide that coordination, they said that these efforts are not yet complete. DOD has designated the Air Force to lead initial development of both larger and smaller test cards that other organizations will need to develop further to meet their individual needs. After the Air Force develops initial cards for both sizes, the breadth and complexity of this acquisition will multiply, as the offices responsible for upgrading hundreds of weapon systems begin their own individual efforts to further develop and test the cards so they work for the unique needs of their specific system. While some common solutions are being developed, Air Force officials said the military services and individual weapon systems will have the freedom to go to the contractors and begin their own development efforts. DOD does not have a developed plan in place to help ensure that common design solutions are employed and that DOD avoids duplication of effort as multiple entities separately mature receiver cards. We previously found that duplication occurs when two or more agencies or programs are engaged in the same activities. In this case, because the individual organizations and program offices are likely to be pursuing individual and uncoordinated receiver card programs at different times with different contractors, DOD is at risk for significant duplication of effort. We previously found that establishing formal mechanisms for coordination and information sharing across DOD programs reduces the risk of gaps and results in more efficient and more effective use of resources. Internal control standards also state that establishing clear responsibilities and roles in achieving objectives is key for effective management. Further, DOD previously reported clear leadership ensures that programs and stakeholders are aligned with common goals. According to MGUE program officials, the MGUE increment 1 program is already capturing all issues observed in receiver test card risk reduction testing and sharing this information through a joint reporting system. However, while non-lead platforms may also report deficiencies in this system, there is no requirement that they do so, nor is there an entity responsible for ensuring data from testing, design, and development is shared between programs. We previously found that the absence of a formal process for coordination results in the potential for duplication, overlap, and fragmentation. DOD therefore risks paying to repeatedly find design solutions to solve common problems because each program office is likely to undertake its own uncoordinated development effort. Some duplicated effort may already be occurring. Air Force officials have expressed concern that work is already being duplicated across the military services in developing embedded GPS systems to be integrated into aircraft. According to multiple DOT&E assessments, the absence of a plan across the wide variety of intended interfaces leaves significant risk in integrating the receiver cards, and therefore fielding cost and schedule risk for DOD. Conclusions GPS is a national asset for civilians and the military service members who depend upon it each day. Any disruption to the system would have severe economic and military consequences. In keeping the system sustained and modernizing it with additional capabilities, DOD has spent billions of dollars more than planned developing five interdependent GPS programs. Developing these technologies is complex work with the collective effort already years behind initial estimates to provide the warfighter with a means to counter known threats, such as jamming, to the current system. It will be many years before M-code receiver cards are fielded at a cost that remains unknown but that will be substantially higher than the estimated $2.5 billion already identified through fiscal year 2021. In the short term, it is unclear when there will be a receiver card ready for production after the end of operational testing, and in the long term DOD risks wasting resources duplicating development efforts on weapon systems with similar requirements. Without better coordination of this effort, DOD risks unnecessary cost increases and schedule delays because there is no established process or place for collecting and sharing development and integration practices and solutions between programs. Recommendations for Executive Action We are making the following recommendation to DOD: The Secretary of Defense should ensure that the Under Secretary of Defense for Acquisition, Technology, and Logistics, as part of M-code receiver card acquisition planning, assign an organization with responsibility for systematically collecting integration test data, lessons learned, and design solutions and making them available to all programs expected to integrate M-code receiver cards. (Recommendation 1) Agency Comments and Our Evaluation We provided a draft of this report to the Department of Defense for review and comment. In its written comments, reproduced in appendix II, DOD concurred with the recommendation. DOD also provided technical comments, which we incorporated as appropriate. We are sending copies of this report to the appropriate congressional committees, the Secretary of Defense, the Secretary of the Air Force, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions concerning this report, please contact me at (202) 512-4841 or by email at chaplainc@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. Key contributors to this report are listed in appendix IV. Appendix I: Objectives, Scope, and Methodology To determine the extent to which there are acquisition risks to sustaining the Global Positioning System (GPS) satellite constellation, we reviewed the Air Force GPS quarterly reports, program acquisition baselines, integrated master schedules, acquisition strategies, software development plans, test plans, and other documents to the extent they existed for GPS III, Next Generation Operational Control System (OCX), and Contingency Operations (COps) programs. We also interviewed officials from the GPS III, OCX, and COps programs; the Air Force Space and Missile Systems Center’s (SMC) GPS Enterprise Integrator office; the prime contractors from all three programs; the Defense Contract Management Agency; the Office of Cost Assessment and Program Evaluation; and the Office of the Director, Operational Test and Evaluation (DOT&E). We also reviewed briefings and other documents from each to evaluate program progress in development. We assessed the status of the currently operational GPS satellite constellation, interviewing officials from the Air Force SMC GPS program office and Air Force Space Command. To assess the risks that a delay in the acquisition and fielding of GPS III satellites could result in the GPS constellation falling below the 24 satellites required by the standard positioning service and precise positioning service performance standards, we employed a methodology very similar to the one we had used to assess constellation performance in 2009, 2010, and 2015. We obtained information dated May 2016 from the Air Force predicting the reliability for 63 GPS satellites—each of the 31 current (on-orbit as of July 2017) and 32 future GPS satellites—as a function of time. Each satellite’s total reliability curve defines the probability that the satellite will still be operational at a given time in the future. It is generated from the product of two reliability curves—a wear- out reliability curve defined by the cumulative normal distribution, and a random reliability curve defined by the cumulative Weibull distribution. For each of the 63 satellites, we obtained the two parameters defining the cumulative normal distribution, and the two parameters defining the cumulative Weibull distribution. For each of the 32 unlaunched satellites we included in our model, we also obtained a parameter defining its probability of successful launch, and its current scheduled launch date. The 32 unlaunched satellites include 10 GPS III satellites currently under contract and 22 GPS III satellites planned for contract award in late 2018; launch of the final GPS III satellite we included in our model is scheduled for October 2031. Using this information, we generated overall reliability curves for each of the 63 GPS satellites. We discussed with Air Force and Aerospace Corporation representatives, in general terms, how each satellite’s normal and Weibull parameters were calculated. However, we did not analyze any of the data used to calculate these Air Force provided parameters. Using the reliability curves for each of the 63 GPS satellites, we developed a Monte Carlo simulation to predict the probability that at least a given number of satellites would be operational as a function of time, based on the GPS launch schedule as of May 2016. We conducted several runs of our simulation—each run consisting of 10,000 trials—and generated “sawtoothed” curves depicting the probability that at least 24 satellites would still be operational as a function of time. We then used our Monte Carlo simulation model to examine the effect of delays to the operational induction of the GPS III satellites into the constellation. We reran the model based on month/year delay scenarios, calculating new probabilities that at least 24 satellites would still be operational as a function of time, determining in terms of month/year the point at which a satellite would be required to enter operations to maintain an uninterrupted maintenance of the 95 percent probability of 24 satellites in operation. The Air Force satellite parameters we used for the Monte Carlo simulation pre-dated the Air Force investigation into navigation payload capacitors and the subsequent decision to launch the first satellite “as is” with questionable parts. Therefore, the reliability parameters for this satellite were not informed by any possible subsequent Air Force consideration of the decision to launch the first GPS III satellite “as is” with these parts. To determine the extent to which the Department of Defense (DOD) faces acquisition challenges developing a new ground system to control the broadcast of a modernized GPS signal, we reviewed Air Force program plans and documentation related to cost, schedule, acquisition strategies, technology development, and major challenges to delivering M-code Early Use (MCEU) and OCX blocks 1 and 2. We interviewed officials from the MCEU and OCX program offices, SMC GPS Enterprise Integrator office, DOT&E, and the prime contractors for the two programs. For OCX, we also reviewed quarterly reviews, monthly program assessments, and slides provided by Raytheon on topics of our request. We also interviewed Office of Performance Assessments and Root Cause Analyses officials regarding root causes of the OCX program’s cost and schedule baseline breach and Defense Contract Management Agency officials charged with oversight of the OCX contractor regarding cost and schedule issues facing the program’s development efforts, major program risks, and technical challenges. To determine the extent to which DOD faces acquisition challenges developing and fielding modernized receiver cards across the department, we reviewed Air Force program plans and documentation related to M-code GPS User Equipment (MGUE) increment 1 cost, schedule, acquisition strategy, and technology development. We interviewed officials at the Air Force SMC GPS program office, MGUE program office, DOT&E, and the three MGUE increment 1 contractors— L3 Technologies, Raytheon, and Rockwell Collins. To identify the military services’ respective development efforts and challenges in integrating MGUE with their lead platforms, we interviewed officials from the lead program offices for the Army’s Defense Advanced GPS Receiver Distributed Device/Stryker, Air Force’s B-2 aircraft, Navy’s DDG-51 Arleigh Burke class destroyer, and Marine Corps Joint Light Tactical Vehicle. Additionally, to understand the extent to which DOD has a plan for implementing M-code for the warfighter, we analyzed DOD Positioning, Navigation, and Timing (PNT) plans and other DOD memorandum on GPS receiver cards. We also held discussions with and received information from officials at Office of the Undersecretary of Defense for Acquisition, Technology, and Logistics; Joint Staff / J-6 Space Branch; and military service officials from the offices responsible for developing M-code receiver card implementation plans. We conducted this performance audit from February 2016 to December 2017 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Comments from the Department of Defense Appendix III: GPS Modernization Cost Increases, Original Baseline vs. Current Estimate Appendix IV: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, David Best, Assistant Director; Jay Tallon, Assistant Director; Karen Richey, Assistant Director; Pete Anderson; Andrew Berglund; Brandon Booth; Brian Bothwell; Patrick Breiding; Erin Carson; Connor Kincaid; Jonathan Mulcare; Sean Sannwaldt; Alyssa Weir; Robin Wilson and Marie P. Ahearn made key contributions to this report.

Why GAO Did This Study GPS provides positioning, navigation, and timing data to civilian and military users who depend on this satellite-based system. Since 2000, DOD—led by the Air Force—has been working to modernize GPS and to keep the current system of satellites—known as the GPS constellation—operational, although these efforts have experienced cost and schedule growth. The National Defense Authorization Act for Fiscal Year 2016 contained a provision that the Air Force provide reports to GAO on GPS acquisition programs and that GAO brief the congressional defense committees. GAO briefed the committees in 2016 and 2017. This report summarizes and expands on information presented in those briefings. This report assesses the extent to which DOD faces acquisition challenges (1) sustaining the GPS constellation; (2) developing a new ground control system; and (3) developing and fielding modernized receivers. GAO analyzed GPS quarterly acquisition reports and data, acquisition strategies, software and test plans, and other documents, and interviewed DOD and contractor officials. What GAO Found The Department of Defense's (DOD) acquisition of the next generation Global Positioning System (GPS) satellites, known as GPS III, faces a number of acquisition challenges, but these challenges do not threaten DOD's ability to continue operating the current GPS system, which DOD refers to as the constellation, in the near term. Projections for how long the current constellation will be fully capable have increased by nearly 2 years to June 2021, affording some buffer to offset any additional satellite delays. While the first GPS III satellite has a known parts problem, six follow-on satellites—which do not—are currently scheduled to be launched by June 2021. DOD is relying on a high-risk acquisition schedule to develop a new ground system, known as OCX, to control the broadcast of a modernized military GPS signal. OCX remains at risk for further delays and cost growth. To mitigate continuing delays to the new ground control system, the Air Force has begun a second new program—Military-code (M-code) Early Use—to deliver an interim, limited broadcast encrypted GPS signal for military use by modifying the current ground system. GAO will continue to monitor OCX progress. DOD has made some progress on initial testing of the receiver cards needed to utilize the M-code signal. However, additional development is necessary to make M-code work with over 700 weapon systems that require it. DOD has begun initial planning for some weapon systems, but more remains to be done to understand the cost and schedule needed to transition to M-code receivers. The preliminary estimate for integrating and testing a fraction of the weapon systems that need the receiver cards is over $2.5 billion through fiscal year 2021 with only 28 fully and 72 partially funded (see figure). The cost will increase by billions when as yet unfunded weapon systems are included. The level of development and procurement effort beyond the initial receiver cards is significant and will require close coordination across DOD. After the Air Force develops initial cards, the breadth and complexity of this acquisition will multiply, as the offices responsible for upgrading hundreds of weapon systems begin their own individual efforts to further develop and test the cards. However, DOD does not have an organization assigned to collect test data, lessons learned, and design solutions so that common design solutions are employed to avoid duplication of effort as multiple entities separately mature receiver cards. DOD therefore risks paying to repeatedly find design solutions to solve common problems because each program office is likely to undertake its own uncoordinated development effort. What GAO Recommends DOD should assign responsibility to an organization to collect test data, lessons learned, and design solutions so they may be shared. DOD concurred with the recommendation.

gao_GAO-19-210

gao_GAO-19-210_0

Background DHS R&D Roles and Responsibilities The Homeland Security Act of 2002, as amended, designates the Under Secretary for Science and Technology as responsible for coordinating all R&D activities of DHS. The Act also provides that nothing in it precludes other department components from carrying out R&D activities as long as the activities are coordinated through S&T. As of September 2018, seven DHS components have budget authority to conduct R&D activities—S&T, the Coast Guard, the CWMD, the Secret Service, the Cybersecurity and Infrastructure Security Agency, TSA, and the Office of the Chief Information Officer within the Office of the Undersecretary for Management. Figure 1 provides an organizational overview of DHS components and offices that are involved in the R&D process as of December 2018. S&T reorganized its structure in September 2018 and currently has three technical divisions responsible for managing R&D programs related to improving border, immigration, and maritime security; supporting first responders; and countering physical and cybersecurity threats among others, as shown in figure 2. Most of S&T’s R&D portfolio consists of applied and developmental R&D, which can be transitioned to use within 3 years, as opposed to longer- term basic research. In addition to conducting projects for its DHS customers, S&T conducts research for other federal agencies and first responders. S&T is also responsible for conducting basic and applied research, and collaborates with other government agencies, academia, the private sector, and others. Overview of Past R&D Management Challenges Questions have been raised about S&T’s ability to demonstrate the impact of its investments—in terms of value, tangible products, and advances toward the homeland security mission. Accordingly, for example, House appropriations committee report language has directed S&T to demonstrate how its R&D efforts are timely, with results relatively well defined and to make investment decisions based on clear and sensible priorities. In 2016, Congress passed the National Defense Authorization Act of 2017 (NDAA) which requires DHS to report annually to Congress on the department’s R&D projects including details such as the project name, the component carrying out the project, associated funding levels, and expected objectives and milestones for each project, among other items. Since DHS began operations in 2003, we have made multiple recommendations designed to improve DHS efforts to manage and oversee R&D efforts, as described later in this report. In September 2012, we reported that DHS did not have a department-wide policy defining R&D or guidance directing its components how to report R&D activities. As a result, DHS did not know its total annual investment in R&D, which limited the department’s ability to oversee components’ R&D efforts and align them with agency-wide R&D goals and priorities. We also reported that DHS’s R&D efforts were fragmented and overlapping, which increased the risk of unnecessary duplication. We recommended that DHS develop policies and guidance for defining, reporting and coordinating R&D activities across the department, and that DHS establish a mechanism to track R&D projects. As of August 2017, DHS had implemented these recommendations by, among other things, issuing guidance defining research and development activities and establishing Integrated Product Teams (IPT) as a primary mechanism for coordinating R&D. These actions and others are described in more detail later in this report. Figure 3 provides a summary of key events related to S&T since its inception. Past DHS Accounting Challenges After the consolidation in 2002 of 22 agencies into a single department, DHS had, until recently, different appropriation structures and budget management practices based on agencies’ funding structures prior to DHS consolidation. In 2018, we found that, with over 70 different appropriations and over 100 formal program, project, or activity accounts, DHS operated for over a decade with significant budget disparities and inconsistencies across its components. The lack of uniformity hindered visibility, inhibited comparisons between programs, and complicated spending decisions, including for R&D-related programs. For example, in 2012 we reported that S&T, the Domestic Nuclear Detection Office, and Coast Guard were the only three DHS components with budget authority to conduct R&D. However, in 2012, we identified an additional $255 million in R&D obligations by other DHS components at that time. Further, we found in 2012 that the Domestic Nuclear Detection Office did not report certain R&D budget data to OMB, and R&D budget accounts included a mix of R&D and non-R&D spending, which further complicated DHS’s ability to identify its total investment in R&D activities. Role of the Integrated Product Teams in R&D Within DHS, IPTs – established in 2015 – are to identify and prioritize technological capability gaps, and identify current or future R&D efforts or other solutions to close the gap, among other things. Specifically, the IPT process consists of three activities: 1) Identifying R&D activities in progress, funded, planned, or recently completed; 2) Prioritizing technological capability gaps and corresponding R&D efforts to address those gaps; and 3) Validating and reporting the gaps. The DHS IPT effort is led by S&T, but the individual IPTs are composed of senior-level officials from across DHS. IPT members prioritize R&D gaps based on departmentwide needs and requirements, and align current and planned R&D efforts to the identified gaps. In prioritizing and evaluating the capability gaps, four pre-defined criteria and rating scales are used and are discussed below: Strategic alignment: assesses the R&D gaps alignment with DHS- level and component-level strategic priorities. Impact: assesses if addressing the R&D gap would result in enhanced risk or threat reduction capability, among other things. Feasibility: assesses the feasibility of addressing the R&D gap, given its technical complexity. Considerations include feasibility related to technology, time, and transition. R&D needs: assesses whether the R&D gap would provide a critical R&D solution in an otherwise unaddressed area. The IPT’s role in coordinating R&D is discussed later in this report. Overview of DHS’s Processes for Assessing R&D Performance GPRA, as updated and expanded by GPRAMA, requires agencies to establish annual performance goals with target levels of performance against which to measure progress towards those goals. In addition, GPRA requires executive agencies to prepare an Annual Performance Report on program performance for the previous fiscal year. DHS has developed goals and targets to assess and communicate R&D performance. As shown in figure 4, DHS’s performance assessment process also includes identifying performance gaps and implementing corrective actions to address unmet performance goals. DHS uses strategic and management performance goals and measures to assess and communicate on the performance of its R&D efforts. In addition, DHS uses milestones to track and communicate progress of its R&D project activities. Milestones: A milestone is a scheduled event signifying the completion of a major deliverable or a phase of work. Milestones can help agencies demonstrate that they have clear and fully developed strategies and are tracking progress to accomplish their goals. Milestones are often used as the basis of an alternative form of performance goal. Milestones related to DHS R&D efforts are reported to Congress and publicly available through the DHS congressional budget justification. Strategic Goals: A type of performance goal used to reflect achievement of missions that are publicly reported in the DHS Annual Performance Report. As part of DHS’s Annual Performance Report, these goals are subject to GPRA and GPRAMA requirements. Management Goals: A type of performance goal used to gauge program results and tie to resource requests that are reported to Congress and publicly available through the DHS congressional budget justification along with the strategic goals. As we previously reported in 1997, experts in research measurement have tried for years to develop indicators that would provide a measure of the results of R&D. However, the very nature of the innovative process makes measuring the performance of science-related projects difficult. For example, a wide range of factors determine if and when a particular R&D project will result in commercial or other benefits. It can also take many years for a research project to achieve results. DHS Has Obligated More Than $1 Billion per Fiscal Year from 2010 through 2017 for Conducting R&D That Aligns with Department Mission Areas across Seven DHS Components R&D Funding Amounts DHS is required to report department-wide R&D-related funding to OMB on an annual basis. DHS uses several mechanisms to report the R&D- related funding, including budget authority (the legal authorization to obligate funds), obligations (binding agreements to make a payment for services), and outlays (payments to liquidate obligations representing amount expended). Further, OMB requires agencies to submit data on R&D programs as part of their annual budget submissions on investments for basic research, applied research, development, R&D facilities construction, and major equipment for R&D using OMB’s definition of R&D. Based on our analysis of OMB’s federal obligations data, we identified R&D-related obligations data for DHS components for fiscal years 2010 through 2017. Figure 5 depicts the R&D related obligations that were reported for fiscal years 2010 through 2017, which, on average, were about $1.3 billion annually or more than $10 billion overall for that time frame. Additionally, S&T obligated nearly 80 percent of all DHS R&D funds for that time period. Types of R&D Activities that S&T Conducts S&T may conduct or fund R&D activities on its own or jointly with other entities. In addition to S&T, six other DHS components currently have budget authority to conduct R&D—the Coast Guard, CWMD, TSA, Secret Service, Cybersecurity and Infrastructure Security Agency, and the Undersecretary for Management. In August 2018, S&T reported that there were at least 132 ongoing R&D projects across the Department. Some R&D projects aim to produce a specific prototype or piece of technology for an end user, while others might be for developing IT systems, conducting specific training, or providing written reports, or knowledge products. According to S&T officials, S&T generally leads or funds R&D projects by providing technology and knowledge products for four homeland security areas: Disaster resilience. Improving community resilience to natural disasters through technology and tools that support planning, decision-making and mitigation efforts; Critical incidents. Improving technological capabilities during all stages of critical incident response; Border security. Improving the nation’s ability to detect, interdict and prosecute illegal activity across air, land and sea. Cybersecurity. Developing technologies, tools and techniques to defend, mitigate, and secure current and future systems, networks and critical infrastructures against cyberattacks. Figure 6 illustrates the types of R&D projects that are either led or funded by S&T for each category. For more in-depth examples and descriptions of S&T projects, please see appendix I. In its efforts to determine how to best support the DHS components and first responders, S&T seeks first to identify the end user’s needs by discussing operational challenges with components and first responders; then develop prototypes or leverage existing technologies to find solutions; and finally to test and evaluate potential solutions to ensure that they meet the end user’s needs and ultimately deploy solutions to the field. The other six DHS components with R&D budget authority typically lead and fund R&D projects tailored to support their specific operational requirements and respective missions. Examples of R&D projects conducted by DHS components other than S&T are listed below in table 1. S&T Implemented R&D Coordination Mechanism and Directives to Strengthen Collaboration Across DHS, but Challenges Remain in Ensuring Participation DHS Established a Process to Coordinate R&D Efforts DHS established its IPT process in August 2015 as the central mechanism to coordinate R&D efforts across the department, in accordance with recommendations we made in 2012. The IPT process works to identify DHS technological capability gaps and coordinate R&D to close the gaps across DHS mission areas. The IPTs consist of senior representatives from operational components. As of October 2018, IPTs are organized according to the department’s identified missions and include the following sub-IPTs, as shown in table 2. Each IPT has an establishing charter document, which formally identifies the IPT component members and responsibilities and lists the corresponding sub-IPTs. IPTs and sub-IPTs are to meet multiple times throughout the year to support the process of identifying and prioritizing R&D capability gaps and R&D efforts. For example, the charter for the “Secure Borders” IPT states that they anticipate meeting at least 2 or 3 times per year, or more frequently to support the annual program planning and budgeting process. Overall, components reported that the IPT process enhanced collaboration and improved visibility into R&D efforts across DHS. Officials from all 10 of the DHS components we interviewed reported the IPT process has been helpful in various ways, including identifying capability gaps, prioritizing and closing the gaps, and providing transparency and insight into other components’ R&D efforts. For example, CBP officials reported that, through the IPT process, they were able to identify R&D projects that the Coast Guard had been pursuing related to maritime security. The R&D projects that Coast Guard was pursuing were also of interest to CBP, and therefore CBP worked through the IPT process to prevent duplicative work and combine some of those efforts. In another example, TSA officials reported that they collaborated with the Secret Service to test explosive screening technologies, and that the IPT process facilitated their ability to collaborate and share information about the screening technologies. In addition to enhancing collaboration, component officials provided their perspectives on how the IPT process prioritizes technology capability gaps that components have identified. For example, TSA officials reported that the gap identification and prioritization process works well, but that funding R&D activities to close the gaps is more challenging because it is influenced heavily by competing budget priorities, emerging threats, and other DHS senior leadership priorities. TSA officials further reported that departmental resource constraints limit the number of identified capability gaps that can be addressed. However, officials from CBP reported that several R&D projects were successfully implemented after CBP had worked with S&T to identify a capability gap and transition a solution to close the gap, such as certain upgrades needed on CBP trucks. S&T officials stated that they have also taken steps to integrate with the department’s Joint Requirements Council and utilize component requirement executives who work with component agencies to provide a basis for requirements and aid the components with the means to track the progress and disposition of each capability gap on a regular basis. DHS Issued R&D Coordination Directives but Faces Challenges in Ensuring IPT Participation In 2012, we found that, among other things, DHS had not developed a policy defining who was responsible for coordinating R&D within the department and what processes should be used to coordinate it. As a result, components did not consistently coordinate with S&T on what R&D was planned or underway, leading to increased risk of unnecessary duplication of R&D efforts. We recommended that DHS develop and implement policies and guidance for overseeing R&D that included, among other things, a description of the department’s process and roles and responsibilities for overseeing and coordinating R&D investments. DHS concurred with our recommendation, and, in response, the Secretary for Homeland Security delegated the authority to coordinate and integrate the department’s R&D, testing, evaluation efforts to the Under Secretary for Science and Technology in 2014. In 2015 and 2016, DHS issued two guidance documents regarding the establishment and progress of the IPT process. These documents specified how DHS, through the IPT process, is to implement processes and mechanisms to coordinate department-wide R&D efforts. Additionally, in January 2017, DHS issued an R&D directive and associated instruction to formalize R&D reporting and coordination among components, as shown in figure 7. The 2017 directive and associated instruction identify the roles and responsibilities, including IPT participation requirements, for key entities involved in R&D across DHS. However, the directive and instruction do not specifically address steps to be taken if components do not adhere to the requirements. For example, the January 2017 DHS instruction states that “to effectively coordinate DHS R&D activities, DHS components are required to follow the DHS IPT process.” However, officials from CWMD stated that they do not participate in the S&T-led IPT sessions because they have their own internal process for identifying and prioritizing capability gaps. S&T officials stated that CWMD’s predecessor organization, the Domestic Nuclear Detection Office, participated in the IPT process until DHS initiated a reorganization of its weapons of mass destruction programs (resulting in the current CWMD). Current non-participation by CWMD, which has the second-largest R&D budget within DHS and obligated approximately 17 percent of DHS R&D funds, or $176 million in fiscal year 2017, poses risk of R&D project information not being shared among components. In August 2018, we reported that DHS’s chemical defense programs and activities were fragmented and not well coordinated across the department, including R&D activities. We recommended that CWMD develop a strategy and implementation plan to help DHS integrate and coordinate its chemical defense programs and activities, among other things. Additionally, in its 2014-2018 Strategic Plan, DHS states that, to anticipate key threats, DHS should, among other things, prioritize R&D activities related to chemical, biological, radiological, and nuclear terrorism. Given these factors, CWMD’s participation in the IPT process is important to ensure that all R&D efforts are fully coordinated thereby mitigating the risk of potential duplication of other DHS R&D efforts. S&T officials recognize that some components might not be complying fully with the departmental directives and associated guidance documents which require participating in the IPT process – the key R&D coordination mechanism within DHS. S&T officials stated that, despite these challenges, they have strong collaborative relationships with the components, and the existing collaboration mechanisms, such as the IPT process, continue to mature and facilitate R&D-related information sharing. However, DHS guidance documents require that components participate in the IPT process. By ensuring that all required components participate in the IPT process, DHS can help S&T maintain visibility of R&D projects in order to fulfill its statutory role of coordinating R&D. S&T Has Taken Steps to Identify and Track DHS’s R&D Project Information, But Efforts Are Disjointed or Do Not Align with DHS Guidance Since 2012, S&T has taken steps to identify and track information related to ongoing R&D projects across DHS, and in 2017, DHS developed a common appropriations structure that standardized R&D budgeting processes across the department. However, S&T’s efforts to identify and track R&D project information have limitations and can result in information that is not comprehensive. We also identified challenges in collecting information related to the achievement of R&D milestones. DHS Developed the Common Appropriations Structure to Better Streamline and Standardize the Budgeting Process across Components In 2017, DHS developed a common appropriations structure that allowed it to calculate and monitor its expenses, including R&D expenses, across the department. Officials from DHS’s OCFO reported that, prior to the new structure, some components categorized their R&D expenses as other types of expenses, such as “salaries and expenses.” These categorizations made it difficult to account for R&D expenses outside of an individual component’s budget management division. Furthermore, OCFO officials reported that components previously utilized inconsistent R&D definitions, which often led to discrepancies in how components would report R&D activities. In our April 2018 report, we found that DHS had operated for over a decade with significant budget disparities and inconsistencies across its components. We found that the lack of uniformity hindered visibility, inhibited comparisons between programs, and complicated spending decisions. According to DHS OCFO officials, the introduction of the common appropriations structure, among other things, has helped improve transparency within DHS and among the components so that R&D can be more readily identified and tracked. DHS is also able to compare R&D funding amounts throughout DHS more easily than in previous years. In addition, of the seven components that have their own R&D funding to report, five indicated that the new structure has improved the department’s ability to identify and report R&D activities. S&T’s Efforts to Maintain an Inventory of R&D Projects across DHS Have Resulted in Reporting That Is Not Comprehensive We identified multiple sources of component R&D project information, each posing its own challenges or limitations. As described below, these challenges and limitations include difficulty in collecting and integrating R&D project information, and reporting that is not comprehensive. DHS’s response to the National Defense Authorization Act of 2017: The NDAA, passed in December 2016, required DHS to provide a list of ongoing R&D projects and accompanying milestone information by January 2017, and annually thereafter, to specified congressional committees. In December 2017, DHS officials reported that they had not yet submitted the report, and anticipated that the response to the NDAA requirement would be completed by January 2018. In August 2018, DHS submitted its response to the committee, then 19 months late. S&T officials stated that the reporting delays were due to challenges in collecting and integrating the data. S&T officials also reported that it used the components’ congressional budget justification documents as a starting point to identify R&D projects to include in its report in response to the NDAA requirement. However, additional details about the R&D projects had to be collected via a “data call” process from the components. S&T officials told us that it was a challenge to have components report information about their R&D projects consistently and systematically. Furthermore, S&T officials identified terminology-related challenges in their R&D data call efforts, including making distinctions between R&D projects, efforts, and activities. S&T officials also reported that, in its current format, they would not be able to easily identify how many projects were added to the NDAA list across years, or if a given R&D project experienced a large increase or decrease in funding. Annual Reports of Coordinated R&D: In response to a 2015 request from the Secretary of Homeland Security that the IPTs identify R&D work being performed across DHS, S&T issued a “Report of Coordinated R&D” in 2016 and 2017. The content for the reports was developed through a “data call” process, and the reports identified R&D activities and projects across DHS. The reports – for 2016 and 2017 – contain tables of R&D project names and the component leading the project, among other things. However, during the course of our review, S&T officials reported that these annual reports should not be considered authoritative lists of R&D projects due to inconsistencies in the project information that components reported which led to the reports not being comprehensive. For example, when we asked about some significant variations in the number of projects between 2016 and 2017, S&T officials told us that one DHS component responded to the data call with a list of R&D activities that included a “wish list” of R&D for their component, and not actual ongoing R&D activities. DHS officials acknowledged that they do not have a mechanism to ensure the comprehensiveness of information reported by the components through the data call process. In addition, two components did not respond to S&T’s request for R&D project information for the 2017 Annual Report of Coordinated R&D. Congressional Budget Justifications: In May 2018, in the absence of a single, comprehensive list of R&D projects across DHS prior to the issuance of its report in response to the NDAA, S&T officials referred us to the R&D projects listed within the seven individual congressional budget justifications for the components that currently have budget authority to conduct R&D. Furthermore, as discussed earlier, S&T officials used the congressional budget justifications as their starting point in developing their response to the NDAA. However, S&T officials stated that there may be differences between the projects listed in the NDAA response and the projects listed in the congressional budget justifications. For example, S&T stated that the report in response to the NDAA includes all “ongoing” projects, regardless of the fiscal year in which they received funds; while the congressional budget justifications include R&D projects for which funding was requested for the given fiscal year. In other words, S&T officials clarified, they included all R&D projects in their response to the NDAA that had project activity, regardless of whether funding was requested in a particular congressional budget justification. S&T’s Project Tracker Database and the S&T Analytical Tracking System: A 2014 House Appropriations Committee report noted that the committee had repeatedly raised questions about S&T’s prioritization of R&D projects and that, without the ability to easily review and compare detailed information on all S&T projects and activities, the Under Secretary for S&T could not effectively carry out S&T’s responsibilities. Accordingly, the Committee directed S&T to develop a method or system for tracking all S&T-funded R&D projects. A November 2016 DHS Directive reiterates this requirement, specifying that the list of projects should be updated on at least a quarterly basis throughout the duration of an R&D project. S&T officials told us that, in response to the committee report, they developed the Project Tracker Database, which was in use at the time of our review, but was transitioned to a new system, the S&T Analytical Tracking System, in September 2018. Neither the S&T Analytical Tracking System nor its predecessor system, the Project Tracker Database, is intended to comprehensively collect information on R&D projects across the department, only for R&D projects managed within S&T. Given the recent implementation of the S&T Analytical Tracking System, it is too soon to tell whether it will improve and streamline S&T’s efforts to collect and analyze R&D-related information within the directorate. In addition, S&T officials stated that none of the above R&D information sources are suited to long-term trend analysis or data aggregation of department-wide R&D project information, and that these sources are disparate across DHS. S&T officials also acknowledged that better aligning R&D project information sources is an important aspect of improving how the department collects information DHS-wide. Standards for Internal Control in the Federal Government call for agencies to maintain quality information that is, among other things, current, accurate, accessible, and provided on a timely basis. Furthermore, the standards call for an agency’s management team to process relevant data from reliable sources and utilize it to make informed decisions. The disparate R&D project information sources that S&T maintains, such as DHS’s response to the National Defense Authorization Act of 2017 and the Annual Reports of Coordinated R&D discussed above, and the manual data-call process it takes to update the sources limits departmental access to current and reliable R&D project information. For example, an internal DHS web-based portal with pre- defined fields could provide component officials with a means for reporting information more consistently and comprehensively. Without complete and readily accessible R&D information, DHS may not have the information it needs to make informed decisions about R&D investments, such as which projects are to be prioritized. By developing a mechanism to address challenges and limitations related to the collection, integration, and comprehensiveness of R&D data across the department, S&T can improve its visibility on R&D efforts across DHS in accordance with its role as DHS’s coordinator of R&D efforts. DHS Has Not Effectively Developed Some R&D Performance Management Information, Posing Challenges to Assess and Report on the Progress of R&D Projects DHS components have processes in place to collect certain indicators of R&D performance, but we found that these processes have limitations. The methods used to assess and report performance and progress of DHS R&D efforts we identified include: Milestone information – used to assess and communicate progress to Congress and agency decision makers on individual R&D projects Strategic and Management Performance Goals – milestone and other information is aggregated to provide summary information on R&D performance by mission area Customer feedback – information gathered by component officials on R&D customer perspectives on the utility of ongoing or completed projects Below is our analysis of the three methods. Milestones Milestones are often used as the basis of an alternative form of performance goal. Performance goals specified in alternative form must be described in a way that makes it possible to discern if progress is being made toward the goal. Milestones related to DHS R&D efforts are reported to Congress and publicly available through the DHS congressional budget justification. A milestone is a scheduled event signifying the completion of a major deliverable or a phase of work, and can be described in a way that makes it possible to discern if progress is being made toward a goal. Milestones can also help agencies demonstrate that they have clear and fully developed strategies and are tracking progress to accomplish their goals. In our analysis of 14 milestones for seven S&T high-priority R&D projects identified in fiscal year 2018 DHS budget justification documents, we found that 3 of the 14 milestones fully adhered to DHS guidance for milestone descriptions. DHS budget development guidance suggests DHS components, which develop milestones for inclusion in congressional budget justification documents, utilize leading practices provided in the guidance. The leading practices state that successful milestones contain the following characteristics: 1. Specific - provide a clear understanding of expected results; 2. Measurable - the result can be reported in quantitative or qualitative 3. Results-Oriented/Relevant - milestone clearly links to results-oriented activities such as strategy, budget, and/or program/project plans; 4. Time-Bound - milestone specifies a beginning and end date for As shown in table 3, we identified that more than half of the milestones (8 of 14) were not specific and 10 of 14 were not results-oriented. Eleven of 14 milestones we analyzed were measurable and time-bound. While our analysis is not generalizable to all fiscal year 2018 R&D milestones, it illustrates areas where the selected milestones do not fully incorporate the DHS guidance. Below is more detail on our assessment of the Specific and Results- Oriented guidance. Specific. Of the 14 milestones we reviewed, eight did not contain specific information that would allow reviewers to have a clear understanding of the result expected in connection with the milestone. For example, one milestone for a cyber-related R&D project states that “testbeds and pilots would be conducted with at least one department or agency.” However, the milestone is not specific enough to ascertain what types of testbeds or pilots are being assessed and how the testbed effort would link to a possible end result. Results-Oriented and Relevant. Ten of the 14 milestones that we reviewed did not clearly link the milestone back to results-oriented activities, such as strategy, budget, or project/program plans. For example, one milestone for a first responder program stated the following: “Transition, commercialize, or make available through open source platforms at least three technologies (e.g., Analyses, models, technology prototypes and/or knowledge prototypes).” It is unclear which technologies would be transitioned or how these technologies would be transitioned and made available. According to DHS OCFO officials, DHS congressional budget justifications, which include milestones, serve to provide explanation and detail for why DHS believes Congress should support the department’s R&D projects. DHS components are instructed by DHS’s budget office to routinely submit their congressional budget justifications for internal DHS review, which is a process and mechanism that results in the supporting justifications for R&D funding requests. DHS OCFO officials also stated that they are not aware of a singular reason for why milestones do not consistently incorporate DHS’s guidance and stated that they have also identified instances in which milestones do not align with the guidance. As our analysis indicates, S&T’s milestones could better incorporate milestone criteria included in DHS’s budget preparation guidance. Without milestone information that more closely aligns with DHS guidance, Congress and DHS decision-makers may not be able to fully assess whether R&D projects are meeting specific goals within assigned time frames or identify what adjustments, if any, may be needed to facilitate the achievement of project goals and the R&D mission overall. Strategic and Management Performance Goals DHS has developed 12 performance goals to assess and report on its R&D efforts, DHS is required to identify department-wide goals in its strategic plan and annual performance report. For fiscal years 2016 through 2018, DHS’s Annual Performance Report included two strategic performance goals related to S&T’s R&D efforts. DHS’s congressional budget justification includes the two strategic performance goals as well as 10 related management performance goals. For a detailed listing of the 12 performance goals, see table 4. Seven of the 10 management performance goals were for S&T R&D efforts and the remaining three were for Domestic Nuclear Detection Office’s R&D efforts, which cover the components that account for 96 percent of DHS’ fiscal year 2017 R&D obligations. DHS has performance goals for mission programs that produce operational results that link directly to the DHS Strategic Plan, according to officials from the OCFO’s Program Analysis and Evaluation division. DHS also uses milestones to track the progress of the other components’ R&D efforts. R&D Customer Feedback DHS components that conduct R&D use various methods to collect and analyze customer feedback to assess their R&D efforts, as shown in table 5. However, DHS is not well positioned to integrate the results because limited customer feedback information is collected and analyzed. Six DHS components that have R&D-related responsibilities evaluate customers’ needs and improve customer satisfaction by listening to customers’ feedback about the quality of deliverables they receive—both good and bad— and making changes necessary to enhance that deliverable. Specifically, officials from S&T, the Coast Guard, CWMD, TSA, the Cybersecurity and Infrastructure Security Agency, and the Secret Service stated they have varying methods in place for gathering customer feedback regarding the progress and the results of R&D activities and deliverables. Below is a summary of these components’ efforts to consider customer feedback. S&T. S&T’s project management guide outlines a process for ensuring customer requirements are being adequately met using a customer survey that can be modified and provided to the customer to complete at each major milestone. In addition, proceedings (e.g., minutes) from regularly scheduled meetings with customer and end user groups may be used to gather information regarding value and operational impact in lieu of a survey. The S&T survey asks customers to rate their overall satisfaction with S&T products and services, along with specific aspects of support, such as providing products in time to meet needs and effectively keeping customers informed. However, out of the 97 R&D activities that S&T reported in fiscal year 2017 and the 110 activities in fiscal year 2016, S&T collected one customer survey form. Coast Guard. The Coast Guard also has a process in place for surveying and interviewing its customers following the completion of an R&D project and officials reported using this information for future R&D planning. The Coast Guard’s survey instrument seeks feedback on: customer satisfaction, timeliness, utility, and communications, among other things. The customer service survey is distributed for feedback on deliverables. At least 6 months after an R&D project is completed, Coast Guard also conducts an in-person interview with project sponsors to collect project transition performance success and feedback information. The surveys that Coast Guard uses to obtain feedback elicit a relatively low number of responses from customers, significantly limiting their usefulness in soliciting feedback data. Specifically, the response rates for fiscal years 2013-2017 were 16%, 17%, 27%, 13%, and 17%, respectively. Experts on customer satisfaction measurement have stated that although survey response rates are never 100 percent, an organization should strive to get its rate as close as possible to that number. They suggest that ideally, organizations can obtain response rates of over 70 percent. CWMD. CWMD does not have a formal mechanism, such as standard processes and procedures, for collecting and analyzing customer feedback. However, CWMD officials stated that certain informal mechanisms are used to collect customer feedback. For example, CWMD officials reported that the CWMD Office of Policy, Plans, Analysis, and Requirements Directorate communicate with customers and gather customer needs and requirements. In addition, as part of these informal mechanisms, internal and external reviews feedback may be obtained from eventual end users of the R&D technology such as operators from CBP, the U.S. Coast Guard, and the TSA, according to CWMD officials. Cybersecurity and Infrastructure Security Agency. The directorate does not have a formal mechanism for collecting and analyzing customer feedback. However, periodic control gates are used to gather customer feedback, according to directorate officials. The input received during these reviews is used to make corrective actions and manage R&D efforts as necessary. For example, according to directorate officials, they conduct a comprehensive review of R&D coordination efforts annually to determine what was effective and what can be improved. TSA. TSA does not have a formal mechanism for collecting and analyzing customer feedback. However, according to TSA officials, informal feedback may be obtained through review of the weekly reports and meetings regarding recent developments and project milestones. In addition, feedback may be obtained during quarterly program management reviews, third party project development, and certification testing. Secret Service. The Secret Service does not have a formal mechanism for collecting and analyzing customer feedback. However, according to Secret Service officials, informal feedback may be obtained in conjunction with other related internal review activities, including program management reviews. To formalize and improve customer feedback processes for R&D efforts, the National Academy of Sciences has stated that feedback from both R&D failures and successes may be communicated to stakeholders and used to modify future investments. Research on leading practices in the area of customer satisfaction suggests that multiple approaches are needed to effectively listen to customers about their perceptions of quality service and needs. The research also points to a need for centrally integrating all customer feedback so that managers can achieve a better understanding of customers’ perceptions and needs. Also, we have previously reported that leading organizations combine quantitative and qualitative listening tools to obtain customer feedback and then centrally integrate the data in one location. Such approaches include the following: Customer satisfaction surveys. We previously reported that most major organizations use tools such as surveys to periodically capture customers’ overall perceptions about their organization and to measure satisfaction with specific transactions soon after they occur. These surveys can be administered through the mail, by telephone, in person, or electronically. Benchmark surveys. Benchmark surveys gather perceptions of performance from the entire market. These surveys usually gather customer perceptions of performance about top competitors in an industry. This allows the company to examine its customer-perceived strengths and weaknesses in the overall marketplace. While continuous improvement may be a result of this listening tool, the real value, according to the research in this area, comes from breakthrough thinking to gain a sustainable advantage. Focus groups. Organizations use focus groups to get better information from customers than survey results provide. In these groups, customers are probed about why they answered survey questions the way they did. Customer interviews. Conducting interviews with customers can provide a way to get very detailed information about their specific needs and problems. Like focus groups, this tool is used by leading customer service organizations to probe survey respondents as to why they answered survey questions a certain way. The National Academy of Sciences have stated that evaluating the relevance and impact of R&D is a key stage of the R&D process and that measuring the impact of R&D activities requires looking to the end users and stakeholders for an evaluation of the impact of a research program, such as through polling or systematic outreach. In addition, Standards for Internal Control in the Federal Government calls for entities to determine an oversight structure to fulfill responsibilities that are set forth by feedback from key stakeholders, among other things. As a result of the limited customer feedback information that is collected and analyzed, DHS is unable to more fully understand its customers’ perceptions and experience to allow it to assess the benefits and performance of its R&D efforts. Moving forward, standard processes and procedures for collecting and analyzing R&D customer feedback would help in assessing R&D efforts. Conclusions Since 2010, DHS has obligated more than $10 billion dollars on R&D to develop technologies to support DHS’s efforts to prevent, mitigate, and recover from terrorist and natural threats. S&T officials indicated that they have strong collaborative relationships with components; however, it is important that required components fully participate in the IPT process in order for S&T to maintain visibility of R&D projects and successfully fulfill its statutory role of coordinating R&D and to help reduce the risk of potential duplication of R&D efforts across the department. Furthermore, S&T faces challenges and limitations related to the collection, integration, and comprehensiveness of information on R&D projects. Without a mechanism that aligns information sources and results in comprehensive and accurate data, among other things, DHS may not have the information it needs to make informed decisions about R&D investments. S&T also does not fully leverage existing guidance when developing milestones for R&D efforts. Without milestone information that more fully aligns with DHS criteria, Congress and DHS decision-makers may not have a full understanding of R&D progress and challenges. Finally, standard processes and procedures for collecting and analyzing R&D customer feedback would help to assess its R&D efforts. Recommendations for Executive Action We are making the following four recommendations to the Deputy Secretary of the Department of Homeland Security: The Deputy Secretary of the Department of Homeland Security should ensure that all components adhere to IPT participation requirements, in accordance with the DHS directives. (Recommendation 1) The Deputy Secretary of the Department of Homeland Security should develop a mechanism that aligns processes and information sources for collecting R&D project data from DHS components to ensure that the information can be collected, integrated and result in a comprehensive accounting of R&D projects DHS-wide. (Recommendation 2) The Deputy Secretary of the Department of Homeland Security should direct OCFO program officials to ensure that S&T take steps to more fully incorporate leading practices, such as those included in DHS’s budget preparation guidance, into R&D milestones. (Recommendation 3) The Deputy Secretary of the Department of Homeland Security should develop standard processes and procedures for collecting and analyzing customer feedback, applicable to components conducting R&D, for improving the usefulness of existing customer feedback mechanisms to assess R&D efforts and for implementing such mechanisms where absent. (Recommendation 4) Agency Comments and our Evaluation We provided a draft of this report to DHS for review and comment. DHS provided written comments which are reproduced in appendix II. In its comments, DHS concurred with our recommendations and described actions planned to address them. S&T, OCFO, CBP, the Cybersecurity and Infrastructure Security Agency, and CWMD also provided technical comments, which we incorporated as appropriate. With regard to our first recommendation, that the Deputy Secretary of the Department of Homeland Security should ensure that all components adhere to IPT participation requirements, in accordance with DHS directives, DHS stated that S&T’s Office of Science & Engineering will revise the relevant DHS directive to require participation in the IPT process by all components. DHS estimated that this effort would be completed by December 31, 2019. This action, if fully implemented, should address the intent of the recommendation. With regard to our second recommendation, that the Deputy Secretary of the Department of Homeland Security should develop a mechanism for collecting R&D project data in order to complete a comprehensive accounting of R&D projects DHS-wide, DHS stated that S&T’s Office of Science & Engineering will revise the relevant DHS directive to include requirements for data collection on all R&D projects across DHS to ensure alignment of the appropriate data elements and existing guidance. DHS estimated that this effort would be completed by December 31, 2019. This action, if fully implemented, should address the intent of the recommendation. With regard to our third recommendation, that the Deputy Secretary of the Department of Homeland Security should direct OCFO program officials to ensure that S&T take steps to more fully incorporate leading practices, such as those included in DHS’s budget preparation guidance, into R&D milestones, DHS stated that the OCFO will continue to work with S&T to incorporate the leading practices and that the OCFO will validate all S&T annual budget submissions and provide S&T feedback, as appropriate. DHS estimated that this effort would be completed by April 30, 2020. This action, if fully implemented, should address the intent of the recommendation. With regard to our fourth recommendation, that the Deputy Secretary of the Department of Homeland Security should standardize processes for collecting and analyzing customer feedback to aid in assessing R&D efforts, DHS stated that S&T’s Office of Science & Engineering will revise the relevant DHS directive to incorporate customer feedback procedures into the IPT process for the recipients of R&D programs. DHS estimated that this effort would be completed by December 31, 2019. This action, if fully implemented, should address the intent of the recommendation. We are sending copies of this report to the appropriate congressional committees and the Secretary of Homeland Security. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff members have any questions about this report, please contact William Russell at (202) 512-8777 or russellw@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. Key contributors to this report are listed in appendix III. Appendix I: Overview of the Science and Technology Directorate’s Research and Development Projects The following appendix provides a general overview of the Department of Homeland Security’s (DHS) Science and Technology Directorate (S&T) research and development (R&D) projects and programs that support the following homeland security mission areas: (1) border security; (2) disaster resilience; (3) critical incidents, and (4) cybersecurity. The examples provided below are illustrative and therefore not intended to provide a comprehensive list of DHS R&D programs or projects. S&T and the U.S. Coast Guard conducted drone demonstrations and tests at two sites in Mississippi to test unmanned aerial systems before they are deployed to the field. The Technical Assessment of Counter Unmanned Aerial Systems Technologies in Cities is designed to help public safety and industry officials identify potential methods for countering nefarious uses of small unmanned aerial systems. The Wall System Design Support Tool Independent Verification and Validation seeks to strengthen the U.S. Border Patrol’s decision analysis model used to identify the areas of the border where a wall would be most beneficial. The Border Research in Instrumented Construction Project is designed to identify cameras, sensors and other technology that can be applied on or near a smart wall via ground, surface/air, subsurface and water to enhance border security and agent safety. The Apex Border Situational Awareness program aims to help U.S. Customs and Border Protection access more data sources, develop decision support tools and share information with partner law enforcement agencies to improve situational awareness. The Integrated Maritime Domain Enterprise is a platform that seeks to bridge disparate data systems to make it easier for DHS components to share information and collaborate. The Adaptive Sensor Analytics Project aims to provide automated data analytics to process satellite imagery, identify patterns of nefarious activity and alert DHS officials. Ground-Based Technologies Program seeks to improve the ability to detect illegal activity at the border through stronger situational awareness, automated detection and alerts, target classification and tools to promote agent safety Air-Based Technologies Program is designed to identify, test and evaluate unmanned and manned aircraft platforms and sensors for law enforcement, search and rescue, and disaster response in both land and maritime environments. The Expert Tracker training program aims to help U.S. Border Patrol agents improve their ability to track movement in rough terrain along the nation’s borders. Port of Entry Based Technology Program seeks to improve illicit cargo detection and legitimate cargo throughput by upgrading legacy scanning systems and linking them to new analysis and information sharing tools that may make the most of personnel resources. The Port of Entry People Screening Program aims to identify, evaluate and implement combinations of process and technology improvements that facilitate the movement of people through the nation’s air, land and sea ports of entry. Autopsy is an open-source digital forensics platform that seeks to help law enforcement determine how electronic devices were used in a crime and recover evidence. Voice Forensics aims to help identify individuals who make hoax rescue calls to the U.S. Coast Guard, which may make it easier to find and prosecute suspects. Child Exploitation Image Analytics seeks to reduce the amount of time it takes to identify and rescue children from exploitation, as well as identify perpetrators, through automated face recognition algorithms and forensic tools. The Tunnel Detection and Surveillance Program is designed to help border officials detect and locate clandestine tunnels, as well as gather forensic data to support investigation and prosecution of drug smuggling activities. The Port of Entry Forensics and Investigations Program aims to help combat transnational crime and investigate child exploitation and human trafficking through open source data and forensic analysis of material collected from suspicious packages and cargo. DHS S&T seeks to help improve community resilience to natural disasters through technology and tools that support planning, decision making and mitigation efforts. The Canada-U.S. Enhanced Resiliency Experiment series aims to use real-world exercises to demonstrate that seamless communication is possible between responders on either side of the northern border during a large-scale emergency. DHS S&T and the Central United States Earthquake Consortium are developing a suite of decision support tools designed to help emergency managers analyze data used when planning, managing, coordinating and communicating during natural disasters The Mutual Aid Resource Planner is a prototype application designed to help jurisdictions develop more accurate resource plans by incorporating custom data on geospatial hazards, risk assessments and potential mutual aid partners. The National Mutual Aid Technology Exercise seeks to test existing mutual aid systems to improve users’ ability to exchange information between systems in real time and develop technical guidance for future use. The Coastal Resilience Center of Excellence aims to conduct targeted research and education to address key challenges facing coastal communities in the United States, including storm surge modeling, pre-disaster planning, communicating risk and more. The Flood Apex Program is designed to help identify and develop technology that can reduce flood-related fatalities and property loss, increase community resilience and improve flood preparation, response and recovery. The Internet of Things Low Cost Flood Inundation Sensors project seeks to develop and test sensor technology that can provide real- time updates on rising water levels. The Kentucky Dam Safety project aims to create technology and processes to better monitor dams and alert communities of potential danger, reducing loss of life and property. The Advanced CIRCulation modeling tool seeks to accurately predict coastal flooding threats to help emergency managers better coordinate evacuation and response. The Hurricane Evacuation -eXtended platform is a decision support tool for emergency managers designed to organize and stage resources for hurricane response. The Simulation-Based Decision Support System for Water Infrastructural Safety Lite™ tool is designed to quickly model the effects of potential dam breaks, helping officials develop accurate emergency response plans and anticipate evacuation needs. The Tunnel Plug is an inflatable device that aims to seal off subway tunnels to prevent water from flowing into the system, minimizing damage to critical transportation systems. The Linking the Oil and Gas Industry to Improve Cybersecurity project seeks to facilitate cooperative research, development, testing and evaluation procedures to improve cybersecurity in petroleum industry digital control systems The Homeowner Flood Insurance Roundtable aims to help reduce future uninsured flood losses by identifying decision support and research and development needs. The Automated National Structures Inventory project is seeking to build a comprehensive list of private and commercial property at risk for flood damage, which may help promote proper insurance and more effective flood protection efforts. The Smart Cities IoT innovation project is designed to help first responders improve their situational awareness through advances in autonomous drone navigation, intelligent building sensors and body- worn interoperability platforms. The Wireless Emergency Alerts Research, Development, Testing and Evaluation program aims to inform changes to the Federal Communications Commission’s alerting system, including increased character length and adding URLs, pictures, videos and geo-targeting capabilities. The System Assessment and Validation for Emergency Responders program seeks to evaluate available responder technology on affordability, usability, and other criteria to help agencies understand which equipment will best fit their needs. The Urban Operational Experimentation program is designed to let responders test new technologies in real-world settings, and may help provide developers with direct feedback on how their products can better meet operational needs. The Enhanced Dynamic Geo-Social Environment training platform is a free virtual tool that aims to allow responders to practice responding to an active shooter incident, whether within a single agency or with multiple jurisdictions and disciplines. The Surface Transportation Explosives Threat Detection program is aiming to develop screening technology that can identify potential threats on people and in their bags without physically interacting with them. The Explosives Detection Canine program is designed to help detection canine teams identify new explosive compounds through non-hazardous training aids and increase their proficiency through realistic self-assessment and training events across the country. The Datacasting project aims to help responders send encrypted video, data files, and other critical information through existing public broadcast television signals, which helps prevent other communication channels from being overwhelmed. The Next-Generation Incident Command System, a web-based platform, seeks to allow responders to share data and request assistance in real-time, and also allows officials to observe and make critical decisions during evolving situations to better support preparation, response, and recovery. The Android Team Awareness Kit, a free app, is designed to help responders visually track team members and assets in real time during an incident, as well as share encrypted data across jurisdictions, disciplines, and components. The Assistant for Understanding Data through Reasoning, Extraction and Synthesis platform aims to help responders overcome information overload by providing actionable insight based on up-to-the-minute sensor data. The First Responder Electronic Jamming Exercise seeks to identify mitigation tactics against intentional or accidental communications jamming, which responders were able to practice implementing in realistic scenarios. The Telephony Denial of Service program is designed to help improve 911 emergency call centers’ ability to defend against attacks through cyber security technologies that can analyze incoming calls and may help determine potential threats in real time. The Finding Individuals for Disaster and Emergency Response is designed to detect human heartbeats under up to 30 feet of rubble, which may help responders more effectively target rescue efforts. The Rapid DNA technology can complete a DNA test within 90 minutes or less from the field, which seeks to help officials identify victims and inform family members in a timely manner. The Forensic Video Exploitation and Analysis tool aims to help responders quickly analyze video to identify potential suspects by allowing users to tag a person to a left-behind item and reconstruct that individual’s path across multiple camera views. The Cyber Risk Economics program seeks to fund applied R&D, knowledge products by gathering stakeholders across government, industry and academia to discuss cyber risk economics capability gaps and needs. Through these stakeholder discussions, along with scholarly cybersecurity economics research literature reviews and authoritative U.S. federal government documents, DHS S&T developed the newly released Cyber Risk Economics Capability Gaps Research Strategy which aims to consider business, legal, technical and behavior factors impacting cyber risk. Appendix II: Comments from the Department of Homeland Security Appendix III: GAO Contact and Staff Acknowledgements GAO Contact William Russell, (202) 512-8777 or russellw@gao.gov. Staff Acknowledgements In addition to the contact named above, Ben Atwater (Assistant Director), Melissa Hargy (Analyst-in-Charge), Nanette Barton and Gary M. Malavenda made key contributions to this report. In addition, key support was provided by Chris P. Currie, Dominick Dale, Michele Fejfar, Richard Hung, Benjamin Licht, John Mingus, Janet Temko-Blinder, and Sarah Veale.

Why GAO Did This Study Conducting R&D on technologies is vital to enhancing the security of the nation. The Homeland Security Act of 2002, as amended, designates S&T as responsible for coordinating all R&D activities of DHS. Questions have been raised about S&T's ability to demonstrate the impact of its R&D investments. Since DHS began operations in 2003, GAO has made recommendations to help improve DHS's efforts to coordinate and oversee R&D. GAO was asked to review DHS's R&D efforts. This report examines (1) how much DHS has obligated for R&D and what types of R&D DHS conducts, (2) to what extent S&T coordinates R&D across DHS, and (3) how, if at all, DHS identifies and tracks R&D efforts. GAO reviewed documentation from DHS related to the conduct, coordination, tracking, and evaluation of R&D projects. GAO interviewed DHS officials with responsibilities related to, among other things, R&D financial reporting, performance evaluation, and the IPT process, including officials from the 10 DHS components that participate in the IPTs. GAO also reviewed DHS R&D budget and obligation data from fiscal years 2010 through 2017. What GAO Found The Department of Homeland Security (DHS) obligated more than $10 billion for research and development (R&D) from fiscal years 2010 through 2017. Seven DHS components have budget authority to conduct R&D, and the Science and Technology Directorate (S&T) obligated nearly 80 percent of all DHS R&D funds during this time period. These components conduct a wide range of R&D, from cybersecurity to border security projects. S&T generally leads or funds R&D projects by providing technology and knowledge products to support four homeland security mission areas: Disaster resilience . Improving community resilience to natural disasters through technology and tools; Critical incidents . Improving response technological capabilities; Border security . Improving the nation's ability to detect, interdict and prosecute illegal activity across air, land and sea. Cybersecurity . Developing technologies and tools to secure systems and critical infrastructures against cyberattacks. S&T strengthened its R&D coordination efforts across DHS, but some challenges remain. In 2015, DHS established an R&D coordination mechanism, to be led by S&T, and in 2017 issued R&D coordination-related guidance. Specifically, to improve coordination, DHS established an Integrated Product Team (IPT) process to serve as the key R&D coordination mechanism within DHS. All ten DHS components that GAO interviewed stated that the IPT process improved visibility into DHS R&D efforts. However, the component that obligated approximately 17 percent of DHS R&D funds in 2017, or $176 million, did not participate in the IPT process in 2018, as required. Nonparticipation poses a risk to R&D coordination efforts across DHS, including R&D project information not being shared among components. Furthermore, ensuring that all required components participate in the IPT process would help S&T maintain visibility of R&D projects in order to fulfill its statutory role of coordinating R&D, and mitigate the risk of potential duplication of effort. S&T, in its coordination role for DHS, uses disparate information sources to identify and track R&D project information and faces challenges to track progress and other information for ongoing R&D projects. For example, R&D project information is stored in multiple information sources leading to difficulty in integrating complete R&D project information and resulting in reporting that is not comprehensive. By developing a mechanism to address these challenges, S&T can further improve its efforts to report and analyze R&D project information, and have improved visibility on R&D efforts across DHS. GAO also identified challenges in collecting information related to R&D performance. Among other things, DHS is not well positioned to integrate the results and share lessons learned because limited R&D customer feedback information is collected and analyzed. Of the seven DHS components with R&D budget authority, two reported having formal customer feedback mechanisms. As a result, DHS is unable to more fully understand its customers' perceptions and experience which would allow DHS to better assess the performance of its R&D efforts. What GAO Recommends GAO is making four recommendations, with which DHS concurred, including that DHS: 1) ensure all components participate in the IPT process, 2) develop a mechanism that aligns R&D project tracking sources, and 3) collect feedback from R&D customers.

gao_GAO-19-43

gao_GAO-19-43_0

Background EXIM is an independent executive branch agency and a wholly owned U.S. government corporation. EXIM is the official export credit agency (ECA) of the United States, and its mission is to support the export of U.S. goods and services, thereby supporting U.S. jobs. EXIM’s charter states that it should not compete with the private sector. Rather, EXIM’s role is to assume the credit and country risks that the private sector is unable or unwilling to accept, while still maintaining a reasonable assurance of repayment. EXIM must operate within the parameters and limits authorized by law, including, for example, statutory mandates that it support small business and promote sub–Saharan African and environmentally beneficial exports. In addition, EXIM is authorized to provide financing on a competitive basis with other ECAs and must submit annual reports to Congress on its actions. EXIM operates under the leadership of a president who also serves as Chairman of EXIM’s Board of Directors. The board is structured to include five members. All positions are appointed for 4-year terms by the President of the United States with the advice and consent of the Senate. The board is responsible for adopting and amending bylaws for the proper management and functioning of EXIM. Furthermore, the board approves EXIM’s financing either directly or through delegated authority. On May 8, 2019, the Senate confirmed a new president and two other board members, ending the lack of a quorum needed to approve transactions over $10 million that had existed since July 20, 2015. EXIM’s organizational structure includes various offices and divisions operating under its president. The Office of Board Authorized Finance is subdivided into business divisions that are responsible for underwriting related to loans and loan guarantees, including processing applications, evaluating the compliance of transactions with credit and other policies, performing financial analyses, negotiating financing terms, coordinating and synthesizing input to credit recommendations from other divisions, and presenting credit recommendations for approvals. EXIM facilitates support for U.S. exports through three major products: (1) loans; (2) loan guarantees, which include working capital guarantees; and (3) export credit insurance. All EXIM obligations carry the full faith and credit of the U.S. government. Based on its mission to support U.S. employment, EXIM currently requires a certain amount of U.S. content for an export contract to receive EXIM financing. EXIM’s loans generally carry fixed-interest rate terms under the Arrangement on Officially Supported Export Credits negotiated among OECD members. EXIM’s loan guarantees cover loans disbursed by private lenders by committing to pay the lenders if the borrower defaults. Both loans and loan guarantees may be classified as short-, medium-, or long-term. From fiscal year 2008 to fiscal year 2017, EXIM was “self-financing” for budgetary purposes—financing its operations from receipts collected from its customers—and operating within the parameters and limits authorized by Congress. However, according to EXIM, because of the lack of quorum on the Board of Directors, in fiscal year 2018 it was unable to approve transactions over $10 million and, as a result, was not able to generate sufficient cash inflows to fully self-finance program and administrative costs. EXIM reported that when it is back to being fully operational, it plans to regain full self-financing status. See figure 1 for additional details on EXIM’s loans and loan guarantees. Short-term loans and loan guarantees: Short-term financing consists of all transactions with repayment terms of less than 1 year, while Working Capital Guarantee program short-term financing may be approved for a single loan or a revolving line of credit that can be renewed for up to 3 years. In general, if the financed eligible product contains at least 50 percent U.S. content, then the entire transaction value is eligible for a working capital guarantee. Generally, for working capital guarantees, EXIM guarantees 90 percent of the loan’s principal and interest if the borrower defaults. Therefore, the lender maintains the risk of the remaining 10 percent. EXIM’s payment of working capital claims is conditional upon transaction participants’ compliance with EXIM requirements such as underwriting policies, deadlines for filing claims, payment of premiums and fees, and submission of proper documentation. EXIM has reported that over 80 percent of its working capital guarantee transactions are approved by lenders with delegated authority, which means that commercial lenders approve the guaranteed loans in accordance with agreed- upon underwriting requirements without first obtaining EXIM approval. If a lender does not have delegated authority, EXIM performs its own underwriting procedures and approves the guaranteed loans. Medium- and long-term loans and loan guarantees: For medium- and long-term loan and loan guarantee transactions, EXIM provides up to 85 percent financing with the remaining 15 percent paid by the borrower or financed separately. The financing could be less than 85 percent depending on the U.S. content. EXIM’s medium- and long- term loan guarantees generally cover 100 percent of the financed amount if the borrower defaults. EXIM’s guarantee to the lender is transferable and unconditional, meaning that EXIM must pay submitted claims regardless of the cause of default. EXIM generally underwrites medium- and long-term loans and loan guarantees for $10 million and less, and EXIM officials with delegated authority approve the transactions. Further, EXIM has provided certain lenders delegated authority to underwrite and approve these guarantees. EXIM underwrites long-term loans and loan guarantees greater than $10 million, and its Board of Directors approves the transactions. EXIM’s Credit Authorizations As noted earlier, EXIM’s authority to approve transactions lapsed from July 1, 2015, to December 4, 2015. Further, from July 20, 2015 to May 8, 2019, EXIM’s Board of Directors lacked a quorum, and, as a result, EXIM was unable to approve transactions greater than $10 million. Consequently, EXIM’s annual authorizations for loans, loan guarantees, and export credit insurance decreased from about $20 billion in 2014 to about $3 billion in 2018, a decrease of about 83 percent. See figure 2 for EXIM’s total authorizations by type and length of term for 2014 through 2018. EXIM’s Policies and Procedures for Extending Credit EXIM’s Manual describes EXIM’s underwriting policies and procedures for each of its products offered, including short-, medium-, and long-term loans and loan guarantees. The Manual describes the responsibilities of EXIM’s divisions (e.g., Transportation, Structured and Project Finance, or Working Capital Finance) involved in the underwriting process. EXIM’s Office of Board Authorized Finance is in the process of streamlining the Manual, which is over 1,400 pages. A goal of this process is to separate procedures from policies, thus allowing for policies and procedures to be continuously reviewed. An EXIM official told us that these steps should improve the agency’s efficiency, transparency, and accountability. The underwriting sections of the Manual are tentatively scheduled for review in 2019. Overview of EXIM’s Underwriting Process EXIM loan officers perform the underwriting for loans and long-term loan guarantees. The underwriting of medium-term or working capital loans guaranteed by EXIM is performed by either EXIM loan officers or qualified lenders with delegated authority, which allows the lender to authorize a loan that EXIM guarantees in accordance with agreed-upon underwriting requirements without first obtaining EXIM approval. When the underwriting and credit decision is delegated to approved lenders, EXIM does not perform the underwriting procedures. EXIM’s underwriting process calls for thorough credit assessments by subject matter experts and loan officers. These assessments evaluate key transactional risks, such as the borrower’s industry, competitive position, operating performance, liquidity position, leverage, and ability to service debt obligations. Frequently, credit enhancements are included in the structure of long-term financing (often in the form of collateral) in order to decrease the risk of a borrower default but also to increase the recovery in the event of a default. A risk rating is assigned to the transaction based on this evaluation which, in turn, determines the transaction fee that a borrower pays and assists in establishing the level of loss reserves EXIM must set aside. The credit assessments undergo multiple levels of internal review. All transactions of EXIM carry some risk; however, transactions approved through delegated authority lenders potentially carry a higher level of inherent risk because third-party financial institutions make the decisions. To mitigate the risk, EXIM reviews medium-term transactions approved by delegated authority lenders before the transactions are executed to assure compliance with EXIM’s delegated authority lending policies. For working capital guarantee delegated authority, EXIM conducts periodic examinations of the lenders, reviewing ongoing transactions and lender compliance with the delegated authority program. The examinations are intended to identify lenders that are not satisfactorily managing the requirements of the delegated authority program. To mitigate the risk for its internal credit process, EXIM developed and documented underwriting processing steps from the time the application is received through the approval of the appropriate credit structure. These steps serve to (1) establish a framework for sound credit decisions, (2) communicate to EXIM employees the requirements governing the extension of credit, and (3) encourage documentation and the consistent application of EXIM’s credit policies and procedures. According to EXIM officials, the underwriting process also serves as EXIM’s primary method for preventing fraud because of the due diligence performed on the proposed transaction. Figure 3 summarizes EXIM’s underwriting process. Application intake. When an application is initially received, it is screened for basic completeness, follow-up on incomplete or unacceptable applications is performed, and it is assigned to a processing division. Application screening. After an application is determined to be complete, it is assigned to the applicable EXIM division that oversees the applicable type of project. For example, an application for the purchase of an aircraft would be assigned to the Transportation Division. Once assigned, a loan officer in that division is to assess the eligibility of the transaction. To ensure compliance with laws and regulations, the loan officer is to obtain and assess various certifications from transaction participants. Loan officers are also required to submit the corporate and individual names and addresses of lenders, borrowers, guarantors, and other transaction participants to the EXIM Library. Library staff are then to conduct a Character, Reputational, and Transaction Integrity (CRTI) review—a procedure designed to provide a level of due diligence over various risks and to help prevent fraud by checking loan participants’ information against 28 databases. Risk assessment and due diligence. Once the transaction is considered minimally eligible for EXIM support, the loan officer is required to perform a series of due diligence activities to determine (1) whether the transaction provides a reasonable assurance of repayment, (2) any potential material issues regarding the transaction or the participants that would preclude EXIM support, and (3) the appropriate risk level and pricing for the transaction. As part of the financial evaluation of the transaction, the loan officer is required to obtain and analyze the borrower’s financial statements, credit reports or rating agency reports, financial projections, and other relevant information. As applicable, the loan officer is required to obtain input from other EXIM staff, such as attorneys or engineers, to conclude on the legal, technical, economic, or environmental risks of the transaction. Based on this due diligence, the loan officer is to assess the transaction for risk and assign an overall risk rating. This rating is used to calculate the exposure fee EXIM will charge the borrower for guaranteeing the transaction. Greater perceived risks result in higher fees. Credit structure. After the risk assessment and due diligence is performed, the loan officer determines the financing terms and conditions to be recommended. The loan officer is generally required to structure the transaction to include a security interest (collateral) in the financed goods or other assets of the borrower. If it is determined that collateral is not necessary, the loan officer is to document the explanation and mitigating factors (e.g., EXIM support is small relative to a borrower’s size). For all aircraft transactions, the loan officer is required to perform an assessment and loan-to-value analysis of the collateral, and the financing terms must include requirements for the borrower to maintain ownership and condition of the collateral. Credit decision. The loan officer is to document the due diligence in a credit or board memo, which includes the loan officer’s recommendation to approve or decline the transaction. These memos and applicable supporting documentation are then to be forwarded to the approving party. The credit memo applicable to working capital or medium-term transactions is to be provided to EXIM officials with individual delegated authority to approve transactions of $10 million and less. Board memos for long-term transactions or transactions greater than $10 million are to be provided to the EXIM Board of Directors for approval. From July 2015 to May 2019, EXIM lacked a quorum on its Board of Directors, and as a result, EXIM was unable to approve new transactions greater than $10 million. Management of Federal Credit Programs Government-wide guidance for federal agencies to follow for the management and operation of federal credit programs, such as loan and loan guarantee programs, include the following: OMB Circular A-129, Policies for Federal Credit Programs and Non- Tax Receivables, revised in January 2013, describes policies and procedures for designing and managing federal credit programs. The guidance addresses various standards for applicant screening, loan documentation, collateral requirements, determining and monitoring lender and servicer eligibility, and lender and borrower stake in full repayment. In addition, it details risk sharing practices that agencies should follow, such as ensuring that lenders and borrowers who participate in federal credit programs have a substantial stake in full repayment in accordance with the loan contract. Treasury’s Bureau of the Fiscal Service’s Managing Federal Receivables provides federal agencies with an overview of standards, guidance, and procedures for successful management of credit activities, including screening applicants for creditworthiness and financial responsibility, and managing, processing, evaluating and documenting loan applications and awards for loan assistance. Furthermore, it details how federal agencies should manage lenders and servicers that participate in federally insured guaranteed loan programs. EXIM’s Process for Updating Its Underwriting Policies and Procedures Was Properly Designed and Implemented We found that EXIM’s process for updating its underwriting policies and procedures was properly designed and implemented. Standards for Internal Control in the Federal Government states that management should design control activities to achieve objectives and respond to risks. Management’s design of internal control establishes and communicates the who, what, when, where, and why of internal control execution to personnel. Management should clearly document internal control in a manner that allows the documentation to be readily available and properly managed and maintained. Further, management should also implement control activities through policies and periodically review policies, procedures, and related control activities for continued relevance and effectiveness in achieving the entity’s objectives or addressing related risks. Underwriting policies and procedures are documented in EXIM’s Manual, which consists of 26 chapters, covering various topics by product (e.g., long-term loans and loan guarantees) or process (e.g., application intake or credit structure). We found that the Manual provides EXIM’s divisions involved in the underwriting process with direction and guidance for making credit decisions and is to be updated at least annually, except for material changes, which are required to be incorporated as soon as possible. The Credit Policy Division (Credit Policy) maintains and manages the process for updating of the Manual and relies on EXIM’s divisions for additions, updates, and revisions to it. Credit Policy maintains an assignment list of the primary officer, the primary reviewer, and the Office of General Counsel (OGC) reviewer, who are responsible for each chapter in the Manual. Each year, the process calls for Credit Policy to send an email to the primary officer and two reviewers assigned to each chapter. This communication requests that the officer review the assigned chapters for any needed changes. After the assigned officer has reviewed the chapter, if there are no contemplated changes, the primary officer assigned to the chapter is required to notify Credit Policy of this determination by email with the concurrence of the respective primary and OGC reviewers. If changes are needed, the assigned officer is required to provide the updated chapter to Credit Policy by email with the concurrence of the primary and OGC reviewers. According to EXIM’s process, when material changes to the Manual are needed, the necessary revisions do not wait for the annual update. Instead, the responsible division is required to incorporate such changes into the applicable chapter(s) of the Manual and submit them to Credit Policy as soon as possible. EXIM officials stated that examples of material changes that would be addressed immediately include recommendations from oversight bodies, such as the EXIM OIG or GAO, and changes resulting from legislative actions, such as updates to EXIM’s charter or changes in compliance procedures related to sanctions. EXIM’s Underwriting Policies and Procedures Were Mostly Consistent with Federal Guidance, but Could Be Enhanced The underwriting policies and procedures in EXIM’s Manual for its loan and loan guarantee transactions were mostly consistent with OMB and Treasury guidance for managing federal credit programs. We evaluated these policies and procedures for (1) applicant screening, (2) loan documentation, (3) collateral requirements, (4) lender and servicer eligibility, and (5) risk sharing practices. As shown in table 1, EXIM’s underwriting policies and procedures for the loan and loan guarantee programs were consistent with 12 of 15 applicable standards for managing federal credit programs and were partially consistent with three. Three other standards were not applicable to EXIM’s underwriting. EXIM’s Policies and Procedures Were Consistent with Federal Guidance Related to Applicant Screening Applicant screening refers to determining an applicant’s eligibility and creditworthiness for a loan or loan guarantee. Federal guidance for applicant screening includes specific standards related to the applicant’s (1) program eligibility, (2) delinquency on federal debt, (3) creditworthiness, (4) delinquent child support, and (5) taxpayer identification number (TIN). As shown in table 2, EXIM’s underwriting policies and procedures were consistent with federal guidance for applicant screening. For all loan and loan guarantee applications, EXIM requires applicants to provide identifying information, such as name, address, phone number, and Dun & Bradstreet Data Universal Numbering System (DUNS) number. Applicants are also required to provide relevant financial information, such as income, assets, cash flows, liabilities, financial statements, and credit reports. EXIM’s underwriting process requires screening of applicants for eligibility, which is partly completed through the CRTI review. As part of the CRTI review, EXIM screens the corporate and individual names and addresses of lenders, borrowers, guarantors, and other transaction participants against 28 databases that include various U.S. government and international debarment and sanctions lists for red flags. If a match is identified, EXIM’s Credit Review and Compliance Division works with the loan officers to determine the legitimacy of the match and, as necessary, works with OGC to determine what additional due diligence measures may be required and whether to continue the underwriting process. In addition to the CRTI review process, loan officers must obtain and use credit reports to assess creditworthiness and identify whether transaction applicants are delinquent on federal tax or nontax debts, including judgment liens against property for a debt to the federal government, and are therefore not eligible to receive federal loans and loan guarantees. EXIM’s policies and procedures contain instructions to suspend application processing and contact OGC for further guidance upon finding federal debt delinquencies or other insufficient or negative information on applicant credit reports. Loan officers must document any issues encountered on applicant credit reports and explain why a transaction is creditworthy if they recommend it for approval. Lastly, OMB Circular A-129 requires agencies to obtain the TIN of all persons doing business with the agency. The working capital guarantee application form requests the TIN for transaction applicants, which an EXIM official stated are used to obtain applicant credit reports. EXIM does not require the TIN for medium- and long-term applications. EXIM officials stated that applicants for medium- and long-term transactions are likely foreign entities and thus would not have federal TINs. However, all applications request the DUNS number which EXIM must use to perform the credit review and CRTI due diligence procedures. EXIM’s Policies and Procedures Were Consistent with Federal Guidance Related to Loan Documentation Federal guidance calls for the maintenance of loan files containing key information used in loan underwriting. As shown in table 3, EXIM’s underwriting policies and procedures were consistent with the federal guidance related to loan documentation. EXIM’s underwriting policies and procedures state that loan officers must maintain a loan file on the transaction applicant and other participants, which includes the completed application, credit bureau reports, credit analysis, certifications, verifications and other legal documents, and loan or service agreements with the debtor, as appropriate. EXIM’s process calls for obtaining debt collection certification statements for the working capital guarantee applications because the applicants are domestic entities. While the debt collection certification statement is not applicable for medium- and long-term applications, because the applicants are foreign entities, EXIM’s executed credit agreements and promissory notes define the terms of the transactions, including defaults and the remedies EXIM may take, such as declaring default and accelerating debt repayment, and pursuing restructuring or recovery actions, including possible litigation. EXIM’s Policies and Procedures Were Consistent with Federal Guidance Related to Collateral Collateral refers to the assets used to secure a loan. For many types of loans, the government can reduce its risk of default and potential losses through well-managed collateral requirements. However, several of the collateral requirements contained in federal guidance relate specifically to real property. Since EXIM’s mission is to support U.S. exports, it does not finance real property and, accordingly, does not accept real property as the primary collateral. As a result, three of the four federal guidance standards were not applicable to EXIM’s underwriting. As shown in table 4, EXIM’s underwriting policies and procedures were consistent with the applicable federal guidance related to collateral. EXIM’s underwriting policies and procedures state that it should have a security interest in the financed export items. The loan officer and a transaction engineer will evaluate the export sales contracts, and this evaluation will be used as the assessment of collateral for the transaction. If using the financed export items as collateral is not possible, the loan officer should secure the EXIM financing with other assets owned by the primary source of repayment that are at least of comparable value to the financed items. Collateral that could be considered includes fixed assets, inventory, accounts receivable, or a third-party guarantee. While OMB Circular A-129 requires a real property appraisal and contains specific criteria defining acceptable appraisals, the standard was not applicable to EXIM’s loans and loan guarantees. According to EXIM officials, EXIM rarely takes real property as collateral because the primary collateral for EXIM’s transactions is the asset financed, and EXIM does not finance real property. Further, EXIM officials stated that the U.S. appraisal standards cannot be applied to foreign real property. However, if real property is taken as collateral, it would be as secondary or additional collateral. When EXIM accepts real property as additional collateral for a transaction, EXIM officials stated that an independent third-party appraisal in accordance with regional practices is obtained. EXIM’s Policies and Procedures Were Not Fully Consistent with Federal Guidance Related to Lender and Servicer Eligibility Federal guidance calls for policies and procedures related to lender and servicer eligibility, monitoring, and recertification. As shown in table 5, EXIM’s policies and procedures were consistent with three and partially consistent with two of five federal standards for lender and servicer eligibility. OMB Circular A-129 calls for agencies to establish specific procedures to continuously review lender and servicer eligibility and decertify lenders and servicers that fail to meet the agency’s standards for continued participation. EXIM’s policies and procedures related to requirements for working capital guarantee delegated authority lenders were consistent with federal guidance. However, for medium-term delegated authority lenders, EXIM has not established documented policies and procedures for (1) determining their eligibility for continued participation in the program and (2) decertifying or taking other appropriate actions for those that do not meet compliance or eligibility standards. EXIM officials told us that currently EXIM has only three medium-term delegated authority lenders: two were renewed for continued participation and one became inactive in 2018. Further, according to EXIM officials, since 2009 only 2.3 percent of all medium- term guarantee authorizations have been delegated authority authorizations ($71 million out of $3.1 billion). EXIM reviews the performance of its primary medium-term lenders quarterly. In these reviews, EXIM officials evaluate the lenders’ portfolio performance, underwriting capabilities, and a set of qualitative factors. However, without documented policies and procedures for determining the eligibility of the medium-term delegated authority lenders’ continued participation in the program and for decertifying such lenders, as appropriate, EXIM may allow lenders who are not qualified to underwrite transactions, thus increasing the risk for improper underwriting and defaults. EXIM officials stated that they are in the process of updating and enhancing the Manual and will include procedures for medium-term delegated authority lender reviews and the consequences of an unfavorable assessment. EXIM’s Policies and Procedures Were Not Fully Consistent with Federal Guidance Related to Risk Sharing Practices OMB Circular A-129 calls for lenders and borrowers who participate in federal credit programs to have a substantial stake in full repayment but also states that the level of guarantee should be no more than necessary to achieve program purposes. As shown in table 6, EXIM’s underwriting policies and procedures were generally consistent with the federal guidance related to certain risk sharing practices for lenders and borrowers to have a stake in full repayment and were partially consistent with the federal guidance related to periodic program reviews. Although OMB Circular A-129 calls for lenders who extend credit to have substantial stake in full repayment and bear at least 20 percent of any loss from a default, it also states that the level of guarantee should be no more than necessary to achieve program purposes. However, consistent with its charter, EXIM is authorized to provide terms that are competitive with those of other ECAs, such as up to 100 percent loan guarantee coverage. EXIM does not require lenders to bear 20 percent of the risk of default. For working capital guarantees, EXIM offers 90 percent guarantee coverage and lenders retain 10 percent risk. For medium- and long-term loan guarantees, EXIM provides up to 85 percent financing with the remaining 15 percent paid by the borrower or financed separately. EXIM financing could be less than 85 percent depending on the U.S. content. According to EXIM, guaranteeing 100 percent of the amount it finances permits it to explore capital markets and is more desirable to banks for large and long-term projects. As a result, the lender may not retain any risk of default in the transaction. According to an OECD official, guaranteeing 100 percent of the financed amount is consistent with other ECAs. For example, the ECAs of Canada, Germany, and the United Kingdom also provide guarantees up to 100 percent of the financed amount on certain products. OMB Circular A-129 states that borrowers should have equity interest in assets financed with credit assistance and substantial capital or equity at risk in their business. However, consistent with its charter, EXIM is authorized to provide terms that are competitive with those of other ECAs. EXIM does not specifically require borrowers to have an equity interest in the transaction or to contribute the minimum cash payment. EXIM’s policies and procedures state that in practice, buyers often secure alternative financing for the cash payment, which is permissible as long as the financing is not officially supported by EXIM or another U.S. government agency. EXIM officials noted that during the analysis of creditworthiness, loan officers examine supporting documents for the alternative financing to assure that it is not guaranteed by EXIM or another U.S. government agency. OMB Circular A-129 states that the agency should periodically review programs in which the government bears more than 80 percent of any loss. The review is intended to evaluate the extent to which credit programs achieve intended objectives and whether the private sector has become able to bear a greater share of the risk. EXIM officials stated that EXIM performs program reviews through annual budget justifications submitted to OMB and annual competitiveness reports submitted to Congress. EXIM officials also stated that there are established timelines for preparing these reviews that must be followed to ensure that EXIM meets deadlines for submitting its budget documentation and the June 30 deadline for the annual competitiveness report. In addition, EXIM employs a detailed summary of the products and terms that other countries’ official ECAs offer. However, EXIM does not have documented policies or procedures related to performing periodic program reviews. As a result, EXIM runs the risk that it will not effectively review its programs to determine whether the private sector could bear a greater share of the risk. Conclusions EXIM’s Manual provides a framework for making credit decisions so that only qualified applicants that demonstrate reasonable assurance of repayment are provided loans or loan guarantees. This framework helps ensure consistent application of procedures for assessing an applicant’s creditworthiness and for overseeing certain delegated authority lenders. However, EXIM’s underwriting process could be improved by additional procedures. For example, the Manual did not address medium-term delegated authority lenders’ eligibility requirements for continued participation and decertification procedures for lenders who fail to meet agency’s standards. Further, EXIM has not documented its process for periodic program reviews to determine whether the private sector could bear a greater share of the risk. Improvements in these areas could help enhance the oversight of lenders and the usefulness of program reviews. Recommendations for Executive Action We are making the following two recommendations to EXIM: The Chief Operating Officer of EXIM should consider establishing documented policies and procedures for (1) determining medium-term delegated authority lenders’ eligibility for continued participation in EXIM’s programs and (2) decertifying or taking other appropriate actions for such lenders that do not meet compliance or eligibility standards. (Recommendation 1) The Chief Operating Officer of EXIM should establish documented policies and procedures for periodically reviewing credit programs in which the government bears more than 80 percent of any loss to determine whether private sector lenders should bear a greater share of the risk. (Recommendation 2) Agency Comments We provided a draft of this report to EXIM for review and comment. In written comments on a draft of this report, which are reproduced in appendix II, EXIM concurred with our two recommendations. EXIM also provided technical comments that we incorporated into the final report, as appropriate. In its written comments, EXIM described planned actions to address our recommendations. Specifically, EXIM stated that it will consider establishing documented policies and procedures for determining medium-term delegated authority lenders' eligibility for continued participation in EXIM's programs and decertifying or taking other appropriate actions for such lenders that do not meet compliance or eligibility standards. Further, EXIM will establish documented policies and procedures for periodically reviewing credit programs in which the government bears more than 80 percent of any loss to determine whether private sector lenders should bear a greater share of the risk. If implemented effectively, EXIM’s planned actions should address the intent of our recommendations. We are sending copies of this report to appropriate congressional committees, the Chairman of the Export-Import Bank, and the EXIM Inspector General. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-3133 or dalkinj@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix III. Appendix I: Objectives, Scope, and Methodology Our objectives were to determine the extent to which Export-Import Bank’s (EXIM) (1) process for updating its underwriting policies and procedures is properly designed and implemented and (2) underwriting policies and procedures for loan and loan guarantee transactions are consistent with guidance for managing federal credit programs. To assess the extent to which EXIM’s process for updating its underwriting policies and procedures was properly designed and implemented, we reviewed EXIM’s policies and procedures for updating its Loan, Guarantee and Insurance Manual (Manual) and interviewed EXIM officials. We assessed EXIM’s process to determine whether it sufficiently communicated the procedures to be performed and documentation to be prepared and was consistent with Standards for Internal Control in the Federal Government. We did not evaluate EXIM’s compliance with its process for updating its underwriting policies and procedures or assess their operating effectiveness. To assess the extent to which EXIM’s underwriting policies and procedures for loan and loan guarantee transactions were consistent with guidance for managing federal credit programs, we reviewed relevant requirements and guidance, including the Office of Management and Budget’s (OMB) Circular A-129, Policies for Federal Credit Programs and Non-Tax Receivables, and the Department of the Treasury’s Bureau of the Fiscal Service’s Managing Federal Receivables: A Guide for Managing Loans and Administrative Debt. Specifically, we focused on OMB Circular A-129’s Section II (C)(1)(a) through (c), Section III (A)(1) through (3), and Section III (C)(1)(a) through (e)), which contain standards pertinent to risk management for loan and loan guarantee programs, including standards for (1) applicant screening (program eligibility, delinquency on federal debt, creditworthiness, delinquent child support, and taxpayer identification number); (2) loan documentation; (3) collateral (appraisal of real property, loan-to-value ratio, liquidation of real property collateral, and asset management standards and systems for real property disposal); (4) lender and servicer eligibility (participation criteria, review of eligibility, fees, decertification, and loan servicers); and (5) risk sharing practices (private lenders stake in full repayment, borrowers stake in full repayment, and program reviews). From the Bureau of the Fiscal Service’s Managing Federal Receivables, we identified key guidance related to credit extension (ch. 3) and management of guaranteed lenders and servicers (ch. 5). We reviewed EXIM’s policies and procedures related to underwriting for the loan and loan guarantee programs contained in its Manual and other documentation, such as its charter. We also discussed EXIM’s policies and procedures related to underwriting with EXIM officials. We compared EXIM’s underwriting processes to federal guidance for managing federal credit programs. As part of this comparison, we assessed whether policies and procedures included in EXIM’s Manual were consistent with federal guidance. However, because of EXIM’s limited lending authority during the period of our audit, we did not verify EXIM’s compliance with its underwriting policies and procedures or assess their operating effectiveness. In areas where we found EXIM’s policies and procedures to be consistent with federal guidance, there may still be opportunities to improve operating effectiveness. Further, guidance for managing federal credit programs includes additional requirements not related to underwriting, which we did not assess. In addition, we reviewed EXIM’s Office of Inspector General (OIG) reports since 2014 related to underwriting issues, various laws applicable to EXIM, and GAO reports related to EXIM. We also reviewed EXIM’s annual reports and competitiveness reports. We also discussed EXIM underwriting process with EXIM OIG officials and export credit financing and risk sharing practices with an official from the Organisation for Economic Co-operation and Development. We conducted this performance audit from January 2017 to May 2019 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Comments from the Export- Import Bank of the United States Appendix III: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Marcia Carlsen (Assistant Director), Dragan Matic (Analyst in Charge), Sarah Lisk, Erika Szatmari, and Jingxiong Wu made key contributions to this report.

Why GAO Did This Study EXIM serves as the official export credit agency of the United States, providing a range of financial products to support the export of U.S. goods and services. Following the 2007–2009 financial crisis, demand for EXIM support increased. However, from July 2015 to May 2019, EXIM lacked a quorum on its Board of Directors and, as a result, was unable to approve medium- and long-term transactions greater than $10 million. The Export-Import Bank Reauthorization Act of 2012 includes a provision for GAO to evaluate EXIM's underwriting process. This report discusses the extent to which EXIM's (1) process for updating its underwriting policies and procedures is properly designed and implemented and (2) underwriting policies and procedures for loan and loan guarantee transactions are consistent with guidance for managing federal credit programs. To address these objectives, GAO evaluated EXIM's underwriting policies and procedures against federal guidance and discussed the underwriting process with EXIM officials. What GAO Found GAO found that Export-Import Bank's (EXIM) process for updating its underwriting policies and procedures was properly designed and implemented. EXIM's Loan, Guarantee and Insurance Manual (Manual) describes EXIM's underwriting policies and procedures for its short-, medium-, and long-term loans and loan guarantees. The Manual describes the responsibilities of divisions and loan officers involved in the underwriting process and is required to be updated at least annually, except for material changes (e.g., changes resulting from legislative actions or compliance with sanctions), which are required to be made as soon as possible. EXIM has initiated a process to streamline the Manual, which consists of over 1,400 pages, by separating the policies and procedures, thus allowing for continuous reviews. The underwriting sections of the Manual are tentatively scheduled for review in 2019. The primary guidance for designing and managing federal credit programs is Office of Management and Budget Circular A-129, Policies for Federal Credit Programs and Non-Tax Receivables . GAO found that EXIM‘s policies and procedures were consistent with three of five areas of federal guidance; two areas related to lender and servicer eligibility and risk sharing practices were partially consistent with federal guidance. Applicant screening. EXIM's policies and procedures were consistent with guidance in that they require applicants to provide relevant financial information and assessments of applicant eligibility and creditworthiness. Loan documentation. EXIM's process was consistent with guidance in that it requires the preparation of loan files, which include the application, credit reports, and related analyses, as well as collateral documentation and loan agreements. Collateral requirements. EXIM's process was consistent with guidance in that it requires a security interest in the financed export items. Lender and servicer eligibility. EXIM established eligibility and decertification procedures for short-term delegated authority lenders that were consistent with guidance. However, it did not establish similar procedures for medium-term delegated authority lenders. Risk sharing practices. EXIM's process was generally consistent with guidance in that EXIM provides loan guarantee terms that officials stated were necessary to achieve program purposes. However, federal guidance also calls for an agency to periodically review programs in which the government bears more than 80 percent of any loss. While EXIM prepares various program reviews, it has not developed procedures to help ensure that its risk sharing practices are routinely reviewed. Without enhancements to its policies and procedures, EXIM may allow lenders that are not qualified to underwrite transactions and runs the risk that it will not effectively review its programs. What GAO Recommends GAO is making two recommendations to enhance EXIM's policies and procedures related to (1) the use of medium-term delegated authority lenders and (2) periodic program reviews. EXIM concurred with GAO's recommendations and described actions planned to address them.

gao_GAO-18-208

gao_GAO-18-208_0

Background VA is responsible for providing a variety of services to veterans and their families (i.e., spouses and children), including health care, disability compensation, and vocational rehabilitation. Within the department, VHA oversees the delivery of health care services, including primary care, specialized care, and related medical and social support at its more than 1500 medical facilities located throughout the country. As of fiscal year 2016, about 9 million veterans were enrolled in the VA health care system, with almost 7 million patients receiving services at its medical facilities each year. Overview of VistA VHA relies on VistA—its health information system—to assist in the daily operations of providing health care to patients. VistA began operation in 1983 as the Decentralized Hospital Computer Program. In 1996, the department changed the name of the system to the Veterans Health Information Systems and Technology Architecture—VistA. The system is comprised of more than 200 different software applications, including 17 pharmacy applications; 11 laboratory applications; 10 eligibility, enrollment, and registration applications; and 12 financial management applications. Most VistA applications are based on an architecture that links servers and personal computer workstations at VA facilities. VistA also has interfaces with applications within other VA systems, as well as selected systems of other federal agencies (e.g., DOD health information systems used to treat injured service members) and private care providers and pharmacies. VistA was developed based on the collaboration of staff in the VA medical facilities and VHA IT personnel, with the intention of providing a system that would meet the clinicians’ needs. Specifically, clinicians and IT personnel in the various medical facilities collaborated to define the system’s requirements and, in certain cases, carried out its development and implementation. In this regard, staff at a medical center could develop and implement applications at the local level to facilitate the potentially different functions at each location. This approach has resulted in about 130 different instances, or variations, of the system being used throughout the department’s medical facilities. VA has made numerous enhancements to the functionality of VistA since 1983. A significant example is the release in 1996 of the Computerized Patient Record System (CPRS), a graphical user interface that enabled the department to provide an individual electronic health record for each VA patient. Specifically, CPRS enables clinicians to enter, review, and continuously update information connected with a patient. Among other things, clinicians can order lab tests, medications, diets, radiology tests, and procedures; record a patient’s allergies or adverse reactions to medications; request and track consults; enter progress notes, diagnoses, and treatments for each encounter; and enter discharge summaries. Another example of the enhancements made to VistA was the department’s implementation of an imaging capability (VistA Imaging) at all of the medical facilities. This capability enabled multimedia data, such as radiology images, to be linked to a patient’s electronic medical record. VistA contains a comprehensive, integrated, electronic health record for each patient that is viewable by all of the department’s clinicians at all of its medical facilities, thus eliminating the need for paper medical records. This capability has been key to the department’s efforts over the last 20 years to share electronic medical records with DOD, and with its work to achieve interoperability, which enables different information systems or components to exchange information and to use the information that has been exchanged. Nevertheless, even with the enhancements and modifications made to VistA over time, the system is more than 30 years old and has become increasingly difficult and costly to maintain. One reason VistA is difficult and costly to maintain is that the system is programmed in the MUMPS programming language, a language for which there is a continually dwindling supply of qualified software developers, according to the department. In 2015, an independent assessment of health IT at VA, conducted by the MITRE Corporation, raised questions regarding the lack of any clear advances made during the past decade with VistA and the increasing amount of time needed for VA to release new capabilities. The study also noted that the standards and terminology used by VistA do not enable interoperability across the multiple systems within VA, or between the department and non-VA facilities, including private sector providers and DOD. Given the concerns identified, the study recommended that VA assess the cost versus benefits of various alternatives for delivering modernized capabilities, such as COTS electronic health record systems, open source systems, and the continued development of VistA. VA Has Pursued Four VistA Modernization Initiatives Since 2001, with Over a Billion Dollars Obligated for Contractors’ Activities During Fiscal Years 2011 through 2016 Since 2001, VA has pursued four efforts to modernize VistA. These efforts—HealtheVet, the integrated Electronic Health Record (iEHR), VistA Evolution, and the Electronic Health Record Modernization (EHRM)—reflect varying approaches that the department has considered to achieve a modernized health care system over the course of nearly two decades. The modernization efforts are described as follows. In 2001, VA undertook its first VistA modernization project, the HealtheVet initiative, with the goals of standardizing the department’s health care system and eliminating the approximately 130 different systems used by its field locations at that time. HealtheVet was scheduled to be fully implemented by 2018 at a total estimated development and deployment cost of about $11 billion. As part of the effort, the department had planned to develop or enhance specific areas of system functionality through six projects, which were to be completed between 2006 and 2012. Specifically, these projects were to provide capabilities to support VA’s Health Data Repository and Patient Financial Services System, as well as the Laboratory, Pharmacy, Imaging, and Scheduling functions. In June 2008, we reported that the department had made progress on the HealtheVet initiative, but noted issues with project planning and governance. In June 2009, the Secretary of Veterans Affairs announced that VA would stop financing failed projects and improve the management of its IT development projects. Subsequently in August 2010, the department reported that it had terminated the HealtheVet initiative. In February 2011, VA began its second VistA modernization initiative, the iEHR program, in conjunction with DOD. The program was intended to replace the two separate electronic health record systems used by the two departments with a single, shared system. In addition, because both departments would be using the same system, this approach was expected to largely sidestep the challenges that had been encountered in trying to achieve interoperability between their two separate systems. Initial plans called for the development of a single, joint iEHR system consisting of 54 clinical capabilities to be delivered in six increments between 2014 and 2017. Among the agreed-upon capabilities to be delivered were those supporting laboratory, anatomic pathology, pharmacy, and immunizations. According to VA and DOD, the single system had an estimated life cycle cost of $29 billion through the end of fiscal year 2029. However, in February 2013, the Secretaries of VA and DOD announced that they would not continue with their joint development of a single electronic health record system. This decision resulted from an assessment of the iEHR program that the secretaries had requested in December 2012 because of their concerns about the program facing challenges in meeting deadlines, costing too much, and taking too long to deliver capabilities. In 2013, the departments abandoned their plan to develop the integrated system and stated that they would again pursue separate modernization efforts. In December 2013, VA initiated its VistA Evolution program as a joint effort of VHA and OI&T that was to be completed by the end of fiscal year 2018. The program was to be comprised of a collection of projects and efforts focused on improving the efficiency and quality of veterans’ health care by modernizing the department’s health information systems, increasing the department’s data exchange and interoperability with DOD and private sector health care partners, and reducing the time it takes to deploy new health information management capabilities. Further, the program was intended to result in lower costs for system upgrades, maintenance, and sustainment. According to the department’s March 2017 cost estimate, VistA Evolution was to have a life cycle cost of about $4 billion through fiscal year 2028. Since initiating VistA Evolution in December 2013, VA has completed a number of key activities that were called for in its plans. For example, the department delivered capabilities, such as the ability for health providers to have an integrated, real-time view of electronic health record data through the Joint Legacy Viewer, as well as the ability for health care providers to view sensitive DOD notes and highlight abnormal test results for patients. VA also initiated work to standardize VistA across the 130 VA facilities and released enhancements to its legacy scheduling, pharmacy, and immunization systems. In addition, the department released the enterprise Health Management Platform, which is a web- based user interface that assembles patient clinical data from all VistA instances and DOD. Although VistA Evolution is ongoing, VA is currently in the process of revising its plan for the program as a result of the department recently announcing its pursuit of a fourth VistA modernization program (discussed below). For example, the department determined that it would no longer pursue additional development or deployment of the enterprise Health Management Platform—a major VistA Evolution component— because the new modernization program is envisioned to provide similar capabilities. In June 2017, the VA Secretary announced a significant shift in the department’s approach to modernizing VistA. Specifically, rather than continue to use VistA, the Secretary stated that the department plans to acquire the same electronic health record system that DOD is implementing. In this regard, DOD has contracted with the Cerner Corporation to provide a new integrated electronic health record system. According to the Secretary, VA has chosen to acquire this same product because it would allow all of VA’s and DOD’s patient data to reside in one system, thus enabling seamless care between the department and DOD without the manual and electronic exchange and reconciliation of data between two separate systems. According to the VA Secretary, this fourth VistA modernization initiative is intended to minimize customization and system differences that currently exist within the department’s medical facilities, and ensure the consistency of processes and practices within VA and DOD. When fully operational, the system is intended to be the single source for patients to access their medical history and for clinicians to use that history in real time at any VA or DOD medical facility, which may result in improved health care outcomes. According to VA’s Chief Technology Officer, Cerner is expected to provide integration, configuration, testing, deployment, hosting, organizational change management, training, sustainment, and licenses necessary to deploy the system in a manner that meets the department’s needs. To expedite the acquisition, in June 2017, the Secretary signed a “Determination and Findings,” for a public interest exception to the requirement for full and open competition, and authorized VA to issue a solicitation directly to the Cerner Corporation. According to the Secretary, VA expects to award a contract to Cerner in early 2018, and deployment of the new system is anticipated to begin 18 months after the contract has been signed. VA’s Executive Director for the Electronic Health Records Modernization System stated that the department intends to deploy the new system incrementally to its medical facilities. Each facility is expected to continue using VistA until the new system has been deployed at that location. VA expects that the new system will be implemented at all VA medical facilities within 7 to 8 years after the first deployment. Figure 1 shows a timeline of the four efforts that VA has pursued to modernize VistA since 2001. VA Obligated About $1.1 Billion for VistA Modernization Contracts from 2011 through 2016 For iEHR and VistA Evolution, the two modernization initiatives for which VA could provide contract data, VA obligated approximately $1.1 billion for contracts with 138 different contractors during fiscal years 2011 through 2016. Specifically, the department obligated approximately $224 million and $880 million, respectively, for contracts associated with these efforts. Of the 138 contractors, 34 performed work supporting both, iEHR and VistA Evolution. The remaining 104 contractors worked exclusively on either iEHR or VistA Evolution. Funding for the 34 contractors that worked on both iEHR and VistA Evolution totaled about $793 million of the $1.1 billion obligated for contracts on the two initiatives. Obligations for contracts awarded to the top 15 of these 34 contractors (which we designated as key contractors) accounted for about $741 million (about 67 percent) of the total obligated for contracts on the two initiatives. The remaining 123 contractors were obligated about $364 million for their contracts. The 15 key contractors were obligated about $564 million and $177 million for VistA Evolution and iEHR contracts, respectively. Table 1 identifies the key contractors and their obligated dollar totals for the two efforts. Additionally, we determined that, of the $741 million obligated to the key contractors, $411 million (about 55 percent) was obligated for contracts supporting the development of new system capabilities, $256 million (about 35 percent) was obligated for contracts supporting project management activities, and $74 million (about 10 percent) was obligated for contracts supporting operations and maintenance for iEHR and VistA Evolution. VA obligated funds to all 15 of the key contractors for system development, 13 of the key contractors for project management, and 12 of the key contractors for operations and maintenance. Figure 2 shows the amounts obligated for each of these areas. Further, based on the key contractors’ documentation for the iEHR program, VA obligated $102 million for development, $65 million for project management, and $10 million for operations and maintenance. For the VistA Evolution Program, VA obligated $309 million for development, $191 million for project management, and $64 million for operations and maintenance. Figure 3 shows the amounts obligated for contracts on the VistA Evolution and iEHR programs for development, project management, and operations and maintenance. In addition, table 2 shows the amounts that each of the 15 key contractors were obligated for the three types of contract activities performed on iEHR and VistA Evolution. VA Recently Announced a New VistA Modernization Initiative; Development of Plans Is in Progress Industry best practices and IT project management principles stress the importance of sound planning for system modernization projects. These plans should identify key aspects of a project, such as the scope, responsible organizations, costs, schedules, and risks. Additionally, planning should begin early in the project’s life cycle and be updated as the project progresses. Since the VA Secretary announced that the department would acquire the same electronic health record system as DOD, VA has begun planning for the transition from VistA Evolution to EHRM. However, the department is still early in its efforts, pending the contract award. In this regard, the department has begun developing plans that are intended to guide the new EHRM program. For example, the department has developed a preliminary description of the organizations that are to be responsible for governing the EHRM program. Further, the VA Secretary announced in congressional testimony in November 2017 that the Executive Director for the Electronic Health Records Modernization System will report directly to the department’s Deputy Secretary. In addition, the department has developed a preliminary timeline for deploying its new electronic health record system to VA’s medical facilities, and a 90-day schedule that depicts key program activities. The department also has begun documenting the EHRM program risks. Beyond the aforementioned planning activities undertaken thus far, the Executive Director stated that the department intends to complete a full suite of planning and acquisition management documents to guide the program, including a life cycle cost estimate and an integrated master schedule to establish key milestones over the life of the project. To this end, the Executive Director told us that VA has awarded program management contracts to support the development of these plans to MITRE Corporation and Booz Allen Hamilton. According to the Executive Director, VA also has begun reviewing the VistA Evolution Roadmap, which is the key plan that the department has used to guide VistA Evolution since 2014. This review is expected to result in an updated plan that is to prioritize any remaining VistA enhancements needed to support the transition from VistA Evolution to the new system. According to the Executive Director, the department intends to complete the development of its plans for EHRM within 90 days after award of the Cerner contract, which is anticipated to occur in early 2018. Further, beyond the development of plans, VA has begun to staff an organizational structure for the modernization initiative, with the Under Secretary of Health and the Assistant Secretary for Information and Technology (VA’s Chief Information Officer) designated as executive sponsors. It has also appointed a Chief Technology Officer from OI&T, and a Chief Medical Officer from VHA, both of whom are to report to the Executive Director. VA’s efforts to develop plans for EHRM and to staff an organization to manage the program encompass key aspects of project planning that are important to ensuring effective management of the department’s latest modernization initiative. However, the department remains early in its modernization planning efforts, many of which are dependent on the system acquisition contract award, which has not yet occurred. The department’s continued dedication to completing and effectively executing the planning activities that it has identified will be essential to helping minimize program risks and guide this latest electronic health record modernization initiative to a successful outcome—one which VA, for almost two decades, has been unable to achieve. Agency Comments and Our Evaluation We provided a draft of this product to VA for comment. Via email, a liaison in VA’s Office of Congressional and Legislative Affairs stated that the department appreciated the opportunity to comment on the draft report and provided technical comments, which we incorporated as appropriate. We are sending copies of this report to the appropriate congressional committees, the Secretary of Veterans Affairs, the Under Secretary for Health, the Chief Information Officer, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov If you or your staffs have any questions on matters discussed in this report, please contact me at (202) 512-9286 or pownerd@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made major contributions to this report are listed in appendix II. Appendix I: Objectives, Scope, and Methodology The objectives for this study were to (1) identify the efforts that the Department of veterans Affairs (VA) has undertaken to modernize VistA, including key contractors, contract costs, and expected contractor activities and (2) to determine the department’s current plans for, and progress to date, in modernizing VistA. To address the first objective, we reviewed VA’s prior budget submissions, in addition to VistA Evolution planning documentation, such as the VistA 4 Product Roadmap, VistA 4 Life Cycle Cost Estimate, VistA Evolution Integrated Master Plan, and VistA Evolution Business Case. We also reviewed meeting minutes for the VistA modernization projects and prior GAO work on efforts to modernize VistA. To determine the contractors, costs, and expected contractor activities for these efforts, we requested data from VA’s Office of Information and Technology (OI&T) and the Veterans Health Administration (VHA) on all contracts, related obligations, and expected activities for the HealtheVet program for fiscal years 2001 through 2010; the integrated Electronic Health Record (iEHR) program for fiscal years 2011 through 2013; and the VistA Evolution program for fiscal years 2014 through 2016. Neither OI&T nor VHA was able to provide contract data related to the HealtheVet program. The department stated that it could not verify any HealtheVet contractors receiving payments because the time frame for the effort falls outside the record retention required by regulations. According to the Federal Acquisition Regulation, government agencies are required to retain contract records for 6 years after the final payment. VA provided contract data for the iEHR and VistA Evolution programs, which included contractor names, obligated amounts of funding, and descriptions of the work that the contractors were to perform. OI&T provided such data for fiscal years 2011 through 2016 and VHA provided data for fiscal years 2012 through 2016. VHA program officials told us that they were unable to provide contract data prior to 2012. In addition, we did not request contract data subsequent to fiscal year 2016 because that is the last fiscal year for which data for a full year were available at the time that we performed our analysis. We assessed the reliability of the contract data we received by reviewing it for missing elements and outliers. We then met with officials responsible for VistA-related contracting to address questions about any missing data and outliers, as well as to obtain additional information about how the data were developed. Further, we supplemented the data by using additional information received from VA and the Federal Procurement Data System. We determined that the data were sufficiently reliable for the purposes of our reporting objective. To determine the key contractors, we first identified all of the contractors that worked on both the iEHR and VistA Evolution modernization efforts. We subsequently ranked the contractors according to the total dollars obligated for contracts that each contractor had been awarded. Further, we designated the top 15 ranked contractors, in terms of dollars obligated, as key contractors. These 15 key contractors received contracts that accounted for about two-thirds of the funds obligated to VistA modernization contracts from fiscal year 2011 through 2016. We then analyzed the information provided from VA on the work to be performed by these key contractors to identify obligations made for contracts to provide different types of work supporting the two modernization initiatives, including systems development, project management, and operations and maintenance. We then calculated the funds that were obligated to each of the key contractors for the types of work performed. To determine current plans for modernizing VistA, we reviewed draft program schedules, organization charts, Congressional testimonies of the VA Secretary, a White House press conference transcript, departmental press releases, and the department’s justification for awarding a non- competitive contract for a commercial off-the-shelf electronic health record system. To determine the progress achieved on the current efforts, we obtained documentation, such as draft schedules and organization charts, and met with program officials in VA’s Electronic Health Record Modernization program office to obtain updated information on the efforts. We conducted this performance audit from September 2016 to January 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: GAO Contact and Staff Acknowledgements GAO Contact David A. Powner, (202) 512-9286 or pownerd@gao.gov. Staff Acknowledgments In addition to the contact named above, Mark Bird (assistant director), Eric Trout (analyst in charge), Chris Businsky, Vern Cumarasegaran, Nancy Glover, Paris Hawkins, Jacqueline Mai, Jennifer Stavros-Turner, Christy Tyson, and Charles Youman made key contributions to this report.

Why GAO Did This Study VA operates the largest health care delivery system in America and VistA is essential to helping deliver the health care services and ensure the quality of health care received by the nation's veterans and their dependents. The system has been in operation for more than 30 years, is costly to maintain, and does not readily support VA's need to electronically exchange health records with the Department of Defense and private health care providers. VA has pursued initiatives to modernize VistA, using government contractors to support its efforts. In June 2017, VA announced a new effort to purchase a commercial solution to replace VistA. GAO was requested to review VA's prior and current efforts to modernize VistA. This review determined (1) VA's efforts to modernize VistA, including key contractors, contract costs, and expected contractor activities and (2) VA's current plans for modernizing VistA and the progress that has been achieved to date. To conduct its study, GAO reviewed VA documentation and its prior work and requested and obtained data on contracts, related obligations, and expected contractor activities for previous efforts. GAO also obtained documentation on plans for VA's current modernization efforts and progress made on the efforts, and interviewed VA officials. GAO has made recommendations to VA aimed at improving its prior modernization efforts. The department concurred with the recommendations and generally took responsive actions. GAO is making no recommendations at this time. VA provided technical comments on this report, which were incorporated as appropriate. What GAO Found The Department of Veterans Affairs (VA) has, since 2001, pursued four separate initiatives to modernize its health information system—the Veterans Health Information Systems and Technology Architecture (VistA). These efforts—HealtheVet, the integrated Electronic Health Record (iEHR), VistA Evolution, and the Electronic Health Record Modernization (EHRM)—reflect varying approaches that the department has considered to achieve a modernized health system over the course of nearly two decades (see figure). This latest effort, the EHRM program, is to include the adoption of the same commercial electronic health record system that the Department of Defense is in the process of acquiring. VA obligated about $1.1 billion to 138 different contractors that worked on iEHR and VistA Evolution (the two efforts for which VA could provide contract data) during fiscal years 2011 through 2016. Funding for the 34 contractors that worked on both efforts totaled about $793 million of the $1.1 billion obligated for contracts on the two initiatives. The top 15 of the contractors that worked on the two efforts (key contractors) accounted for approximately $741 million—$411 million for the development of new system capabilities, $256 million for project management activities, and $74 million for operations and maintenance for iEHR and VistA Evolution. VA has begun planning for the transition from VistA Evolution to EHRM. However, the department is still early in its efforts and has begun developing plans that are intended to guide the new EHRM program. According to the EHRM Executive Director, the department intends to complete development of its plans for EHRM within 90 days after awarding the contract for its new system, which is planned to occur in early 2018. VA has also begun to staff the EHRM program's leadership positions. The department's dedication to completing and effectively executing the planning activities that it has identified will be essential to helping minimize program risks and expeditiously guide this latest electronic health record modernization initiative to a successful outcome—which VA, for almost two decades, has been unable to achieve.

gao_GAO-18-445

gao_GAO-18-445_0

Background NIST carries out its measurement services and documentary-standards development activities across several agency laboratories. NIST’s standards activities include participation in private-sector standards development organizations that conduct most standards development in the United States, and federal law and guidance provide direction to agencies when they participate in this process. NIST’s Management of Measurement Services and Documentary Standards Activities NIST’s work underlies much of our nation’s business and public infrastructure, from helping to ensure the quality of air and water to helping to ensure the security of online financial transactions. This work includes providing measurement services, such as calibrations of equipment and reference materials used to ensure the accuracy and reliability of a wide range of scientific and industrial devices, and support for the development of documentary standards by the private sector. As of July 2017, NIST employed approximately 3,500 federal personnel and hosted 4,000 associates, who include guest researchers and collaborators, student interns, facility users, and contractors at its locations in Gaithersburg, MD and Boulder, CO. Under its Associate Director for Laboratory Programs, NIST’s activities span seven laboratory programs that cover a wide range of subject matter, such as bioscience and health, energy, manufacturing, and public safety and security (see figure 1). The seven laboratories are divided into divisions and groups of scientists and engineers who perform research in a certain field or discipline, and may also provide measurement services or participate in standards activities. In addition, NIST has three offices that primarily deal with measurement services: (1) the Office of Reference Materials within the Material Measurement Laboratory; (2) Calibrations Services within the Physical Measurement Laboratory; and (3) the Office of Weights and Measures also within the Physical Measurement Laboratory. Further, the Standards Coordination Office, which is also under the Associate Director for Laboratory Programs, conducts standards-related activities and provides guidance to NIST staff on participation in documentary standards activities. In 2010, NIST reorganized its laboratory structure, in part to improve the agency’s provision of measurement services. Under the new structure, the Physical Measurement Laboratory includes staff that handles most of the agency’s measurement standards and calibrations. The Material Measurement Laboratory includes staff supporting materials science and produces most of the agency’s standard reference materials. According to the NIST Director at the time of the reorganization, managing related research and measurement services together would allow the agency to improve its services. NIST’s measurement services encompass calibrations, standard reference materials, and standard reference data, among others. NIST provides calibration services for about 700 different types of devices and has over 1,200 different types of reference materials available (see figure 2). For example, NIST performs calibrations on many different types of thermometers for both scientific and industrial uses. For customers that have unique calibration needs, NIST can perform special tests tailored to their specific circumstances. NIST also performs detailed analysis of certain materials to precisely characterize their properties and makeup and provides these reference materials for use by others. For example, NIST produces a number of food-related reference materials that allow companies to accurately determine the nutritional content of their products. NIST has established a formal quality-control system covering the calibrations, special tests, and standard reference materials provided by the agency. The NIST quality system is intended to provide customers with confidence in the quality of NIST’s measurement services and create an environment of continual improvement for NIST management and staff. The quality system is described in policies and procedures governing the agency’s measurement services. Specifically, the NIST Quality Manual for Measurement Services, NIST-QM-I, contains NIST- wide policies and procedures and additional sub-level quality documentation contains policies and procedures established and maintained by each Division or Office to meet its technical needs. The system is overseen by the NIST Quality Manager, a position within the Standards Coordination Office, and the NIST Measurement Services Council, comprised of the Quality Manager and other agency officials, who report to the Associate Director for Laboratory Programs. Private sector calibration and testing companies may use NIST’s measurement services to provide NIST-traceable services, meaning that the accuracy and precision of the private company’s service has been documented and compared to NIST’s capabilities. This process allows these companies to provide services to consumers who do not need the high level of certainty provided by NIST while still providing assurance that their measurements are sufficient for their needs. As shown in figure 3, NIST performed calibrations on about 13,000 individual devices per year and provided about 30,000 reference materials per year or more from fiscal year 2006 to fiscal year 2016. NIST also provides standard reference data—such as detailed technical data on various elements, materials, and chemicals—and keeps time with its atomic clock in Boulder, CO, and broadcasts it. NIST also accredits public and private- sector laboratories to perform calibrations and other tests through the National Voluntary Laboratory Accreditation Program. Such accreditation shows, among other things, that the measurement services provided by these labs comports with certain federal and international requirements for calibration and testing. NIST’s standards activities support the development and use of standards to enhance the economic and technological competitiveness of the United States. There are various types of standards including measurement standards that define specific units, such as the kilogram, and documentary standards that describe the performance or design of a particular product, process, or test. NIST develops and refines numerous measurement standards and collaborates with other national metrology institutes across the world through the General Conference on Weights and Measures, the Bureau International des Poids et Mesures, and other organizations. This work includes supporting the International System of Units, which includes the kilogram, meter, second, and other units of measurement that form the basis for NIST’s calibration services. Measurement standards ensure that these units are consistently used and applied around the world. Documentary standards, in comparison, can specify how a product is designed or made, or they may establish performance standards that define the product by function rather than material. For example, documentary standards define Wi-Fi radios, certain aspects of building codes, and safe design for children’s toys, among other things. Both of these standards help define the properties and functions of today’s products and provide businesses and consumers with confidence that products will work as expected before purchase. The U.S. Documentary Standards-Setting Process and NIST Participation In the United States, documentary standards are generally developed by the private sector through an open, consensus-based process overseen by various SDOs. Private sector companies in the United States choose when it is in their interest to participate in standards development. Many SDOs follow similar processes in the development of standards, and generally adhere to certain principles, including openness, balance of interests, and consensus. Specifically, once an SDO agrees to develop a new or revised standard, a committee is formed of representatives with subject-matter expertise from companies, nonprofit organizations, and government agencies. The representatives serve on a voluntary basis, and the committee drafts the standard. SDOs may have certain requirements for participants, such as payment of membership dues, to fully participate. In the process of creating or revising documentary standards, certain committee members will take on leadership roles, such as chairing committee meetings or leading writing of draft standards or other documents. Generally, a committee will use a consensus-based process to vote on whether to approve the draft standard. For example, to approve a draft standard, some SDOs require a supermajority, at least two-thirds, of the members who cast ballots as well as resolution of any negative comments. Documentary standards define the technical aspects or capabilities of materials, devices, machines, and other products to ensure their performance and interoperability. For example, in 1990, the IEEE Standards Association began work to develop a documentary standard for allowing devices to connect wirelessly to the internet. The IEEE wireless networking working group, designated as 802.11, approved its first standard in 1997 and has since approved a series of amendments and improvements to the standard. In 1999, a group of companies formed the Wi-Fi Alliance to help drive usage of the 802.11 standard and provide consumers with information on products that implement the standard. The Wi-Fi Alliance coined the brand Wi-Fi and developed certification procedures to show that devices using the 802.11 standard from different vendors are interoperable and provide a consistent user experience. Devices that comply with the standard are able to wirelessly transfer data within a local area. Within 2 years of the standard’s initial approval, the first devices using the standard were available to consumers and 21 years later wireless networks have become commonplace in libraries, coffee shops, and homes around the world. Originally intended for linking home or office computers, the standard has been implemented in a growing array of devices including lightbulbs and other household items. Hundreds of companies now incorporate Wi-Fi into their products, leveraging the ubiquity of the standard to improve the value of their products and give consumers options for meeting their networking needs. Several, large private sector organizations help create documentary standards in the United States. The American National Standards Institute (ANSI) is a membership organization that accredits numerous U.S. SDOs that oversee the creation, promulgation, and use of over 10,000 American National Standards. Other U.S.-based organizations that develop standards for domestic and international use include ASTM International, IEEE, and the National Fire Protection Association. ANSI also serves as the U.S. representative to two Geneva-based international organizations that support the creation of global standards, the International Organization for Standardization (ISO) and the International Electrotechnical Commission (IEC). When followed, international standards may reduce technical barriers to trade by reducing conflict among domestic standards in various nations and allowing companies to produce a single product for multiple markets. For example, ISO encompasses 163 national standards organizations and is a major source of international standards. In 2000, NIST and ANSI signed a memorandum of understanding to, among other things, improve communication and coordination among the private and public sector on voluntary standards issues. Staff across many of NIST’s laboratories participate in documentary- standards development activities. NIST policy encourages staff participation in domestic- and international-standards activities whenever such participation is in the public interest and is compatible with NIST’s mission, policies, positions, priorities, and available resources. In 2016, NIST reported that staff participated in 114 SDOs. In limited policy areas, where a national priority has been identified in statute, regulation, or administration policy, active engagement or a convening role by the federal government may be needed to accelerate standards development and implementation. Federal Agencies’ Documentary Standards Activities and Relevant Law and Guidance Federal agencies may use or help develop documentary standards for several reasons, including (1) to procure goods or services, (2) to incorporate standards into agency regulations, or (3) to improve agency operations or further agency policy goals. For example, the General Services Administration uses standards to specify packaging, marking, and labeling of products purchased for government use and for descriptions of the products themselves; the Consumer Product Safety Commission has incorporated various consensus standards into its regulations of consumer products; and the Department of Energy uses a number of consensus standards to help operate its contractor-run laboratories, among other uses. As a result, a number of federal agencies participate in a range of standards development activities that span many different areas of national need. Federal law and guidance provide that where possible, agencies are to use voluntary, private sector standards instead of creating their own unique standards and are to establish certain responsibilities in overseeing and coordinating these efforts. The National Technology Transfer and Advancement Act of 1995 (NTTAA) states that federal agencies are generally to use technical standards developed or adopted by voluntary-consensus standards bodies, and in doing so are to consult with voluntary, private-sector consensus standards bodies and participate with such bodies in the development of technical standards when such participation meets certain conditions. NTTAA, as amended, also provides that it is a function of NIST to coordinate the use of private sector standards by federal agencies emphasizing where possible the use of standards developed by private-sector, consensus organizations. Furthermore, the Trade Agreements Act of 1979 (1979 act) directs the Secretary of Commerce to keep adequately informed regarding international standards-related activities and identify those that may substantially affect the commerce of the United States. The Secretary is also to monitor the adequacy of U.S. representation in private international standards activities. The 1979 act says that the representation of U.S. interests before any private international standards organizations is to be carried out by a private person recognized as an organization member. Further, the 1979 act establishes a process for the Secretary to follow if the Secretary has reason to believe that such participation will not result in the adequate representation of U.S. interests or if there is no current organization member. These tasks have been delegated to NIST. NIST’s memorandum of understanding with ANSI also describes NIST’s role under the NTTAA, OMB Circular A-119, and the 1979 act to, among other things, ensure adequate representation of U.S. interests in all relevant international standards organizations and to coordinate federal activities in voluntary standards. In addition, Office of Management and Budget (OMB) Circular A-119, as revised in 2016, sets forth the policy for federal participation in the development and use of voluntary consensus standards. Federal representatives are encouraged to participate actively in standards development activities and to be fully involved in discussions and technical debates, register opinions, and serve in leadership positions if selected, among other things. A-119 directs the Secretary of Commerce, who has delegated this responsibility to NIST, to foster implementation of the Circular. Further, A-119 provides for a NIST-chaired interagency group called the Interagency Committee on Standards Policy (ICSP). The ICSP is composed of agency standards executives—senior-level officials who are broadly engaged in the agencies’ standards activities. A-119 directs standards executives to coordinate their agencies’ views when they participate in the same standards activities so as to present, whenever feasible, a unified position and, when not feasible, mutual recognition of differences. A-119 directs the ICSP to coordinate with certain entities with a view to encouraging more strategic and coordinated federal participation in the development and use of standards in regard to regulatory policy. According to the ICSP charter, the ICSP has the objective to promote effective and consistent standards policies in furtherance of U.S. domestic and foreign goals and, to this end, to foster cooperative participation by the federal government and U.S. industry and other private organizations in standards activities, and its purpose is to ensure effective federal participation in domestic- and international- standards activities. Further, in 2012 the Executive Office of the President (EOP) issued a memo for federal agencies to clarify principles guiding federal government engagement in standards activities that can help address national priorities. According to the memo and A-119, federal engagement in standards activities should be guided by the following strategic objectives: Produce timely, effective standards and efficient conformity assessment schemes that are essential to addressing an identified need. Achieve cost-efficient, timely, and effective solutions to legitimate regulatory, procurement, and policy objectives. Promote standards and standardization systems that promote and sustain innovation and foster competition. Enhance U.S. growth and competitiveness and ensure non- discrimination, consistent with international obligations. Facilitate international trade and avoid the creation of unnecessary obstacles to trade. To address these strategic objectives, the memo notes that the federal government works with the private sector to address common standards- related needs, while taking on a more active role where necessary to ensure a rapid, coherent response to national challenges. The memo also identifies responsibilities for agencies, such as periodically reviewing their standards activities to identify gaps in representation for mission-critical areas as part of their long-range planning and ensuring effective intra- and inter-agency coordination of engagement in standards development activities. NIST Faces Several Challenges in Providing Measurement Services and Supporting Documentary- Standards Development We identified three areas where NIST faces challenges in providing measurement services and supporting documentary-standards development, based on our literature review, NIST stakeholder focus groups, and interviews with stakeholders and agency officials. First, the breadth of U.S. industry and the number of SDOs, among other factors, make identifying and prioritizing measurement service and standards needs and communicating with stakeholders about NIST’s services challenging. Second, ensuring adequate U.S. representation in international standards activities can be challenging due to the number of activities and other factors. Third, the involvement of multiple agencies and interdisciplinary issues makes coordinating among federal agencies on documentary standards challenging. NIST Faces Challenges Identifying and Prioritizing Measurement Services and Documentary- Standards Activities, and Communicating with Stakeholders Can Be Challenging The breadth of U.S. industries and commercial sectors with measurement-service and documentary-standards needs, and other factors, make it challenging to identify and prioritize among these needs, and make it challenging for NIST to ensure stakeholders are aware of the agency’s services. Identifying measurement-service and documentary-standards needs: Identifying measurement-service and documentary-standards support needs can be challenging, according to participants in all five of our focus groups and other NIST stakeholders we interviewed. Participants and stakeholders identified several factors that contribute to these challenges, including difficulty identifying needs across the breadth of U.S. industries and standards development activities, and difficulties presented by emerging, crosscutting, or interdisciplinary technology areas. The breadth of U.S. industries and standards development activities can make it difficult to identify their measurement service needs. NIST’s potential customer base covers the entirety of the U.S. manufacturing sector and many service sectors, including small- and medium-sized enterprises, as well as federal agencies and state and local governments. Identifying needs across the full range of this customer base can be a challenge, according to participants in all five of our focus groups and other stakeholders. Further, NIST officials noted that even within an industry sector, stakeholders may have differing views on the industry’s measurement service needs, which can make it harder to determine whether or how NIST should take action to meet those needs. Similarly, the diversity of documentary standards activities across many SDOs may make it difficult to identify when industry needs NIST staff participation in documentary standards efforts. There are no restrictions on which organizations may develop standards, and therefore, the total number of SDOs is not precisely known. However, ANSI estimates that there are hundreds of such bodies in the United States, and NIST has reported participating in 114 separate SDOs. Participants in four of our focus groups and two agency standards executives said that it can be difficult to keep track of SDOs or standards development activities, and NIST standards officials noted that the breadth of active SDOs and volume of their activities was an ongoing challenge. Similarly, three agency standards executives we interviewed said that identifying standards activities of interest to their agencies is challenging due to the large number of activities. Furthermore, emerging, crosscutting, or interdisciplinary technology areas can be a challenge, according to participants in all five of our focus groups. For example, participants in three focus groups discussed the difficulties faced by organizations that work in areas that combine multiple areas of technical expertise. A participant in one focus group cited electronic health records as an example of an interdisciplinary technology, as it includes biomedical research, public health research, and information technology. Another participant cited increasingly high- tech development in biological devices that involve physics, engineering, and mathematics. Participants said that organizations need to coordinate across disciplines and break down communication barriers to address these challenges. Additionally, representatives from one SDO we interviewed as well as an agency standards executive we spoke with highlighted the difficulty associated with predicting the trajectory of future change in emerging technologies. NIST officials noted that taking action to support the measurement service and standards needs of emerging technologies may be more challenging where there is a lack of industry consensus on how a technology will develop. Prioritizing among needs: Prioritizing among different measurement services can also be challenging. Participants in all of our five focus groups said that NIST must prioritize among measurement service needs because it does not have the resources to provide services for all industry needs. Participants in three of our five focus groups described challenges balancing between continuing older measurement services that serve current needs and creating new services. Further, a 2017 review of the activities of NIST’s Material Measurement Laboratory by the National Academies found that stakeholders have high demands for the laboratory and that it faces challenges balancing between maintaining ongoing efforts and initiating new efforts. A participant in one focus group also said that it can be difficult to prioritize between services that have broad use and those that are vital to narrower customer bases. For example, NIST performs calibrations for thermometers across a wide range of temperatures for use in many different sectors. On the other hand, NIST’s million-pound deadweight machine provides calibrations for very large force gauges used by aerospace manufacturers and the U.S. military. Focus group participants and NIST officials said that the volume of services provided may not reflect the value of the service to the industry, because a single calibration can support many millions of dollars of economic activity. When prioritizing staff participation in documentary standards activities, NIST faces a similar challenge. Specifically, the abundance of ongoing standards development activities means NIST staff may have to choose among several standards development activities in their areas of expertise. While some staff may have expertise that is closely linked to a small number of SDOs and activities, others may have expertise in foundational technologies that have relevance to numerous activities. Further, a participant in one focus group and two agency standards executives we interviewed stated that standards need to be revised from time to time, for example, to incorporate new technologies, and these revisions may compete for time and attention against new standards efforts in related areas. While individual SDOs can plan for and prioritize among their new standards efforts and revisions, NIST staff who participate in standards development efforts across a number of SDOs may still have to choose among contemporaneous efforts. Communicating with stakeholders: Communicating with stakeholders about NIST’s measurement services can be challenging, according to participants in all five of our focus groups and other stakeholders. Specifically, the breadth of potential users of NIST’s measurement services makes it more difficult for NIST to communicate with industry about its needs and NIST’s services. Participants in four of the five focus groups said that it can be difficult for potential users to be aware of and understand the services NIST provides relevant to their needs. For example, participants in one focus group described concerns regarding how well they, and industry stakeholders generally, understand the extent of NIST’s capabilities within their areas of expertise. Participants in this focus group cited benefits of having NIST and industry staff perform site visits to elucidate each other’s needs and capabilities. One NIST calibrations official we interviewed said that some commercial sectors, such as the automotive industry, may be underserved by NIST’s services due to a limited understanding of how NIST could help companies remain innovative and competitive. However, participants in one focus group said that NIST’s engagement with the industrial community is generally quite strong. Other stakeholders suggested that NIST faces an increasingly difficult task educating potential customers about its services because those customers may have less technical expertise today than they did in the past. Participants in one focus group, officials from the Department of Energy, and a NIST reference material official said that the portion of the nation’s workforce trained in measurement and standards issues is shrinking and that industry representatives now have less experience in these matters than they used to. Accordingly, NIST now communicates with stakeholders who have less expertise about its measurement services. Ensuring Adequate U.S. Representation in International Documentary-Standards Activities Can Be Challenging According to NIST standards officials, focus group participants, other stakeholders, and a NIST report on U.S. representation in international documentary-standards activities, ensuring adequate U.S. representation in these activities can be challenging. Several factors, such as the large number of international standards activities occurring across numerous industry sectors, underlie this challenge and make it difficult for NIST to ensure adequate U.S. representation. First, the breadth of the global economy and the volume of international documentary-standards development activities make ensuring adequate U.S. representation challenging, according to NIST standards officials, stakeholders, participants in two focus groups, and literature we reviewed. For example, NIST standards officials, two agency standards executives, participants in one focus group, and literature we reviewed said that the large number of SDOs and volume of international standards activities presents a challenge to NIST. A participant in a different focus group also said that in some cases, industry is reliant on NIST to provide them with information on international standards activities relevant to them. As the number of activities increase, it can be difficult to maintain a comprehensive understanding of what is happening in various industry sectors and standards areas. Further adding to this challenge, several sources of information we collected identified a significant increase in the number of international standards activities or the relative participation of other countries in these activities, for example: NIST officials and participants in one focus group said that international SDOs, such as the ISO, are expanding their efforts to create global standards. Participants said that ISO’s efforts could conflict with existing standards that U.S. industry uses. Two stakeholders we interviewed said that U.S. industry also faces increasing competition from other countries, such as China, which, in some cases, is overwhelming the ability of U.S. industry to participate. Further, according to a 2012 testimony to Congress by the director of NIST’s Standards Coordination Office, other countries have made significant investments in their standards efforts and have attempted to increase their participation in international standards activities. According to the testimony, other countries increasingly view standards as a tool to increase their international competitiveness and are developing strategies and tactics to play a greater role in standardization, such as increasing their participation and leadership in international standards bodies. A 2014 NIST report on U.S. representation in international SDOs showed that the United States fell from first in 2005 to second in 2012 in the number of experts participating in one international SDO, the IEC, which produces standards for electric and electronic products, systems, and services. Specifically, the number of technical experts from the top ten countries that participate in the IEC was 5,528 in 2005 and 9,199 in 2012—an increase of 66 percent. However, participants from outside the United States were responsible for 85 percent of the increase. Additionally, the report showed that the United States fell from third in 1999 to fourth in 2012 in the number of IEC standards proposals submitted. The report showed that the number of new IEC standards proposed by all countries was 70 proposals in 1999 and 124 proposals in 2012—an increase of 77 percent. However, the percentage of U.S. country proposals out of all country proposals fell from 19 percent in 1999 to 14 percent in 2012. NIST standards officials said that there could be additional factors driving changes in U.S. stakeholder participation in international SDO activities. For example, NIST officials said that while U.S. stakeholder participation in ISO and IEC may have declined in some cases, some U.S. stakeholders have increased participation in other international SDOs whose standards are better suited for their industry. Second, what constitutes adequate representation is currently unclear, according to NIST’s 2014 report on U.S. representation in ISO and IEC activities and NIST standards officials we interviewed. According to NIST’s 2014 report, there are no guidelines or definitions given for what is deemed to be adequate representation of U.S. interests in international standards activities. Further, NIST standards officials said that it was not clear what circumstances would lead NIST to use the process established under the 1979 act if U.S. representation in an international SDO may be potentially inadequate. NIST officials also said that defining what would constitute adequate U.S. representation at international SDOs and collecting the information to help assess the adequacy of U.S. representation would be difficult and the definition and metrics could vary by industrial sector. Third, according to participants in two focus groups, the large number of companies and other stakeholders that could be involved in or have an interest in various international SDO activities under the U.S. system of private-sector-led standards development can make ensuring adequate U.S. representation a challenge. For example, participants in two focus groups said that NIST would need to consult with numerous industry stakeholders or SDOs to facilitate representation in situations where U.S. representation was inadequate. Further, as we mentioned above, documentary standards needs in emerging, crosscutting, or interdisciplinary technology areas can be a challenge. Literature we reviewed highlighted the need for NIST or other agencies to help bring together industries or other stakeholders that may not have a history of collaborating to resolve standards issues. Coordinating Federal Agencies’ Participation in Developing Documentary Standards Can Be Challenging Fulfilling NIST’s role to work with other agencies to coordinate use of and participation in standards activities under the NTTAA, as chair of the ICSP, and in implementing OMB Circular A-119 is challenging due to (1) the involvement of multiple federal agencies in documentary standards activities, and (2) increasingly interdisciplinary technology areas. Multiple agency involvement: Multiple federal agencies are involved in documentary standards activities, a situation that can make coordinating agencies’ activities challenging, according to participants in all five of our focus groups, and some stakeholders and agency standards executives we interviewed. Because multiple agencies are involved in documentary standards, agency efforts can be fragmented. Fragmentation refers to those circumstances in which more than one federal agency (or more than one organization within an agency) is involved in the same broad area of national need and opportunities exist to improve service delivery. We have previously reported that fragmentation of federal efforts occurs in a number of areas and can lead to challenges to effective coordination. As we mentioned earlier, federal agencies may use or help develop documentary standards for several reasons, including (1) to procure goods or services, (2) to incorporate standards into agency regulations, or (3) to improve agency operations or further agency policy goals. As a result, a number of federal agencies participate in a range of standards development activities that span many different areas of national need. Further, while some documentary standards issues may only affect the mission or activities of a limited number of agencies, other issues may affect many agencies. Participation by multiple federal agencies in documentary standards activities can be beneficial, according to some focus group participants, stakeholders, and agency standards executives. Some focus group participants, stakeholders, and agency standards executives identified examples of federal participation in which agencies could leverage their different strengths and expertise. For example, participants in four focus groups and some stakeholders we interviewed noted NIST’s unique role as a non-regulatory and neutral agency in facilitating the development of standards. These participants and stakeholders said that, in combination with NIST’s technical capability, this role allowed NIST to gain trust and cooperation from industry in advancing standards development, whereas industry may view regulatory agencies as less neutral. Participants in one focus group said that this role was also helpful to regulatory agencies because these agencies, such as the Food and Drug Administration, would not be able to work as closely with industry in regard to solving technical standards problems or assisting industry because of their regulatory role. Further, two agency standards executives said that some standards activities benefit from the expertise of multiple agencies. For example, one agency standards executive said that evaluating whether standards or product specifications in other countries constituted a barrier to trade required the expertise and participation of different agencies. Two stakeholders we spoke to also said that participation by all relevant federal agencies in standards activities is beneficial because the agencies can provide technical expertise and are important stakeholders for standards efforts because agencies regulate industry, develop policy, and procure goods from the private sector. At the same time, participants in all five of our focus groups, some stakeholders, agency standards executives, and NIST officials we interviewed cited challenges in coordinating documentary standards among multiple federal agencies, for example: Some stakeholders, agency standards executives, and another federal standards official we interviewed said that communication between federal agencies on their standards activities can be a challenge. For instance, three agency standards executives and one stakeholder said that it can be difficult to identify when other agencies are working on the same standards areas, and two of the standards executives said it can be difficult to know who to contact in other agencies to coordinate efforts. Three agency standards executives said that it can be difficult for standards executives to be fully aware of all standards activities in their department or agency. One standards executive also said that some standards executives have split responsibilities and are not full time, a situation that may make it difficult for these executives to devote sufficient time to understanding their agencies’ standards activities, particularly in large agencies. Further, NIST officials said that there is an uneven level of interest and focus on standards as a policy issue among federal agencies, generally. Participants in three focus groups cited differing priorities and interests among federal agencies as a challenge to coordinating on standards activities. For example, participants in one focus group said that different interests and priorities among financial regulatory agencies posed a challenge to coordination on cybersecurity standards. As we reported in December 2015, NIST undertook a collaborative process that involved federal agencies as well as nonfederal stakeholders in developing a cybersecurity framework in response to executive order and legislative requirements. Participants in one focus group noted NIST’s efforts to work with a variety of public and private-sector stakeholders but also said that financial regulatory agencies have their own cybersecurity regulations that may not align with NIST’s framework. In a February 2018 report on implementation of the cybersecurity framework, we noted the complex regulatory and cybersecurity environment of the financial sector and noted that sector representatives said that agencies’ differing cybersecurity requirements led to competition among various cybersecurity frameworks. NIST officials also said that it can be challenging for federal agencies to harmonize their views on standards because they each have individual missions and priorities that may lead to varying views. Similarly, a 2011 National Science and Technology Council report cited a lack of coordination among agencies with interests in standards activities as having a negative impact on government effectiveness. The report noted that agency objectives may not always be aligned and that they may be providing redundant support or competing with one another. An additional complexity to coordination of federal agencies’ documentary standards activities is that some standards issues may have multiple venues for interagency coordination. NIST officials said that agencies participated in the U.S. private-sector-led standards system and that there were a number of different organizations and groups through which federal agencies shared information, depending on the standards activity. There are at least four groups including the ICSP that provide interagency coordination on standards issues generally. According to NIST officials, interagency coordination also occurs through the National Science and Technology Council’s committees and subcommittees. Further, individual documentary standards areas may have additional interagency coordination venues. For instance, interagency coordination on cybersecurity standards also occurs through the Interagency International Cybersecurity Standardization Working Group, according to a NIST 2018 draft report. This group was established by the National Security Council’s Cyber Interagency Policy Committee to coordinate on major issues in international cybersecurity standardization and enhance U.S. federal agencies’ participation in these efforts. Furthermore, agencies may coordinate amongst themselves informally on specific standards interests, according to two standards executives. Interdisciplinary technology areas: Documentary standards development issues have become increasingly interdisciplinary— potentially creating challenges to coordinating agencies’ standards activities, according to our prior work, literature we reviewed, focus groups we conducted, and stakeholders we interviewed. As described above, technology areas, including emerging areas of technology such as electronic health records, can cut across disciplines. According to literature we reviewed, our prior work, and stakeholders we interviewed, interdisciplinary standards can be more difficult to develop or implement because they can be complex and involve a broader range of industry and government stakeholders with potentially different interests and needs. Standards areas are also becoming increasingly interdisciplinary, according to literature we reviewed, one stakeholder, and participants in one focus group. In addition, participants in all five focus groups and some stakeholders we interviewed said that it can be challenging to facilitate interagency coordination on interdisciplinary standards areas. For example, some participants and some stakeholders said that these standards areas can involve the need for collaboration among multiple agencies that can have different roles and responsibilities, priorities, or levels of expertise. Further, two stakeholders said that it could be a challenge for federal agencies to identify these areas. Our prior work, participants in two of our focus groups, and an agency standards executive identified several examples of interdisciplinary standards areas that present challenges to interagency coordination: In February 2018, we reported that protecting the nation against complex and growing cybersecurity threats required coordination between 10 different agencies, 9 of which had responsibility for coordinating implementation of NIST’s cybersecurity framework across 16 different critical infrastructure sectors. In November 2016, we reported that improved interagency coordination could help to address challenges SDOs face in using forward-looking climate information—an interdisciplinary standards area that requires expertise from multiple agencies—and we made a related recommendation. One focus group participant identified open source software and a participant in a different focus group identified machine learning and artificial intelligence as interdisciplinary standards areas needing increased coordination among federal agencies. One agency standards executive also said that federal coordination could be beneficial for developing standards for “Big Data” because multiple federal agencies have expertise in Big Data that is not being leveraged to create standards that could facilitate appropriate use of Big Data-related technology and techniques. Establishing a private-public partnership to coordinate standards development with different stakeholders. Implementing a testing framework. According to NIST officials, developing documentary standards for interdisciplinary technologies can be more resource intensive because of the need to pull together expertise from different disciplines and potential competition among SDOs in developing a standard. However, NIST officials also noted that standards development for interdisciplinary technologies may not always be more challenging than other types of standards development efforts, when an SDO has willing participants with the necessary expertise. For example, according to a 2010 VCAT report, NIST established a public-private Smart Grid Interoperability Panel to identify, prioritize, and address new and emerging requirements for this interdisciplinary standards area involving many stakeholders and agencies. According to VCAT, the panel allowed for a wide range of participating stakeholders and served as an effective way to determine and incorporate the different needs and interests of participants in a framework that enabled further development of smart grid standards. NIST Has Taken Steps to Address Challenges in Providing Measurement Services and Supporting Documentary Standards, but Some Efforts Could Be Improved NIST’s Efforts Help Address Challenges to Identifying and Prioritizing Needs but Could More Fully Align with Federal Guidance and NIST Policy NIST works to address the challenges it faces in providing measurement services and supporting documentary-standards development in a variety of ways, but opportunities exist to improve some efforts. First, NIST’s efforts help address challenges to identifying and prioritizing measurement-service and documentary-standards needs, but some efforts do not fully align with federal guidance or NIST policy. Second, NIST’s efforts help support U.S. representation in international standards organizations but may not fully implement its role and address the challenge it faces. Third, NIST’s efforts support federal agency coordination on standards issues but do not fully align with selected leading collaboration practices. NIST takes a variety of steps to identify agency stakeholders’ measurement-service and documentary-standards needs, has procedures to support the prioritization of measurement services and standards activities in the agency, and has developed mechanisms for communicating with stakeholders. While these steps help NIST address challenges to identifying and prioritizing needs and communicating with stakeholders, some aspects of these efforts do not fully align with federal guidance or NIST policy. Identifying measurement-service and documentary-standards needs: NIST identifies stakeholders’ measurement-service or documentary-standards needs in various ways, including discussions with industry by NIST technical staff, attendance at trade shows and scientific professional society meetings, workshops hosted by NIST on technology areas of interest, and participation in planning activities of SDOs, among other ways. However, NIST does not regularly perform a comprehensive assessment of its measurement services and standards activities to identify and address any gaps between the agency’s efforts and industry needs. Genome in a Bottle The Genome in a Bottle consortium is one of several ongoing collaborations among the National Institute of Standards and Technology (NIST), Stanford University, and various industry and government partners that focus on measurements and standards supporting the newest developments in biology. Genome sequencing involves determining the chemical building blocks of deoxyribonucleic acid (DNA) or ribonucleic acid (RNA) and can give insights into the genes carried by an individual and how and when they are activated. Since the completion of the Human Genome Project in 2003 which first sequenced the whole genome of a human, scientists have worked to make whole human genome sequencing faster and less expensive. Genome in a Bottle aims to develop the tools needed to allow clinical use of whole human genome sequencing. These tools include reference materials that allow laboratories to ensure the reliability and accuracy of their sequencing equipment, increasing laboratories’ capability to perform genetic testing, medical diagnoses, and customize drug therapies. NIST’s primary method for assessing stakeholders’ needs is through outreach by individual technical staff and their expertise in relevant disciplines and related industries. Measurement services officials said that NIST’s staff work closely with their respective industry stakeholders and others to understand their measurement service needs. They stated that NIST staff engage with industry through direct contact at conferences and trade shows, company and NIST laboratory visits, training, NIST-led workshops, through their ongoing research activities, and other activities. NIST also collaborates with other metrology organizations to identify measurement service needs and advance measurement science. For example, measurement services officials described NIST’s participation in international organizations that develop strategic plans for calibrations and measurement standards, such as the Bureau International des Poids et Mesures and Inter-American Metrology System. These organizations allow national metrology institutes, like NIST, to collaborate with their peers and other stakeholders to improve the world’s measurement standards and services. NIST also collaborates with government, industry, and research institutions on emerging issues through various collaboration mechanisms. For example, NIST formed the Genome in a Bottle Consortium in 2011. It provides an open forum for discussion and planning for reference materials and other measurement infrastructure needed to use human genomic sequencing in clinical settings. Similarly, NIST’s Advanced Materials Center of Excellence allows the agency to work with universities, a government lab, and others to address research and development needs related to designing novel materials for manufacturing. Furthermore, since 1905 NIST has participated in activities of the National Conference on Weights and Measures. Recent activities of this group include developing measurement practices and measurement standards to ensure that ride-sharing companies accurately measure time and distance charges. The National Institute of Standards and Technology (NIST) began an active mercury reduction campaign in 2007 and stopped calibrating mercury thermometers entirely on March 1, 2011. NIST's Temperature and Humidity Group is actively participating in several U.S. and international phasing out efforts to identify alternative thermometers for a broad range of measurement applications, and to coordinate efforts to replace mercury- based instruments. For example, the Minamata Convention on Mercury is a global treaty to protect human health and the environment from the adverse effects of mercury and includes a phase out of the use of mercury in products and processes. According to NIST officials, NIST worked closely with one standards development organization, ASTM International, to develop a new standard for the manufacture and selection of digital thermometers. This standard describes three different types of digital-thermometer sensors and defines different classes of devices based on accuracy and, according to NIST, allowed ASTM to revise over 750 additional standards to replace required mercury thermometers with digital thermometers. NIST provides calibrations for all three types of sensors to the worldwide measurement standard, the International Temperature Scale. Furthermore, NIST examines trends in the measurement services it provides to better understand industry needs. In particular, NIST conducts individualized testing for companies known as special tests that can give the agency insight into industry’s needs. Special tests comprise calibrations and related measurements that are unique to the customer and are not part of NIST’s regular catalogue. A company may request a special test, for example, to evaluate a prototype product or measurement technology. According to NIST officials, NIST uses special tests as a way to meet industry needs and also understand what kinds of measurement services industry may need more of in the future. Similarly, NIST standards officials also described staff expertise as important for identifying stakeholders’ needs for support in the development of documentary standards. In particular, standards officials described staff participation in roadmapping activities—used to identify and plan for future standards activities in certain fields—sponsored by NIST or SDOs as important opportunities for staff to assess the standards landscape and identify needs. For example, NIST officials noted the importance of NIST participation in ANSI’s standards panels and collaboratives, some of which are co-led by NIST staff, for identifying standards needs. NIST also participates in SDO administrative groups, such as the ANSI Government Members Forum, that can alert NIST to important international and domestic standards activities. Through participation in SDOs, NIST standards officials said that NIST may obtain information on international standards activities in which U.S. industry representation is needed. NIST officials said that developing new documentary standards can take from a year and a half up to a decade to complete, and accordingly, NIST considers what the standards industry is likely to need in the next 1 to 5 years. Various thermometers at NIST’s temperature calibration lab. In addition, NIST gathers information on how its efforts align with stakeholders’ needs through feedback from industry customers and external reviews by the National Academies, VCAT, and others. For example, NIST asks the users of its calibration services and standard reference materials to respond to customer satisfaction surveys. NIST measurement services officials said that while a small percentage of customers respond to the surveys, the information gathered provides useful input on what new services customers need. By 2018, NIST implemented new information systems to track its measurement services’ sales and customers, and is evaluating if the agency’s outreach to these customers can be improved using the new systems. According to NIST officials, NIST also receives feedback when stakeholders contact the agency through phone, email, or the NIST website. Further, the National Academies evaluated NIST’s Material Measurement Laboratory in 2017 and has reviewed every NIST laboratory since the 2010 reorganization. In a 2017 report, the National Academies recommended that NIST’s Material Measurement Laboratory develop a strategy to balance between existing product support and the research, production, and certification of new standard reference materials. Recent VCAT reports have also considered how well NIST identifies measurement-service and documentary-standards needs of its customers and assessed the agency’s services: In a 2009 report, VCAT examined NIST’s participation in standards development in three specific areas and found that NIST’s technical expertise, its reputation as an unbiased and neutral party, and its extensive participation in standards activities strongly position NIST to address the standards-related challenges of the 21st century. In a 2010 report, the VCAT found that NIST’s analysis and planning practices for its measurement services tended to be driven by bottom- up initiatives more than high level strategy and in 2010 and 2011 reports the VCAT recommended, among other things, that NIST perform additional assessments of its measurement services. In 2012 and 2013 reports, the VCAT found that NIST’s participation in standards activities has helped the agency identify industry needs related to advanced manufacturing. Additionally, in 2015, NIST received a peer review of its measurement services by experts from other national metrology institutes. According to the peer review summary, most of the eight individuals from other national metrology institutes who reviewed NIST’s measurement services found that the agency covered major needs, and the reviewers gave NIST additional feedback on areas for expansion the agency should consider. Participants in three of our five focus groups said that when NIST focuses on a specific area, its efforts to understand industry needs can be very effective. For example, participants in one group said that NIST creates an open environment for discussions with industry and companies feel comfortable approaching the agency with their needs. Participants in another group said NIST’s regular contact with and surveys of state metrology labs help the agency understand their needs. However, participants in all five of the focus groups said that NIST’s capacity for outreach is limited. For example, participants in one focus group said that NIST’s outreach efforts can be driven by the personal relationships NIST staff develop with stakeholders and therefore do not scale to the large size of U.S. industry. NIST officials said that its measurement services and documentary-standards support activities serve different populations of stakeholders and that identifying the needs of NIST’s measurement services’ stakeholders is more manageable than with documentary standards. Specifically, by working with measurement equipment manufacturers, NIST officials said that understanding the needs of its measurement services’ stakeholders was manageable. However, NIST officials said it is more difficult to know the measurement-service needs of industry stakeholders that work with equipment manufacturers rather than with NIST directly. NIST’s efforts to identify industry needs are supported by agency policy, and NIST has controls in place to evaluate the efficacy of the measurement services it provides. NIST policy directs staff to consider stakeholders’ measurement service needs and assign responsibility for assessing measurement services to agency management. Specifically, the NIST Quality Manual, which contains the agency’s policies and procedures governing its measurement services, describes meeting and anticipating the needs of measurement services’ users as a goal for the agency. The manual encourages staff to identify improvements to measurement services and assigns ultimate responsibility for providing services that meet industry, academia, and other government agency needs to the Associate Director for Laboratory Programs. Further, the Quality Manual requires multiple levels of review of the agency’s measurement services, including internal audits at the division level, quarterly management review by measurement services officials, and peer-review by a team of experts from other NIST divisions. The assessments are to provide NIST with assurance that its measurement services, and especially the calibration and measurement capabilities, continue to be in compliance with its quality management system. Further, NIST officials told us that the agency was considering measurement services as part of its strategic-planning efforts, but those efforts were preliminary at the time of our review. For documentary standards, NIST’s policies for staff participation in standards development encourages staff participation in domestic- and international-standards development activities whenever such participation is in the public interest and is compatible with NIST’s mission, policies, positions, priorities, and available resources. NIST’s standards participation policy also provides that the Associate Director for Laboratory Programs conduct periodic reviews of the effectiveness of NIST’s participation in documentary standards activities, with support from the Standards Coordination Office. Additionally, the policy directs NIST managers to annually review records of SDO participation by staff in their divisions and calls for laboratory and division managers to periodically review activities to identify gaps in representation for mission- critical areas. Further, NIST officials said that across both measurement services and documentary-standards support activities, the efforts of its staff to meet stakeholders’ needs are assessed via employee performance reviews, among other means. NIST’s multi-level assessments of its measurement services and documentary standard development activities help ensure their quality and help to identify stakeholders’ needs; however, these assessments do not comprehensively identify and assess gaps in NIST’s services or how well they align with stakeholder needs. For example, NIST’s efforts to identify measurement service needs for individual technology areas or industries, or to evaluate the services provided by its labs—both areas of strength for NIST—may not identify gaps in service needs for technology areas not evaluated or that cut across NIST’s labs. Officials working on calibration services and reference materials told us that NIST has not performed a comprehensive assessment of how well its services address industry needs since a 2006 assessment of the U.S. measurement system. NIST measurement services officials raised concerns about the value of this type of review, describing the agency’s 2006 assessment as time consuming and ultimately of limited use in identifying unmet measurement needs. However, members of the 2015 peer review of measurement services said that NIST would benefit from strategic assessments to identify and assess gaps in programs and a calibration official told us that it is difficult for NIST to recognize if it is not effectively reaching stakeholders. Similarly, NIST officials told us that the Associate Director for Laboratory Programs does not perform a periodic review of the effectiveness of NIST’s standards participation, despite the agency’s standards participation policy calling for such a review. One standards official said NIST generally does not comprehensively assess standards needs because of the number and diverse nature of standards activities. Federal standards for internal control direct management to use quality information to determine if the agency is meeting its objectives and to identify, analyze, and respond to significant changes that could present risks to achieving its objectives. In addition, the 2012 EOP memo on standards activities in areas of national priority states that agencies should periodically review their standards activities to identify gaps in representation for mission-critical areas as part of their long-range planning. Revising NIST’s policies to provide for periodic comprehensive management reviews of NIST’s measurement services would augment NIST’s ongoing efforts to assess how well its services align with stakeholder needs and identify any gaps. Conducting comprehensive reviews of NIST’s measurement services and documentary standards activities would provide NIST with greater confidence that its activities align with stakeholders’ needs, consistent with internal control standards. Conducting such reviews would also help NIST address the recommendations made by its recent external reviews and could be used to support NIST’s efforts to develop the strategic plan called for by the American Innovation and Competitiveness Act. Prioritizing among needs: NIST has a process for deciding when new measurement services are warranted; however, decision-making about documentary-standards development activities is decentralized, and NIST management and staff may not have clear guidelines or sufficient information to support decision-making about new standards activities. NIST has processes to guide decision-making about measurement services; this guidance helps address the challenges focus group participants identified given that NIST cannot address all needs. Before choosing to develop a new measurement service, such as a new standard reference material or calibration service, NIST’s Office of Reference Materials and Calibration Services office, respectively, consider the need for and priority of the service. One NIST official said that because NIST cannot cover all measurement services that may be needed by the private sector, the measurement services program focuses on the areas where NIST may have the most impact. NIST has procedures in place to evaluate proposals for new services. For example, before NIST develops a new reference material, Office of Reference Materials and relevant laboratory staff annually review proposals for new materials by evaluating factors such as the potential user base for the material, related legislative or regulatory requirements, and whether the material could be produced by other organizations, such as private companies. The Office of Reference Materials also considers these factors when considering extending reference materials it already provides. Further, because developing a new reference material can be time consuming, NIST is currently evaluating the creation of a suite of reference materials called “research-grade materials” that could address high priority areas with a lower level of precision than NIST’s standard reference materials. According to NIST officials, research-grade materials are one way of providing this type of measurement service faster to meet the needs of U.S. industry. Similarly, new calibration proposals are reviewed by Calibration Services management on a quarterly basis and are evaluated on factors such as stakeholder need and potential impact. In addition, NIST extends its reach through its work with private sector test and calibration companies that also serve the needs of U.S. industry. For example, the National Voluntary Laboratory Accreditation Program allows NIST staff to directly interact with test and calibration laboratories and provides opportunities to share NIST’s expertise and improve services offered by these laboratories. Further, a focus group participant and government laboratory officials described industry association meetings as important opportunities to find out about cutting-edge capabilities and potential future measurement capabilities offered by NIST and others. NIST does not have a similar formalized process to support consistent decision-making across NIST laboratories and divisions about participation in new documentary standards activities. As described previously, NIST’s policy for staff participation in standards development encourages staff participation in domestic- and international-standards development activities. Additionally, NIST guidance directs staff to participate in SDO activities based on their unit’s mission and goals, and the technical competence required, among other factors, and advises that staff may choose to accept leadership positions in these activities, such as the secretary or chair of a standards committee. OMB Circular A-119 and the 2012 EOP memo on standards also encourage agencies to play a variety of roles in the standards process, such as serving as chairpersons or other official capacities. Focus group participants had mixed opinions on when NIST staff should take on leadership roles. Participants in four of our focus groups said that NIST staff are particularly suited to leadership roles, and some attributed this conclusion to the technical expertise of NIST staff or their ability to act neutrally among competing companies. However, participants in two focus groups said that NIST is better suited to a technical advisor role. However, NIST policy and guidance do not describe when it is appropriate for staff to take leadership roles in standards development activities. Individual staff in consultation with their supervisors determine what standards activities, if any, they should participate in and their appropriate role in the standards development activity. According to NIST officials, other levels of management may be involved in the decision- making process to varying degrees depending upon whether participation in an SDO activity aligns with a NIST priority or where involvement entails international travel, among other factors. A guidance document for staff encourages them to attend additional training provided by the Standards Coordination Office, and standards officials we interviewed told us that the training and informal guidance provided by the office could help staff in leadership positions; however, such roles may entail additional time commitments. Although some stakeholders have expressed interest in increasing NIST’s participation and leadership in standards activities, doing so could entail tradeoffs between these activities and other NIST priorities. Without clear agency guidance on staff participation in standards development activities, such as the factors staff could consider when deciding to take on leadership or other more active roles, NIST cannot be assured that decisions on the time staff commit to standards activities are being made consistently across the agency and in accordance with agency priorities. Further, NIST’s ability to ensure participation is appropriately prioritized across the many documentary-standards development activities in which its staff could be involved is limited by incomplete information. Staff are directed in NIST policy to record their participation in standards activities in a centralized database, including a description of the SDO, specific activity, and role of the NIST participant. According to a NIST standards official, the database may be used by laboratory managers to assess the standards activities of their staff. However, NIST’s database does not contain information regarding NIST staffs’ time commitment, information that could be used by management to assess the resources required for participation in these activities. NIST guidelines also direct staff to document their goals and time commitments for standards activities in their individual performance plans, but the data are not included in the standards participation database. According to NIST officials, determining the time spent on documentary standards can be difficult. Specifically, staff participate in standards development within their areas of expertise and often this work is closely related to their research activities at NIST. Because of this confluence, the amount of time staff spend on a particular standards activity may be unclear. The time spent at meetings or directly drafting or responding to standards documents will also depend on the amount of consensus on the standards committee, consensus that may not be known ahead of time. NIST standards officials told us that the self-reported data currently in the database are sufficient for laboratory management to identify what activities staff are participating in, and that management can then ask individual staff for additional information. However, NIST does not have data at an aggregate level on the time staff commit to or expect to commit to these activities. Standards for internal control require agencies to use quality information to achieve the entity’s objectives, such as by using relevant data from reliable internal and external sources in a timely manner based on the identified information requirements. While staff document their roles in documentary standards activities, without information on the estimated amount of time staff commit to these activities, NIST management may not have the information needed to comprehensively assess how staff distribute their limited time and attention. Although it may not be feasible to determine the exact amount of time spent on documentary standards activities, information on estimated amount of time could help inform staff decision-making on when to accept leadership roles in standards development activities and could inform management on trends in time commitments to these activities across the agency’s laboratories and divisions. Communicating with stakeholders: NIST takes a number of steps to address the challenges it faces communicating with its diverse stakeholders about its measurement services and documentary standards activities. NIST measurement services officials described the primary goal of their stakeholder outreach efforts as informing potential customers of the services NIST provides. The officials described multiple avenues for reaching potential customers of NIST’s measurement services, including: attendance at workshops, trade shows, and professional societies; NIST’s measurement services websites; email and newsletter correspondence with current customers; direct contacts between individual staff and stakeholders; and research papers and other scholarly activities. For example, NIST distributes a newsletter to customers that includes information on upcoming changes to the agency’s standard reference materials. NIST has also taken steps to better target its stakeholder communication. For example, NIST measurement services officials described an effort to evaluate customer interest in NIST’s standard reference materials, as expressed through contact with NIST staff at trade shows. As a result of this analysis, NIST reduced the number of trade shows at which it advertised these materials—focusing on those trade shows that were identified as having the greatest number of interested attendees. More broadly, by 2018 NIST implemented new information systems supporting its measurement services sales, inventory, and customer relationship management. Measurement services officials described efforts currently underway to take advantage of these systems to better target customers by, for example, providing email notifications to customers of new materials or improved measurements of current materials. NIST’s Efforts Help Support U.S. Representation in International Documentary Standards Activities but May Not Fully Address the Challenge NIST’s efforts help support U.S. representation in international standards organizations but may not fully implement its role and address the challenges it faces. NIST works to support U.S. industry’s efforts to ensure that its interests are adequately represented in international standards activities. For example, NIST officials said that NIST staff participate broadly in international standards activities that are aligned with NIST’s priorities and share their technical expertise in various committees. Through its participation in SDOs, NIST may obtain information on international standards activities in which U.S. industry representation is needed. NIST also shares information on international standards activities with U.S. industry. NIST hosts the World Trade Organization Inquiry Point, a U.S. government website that serves as a communications hub for information on international standards and related issues. Through the website, U.S. industry and other stakeholders receive notifications of standards-related regulations and procedures, as well as the basis and objective for proposed measures, among other information provided. The website also provides a mechanism to circulate comments on proposed measures. Further, we noted above that NIST participates in ANSI standards panels and collaboratives and the ANSI Government Members Forum. NIST’s participation in these bodies can alert NIST to important international standards activities. NIST officials said that when NIST has become aware of concerns about U.S. representation at an international standards activity within a federal government area of responsibility, it has led efforts to ensure adequate representation in those activities. For example, according to NIST officials, NIST: Worked with the U.S. Patent and Trademark Office to help identify another organization to represent U.S. interests in a standards activity that affected U.S. intellectual property after the original organization decided not to continue participating. Established and administered the U.S. technical advisory group for a new ISO technical committee on biotechnology after industry and many federal agencies chose not to participate. Took on a leadership role to represent U.S. interests in a standards activity at the International Telecommunication Union when no U.S. telecommunications companies took on responsibility for representing U.S. industry in the activity. NIST has also issued three reports on U.S. representation in international SDOs. In June 2014 NIST issued its most recent report on U.S. representation in two international SDOs, the ISO and IEC—its two prior reports were published in 2000 and 1988. The 2014 report describes U.S. representation in ISO and IEC activities from 1966 through 2012. As noted previously, the report also describes U.S. memberships and roles in ISO and IEC standards development committees, and includes data comparing U.S. representation on these committees with representation from other countries. While NIST has helped support U.S. representation in international documentary standards activities, NIST has not developed a mechanism to implement the role delegated to it under the 1979 act to address circumstances when U.S. representation at international standards organizations may be inadequate. As noted previously, the 1979 act directs the Secretary of Commerce to coordinate with the U.S. Trade Representative (USTR) and keep adequately informed regarding international standards-related activities and identify those that may substantially affect the commerce of the United States. The Secretary is also to monitor the adequacy of U.S. representation in private international standards activities. Further, the 1979 act establishes a process for the Secretary to follow to address circumstances in which U.S. representation may be inadequate, specifically: If the Secretary, after an inquiry, has reason to believe that the participation by an organization member in the proceedings of a private international standards organization will not result in the adequate representation of United States interests that are, or may be, affected by the activities of such organization (particularly with regard to the potential impact of such activity on the international trade of the United States) the Secretary shall immediately notify the organization member concerned. The organization member has a 90-day period following the Secretary’s notification to demonstrate its willingness and ability to represent adequately U.S. interests. If the organization member demonstrates willingness and ability, the Secretary should take no further action. If the organization member either does not respond or does not demonstrate the requisite willingness or ability to represent U.S. interests or there is no organization member of the private international standards organization—the Secretary is to make arrangements to provide for the adequate representation of U.S. interests. Although NIST has reported on the extent to which the U.S. participates in some Geneva-based international standards development activities, these reports do not assess the adequacy of this participation. NIST officials we interviewed said that the agency does not assess whether U.S. interests are adequately represented and does not have definitions of or guidelines for what constitutes adequate representation. NIST officials also told us that the agency has not evaluated the circumstances under which it would follow the procedures under the act for addressing inadequate representation. Federal standards for internal control indicate that management should identify, analyze, and respond to risks related to achieving an agency’s objectives. As noted previously, the large number of international standards activities occurring across numerous industry sectors, among other factors, present challenges to ensuring adequate U.S. representation in international standards activities. In commenting on a draft of this report, Commerce stated that a determination to follow the statutory process in the 1979 act would carry significant risk of being perceived by the national and international standards community as a U.S. government change of policy relating to the nation’s private-sector-led standards system. However, the memorandum of understanding between NIST and ANSI states that NIST’s role, under the NTTAA, OMB A-119, and the 1979 act is “to ensure adequate representation of U.S. interests in all relevant international standards organizations.” Further, NIST has previously taken action in some cases in international standards activities within a federal government area of responsibility, as described above. Additionally, the ongoing contacts between NIST and SDOs and staff participation in standards activities can help NIST keep adequately informed on international-standards-related activities. Without a mechanism to identify and respond to circumstances when U.S. representation at international SDOs may be inadequate, such as guidelines for what constitutes adequate representation and when and how to follow the process under the 1979 act, NIST may miss opportunities to take action in furtherance of its mission to support U.S. competitiveness by helping to ensure adequate U.S. representation in international standards activities. Alternatively, given Commerce’s concerns about the statutory process in the 1979 act, NIST could develop a legislative proposal that allows NIST to ensure adequate U.S. participation in international standards activities while addressing those concerns. NIST’s Efforts Help Support Interagency Coordination, but Opportunities Exist to Better Follow Leading Practices for Enhancing Collaboration NIST supports coordination among federal agencies on documentary standards issues as chair of the ICSP, as well as additional coordination efforts outside of the ICSP. However, aspects of the ICSP’s efforts do not fully align with selected leading practices for enhancing and sustaining interagency collaboration. These practices can help agencies manage fragmentation and other coordination challenges. NIST has taken several steps to address the challenge of interagency coordination on documentary standards issues through its efforts as chair of the ICSP: NIST and member agencies have a charter that outlines the purpose, functions, and membership of the ICSP, among other information. According to the charter, the ICSP was established to advise the Secretary of Commerce and the heads of other federal agencies in matters relating to standards policy. The ICSP’s purpose under the charter is to ensure effective participation by the federal government in domestic- and international-standards activities, among other things. NIST officials we interviewed said that the ICSP meets three to four times per year and that the purpose of the ICSP is to promote effective participation by federal agencies in the standards process— when it is within an agency’s mission and in the public interest—but not to force their participation. According to NIST officials, NIST tries to demonstrate the benefits of participation and encourages other agencies to participate actively in relevant standards activities. NIST chairs and supports ICSP activities, including providing administrative services, organizing meetings, and developing agendas and reports. We previously reported in 2012 that designating a lead agency can assist in driving accountability and providing for continuity of leadership for a collaborative effort. According to NIST officials we interviewed, ICSP meetings are open to agencies outside of the member agencies. Further, NIST officials said they routinely invite staff from non-member agencies when the committee plans to discuss items of particular interest to them. NIST officials said that the ICSP network allows NIST to provide a knowledge base for other federal agencies, and to help federal staff understand standards policy and participation in SDOs, among other things. According to ICSP members and NIST officials, NIST facilitates the sharing of best practices on broad standards topics affecting multiple agencies through the ICSP. As of June 2018, 30 federal agencies have identified participants to the ICSP, while 5 agencies have vacant positions on the committee. NIST officials provided several examples of its information-sharing activities: NIST led ICSP efforts to facilitate discussion on, and manage revisions to, key guidance regarding federal agencies’ standards efforts in OMB Circular A-119. NIST led ICSP efforts to promote awareness on and share information related to the development of corporate social responsibility standards in ISO. NIST created an ICSP working group on conformity assessment to help address issues that were frequently being raised during ICSP meetings. NIST invites speakers to share information with ICSP members on various standards-related issues. For example, NIST invited officials from ANSI to present information on standards areas of agency interest and also invited the members of the SDO leading efforts on smart grid standards to brief agencies on the SDO’s efforts. According to six agency standards executives we interviewed, the ICSP helps members to share information, including best practices and to have a general awareness of pertinent documentary standards topics. Three agency standards executives said that the ICSP helps standards executives to know each other on a personal basis so that they know whom to contact to coordinate on standards activities. Bringing agency standards executives together through the ICSP can also spur coordination among agencies if a topic of mutual interest is identified. For example, two agency standards executives said that questions and suggestions raised by agencies at the ICSP led to coordination with other agencies on a standards area of mutual interest. NIST also coordinates with individual agencies on documentary standards activities outside of the ICSP. For example, according to participants in four focus groups, relevant NIST and VCAT reports, and congressional testimonies we reviewed, NIST is particularly strong in bringing relevant federal agencies and other stakeholders together to develop standards and frameworks for individual interdisciplinary technology areas. Such coordination can occur through a variety of methods or groups, such as workshops, that address standards-related issues. Additionally, four agencies’ standards executives said that they coordinated extensively with NIST on specific standards activities. Two of the agency standards executives described how collaboration with NIST on research helped inform their agencies’ standards activities. NIST also offers training, such as NIST’s standards boot camp, according to federal standards executives. Five agency standards executives said staff from their agencies attended the training and four standards executives said the training had improved their staff’s competence in standards. While coordination between NIST and other federal agencies on documentary standards issues occurs in a variety of ways, the ICSP is the primary body established to facilitate interagency coordination on standards policy, according to NIST standards officials. However, some of the ICSP’s efforts to support coordination among federal agencies on standards issues do not fully align with selected leading practices for interagency collaboration we identified in our previous work. Specifically, the ICSP charter has not been updated; ICSP member agencies’ roles and responsibilities have not been fully clarified; and the ICSP may not include relevant members to carry out its functions. Additionally, we reported in 2015 that while collaborative mechanisms differ in complexity and scope, following leading practices can help manage fragmentation and other coordination challenges. ICSP charter: The ICSP charter has not been updated since it was signed in October 2000. According to the charter, the need for and mission of the ICSP was to be reexamined 3 years after the charter was created. However, NIST officials said that the ICSP charter has not been reexamined. We reported in 2012 that agencies that articulate their agreements in formal documents can strengthen their commitment to working collaboratively. We also reported that written agreements are most effective when they are regularly updated and monitored. Further, updating written agreements, such as the ICSP charter, can be an opportunity for members to define common goals and purpose. In addition, focus group participants, stakeholders we interviewed, and literature we reviewed described broad changes in documentary standards that have led to new challenges. For example, we previously noted the challenges related to emerging interdisciplinary standards issues and the increase in the number of international standards activities, and both of these areas can benefit from federal coordination. Further, the ICSP charter has not been updated to reflect the 2016 revisions to OMB Circular A-119 or the guidelines provided to agencies in the 2012 EOP memo on engagement in standards activities to address national priorities. For example, the revised A-119 notes several executive orders relating to review and coordination that were not in existence at the time of the charter’s creation. Additionally, the EOP memo that outlines agency responsibilities for standards areas of national priority was also not in existence at the time of the ICSP charter’s creation. The EOP memo calls on agencies to ensure effective intra and interagency coordination of engagement in standards development activities. Without reexamining and updating the ICSP charter, as necessary, NIST and other ICSP member agencies cannot be assured that their collaborative efforts are best structured to address current standards challenges. ICSP member roles and responsibilities: ICSP member agencies’ roles and responsibilities have not been fully clarified to an extent that would help the ICSP fulfil its purpose, objectives, and functions to gather information and make recommendations to the Secretary of Commerce to strengthen standards policy and coordination. The ICSP charter outlines two basic functions for the committee: (1) gathering, analyzing, and maintaining current information about standards and other specified related information and (2) on the basis of such information, and when appropriate, making recommendations to the Secretary of Commerce to achieve various standards-related objectives, such as strengthening coordination of standards-related policies and activities among federal agencies. The charter also specifies that the ICSP may create task groups as appropriate. However, we found several areas in which the ICSP charter could more fully clarify member agencies’ roles and responsibilities in regard to implementing its functions, purpose, and objectives, for example: The ICSP charter does not fully clarify the ICSP role and member agencies’ responsibilities for identifying and coordinating on interdisciplinary standards issues that cross agency boundaries. Five agency standards executives we interviewed as well as NIST standards officials described the ICSP as primarily an information- sharing or networking body, with little role in establishing federal positions on standards activities or policy, or making joint policy decisions with respect to specific standards issues. While information sharing is an important component of interagency coordination, OMB Circular A-119 gives agency standards executives responsibility for consulting with other relevant agencies on standards issues to avoid, to the extent practicable, expressing inconsistent views on standards issues. Furthermore, the 2012 EOP memo specifies that agencies should periodically review their standards activities to identify gaps in representation for mission-critical areas and should ensure effective coordination of engagement in standards development activities. NIST officials and one standards executive said that the ICSP could identify emerging or interdisciplinary standards issues that may require more active federal roles and coordination; however, the charter does not specify the ICSP’s role and member responsibilities regarding interdisciplinary standards areas that may cut across agencies. The ICSP charter does not fully identify member agencies’ responsibilities for coordinating on international standards issues. OMB Circular A-119 and the ICSP charter specify that the ICSP has a role in coordinating federal agencies’ international standards activities. However, the ICSP charter does not fully identify member agencies’ responsibilities toward fulfilling this role. NIST standards officials and one agency standards executive said that the ICSP does not typically address international standards issues, or policy, or coordinate federal positions on international standards. While NIST officials said other USTR-led efforts could help coordinate agencies’ international standards activities, a USTR official told us that USTR does not have the technical expertise needed to effectively coordinate multiple agencies’ views on standards. Further, the USTR official, one NIST standards official, and representatives from ANSI said that different agencies’ positions are taken into account as part of the standards development process at SDOs that have an open process, such as ANSI. However, as mentioned previously, A-119 specifies that the ICSP is to coordinate with other interagency entities with a view to encouraging more strategic and coordinated federal participation in the development and use of standards. Further, the ICSP charter also specifies that the ICSP is to ensure effective federal participation in international standards activities. The ICSP charter does not fully identify member agencies’ responsibilities for developing joint recommendations to the Secretary of Commerce. The charter describes eight areas in which the ICSP shall make recommendations when appropriate, including to strengthen agency coordination and to improve the efficiency of standardization efforts within the federal government. Further, the charter specifies an administrative process for voting on ICSP recommendations. However, according to NIST officials, the ICSP has never made a recommendation to the Secretary of Commerce to address a standards-related issue. Further, NIST officials said that the ICSP agencies have not shown interest in acting jointly. NIST officials said that there may be circumstances in which making such a recommendation would be appropriate, for example, to strengthen interagency coordination on interdisciplinary standards issues, although the officials said that the ICSP would try to address issues at a lower level before elevating them to the Secretary of Commerce. Further clarifying agencies’ responsibilities may help ensure that the ICSP is able to meet this function of its charter. The ICSP charter does not fully identify member agencies’ roles and responsibilities for creating and participating in ICSP’s task groups. We have previously reported that task groups can be an effective mechanism for agencies to collaborate on joint challenges. NIST standards officials said that one task group has been created and that task groups may be appropriate either (1) when a standards issue may require more focused and sustained monitoring to understand possible effects on U.S. government activities and missions or (2) when an ICSP member suggests the need for a task group and there is a consensus among membership. However, these reasons are not specified in the charter or other documents available on the ICSP website. Further, while NIST officials said that no member has requested the creation of a task group, two standards executives identified standards coordination issues that they thought may benefit from the creation of an ICSP task group. The charter specifies an administrative process for voting to create an ICSP task group but does not specify what the role of task groups are, why they would be created, or ICSP member agencies’ responsibilities in determining the need for and participating in task groups. We have previously reported that to achieve a common outcome, participating agencies should consider clarifying roles and responsibilities. By agreeing on and clearly defining the roles and responsibilities of the members as well as documenting those decisions, collaborating agencies can clarify which agency will do what, organize their joint and individual efforts, and facilitate decision-making. Without ensuring that member agencies’ roles and responsibilities have been fully clarified, NIST and the ICSP may miss opportunities to strengthen agencies’ coordination on standards issues, and better ensure effective coordination related to standards activities. Further, without fully clarifying federal agencies’ roles and responsibilities, the ICSP may also miss opportunities to address standards challenges noted above, which limit its ability to support U.S. competitiveness. ICSP membership: The ICSP may not include all relevant agencies as members or invited observers. The ICSP is comprised of certain specified agencies that are represented by their standards executives as described in OMB Circular A-119. A-119 provides that federal standards executives should be broadly engaged in the agency’s standards-related activities so as to ensure intra-agency coordination and have sufficient authority to ensure compliance with Circular A-119. In addition, the ICSP charter allows the Secretary of Commerce to invite additional members— a role which has been delegated to NIST as chair of the ICSP. Consistent with leading collaboration practices, it is important to ensure that all relevant agencies are included in a collaborative effort. In addition, participants should also have full knowledge of relevant resources in their agency, and the skills and abilities to commit relevant resources and contribute to the outcomes of the collaborative effort, among other attributes. While the chair of the ICSP said that agency standards executives are engaged in understanding their standards activities and that participation in the ICSP is strong, as we noted above, it can be difficult for standards executives to be fully aware of all standards activities in their department or agency, particularly in large agencies. In addition, an agency standards executive, a NIST stakeholder, and NIST standards officials raised concerns about whether standards executives have sufficient time to understand all their agencies’ activities and needs, given their other duties, or whether standards executives have the authority to fully coordinate on standards activities. In some cases, agency officials other than the standards executive may have greater knowledge and expertise about specific standards issues. For example, sub-component offices and agencies may have numerous standards-related activities, such as the Food and Drug Administration within Health and Human Services. According to interviews with two agency standards executives, some sub- component agencies may also have more knowledge and expertise in significant standards areas, and some of these areas can affect multiple agencies. According to one NIST standards official, NIST invites additional agencies when it learns that an agency could potentially add value but has not conducted a comprehensive assessment of ICSP membership within the last 5 years. By assessing whether relevant agencies and offices have been invited to participate as members or observers, the ICSP would have greater assurance of its ability to ensure effective participation by the federal government in domestic- and international-standards activities. Further, having relevant parties involved at the ICSP could enhance the ICSP’s efforts by ensuring the viewpoints of all relevant agencies are considered. Conclusions In NIST’s role as the nation’s measurement science laboratory, NIST works to improve how we design, build, and test the technologies around us. Further, NIST’s measurement services and support of documentary standards development can directly affect innovation and the nation’s economy by helping companies produce better products and compete in the global economy. However, factors such as the breadth of industry needs, number of domestic- and international-standards development activities, and the fragmented nature of federal agencies’ involvement in documentary standards development create challenges to NIST’s ability to fulfill its mission of promoting U.S. innovation and industrial competitiveness. NIST has taken steps to address these challenges. For example, NIST’s expert scientists and engineers maintain close contact with industry through a variety of mechanisms, and use their expertise to help identify industry needs and to communicate about NIST’s services. NIST has also established procedures to help the agency prioritize and evaluate the effectiveness of its measurement services and ensure that supervisors and laboratory management oversee agency staff participation in documentary standards activities. However, NIST has not comprehensively evaluated the extent to which its efforts align with stakeholder needs. Conducting comprehensive reviews of its activities would provide NIST with greater confidence that its activities align with stakeholder needs and may help identify areas not currently served by NIST. Similarly, NIST could gain confidence in the effectiveness of its participation in documentary-standards development activities: by completing the comprehensive review called for in NIST policy, by improving its guidance to staff, and by taking steps to enhance agency management’s understanding of the time devoted to NIST’s current efforts. NIST staff’s expertise and lack of bias make them valuable contributors in documentary-standards development activities, but these individuals have limited time for such activities. Without additional guidance regarding the factors staff could consider when deciding to take more active roles in standards development activities, NIST cannot be assured that decisions on when to participate in such activities are made consistently. Further, while determining the exact amount of time spent on any one standards activity may not be possible due to overlap with other employee duties, examining to the extent possible the aggregate amount of time NIST staff expect to commit to standards development activities could help NIST management assess the agency’s participation in these activities. Further, NIST has a role, delegated by the Secretary of Commerce, to ensure U.S. interests are adequately represented at private international standards organizations but does not currently have a mechanism to use the process under the Trade Agreements Act of 1979 to identify or respond to circumstances when U.S. representation in international standards activities may be inadequate. Developing a mechanism could help ensure that NIST does not miss opportunities to ensure that the United States is adequately represented in international standards activities. Alternatively, NIST could develop a legislative proposal that allows NIST to ensure adequate U.S. participation in international standards activities while addressing any concerns. Finally, as chair of the ICSP, NIST provides leadership to enhance interagency coordination on documentary standards issues. However, some of the ICSP’s efforts do not fully align with selected leading practices for enhancing and sustaining interagency collaboration, as identified in our previous work. Specifically, the ICSP charter has not been updated since 2000 and certain roles and responsibilities for the ICSP and its members are unclear. Further, the ICSP membership may not include relevant members or observers to ensure effective participation by the federal government in domestic- and international- standards activities. Updating the ICSP charter to affirm its mission and to delineate appropriate roles and responsibilities of participating agencies could strengthen interagency coordination through the ICSP on standards development issues and help the ICSP fulfill its role as envisioned under OMB guidance. Moreover, assessing whether relevant agencies or sub- component offices are invited to participate in the ICSP could provide the ICSP with better assurance of its ability to effectively coordinate agencies’ standards activities. Recommendations for Executive Action We are making a total of seven recommendations to NIST, specifically: The NIST Associate Director for Laboratory Programs should: update NIST policy to include periodic comprehensive management review of the agency’s measurement services to assess gaps and ensure alignment with stakeholders’ needs, and take steps to ensure that the Associate Director completes the review of NIST’s standards development activities (Recommendation 1). The NIST Standards Coordination Office Director should: update NIST policy for staff participation in standards development activities to provide additional guidance, such as the factors staff could consider when deciding to take more active roles, including leading efforts to develop standards (Recommendation 2); and assess the feasibility of collecting aggregate data on the estimated amount of time staff spend on documentary standards activities (Recommendation 3). The Director of NIST should establish a mechanism—such as guidelines for what constitutes adequate U.S. representation—to assess whether U.S. representation in international SDOs is adequate, and when to follow the statutory process for addressing inadequate U.S. representation. If NIST determines that it is unable to implement the process described in the 1979 act without conflicting with current standards policy, the Director of NIST should develop a legislative proposal to address those concerns (Recommendation 4). The NIST Standards Coordination Office Director, working with other ICSP member agencies, should: review and, as necessary, update the ICSP charter (Recommendation 5); clearly define ICSP roles and member agencies’ responsibilities, such as for identifying and coordinating on interdisciplinary documentary- standards issues and for making recommendations, as appropriate, to the Secretary of Commerce (Recommendation 6); and, assess whether additional agencies or sub-component offices should be invited to participate as ICSP members or observers (Recommendation 7). Agency Comments and Our Evaluation We provided a draft of this report to the Departments of Commerce, Defense, Energy, and Homeland Security, and Health and Human Services, as well as the Consumer Product Safety Commission, Environmental Protection Agency, General Services Administration, and USTR for their reviews and comments. Commerce provided written comments, which are reproduced in appendix III, in which it generally concurred with six recommendations and disagreed with one. USTR provided technical comments, which we incorporated as appropriate. The remaining seven agencies informed us that they had no comments. Commerce agreed with our recommendation regarding comprehensive reviews of the agency’s measurement services and standards participation, and stated that it will include requiring management review of its measurement services in a future agency order and will have the Standards Coordination Office support the Associate Director’s review of staff participation in standards activities. Commerce also agreed with our recommendations regarding guidance on staff participation in standards activities and assessing the feasibility of collecting data on the time spent on documentary standards activities. Specifically, Commerce said that it will consider updates to guidance to staff and will report on the feasibility of collecting data on the time spent on standards activities. Commerce also agreed with recommendations on improving interagency coordination in the ICSP. It stated that the chair of the ICSP will review the charter and recommend any updates to the committee and will work with existing ICSP members, alternates, and observers to identify other agencies or sub-component offices that may be invited to participate. While Commerce agreed with our recommendation on roles and responsibilities of the ICSP and its members, it stated that the roles and responsibilities of standards executives are effectively stated in OMB A- 119. A-119 describes the roles and responsibilities of standards executives in general, such as their responsibilities to promote effective use of agency resources and participation in standards bodies. However, A-119 does not address specific roles and responsibilities with respect to the activities of the ICSP, such as defining the role of the ICSP in establishing federal positions on standards issues. In our 2012 report on interagency collaborative mechanisms, we state that agencies working together to define and agree on their respective roles and responsibilities can help clarify who will do what and identify how to organize individual and joint efforts. We believe that additional efforts by ICSP members to clarify their roles and responsibilities within the framework of the ICSP will improve the effectiveness of the ICSP as a coordinating body. Commerce disagreed with the recommendation on ensuring adequate U.S. representation in international SDOs, stating that any determination to follow statutory process would itself carry significant risk of being perceived by the national and international standards community as a U.S. government change of policy relating to our private-sector-led standards system. However, we do not perceive a conflict between the private sector leading U.S. standards development, and NIST developing a mechanism to respond to any instances where U.S. representation in international standards efforts are inadequate. NIST has already taken steps, in some cases, to identify or encourage private sector participation in international standards efforts that lacked U.S. participation. Further, the 1979 act does not define what constitutes adequate representation; NIST can develop criteria to allow the agency to take action when appropriate. In carrying out the 1979 act, the agency could continue to encourage private-sector stakeholders to address any areas determined to be inadequate, allowing NIST to step in only as necessary when the relevant private-sector entities are not adequately representing, and are not willing and able to adequately represent, U.S. interests. We continue to believe that a mechanism to identify and respond to circumstances when U.S. representation at international SDOs may be inadequate would allow NIST to consistently take action in furtherance of its mission to support U.S. competitiveness. However, in consideration of the concerns raised by Commerce, we have clarified our recommendation that NIST either develop a mechanism to carry out the process described by the 1979 act, or develop a legislative proposal to address any concerns arising from the implementation of the act. We are sending copies of this report to the appropriate congressional committees, the Secretary of Commerce, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-3841 or neumannj@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix IV. Appendix I: Objectives, Scope, and Methodology This report examines (1) the challenges the National Institute of Standards and Technology (NIST) faces in providing measurement services and supporting documentary-standards development and (2) the extent to which NIST has taken steps to address any challenges and how those steps align with relevant federal guidance and policy. To identify any challenges NIST may face in providing measurement services and supporting documentary-standards development, we began by performing a literature review, including reports on NIST from the National Academies of Sciences, Engineering, and Medicine, NIST’s Visiting Committee on Advanced Technology (VCAT), congressional committee hearings on NIST’s measurement services, and other sources. We reviewed these sources to identify statements regarding any challenges to NIST’s current measurement services and documentary standards activities or statements recommending improvements to these services or activities. For purposes of our analysis, we included NIST’s efforts to support measurement standards as a component of NIST’s measurement services. We supplemented our literature review by holding focus groups with NIST stakeholders including: (1) researchers and (2) representatives working with industry, including commercial entities and states’ metrology laboratories. To ensure our focus groups contained a diverse group of stakeholders and viewpoints on NIST’s measurement services and documentary standards activities, we included participants from a variety of backgrounds. We selected researchers to participate in our focus groups from university scientists in engineering and the physical and biological sciences. We selected industry participants for the focus groups to reflect a range of industrial sectors, including (1) sectors that comprise greater than 1 percent of U.S. gross domestic product, according to Bureau of Economic Analysis data and (2) sectors that the Department of Commerce’s International Trade Agency has identified as U.S. export opportunities. Across these sectors we selected industry participants from the following categories: broadcasting and telecommunications; chemical products and pharmaceuticals; computer and electronic products and related services; building products and construction; finance and insurance; food and beverage and tobacco products; health care and social assistance; transportation; utilities and energy; defense products; and other manufacturing. We also included representatives from the National Conference on Weights and Measures, an organization of commercial entities and state metrology laboratories, among others, that addresses measurement of commercial products. After developing our focus group structure and determining our participant categories, we obtained feedback on our approach during discussions with NIST officials and with representatives from standards development organizations (SDO) selected from those that NIST most often collaborates with—the American National Standards Institute, ASTM International, and IEEE. We then used a snowball approach to identify and invite individuals from across our participant categories. Starting with individuals from several SDOs and the VCAT, we asked for suggestions of individuals knowledgeable in measurement services and standards needs of our participant categories. As we received responses and contacted those individuals, we asked them to recommend additional participants. Through this process, we identified and invited nearly 100 individuals to participate in our focus groups, and 58 individuals agreed to participate. To organize our focus groups we asked the individuals who agreed to participate to describe their expertise regarding measurement services and standards development and familiarity with the industrial sectors we identified. We then selected individuals for each focus group based on availability and to include a mix of expertise. We conducted 3 focus groups for representatives from industry and 2 focus groups for researchers. Each focus group included from 5 to 8 individuals. In total, our focus groups included 31 stakeholders. We reviewed transcripts of the focus groups to identify the challenges NIST faces in providing measurement services and supporting documentary-standards development. We also collected information on the challenges that NIST faces during 36 interviews, including 17 interviews with current and former NIST officials, 10 interviews with officials from other federal agencies, and 9 interviews with representatives from SDOs and other stakeholders. The 10 interviews we conducted with other federal agency officials included 8 agency standards executives—senior level officials with knowledge of, and experience in, standards-related issues at their agencies and who are responsible for coordinating their agency’s participation in SDOs, among other responsibilities—or their alternates on the ICSP. The agencies whose standards executives or other officials we interviewed included the Departments of Defense, Energy, and Homeland Security, and Health and Human Services, as well as the Consumer Product Safety Commission, Environmental Protection Agency, General Services Administration, and Office of the U.S. Trade Representative. GAO reviewed our interview notes to identify challenges and NIST’s efforts to address these challenges. To evaluate the steps NIST has taken to address challenges in providing measurement services and supporting documentary-standards development, we drew upon our focus groups, interviews with NIST staff, and reviews of NIST documentation that described the agency’s measurement services and standards activities, such as agency policies, orders, and publications. We also conducted a review of existing literature, relevant laws, NIST policy, and other guidance documents to identify federal requirements and guidance. For example, we reviewed the National Technology Transfer and Advancement Act of 1995, Office of Management and Budget’s Circular A-119, and Executive Office of the President’s Memo on Principles for Federal Engagement in Standards Activities to Address National Priorities, among other sources. We compared the steps NIST has taken to address the challenges it faces in providing measurement services and supporting standards development to these policies and guidance. To evaluate NIST’s current steps, we also considered our prior work on federal standards for internal control and on interagency collaboration. Internal control is a process created by an agency’s management and other personnel that provides reasonable assurance that the objectives of the agency will be achieved and comprises the plans, methods, policies, and procedures used to fulfill the mission and objectives of the agency. Standards for Internal Control in the Federal Government (known as the Green Book), provide the overall framework for establishing and maintaining an effective internal control system and require that agencies perform and document certain actions to establish an effective internal control system. These requirements include: that management should identify, analyze, and respond to risks related to achieving the defined objectives; that management should use quality information to achieve the entity’s objectives; and that management documents the results of evaluations to identify internal control issues. Our work on interagency collaboration describes leading practices agencies can engage in to enhance and sustain collaborative efforts and describes seven key features to consider to implement these practices. We selected the following five features relevant to NIST’s leadership of the Interagency Committee on Standards Policy (ICSP) for review: Outcomes and Accountability: Have short-term and long-term outcomes been clearly defined? Is there a way to track and monitor their progress? Leadership: How will leadership be sustained over the long-term? If leadership is shared, have roles and responsibilities been clearly identified and agreed upon? Clarity of Roles and Responsibilities: Have participating agencies clarified roles and responsibilities? Participants: Have all relevant participants been included? Do they have the ability to commit resources for their agency? Written Guidance and Agreements: If appropriate, have participating agencies documented their agreement regarding how they will be collaborating? Have they developed ways to continually update and monitor these agreements? We did not review ICSP collaboration with respect to key features regarding resources or bridging organizational culture because we did not fully examine the activities of all agencies participating in the ICSP. We conducted this performance audit from July 2016 to July 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Focus Group Participants The following individuals participated in GAO’s five focus groups: Allen Adler, Former Vice President of Enterprise Technology Strategy, Boeing Kathleen Almand, Vice President for Research, Data, and Analytics, National Fire Protection Association Karin Athanas, Government and Regulatory Affairs Manager, American Association for Laboratory Accreditation (A2LA) Robert Austin, Professor of Physics, Princeton University Karl Bly, Quality Assurance Director, Vermont Thread Gage Jerry Buendel, Weights and Measures Program Manager, Washington State Department of Agriculture Rita Colwell, Professor, University of Maryland, and Johns Hopkins University School of Public Health Ross Corotis, Professor of Engineering, University of Colorado Boulder Denyette DePierro, Vice President and Senior Counsel, Center for Payments and Cybersecurity, American Bankers Association Don Detmer, Professor of Public Health Sciences, University of Virginia Gail Folena-Wasserman, Senior Vice President, Biopharmaceutical Development, MedImmune Ruben G. Carbonell, Professor of Chemical Engineering, North Carolina State University and Chief Technology Officer, National Institute for Innovation in Manufacturing Biopharmaceuticals (NIIMBL) Christopher Guay, Regulatory Fellow, Procter and Gamble Kelvin H. Lee, Professor of Chemical and Biomolecular Engineering, University of Delaware and Director, National Institute for Innovation in Manufacturing Biopharmaceuticals (NIIMBL) Hani Haider, Director of Orthopaedics Biomechanics & Advanced Surgical Technologies Laboratory, University of Nebraska Jennie Hwang, CEO and Principal, H-Technologies Group Walter Jager, Principal, Environmentally Conscious Design (ECD) Compliance Karen Kafadar, Professor and Chair of Department of Statistics, University of Virginia Dave Kreitlow, Operations Manager, MTS Systems Corporation Zhiyong Ma, Vice President and Director of Technology and Manufacturing Labs, Intel Kristin Macey, Director of Division of Measurement Standards, California Department of Food and Agriculture Josh Magri, Vice President and Counsel for Regulation and Developing Technologies, Financial Services Roundtable Dave Maisch, Director of Engineering and Industrial Affairs, PMC Lone Star Tod Sizer, Vice President of Mobile Radio Research Laboratory, Nokia Bell Labs Clifford Spiegelman, Distinguished Professor of Statistics, Texas A&M University Lonnie Spires, President and CEO, American Association for Laboratory Accreditation (A2LA) Appendix III: Comments from the Department of Commerce Appendix IV: GAO Contact and Staff Acknowledgements GAO Contact Staff Acknowledgements In addition to the contact named above, Chris Murray (Assistant Director), Tind Shepper Ryen (Analyst-in-Charge), John Delicath, Justin Fisher, Eli Harpst, Tricia Moye, Danny Royer, Andrew Stavisky, and Sarah Veale made key contributions to this report.

Why GAO Did This Study The U.S. Department of Commerce's NIST provides measurement services and supports standards that promote U.S. competitiveness. For example, NIST provides calibrations for manufacturing equipment and reference materials used in testing. NIST also supports private sector organizations in developing standards to help ensure product performance, among other things, such as Wi-Fi. In recent years, NIST has sought to improve the delivery of its services and documentary standards activities. GAO was asked to review NIST measurement services and standards-support activities. This report examines (1) the challenges NIST faces in providing measurement services and supporting documentary standards development, and (2) the extent to which NIST has taken steps to address these challenges and how those steps align with federal guidance and policy. GAO analyzed testimony, reports, laws, and policies; conducted focus groups with academics and industry representatives; and interviewed various stakeholders. What GAO Found The National Institute of Standards and Technology (NIST) faces challenges in providing measurement services and supporting private sector development of specifications for products' designs or performance—referred to as “documentary standards.” Based on reviews of relevant testimony, reports, and other documents; interviews with stakeholders; and focus groups with academics and industry representatives, GAO identified challenges including: Identifying and prioritizing what measurement services, such as calibrating large force-measurement tools used by aerospace manufacturers, or what documentary standards activities, such as serving as a technical advisor on fire safety standards, are most needed by U.S. industry, and Coordinating with other federal agencies on standards development issues. NIST has taken steps to address these challenges, including industry outreach and reviews of measurement services and standards activities. However, some efforts do not fully align with federal guidance or NIST policy. For example, NIST's measurement-services and standards-activity reviews have not included a comprehensive examination of how these services and standards activities align with stakeholder needs. Federal internal control standards call for managers to use quality information to determine if the agency is meeting its objectives. Comprehensively reviewing NIST's measurement services and documentary-standards activities would provide NIST with greater confidence that its services and activities align with stakeholders' needs. GAO also found that NIST coordinates with other agencies on standards development and related activities, but that some efforts do not fully align with specific leading practices GAO has previously identified for enhancing and sustaining interagency collaboration. For example, NIST and other agencies coordinate on standards activities through a NIST-chaired interagency committee. However, GAO found that the committee has not updated its charter since 2000—contrary to leading practices to update and monitor collaborative agreements. GAO also found that NIST has not worked with other committee members to fully clarify agencies' roles and responsibilities. Without ensuring that member agencies' roles and responsibilities are current and fully clarified, NIST and other agencies may miss opportunities to strengthen coordination. What GAO Recommends GAO is making seven recommendations, including that NIST comprehensively review measurement services and documentary-standards activities, and work with other agencies to take steps to strengthen interagency coordination. The Department of Commerce agreed with six recommendations and disagreed with one, citing risks to the private-sector-led U.S. standards system. GAO clarified its recommendation and continues to believe this action is needed, as discussed in the report.

gao_GAO-18-241

gao_GAO-18-241_0

Background This section describes DOE’s tank waste treatment approach at Hanford and DOE’s quality assurance framework and requirements. DOE’s Tank Waste Treatment Approach Cleanup of the Hanford Site is governed by two main compliance agreements: (1) the 1989 Hanford Federal Facility Agreement and Consent Order, or Tri-Party Agreement, an agreement between DOE, the Washington State Department of Ecology, and the Environmental Protection Agency, and (2) a 2010 consent decree. The Tri-Party Agreement was signed in May 1989 and lays out a series of legally enforceable milestones for completing major activities in Hanford’s waste treatment and cleanup process. The Tri-Party Agreement has been amended a number of times to establish additional enforceable milestones for certain WTP construction and tank waste retrieval activities, among other things. Under the Tri-Party Agreement, DOE must complete waste treatment at the Hanford Site by 2047. The overall mission of the WTP is to treat and immobilize a large part of 54 million gallons of radioactive and chemical waste stored in 177 underground storage tanks. The WTP is the most technically complex and largest construction project within DOE’s Office of Environmental Management, occupying 65 acres of the Hanford Site. Some of DOE’s tank waste is highly radioactive material—known as high-level waste— mixed with hazardous waste. Under current law, this waste must be vitrified—a process in which the waste is immobilized in glass—prior to disposal. Low-activity waste is DOE’s term for the portion of the tank waste at Hanford with low levels of radioactivity. Low-activity waste is primarily the liquid portion of the tank waste that remains after as much radioactive material as technically and economically practical has been removed. The WTP consists of the following set of facilities that are designed to separate waste into low-activity and high-level waste streams and, once completed, treat these waste streams in separate facilities using vitrification. Pretreatment Facility. This facility is to receive the waste from the tanks and separate it into high-level and low-activity waste streams. Low-Activity Waste Facility. This facility is to receive the low-activity waste from the Pretreatment facility and immobilize it by vitrification. The canisters of vitrified waste will be permanently disposed of at another facility at Hanford. High-Level Waste Facility. This facility is to receive the high-level waste from the Pretreatment Facility and immobilize it by vitrification. The canisters of vitrified waste will be stored on-site until a final repository is established. Effluent Management Facility. The Effluent Management Facility is being built to evaporate much of the secondary waste produced during low-activity waste processing and vitrification at the Low- Activity Waste Facility. Analytical Laboratory. This facility will conduct analyses as needed, such as testing samples of the vitrified waste to ensure that it meets certain criteria and regulatory requirements for disposal. Balance of Facilities. These facilities consist of the 22 support facilities that make up the plant infrastructure, such as cooling water systems and silos that hold vitrifying materials. In part because of the 2012 work stoppage at the WTP’s Pretreatment and High-Level Waste Facilities, in 2012 DOE adopted a phased waste treatment strategy through which the department aims to begin treating some of the low-activity waste before resolving all WTP technical issues. During the first phase of this strategy, DOE plans to implement a Direct Feed Low-Activity Waste (DFLAW) approach to transfer some low- activity waste from the tanks to the WTP’s Low-Activity Waste Facility for vitrification before the Pretreatment Facility is completed. The approach relies on construction of a new facility—the Low-Activity Waste Pretreatment System—designed to remove highly radioactive particles from liquid tank waste before sending the waste stream to the Low- Activity Waste Facility. During later phases, DOE intends to complete the WTP Pretreatment Facility and High-Level Waste Facilities. DOE also plans to construct a Tank Waste Characterization and Staging Facility under a different contract to stage, mix, sample, and characterize high- level waste from the tanks prior to delivery to the Pretreatment Facility. Figure 1 illustrates WTP and other facilities planned for Hanford tank waste treatment. DOE’s Quality Assurance Framework and Requirements A set of federal regulations, DOE orders, and ORP procedures collectively make up DOE’s quality assurance framework that aims to ensure that all WTP quality assurance problems can be identified and that identified problems do not recur. DOE’s quality assurance regulations require DOE contractors to establish DOE-approved quality assurance programs. The regulations specify that under an approved program, the contractor’s quality assurance program must, among other things, (1) establish and implement processes to detect and prevent quality problems; (2) identify, control, and correct items, services, and processes that do not meet established requirements; (3) procure items and services that meet established requirements and perform as specified; (4) plan and conduct independent assessments to measure item and service quality, to measure the adequacy of work performance, and to promote improvement; and (5) maintain items to prevent damage, loss, or deterioration. In addition, DOE Order 226.1B requires that DOE’s organizations and contractors implement oversight processes that ensure that relevant quality assurance problems are evaluated and corrected on a timely basis to prevent recurrence. The WTP contract requires compliance with these regulations and requirements. The WTP contract specifies that as the owner of the WTP project, DOE is responsible for providing quality assurance oversight of the WTP. ORP’s Quality Assurance Division provides such oversight, for example, by doing the following: Reviewing a sampling of the contractor’s documentation on the WTP’s engineering, procurement, and construction. Conducting audits and assessments to ensure that the contractor’s work complies with applicable quality assurance requirements. Assessing the effectiveness of the contractor’s Corrective Action Management Program, which involves identifying, documenting, planning, addressing, and tracking actions required to resolve or correct problems. Both the contractor’s and ORP’s quality assurance programs require that corrective actions to address significant problems with the quality of the work must include a determination of the extent to which the problematic conditions exist (known as an extent-of-condition review) as well as the underlying causes of those conditions. If corrective actions do not address the conditions, ORP’s quality assurance policy allows the office to call for a suspension of work. ORP’s stop work procedure includes the process ORP is to follow when the Quality Assurance Division Director, in consultation with ORP management, determines that work needs to be suspended as a result of the occurrence or reoccurrence of significant quality assurance problems. ORP updated this procedure in February 2016 to describe the type of quality assurance deficiencies that should trigger consideration of work stoppage. According to the updated procedure, characteristics of a deficiency that can trigger an order to stop work include, but are not limited to, problems that will result in $25 million or more in loss of productivity, construction rework, or environmental damage or a significant quality problem that if left uncorrected can result in construction delays or create adverse safety conditions. Until February 2016, ORP did not have precise criteria describing the conditions under which it should evaluate work for possible stoppage, according to a DOE headquarters report. ORP Has Taken Several Actions to Identify and Address Quality Assurance Problems at the WTP, but All Planned Actions Have Not Been Completed ORP has taken several actions to identify and address quality assurance problems at the WTP, but all planned actions have not been completed. In 2013 ORP conducted a comprehensive audit, which resulted in several actions, including when the office had the contractor begin implementing a Managed Improvement Plan (MIP) in 2014. The MIP is intended to ensure that the WTP could operate in compliance with DOE-approved safety and quality requirements. Implementation of the MIP was to be completed by April 2016. Although the contractor reported that the implementation was complete, some of the plan’s corrective measures have not been fully implemented, according to contractor documents we reviewed and quality assurance experts we spoke to. In addition, ORP’s effort to verify the extent to which the contractor has implemented MIP corrective measures is not scheduled to be complete until at least December 2018. Actions Taken to Identify and Address Quality Assurance Problems at the WTP ORP has taken several actions to identify and address quality assurance problems at the WTP. After the partial work stoppage in 2012, ORP conducted an audit in 2013 to evaluate the adequacy, implementation, and effectiveness of the contractor’s quality assurance program. The audit found that the contractor’s quality assurance program was generally adequate. However, it also found that the contractor’s quality assurance program was not fully effective in several areas. In response to the audit, ORP and the WTP contractor took the following actions: Developed compensatory measures. At ORP’s request, in 2013, the contractor started implementing “compensatory measures” to ensure that ongoing WTP work during a 2-year performance improvement period would meet DOE quality and safety requirements. For example, in September 2013, the contractor implemented a measure requiring senior management review of all condition reports and their associated levels of significance. According to ORP officials, the compensatory measures were intended to be additional, temporary internal controls to ensure that work at the WTP did not result in new or recurring quality assurance problems. Initiated the MIP. To systematically integrate compensatory measures, the contractor developed the MIP to address all quality assurance problems identified in the two Priority Level One findings and the seven Priority Level One findings associated with engineering and nuclear safety. In August 2014, the contractor started implementing the MIP. The MIP is a set of 52 corrective measures intended to establish processes, procedures, and metrics to produce an overall quality program that ensures that the WTP can safely operate in compliance with DOE-approved nuclear safety requirements, according to the contractor. The measures include the following: Actions to enhance external independent oversight. This measure calls for the contractor to conduct assessments using external subject matter experts to evaluate the ability of the contractor’s quality assurance program to identify precursors to potential problems and their causes. This measure responds to the 2013 audit in which DOE concluded that the contractor’s quality assurance program could not ensure compliance with requirements. Specifically, the audit found that the contractor’s quality assurance program was not fully effective in several areas, including, but not limited to, design, software quality, procurement, and ensuring that identified problems are corrected. Actions to ensure that procured items and services meet requirements and perform as specified. This measure is intended to ensure that the contractor’s processes and procedures to identify and ensure the quality of technical products meet requirements. The nuclear industry uses “commercial grade dedication” to refer to the process by which the contractor or subcontractor verifies that an item (e.g., an electric switch) or service (e.g., design of an electrical system) can meet commercial quality and safety requirements and be approved for use in a nuclear facility. It requires the contractor to perform source verification, perform inspections and tests, and assess the processes that control the quality of purchased items and services to help ensure that critical components of procured items and services are designed, fabricated, assembled, installed, and tested with appropriate documentation to support their compliance with WTP safety requirements. This measure also responds to DOE’s 2013 audit, which found that the contractor had inadequate control over the quality of purchased items and services. Actions to control and correct items and processes that do not meet requirements. This measure is intended to allow the contractor to identify and ensure that materials and equipment that have been received, and that will be received in the future, meet requirements. The contractor is to conduct comprehensive reviews of previously received material and equipment, as well as all future deliveries, to help ensure the verification, accuracy, and completeness of documentation for materials and equipment received from suppliers. This measure also responds to DOE’s 2013 audit, which found that the contractor had received components that did not comply with safety requirements. Performed targeted audits to test compensatory measures and the implementation of the MIP. To assess the effectiveness of the compensatory measures and the MIP, ORP performed targeted audits. For example, to assess the extent to which the contractor has addressed quality assurance program deficiencies, in early 2017 ORP’s Quality Assurance Division conducted a “vertical slice audit.” This audit reviewed engineering, procurement, and construction of a key system that will be needed for initial WTP operations. Because of the long-standing quality assurance problems at the WTP, DOE required ORP to closely monitor the contractor’s implementation of the MIP. Specifically, as a result of a DOE Office of Enforcement investigation into the contractor’s quality assurance and corrective action management programs, DOE entered into a Consent Order with the contractor in 2015. The Consent Order required the contractor to complete the actions identified in the MIP to the extent necessary to restore quality assurance program to full effectiveness by April 30, 2016. The Consent Order does not preclude DOE from reopening the investigation or issuing an enforcement action if there is a recurrence of nuclear safety deficiencies similar to those identified in the Consent Order or the if contractor fails to complete actions required by the Consent Order in a timely and effective manner to prevent recurrence of the identified issues. Corrective Measures to Address Quality Assurance Problems at the WTP Have Not Been Fully Implemented The contractor has not fully implemented corrective measures for all identified quality assurance problems, according to contractor documents we reviewed. In August 2017, the contractor reported that it had finished its actions to implement the MIP. However, according to the contractor’s MIP status update accompanying the contractor’s report, 13 of the 52 corrective measures specified in the MIP had not been fully implemented. Our review of these 13 MIP corrective measures we found that 9 were intended to exclusively or partially address weaknesses in the contractor’s quality assurance program. For example, the two corrective measures to ensure that WTP facilities’ computer software meets requirements were not complete, according to the MIP status update. These corrective measures included improving the software procurement process and revising the quality assurance manual. In addition, of the 39 measures that the contractor considers complete, some do not appear to be fully implemented, according to one ORP quality assurance expert that we spoke to. For example, one ORP quality assurance expert disagreed with the contractor’s assessment that a corrective measure for documentation pertaining to radiographic film— which is needed for conducting quality assurance reviews of certain equipment—was fully implemented. This corrective measure calls for the contractor to review purchase orders for radiographic film and then store the radiographic film as documentation of compliance with nuclear quality standards. According to the expert, radiographic film reviews are still not consistently conducted, and radiographic film documentation is still not consistently stored. In cases where such documentation is incomplete or missing, the contractor is at times forced to re-create the documentation at considerable cost to DOE. According to ORP’s MIP oversight plan, it will take the office until at least December 2018 to verify the extent to which the contractor has implemented each of the 52 MIP corrective measures. ORP Has Not Ensured That All Quality Assurance Problems Have Been Identified, and Some Previously Identified Problems Are Recurring According to DOE documents we reviewed and ORP quality assurance experts we spoke with, ORP’s actions have not ensured that all quality assurance problems have been identified at the WTP, and some previously identified problems are recurring. Specifically, according to DOE documents and the experts we spoke with, ORP’s oversight has not ensured that the contractor has identified all quality assurance problems in structures, systems, and components that were completed and installed before the 2012 work stoppage or identified all such problems in newer structures, systems, and components needed for initial WTP operations. In addition, according to the documents we reviewed and experts we interviewed, previously identified quality assurance problems are recurring. ORP Oversight Has Not Ensured That the Contractor Has Identified All Quality Assurance Problems Recent DOE reviews have found that ORP has not ensured that all quality assurance problems have been identified at the WTP. First, a 2016 DOE Office of Enterprise Assessment report found quality assurance deficiencies that neither ORP nor the contractor had identified at the time the work was conducted. The report identified numerous construction deficiencies, procurement and supplier deficiencies, engineering errors, maintenance issues, and materials with expired shelf lives. For example, the report identified welding deficiencies on tanks designed to hold nuclear waste that were identified in a WTP facility several years after the tanks were installed. The report concluded that the contractor is aware that significant quality assurance problems likely exist in older structures, systems, and components. This report noted that much of the equipment in older structures, systems, and components was manufactured and delivered to the project from 5 to 10 years ago—and some of this equipment was supplied by vendors or manufacturers that are no longer in business—which could lead to costly rework. Second, a 2015 DOE Inspector General report found that the contractor had procured $4 billion in parts and materials through fiscal year 2014, but ORP and the contractor had not always identified problems with the quality of procured items in a timely manner. For example, the report found that in about 45 percent of the nearly 1,400 procurement problems reviewed, the contractor did not identify the problems until at least 2 years after the items arrived on site. The report also found that in many cases the contractor canceled its efforts to recover the costs to resolve the problems because of the length of time that had passed. The report concluded that these problems were caused by weaknesses in the contractor’s quality assurance program and that the contractor’s procedures to prevent or identify problems with procured items were not always followed effectively. The findings of these reports are consistent with the views of ORP quality assurance experts we spoke with who stated that ORP oversight has not ensured that the contractor has identified all quality assurance problems in structures, systems, and components—particularly those that were completed and installed before the 2012 work stoppage. These quality assurance experts said that because quality assurance problems have not been identified, they expect significant rework will be needed for work that was completed before 2012. Specifically, most of the ORP quality assurance experts (seven of the nine) told us that they expect rework will be needed for existing WTP facilities, such as the Pretreatment and High- Level Waste Facilities. One of these seven quality assurance experts noted that the contractor does not have a complete record of the documentation for key systems and equipment, which is required for demonstrating compliance with nuclear safety standards and eventual permitting of WTP facilities for operation. According to this expert, the extent of this shortcoming is not known, but fixing it—that is, creating a complete record of required documentation—may lead to years of delays. ORP Quality Assurance Division officials told us that because ORP’s focus is on ensuring that facilities needed for initial operations will be ready to operate by December 2023, they have not been directed by ORP management to focus on identifying all quality assurance problems for work completed before 2012 for facilities needed for later phases of WTP operations, such as structures, systems, and components of the Pretreatment and High-Level Waste Facilities. In addition, they stated that there may be significant changes to these facilities needed for the WTP’s later phases, making it unnecessary for them to review the extent of quality assurance problems until it is known what parts of the facilities will remain and which parts will not. However, similar problems appear to exist in WTP facilities needed for initial operations. ORP quality assurance experts that we interviewed also stated that ORP oversight has not always ensured that all quality assurance problems in facilities needed for the initial WTP operations, or DFLAW, have been identified. Five experts told us that issues such as identifying problematic items, services, and processes had not been fully resolved. Specifically, these ORP quality assurance experts told us that when quality assurance problems are identified in structures, systems, or components needed for DFLAW, ORP does not always ensure that the contractor identifies the extent to which such problems may exist in other areas affected by the same structures, systems, or components. For example, an ORP quality assurance expert cited an instance in which an ORP quality assurance team reviewed a sample of 25 procurement “packages” (out of thousands) for a DFLAW facility and identified 143 problems—significantly more problems than the team expected for such a small sample. Consistent with ORP quality assurance requirements, this ORP quality assurance expert recommended to ORP upper management that the contractor determine the extent to which such problems could affect other structures, systems, and components needed for DFLAW. However, according to an ORP memo, ORP upper management did not require the contractor to implement this recommendation, instead citing “extenuating circumstances” and requiring a lesser corrective action than what was recommended. Three ORP quality assurance experts told us that they believe that because problems have not been comprehensively assessed, there may be equipment and systems within DFLAW that will fail to meet their intended functions. We also found that although ORP conducted its vertical slice audit in 2017 to test its compensatory measures and the MIP to improve quality assurance, the audit report notes that it was focused on only one system within the Low-Activity Waste Facility. According to ORP officials, there are numerous structures, systems, and components in facilities needed for DFLAW that have not been audited or reviewed in a manner similar to the vertical slice audit. Both the contractor’s and ORP’s quality assurance programs require that corrective actions to address significant problems with the quality of the work include a determination of the extent to which the problematic conditions exist as well as the underlying causes of those conditions. Until ORP requires the WTP contractor to determine the full extent to which problems exist in all WTP structures, systems, and components, DOE lacks a comprehensive understanding of all potential quality assurance problems at all WTP facilities. Previously Identified Quality Assurance Problems Are Recurring DOE requires its program offices, such as ORP, and contractors to have oversight processes to ensure that quality assurance problems are evaluated and corrected in a timely basis to prevent recurrence. However, several DOE documents we reviewed show that previously identified quality assurance problems have recurred in recent years, including the following: In 2015, an ORP audit report identified recurring weaknesses in quality assurance for the contractor’s process for procuring commercial items for use in a nuclear facility. For example, ORP found that the contractor’s internal controls for this process were not consistently performed; did not consistently comply with procedural requirements; and, in many cases, did not establish reasonable assurance that procured systems, services, and components acquired from 2010 to 2014 would perform their intended safety functions. In a 2015 report on the design and operability of key systems and components for the Low-Activity Waste Facility, ORP found that the quality of computer systems software was not in full compliance with DOE requirements, leading to conditions where personnel and the environment may not be adequately protected. ORP had identified a similar problem in 2008, when it found that the contractor’s computer programs used in engineering calculations were not always verified to show that they produced correct solutions within defined limits for all parameters, as required by the contractor’s quality assurance manual. ORP had also previously identified WTP computer software quality problems in 2010 when it issued a Priority Level Two finding on software procedures and another Priority Level Two finding on software testing. In 2017, ORP’s Quality Assurance Division issued a report that examined the contractor’s quality assurance program and found problems in quality assurance areas that had been previously identified. The report noted that in 6 of 19 quality assurance program areas, the contractor’s performance was marginal—and in need of improvement—or indeterminate. These 6 areas included identifying, controlling, and correcting items, services, and processes that do not meet established requirements; maintaining items to prevent damage, loss, or deterioration; and procuring items and services that meet established requirements and perform as specified. ORP quality assurance experts that we spoke with also stated that previously identified quality assurance problems are recurring, including some in areas where the contractor had implemented corrective measures. These quality assurance experts told us that quality assurance problems are recurring in several key areas, including those areas identified in the documents described above: (1) procurement of items and services that do not meet established requirements or perform as specified; (2) software that does not meet established requirements; and (3) a maintenance program that does not prevent damage, loss, or deterioration of WTP structures, systems, and components. For example, see the following. Procurement of items and services that do not meet requirements or perform as specified. Four out of the five ORP quality assurance experts we interviewed who had recent experience with the procurement of items and services told us that problems with procured items and services that do not meet established requirements or perform as specified are not fully resolved. One of these ORP quality assurance experts stated that an ORP team recently reviewed a random sample of 45 of the roughly 30,000 procurements the contractor had made for the WTP and identified a number of instances where materials did not meet requirements, which resulted in one Priority Level Two finding—which represents a serious issue that indicates an adverse condition, such as a noncompliance or breakdown of a management system—and five Priority Level Three findings. The expert noted that this was many more deficiencies than the team expected for such a small sample. Settlement of Allegations of Contractors Knowingly Mischarging Costs at the Waste Treatment and Immobilization Plant (WTP) In November 2016, the WTP contractor and certain subcontractors agreed to pay $125 million to resolve allegations under the False Claims Act that they made false statements and claims to the Department of Energy (DOE) by charging DOE for deficient nuclear quality materials, services, and testing that were provided to the WTP at DOE’s Hanford Site. The contract required materials, testing, and services to meet certain nuclear quality standards. The Department of Justice alleged that the defendants violated the False Claims Act by charging the government the cost of complying with these standards when they failed to do so. In particular, the Department of Justice alleged that the defendants improperly billed the government for materials and services from vendors that did not meet quality control requirements, for piping and waste vessels that did not meet quality standards, and for testing from vendors that did not have compliant quality programs. As part of the settlement, the contractors admitted no wrongdoing, and the United States did not concede that its claims were not well founded. Software that does not meet requirements. ORP quality assurance experts told us that problems are recurring in certain areas where items and processes do not meet requirements, such as computer software quality assurance, despite the contractor developing two MIP corrective measures in this area. Two ORP quality assurance experts reported that problems with software quality are recurring. One ORP quality assurance expert added that the contractor often fails to develop software quality documentation that is needed to demonstrate compliance with quality requirements when permitting facilities for operation. As a result, the contractor will have to re-create this documentation at some cost. A maintenance program that does not prevent damage, loss, or deterioration. Each of the three ORP quality assurance experts with knowledge in this area told us that the contractor had not established a fully effective WTP maintenance program, particularly for the Pretreatment and High-Level Waste Facilities, and as a result, structures, systems, and components at these facilities have deteriorated and been damaged. Such statements are consistent with findings of the Defense Nuclear Facilities Safety Board, which reported in April 2016 that systems and components stored in an outdoor storage yard were not properly covered and showed signs of being affected by water, sand, or animals. In March 2016, ORP reported significant water intrusion into several areas of the High- Level Waste Facility. As a result, some of the facility’s structures, systems, and components had deteriorated and will require costly rework. The contractor notified DOE in April 2017 that because DOE’s focus is on completing facilities needed for initial WTP operations, it would submit a proposal to change the WTP contract to account for the increased scope, cost, and schedule of long-term maintenance, storage, and management of procured and partially installed structures, systems, and components at those facilities not needed for initial WTP operations. Consistent with its quality assurance procedures, ORP can use its authorities—such as those under the Consent Order and its quality assurance policy—to stop work if corrective measures do not prevent quality assurance problems from recurring. However, ORP has not used such authorities. ORP senior officials told us that they did not consider it necessary to stop work because of the recurrence of problems in certain areas because they plan to evaluate the extent of the contractor’s implementation of MIP corrective measures over the next year and have allowed work to continue because they believe that the contractor’s quality assurance program is generally adequate. Without directing ORP to use its authorities to stop work in areas where quality assurance problems are recurring until it can verify that the problems are corrected and will not recur, DOE may face future rework that could increase costs and schedule delays for the WTP. ORP’s Organizational Structure May Not Provide Sufficient Independence for Effective Oversight of the WTP Contractor’s Quality Assurance Program A 2017 assessment from DOE headquarters and our interviews with nine ORP quality assurance experts suggest that ORP’s organizational structure does not provide the quality assurance function with sufficient independence from upper management—which includes the ORP Manager and the WTP Federal Project Director—to effectively oversee the contractor’s quality assurance program. Our prior work has found that to be independent, an oversight organization should be structurally distinct and separate from program offices responsible for achieving the program’s mission to avoid management interference or conflict between program office mission objectives and safety. At ORP, however, the Quality Assurance Division is not fully separate and independent from the upper management of the WTP project, which manages cost and schedule performance. We believe that such a structure has the potential to create a conflict of interest. Specifically, we found that ORP’s Quality Assurance Division performs assessments of the contractor’s quality assurance program, among other things, and reports its findings to ORP upper management, including the ORP Manager, who has the discretion to determine whether and to what extent to require the contractor to take action in response to findings. When quality assurance issues are identified, ORP upper management must balance its mission of meeting cost and schedule targets with its responsibility to ensure that nuclear safety and quality standards are met. However, these are two potentially conflicting responsibilities because meeting WTP cost and schedule targets may be threatened if serious quality assurance problems are identified. A February 2017 external assessment from DOE headquarters noted that ORP’s Quality Assurance Division’s effectiveness has been limited because, in some instances, its findings have been mischaracterized by ORP upper management, and in others, ORP upper management has not used this division effectively to evaluate the extent of potential quality assurance problems. This assessment found that ORP had not performed adequate oversight of the contractor’s MIP and that some critical quality assurance areas were not receiving the necessary scrutiny from ORP. Further, the assessment found that ORP management sometimes mischaracterized the seriousness of the Quality Assurance Division’s findings and, as a result, did not require the contractor to conduct extent-of-condition review for significant quality assurance problems. While this assessment stated that ORP had an effective quality assurance program, it concluded that three of the eight quality assurance areas the assessment team reviewed were not fully effective, including ORP’s ability to conduct assessments of the contractor’s quality assurance program. A Cautionary Tale: Quality Assurance Problems Doom Commercial Nuclear Power Plant In the commercial nuclear industry, there is a notable example of a construction project that faced significant quality assurance challenges. In the 1970s and early 1980s, Cincinnati Gas & Electric attempted to construct a commercial nuclear power plant, known as the Zimmer Plant, near Moscow, Ohio. After 10 years of construction and more than $2 billion spent, the company abandoned its effort to construct the plant. An independent review mandated by the Nuclear Regulatory Commission in 1982 concluded that several issues impeded successful construction of the Zimmer Plant as a commercial nuclear power plant. These issues included (1) the company’s failure to elevate its commitment to quality and quality assurance to an equal status with cost and schedule, (2) the regulator’s failure to hold the company accountable for quality in design and construction, and (3) the company’s inadequate quality assurance procedures. To recoup some of the $2 billion spent in attempting to construct this commercial nuclear power plant, Cincinnati Gas & Electric later converted facilities built at the site for use in a coal-fired power plant. management and the contractor place cost and schedule performance above identifying and resolving quality assurance issues. One quality assurance expert specified that ORP’s culture does not encourage staff to identify quality assurance problems or ineffective corrective measures. This expert said that people who discover problems are not rewarded; rather, their findings are met with resistance, which has created a culture where quality assurance staff are hesitant to identify quality assurance problems or problems with corrective measures. This expert added that quality assurance is subordinate to cost and schedule—that is, senior managers responsible for approving quality assurance findings are more concerned with whether WTP construction meets schedule milestones than identifying and resolving quality assurance issues. This expert compared the WTP to the Zimmer Power Plant—a power plant in Ohio that was designed to be a nuclear power plant but that was never licensed because of unresolved quality assurance problems and a focus on schedule over construction quality. As stated earlier, in October 2008, we identified key elements that any nuclear safety oversight organization should have in order for it to provide effective independent oversight. For example, we found that an organization should be structurally distinct and separate from DOE program offices to avoid management interference or conflict between program office mission objectives, such as cost and schedule performance and safety. We also found that the organization should have sufficient authority to require program offices to effectively address its findings and recommendations. ORP’s Assistant Manager for Technical and Regulatory Support and ORP senior quality assurance staff told us that ORP’s organizational structure ensures that the quality assurance function is sufficiently independent of ORP management. These officials and the ORP Quality Assurance Program Description state that the Quality Assurance Division is structured to report directly to the ORP Assistant Manager for Technical and Regulatory Support and the ORP Manager. They also cited the ORP Quality Assurance Program policy, which states that the Quality Assurance Division has the authority and overall responsibility to independently audit the contractor’s quality assurance program to verify the achievement of quality. According to these officials, this organizational structure ensures independence from cost and schedule considerations and ensures objectivity in quality assurance evaluations, and they added that the ORP Manager evaluates differing opinions without any hindrances or organizational bias. Given that some previously identified problems are recurring at the WTP, including some in areas where the contractor had implemented corrective measures, and given the findings of the 2017 headquarters assessment and the statements of ORP’s quality assurance experts outlined above, we are concerned that ORP’s organizational structure may not entirely ensure that the Quality Assurance Division meets key elements for a nuclear safety oversight organization to provide effective independent oversight. According to ORP reports and officials, in ORP’s current organizational structure, upper level management retains discretion in how to resolve quality assurance problems. As a result, the Quality Assurance Division does not have sufficient authority to ensure that its findings are addressed and its recommendations are implemented. By revising ORP’s organizational structure so that the quality assurance function is independent of ORP upper-level management, DOE can have better assurance that compliance with nuclear safety requirements will not be subordinated to meeting cost and schedule targets. Conclusions For years DOE has faced quality assurance problems at the WTP. Upon learning in 2012 that it could not verify that engineering, procurement, and construction at the WTP met nuclear safety and quality requirements, ORP directed the contractor to implement quality assurance corrective measures to ensure that problems would be identified and prevented from recurring. However, 5 years later, the contractor has not fully implemented all planned corrective measures. Moreover, in some areas where the contractor has stated that corrective measures are now in place, ORP continues to encounter quality assurance problems similar to those it encountered in the past. When and where problems have recurred, ORP has not always required the contractor to determine the extent to which the problems may affect all parts of the WTP. By directing ORP to require the WTP contractor, where quality assurance problems have been identified, to determine the full extent to which problems exist in all WTP structures, systems, and components, DOE will gain a comprehensive understanding of all quality assurance problems at all WTP facilities. In addition, ORP has not always used its authorities to stop work when problems are detected before they are fully corrected. Without directing ORP to use its authorities to stop work in areas where quality assurance problems are recurring until it can verify that the problems are corrected and will not recur, DOE may face future rework that could increase costs and schedule delays for the WTP. Also of concern is the potential lack of sufficient independence of ORP’s Quality Assurance Division from ORP’s upper management. This has resulted in ORP upper management not always allowing its own experts to fully examine the contractor’s work even when problems have recurred. At other times, this has resulted in the significance of identified problems—and strength of associated corrective measures—being reduced. DOE’s ability to effectively self-regulate a high-hazard nuclear facility not only depends on vigorous oversight of the contractor by the program office but also on active oversight by an independent group. The WTP is the largest and most technically complex cleanup project managed by DOE, and we recognize that meeting its cost and schedule targets places immense pressure on ORP upper management. However, meeting those targets is further threatened when quality assurance problems are downgraded. By revising ORP’s organizational structure so that the quality assurance function is independent of ORP upper management, DOE can have better assurance that compliance with nuclear safety requirements will not be subordinated to meeting cost and schedule targets. Recommendations for Executive Action We are making the following three recommendations to DOE: The Secretary of Energy should direct ORP to require the WTP contractor to determine the full extent to which problems exist in all WTP structures, systems, and components. The Secretary of Energy should direct ORP to use its authorities to stop work in areas where quality assurance problems are recurring until ORP’s Quality Assurance Division can verify that the problems are corrected and will not recur. The Secretary of Energy should revise ORP’s organizational structure so that the quality assurance function is independent of ORP upper management. Agency Comments and Our Evaluation We provided DOE with a draft of this report for its review and comment. In its written comments, reproduced in appendix I, DOE generally agreed with the findings in the report and its recommendations. DOE agreed with our first two recommendations and described actions it has under way and planned to address them. In addition, DOE agreed with our third recommendation—to revise ORP’s organizational structure so that the quality assurance function is independent of ORP upper management—in principle. While DOE states that it believes that the current ORP quality assurance reporting relationship meets all established requirements, it also states that the report identifies instances that indicate that ORP could be strengthened to improve the effectiveness and independence of its quality assurance functions. In response to our recommendation, DOE plans to direct ORP to assess the quality assurance functional reporting lines, responsibilities, and processes to enhance the independence of the quality function from cost and schedule influences and to strengthen and clarify quality assurance reporting to the ORP Manager. This planned action is a positive first step toward implementing our recommendation. We are sending copies of this report to the appropriate congressional committees; the Secretary of Energy; and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff members have any questions about this report, please contact me at (202) 512-3841 or trimbled@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff members who made major contributions to this report are listed in appendix II. Appendix I: Comments from the Department of Energy Appendix II: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Nathan Anderson (Assistant Director), Mark Braza, Scott Fletcher, Ellen Fried, Richard Johnson, Paul Kazemersky, and Peter Ruedel made key contributions to this report.

Why GAO Did This Study DOE and its contractor are building the WTP—which consists of multiple facilities—to treat a large portion of nuclear waste at Hanford. The project has faced persistent challenges, including quality assurance problems that have delayed it by decades and more than tripled its costs, to nearly $17 billion. DOE's quality assurance framework aims to ensure that all problems are identified and do not recur. Senate Report 114-49 accompanying the National Defense Authorization Act for Fiscal Year 2016 included a provision for GAO to carry out an ongoing evaluation of the WTP. This first report examines (1) the actions DOE has taken to identify and address WTP quality assurance problems, (2) the extent to which DOE has ensured that quality assurance problems have been identified and do not recur, and (3) the extent to which DOE's organizational structure at ORP provides the Quality Assurance Division with independence to effectively oversee the contractor's quality assurance program. GAO reviewed DOE documents and obtained the insights of ORP's internal experts on WTP quality assurance efforts and outcomes. What GAO Found The Department of Energy (DOE) has taken several actions to identify and address quality assurance problems at the Waste Treatment and Immobilization Plant (WTP) at its Hanford site in Washington. Among the actions taken is the implementation of the Managed Improvement Plan by DOE's Office of River Protection (ORP) and the WTP contactor. The plan is intended to ensure that the WTP can operate in compliance with DOE-approved safety and quality requirements. The contractor has stated that the plan is fully implemented, but GAO found that a number of key activities may be incomplete and ORP officials will not be able to verify the extent of implementation until December 2018. According to DOE documents that GAO reviewed and ORP quality assurance experts GAO spoke with, ORP has not ensured that all WTP quality assurance problems have been identified and some previously identified problems are recurring. For example, a 2016 DOE report found quality assurance problems, such as engineering errors and construction deficiencies, that neither ORP nor the contractor had identified when the work was conducted. ORP quality assurance experts GAO spoke with reiterated the issues identified in reports. In addition, DOE audits have found that previously identified quality assurance problems have recurred in key areas, such as the procurement of items that do not meet requirements or perform as specified. These problems were also raised by several of the ORP quality assurance experts GAO interviewed. According to these experts, such recurring problems may lead to significant rework at WTP facilities in the future if work is not stopped and the issues addressed. ORP's quality assurance framework requires the contractor to determine the extent to which quality assurance problems exist in all WTP structures, systems, and components when such problems are identified, and allows ORP to stop work at a facility if recurring issues arise. However, ORP has neither directed the contractor to make this determination nor stopped work when problems recur because it has confidence in the Managed Improvement Plan. ORP's organizational structure may not provide its Quality Assurance Division with sufficient independence from the office's upper management to oversee the contractor's quality assurance program effectively. GAO has previously found that an oversight organization should be structurally distinct and separate from program offices responsible for cost and schedule performance to avoid conflict between mission objectives and safety. However, a 2017 DOE headquarters assessment found that ORP's Quality Assurance Division's effectiveness has been limited. This is because in some cases ORP upper management had mischaracterized its findings, and in other instances, ORP upper management had not used this division to evaluate the extent of potential quality assurance problems. ORP quality assurance experts GAO spoke to were also concerned that ORP's organizational structure does not always ensure the independence of the division. For example, two of these experts described instances when ORP upper management had downgraded the division's findings so that the contractor could take less stringent corrective measures. By providing the Quality Assurance Division adequate independence, DOE can better ensure that compliance with nuclear safety requirements will not be subordinated to other project management goals, such as meeting cost and schedule targets. What GAO Recommends GAO recommends that DOE direct the WTP contractor to determine the extent of problems in WTP structures, systems, and components and order work stops when problems recur, and DOE should direct ORP to revise its organizational structure to ensure the independence of the Quality Assurance Division. DOE generally agreed with GAO's recommendations.

gao_GAO-18-189T

gao_GAO-18-189T_0

VA’s Efforts to Align its Facilities Are Affected by Several Factors and Are Impeded by Limitations in Its Capital-planning Processes Facility Alignment Is Challenged by Shifting Veterans’ Populations, Evolving Care Standards, Aging Infrastructure, and Limited Stakeholder Involvement Geographic shifts in the veterans’ population, changes in health care delivery, an aging infrastructure, and limited stakeholder involvement affect VA’s efforts to align its services and real property portfolio to meet the needs of veterans. For example, there has been a shift over time from inpatient to outpatient care. This shift will likely result in underutilized space once used for inpatient care. In such instances, it is often difficult and costly for VA to modernize, renovate, and retrofit these older facilities. In June 2017, VA reported that its facility inventory includes 430 vacant or mostly vacant buildings that are, on average, more than 60 years old, and an additional 784 buildings that are underutilized. The historic status of some VA facilities adds to the complexity of converting or disposing of them. In 2014, VA reported holding 2,957 historic buildings, structures, or land parcels—the third most in the federal government after DOD and the Department of the Interior. In some instances, it may be more expensive to renovate than to demolish and rebuild outdated facilities. In other cases, however, there may not be an option to demolish if these buildings are designated as historic. For example, planning officials at four medical facilities in our review told us that state historic preservation efforts prevented the VA from demolishing vacant buildings, even though these buildings require upkeep costs and pose potential safety hazards. (See fig. 1.) VA has also encountered challenges to its facility alignment efforts, in part, because it has not consistently followed best practices for effectively engaging stakeholders. VA may align its facilities to meet veterans’ needs by expanding or consolidating facilities or services. Stakeholders— including veterans; local, state, and federal officials; Veterans Service Organizations; historic preservation groups; VA staff; and Congress— often view changes as working against their interests or those of their constituents, especially when services are eliminated or shifted from one location to another. We found that VA has not consistently engaged with stakeholders, and, in some cases, this inconsistency resulted in adversarial relationships that reduced VA’s ability to better align facilities with the needs of the veteran population. In our April 2017 report, we recommended that VA improve stakeholder communication guidance and evaluate its efforts. VA agreed with our recommendations and outlined a plan to implement them. Limitations in VA’s Capital- planning Processes Impede Its Alignment of Facilities SCIP Process Two of the planning processes VA uses to align its facilities—VA’s Strategic Capital Investment Planning (SCIP) and the VA Integrated Planning (VAIP)—have limitations. VA relies on the SCIP process to plan and prioritize capital projects system-wide, but SCIP’s limitations—including subjective narratives, long timeframes, and restricted access to information—undermine VA’s ability to achieve its goals. For example, the time between when planning officials at VA medical facilities begin developing the SCIP narratives and when they are notified that a project is funded has taken between 17 and 23 months over the past 6 fiscal-year’s SCIP submissions. (See fig. 2.) As such, VA routinely asks its facility planners to submit their next year’s planned project narratives before knowing if their project submissions from the previous year have been funded. An official from the office that oversees SCIP told us that the timing of the budgeting process, which is outside VA’s control, contributes to these delays. While these aspects are outside of VA’s control, VA has chosen to wait about 6 to 10 months to report the results of the SCIP scoring process to the medical facilities. This situation makes it difficult for local officials to understand the likelihood that their projects will receive funding. A VA official said that for future SCIP cycles, VA plans to release the scoring results for minor construction and non-recurring maintenance projects to local officials earlier in the process. At the time of our review, however, the official did not have a time frame for when VA would do this. Although VA acknowledges many of these limitations, it has taken little action in response. Federal standards for internal control state that agencies should evaluate and determine appropriate corrective action for identified limitations on a timely basis. If VA does not address known limitations with the SCIP process, it will not have reasonable assurance that SCIP can be used to accurately identify the capital necessary to address VA’s service and infrastructure gaps. In our April 2017 report, we recommended that VA address identified limitations to the SCIP process, including limitations to scoring and approval, and access to information. VA concurred with the recommendation to the extent the limitations were within its control. While VA has taken some actions, the recommendation remains open. VAIP Process The VAIP process produces a market-level health services delivery plan for each Veterans Integrated Service Network (VISN) and a facility master plan for each medical facility. VA has estimated the entire process to create plans for VISNs and facilities to cost $108 million when fully complete. However, the VAIP process’s facility master plans assume all future growth in services will be provided directly through VA facilities. This assumption is not accurate given that (1) VA obligated about $10.1 billion to purchase care from non-VA providers in fiscal year 2015 and (2) VA can provide care directly through its medical facilities or purchase health care services from non-VA providers through both the Non-VA Medical Care Program (referred to as “care in the community” by VA) and clinical contracts. The Office of Management and Budget’s acquisition guidance notes that investments in major capital assets should be made only if no alternative private sector source can support the function at a lower cost. In our April 2017 report, we recommended that VA assess the value of the VAIP’s facility master plans as a facility-planning tool, and based on conclusions from the review, to either (1) discontinue the development of VAIP’s facility master plans or (2) address the limitations of VAIP’s facility master plans. VA concurred with the recommendation, and in August 2017, VA noted that it has discontinued its VAIP facility master plans while VA pursues a national realignment strategy, after which it plans to adjust its future facility master plans to incorporate pertinent information, including care in the community realignment opportunities. Key Elements and Challenges Affecting DOD and the Commission in BRAC 2005 Key Elements That DOD Used to Develop Its 2005 BRAC Recommendations That Could Benefit VA Asset and Infrastructure Review As Congress evaluates proposed legislation for disposing of or realigning VA property, it may wish to consider seven elements DOD relied on as it developed its recommendations for the BRAC Commission. Establish goals for the process. The Secretary of Defense emphasized the importance of transforming the military to make it more efficient as part of the 2005 BRAC round. Other goals for the 2005 BRAC process included fostering jointness among the four military services, reducing excess infrastructure, and producing savings. Prior rounds focused more on reducing excess infrastructure and producing savings. Develop criteria for evaluating closures and realignments. DOD proposed selection criteria, which were made available for public comment via the Federal Register. Ultimately, Congress enacted the final BRAC selection criteria in law with minor modification and specified that four selection criteria, known as the “military value criteria,” were to be given priority in developing closure and realignment recommendations. Further, Congress required that the Secretary of Defense develop and submit to Congress a force structure plan that described the estimated size of major military units needed to address probable threats to national security for the 20- year period beginning in 2005, along with a comprehensive inventory of global military installations. In authorizing the 2005 BRAC round, Congress specified that the Secretary of Defense publish a list of recommendations for the closure and realignment of military installations inside the United States based on the statutorily-required 20-year force structure plan and infrastructure inventory, and on the final selection criteria. Estimate costs and savings to implement closure and realignment recommendations. To address the cost and savings criteria, DOD developed and used the Cost of Base Realignment Actions (COBRA) model, a quantitative tool that DOD has used since the 1988 BRAC round to provide consistency in potential cost, savings, and return-on-investment estimates for closure and realignment options. We found the COBRA model to be a generally reasonable estimator for comparing potential costs and savings among alternatives. (See fig. 3.) As with any model, the quality of the output from COBRA was a direct function of the data DOD included in the model. Also, DOD’s COBRA model relied to a large extent on standard factors and averages and did not represent budget quality estimates that were developed once BRAC decisions were made and detailed implementation plans were developed. Nonetheless, the financial information provided important input into the selection process as decision makers weighed the financial implications—along with military value criteria and other considerations—in arriving at final decisions about the suitability of various closure and realignment options. Establish an organizational structure. The Office of the Secretary of Defense emphasized the need for joint cross-service groups to analyze common business-oriented functions. For the 2005 BRAC round, as for the 1993 and 1995 rounds, these joint cross-service groups performed analyses and developed closure and realignment options in addition to those developed by the military departments. Our evaluation of DOD’s 1995 BRAC round found that few cross- service recommendations were made, in part because of the lack of high-level leadership to encourage consolidations across the departments’ functions. In the 1995 BRAC round, the joint cross- service groups submitted options through the military services for approval, but few were approved. The number of approved recommendations that the joint cross-service groups developed significantly increased in the 2005 BRAC round. This increase was, in part, because high-level leadership ensured that the options were approved not by the military departments but rather by a DOD senior- level group, known as the Infrastructure Steering Group. As shown in figure 4, the Infrastructure Steering Group was placed organizationally on par with the military departments. Establish a common analytical framework. To ensure that the selection criteria were consistently applied, the Office of the Secretary of Defense, the military departments, and the seven joint cross- service groups first performed a capacity analysis of facilities and functions. Before developing the candidate recommendations, DOD’s capacity analysis relied on data calls to hundreds of locations to obtain certified data to assess such factors as maximum potential capacity, current capacity, current usage, and excess capacity. Then, the military departments and joint cross-service groups performed a military value analysis for the facilities and functions based on primary military value criteria, which included a facility’s or function’s current and future mission capabilities, physical condition, ability to accommodate future needs, and cost of operations. Develop BRAC oversight mechanisms to improve accountability for implementation. In the 2005 BRAC round, the Office of the Secretary of Defense for the first time required the military departments to develop business plans to better inform the Office of the Secretary of Defense of the status of implementation and financial details for each of the BRAC 2005 recommendations. These business plans included: (1) information such as a listing of all actions needed to implement each recommendation; (2) schedules for personnel relocations between installations; and (3) updated cost and savings estimates by DOD based on current information. This approach permitted senior-level intervention if warranted to ensure completion of the BRAC recommendations by the statutory completion date. Involve the audit community to better ensure data accuracy. The DOD Inspector General and military department audit agencies played key roles in identifying data limitations, pointing out needed corrections, and improving the accuracy of the data used in the process. In their oversight roles, the audit organizations, which had access to relevant information and officials as the process evolved, helped to improve the accuracy of the data used in the BRAC process and thus strengthened the quality and integrity of the data used to develop closure and realignment recommendations. For example, the auditors worked to ensure certified information was used for BRAC analysis and reviewed other facets of the process, including the various internal control plans, the COBRA model, and other modeling and analytical tools that were used in the development of recommendations. Key Challenges Affecting DOD and the Commission in BRAC 2005 We identified two key challenges that affected DOD’s implementation of BRAC 2005 and would need to be addressed for VA to adopt a BRAC- like process for its asset and infrastructure review. Some transformational-type BRAC recommendations required sustained senior leadership attention and a high level of coordination among many stakeholders to complete by the required date. Implementation of some transformational BRAC recommendations—especially those where a multitude of organizations had roles to play to ensure the achievement of the goals of the recommendation—illustrated the need to involve key stakeholders and effective planning. For example, the Defense Logistics Agency committed sustained high-level leadership and included relevant stakeholders to address implementation challenges faced with the potential for disruptions to depot operations during implementation of the BRAC consolidation recommendation. To implement the BRAC recommendations, the agency had to develop strategic agreements with the services that ensured that all stakeholders agreed on its plans for implementation, and had to address certain human capital and information technology challenges. Large number of actions and interdependent recommendations complicated the implementation process. The large number and variety of BRAC actions presented challenges during implementation. The BRAC 2005 round had more individual actions (813) than the four prior rounds combined (387). The executive staff of the Commission told us that it was more difficult to assess the costs and the amount of time for the savings to offset the implementation costs since many of the recommendations contained multiple interdependent actions, all of which needed to be reviewed. Specifically, many of the BRAC 2005 recommendations were interdependent and had to be completed in a sequential fashion within the statutory implementation period. In cases where interdependent recommendations required multiple relocations of large numbers of personnel, delays in completing one BRAC recommendation had a cascading effect on the implementation of other recommendations. Specifically, DOD had to synchronize the relocations of over 123,000 people with about $24.7 billion in new construction or renovation. Commission officials told us that in prior BRAC rounds each base was handled by a single integrated recommendation. However, in BRAC 2005, many installations were simultaneously affected by multiple interconnected BRAC recommendations. Given the complexity of interdependent recommendations, the Office of the Secretary of Defense required the military departments and defense agencies to provide periodic updates on implementation challenges and progress. Chairman Roe, Ranking Member Walz, and Members of the Committee, this concludes our prepared statement. We are happy to answer any questions related to our work on VA’s efforts to align its medical facilities and services or on DOD’s BRAC process. GAO Contact and Staff Acknowledgments If you or your staff members have any questions concerning this testimony, please contact David Wise at (202) 512-2834 or wised@gao.gov regarding federal real property, or Brian Lepore at (202) 512-4523 or leporeb@gao.gov regarding the BRAC process. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this statement. Other individuals who made key contributions to this testimony include Keith Cunningham, Assistant Director; Gina Hoffman, Assistant Director; Tracy Barnes; Jeff Mayhew; Kevin Newak; Richard Powelson; Malika Rice; Jodie Sandel; Eric Schwab; Amelia M. Weathers; and Crystal Wesco. This is a work of the U.S. government and is not subject to copyright protection in the United States. The published product may be reproduced and distributed in its entirety without further permission from GAO. However, because this work may contain copyrighted images or other material, permission from the copyright holder may be necessary if you wish to reproduce this material separately.

Why GAO Did This Study VA operates one of the largest health care systems in the United States, utilizing more than 6,000 federally owned and 1,500 leased buildings. DOD has repeatedly applied the BRAC process to reduce the amount of unneeded property that it owns and leases and to save billions of dollars that could be applied to higher priority defense needs. This statement is based on GAO's April 2017 report related to VA facility alignment ( GAO-17-349 ) and numerous GAO reports related to the BRAC process as summarized in a June 2011 testimony ( GAO-11-704T ) and a March 2012 testimony ( GAO-12-513T ). This statement addresses (1) the factors that affect VA's facility alignment and the extent to which VA's capital-planning process facilitates the alignment of facilities with the veterans' population, and (2) the key elements and challenges affecting DOD and the Commission in BRAC 2005. Detailed information on our scope and methodologies for this work can be found in these published products, cited throughout this testimony. What GAO Found Geographic shifts in the veterans' population, changes in health care delivery, aging infrastructure, and limited stakeholder involvement affect the Department of Veterans Affairs' (VA) efforts to align its services and real property portfolio to meet the needs of veterans. For example, a shift over time from inpatient to outpatient care will likely result in underutilized space once used for inpatient care. Further, the historic status of some VA facilities adds to the complexity of converting or disposing of them. In such instances, it is often difficult and costly for VA to modernize, renovate, and retrofit these older facilities. GAO reported that two of the planning processes VA uses to align its facilities—VA's Strategic Capital Investment Planning (SCIP) and the VA Integrated Planning (VAIP)—have limitations that undermine VA's efforts to achieve its goals. Specifically: VA relies on the SCIP process to plan and prioritize capital projects, but VA routinely asks its facility planners to submit their next year's planned project narratives before knowing if their previous submissions have been funded. The overlapping budget cycle, which is outside of VA's control, combined with other SCIP limitations—including subjective narratives, long time frames, and restricted access to information—make it difficult for VA to rely on SCIP to accurately identify the capital necessary to address its service and infrastructure gaps. VA concurred that it needs to address SCIP limitations that are within its control, as GAO recommended; VA has made some progress in implementing the recommendation has made some progress in implementing the recommendation.\ The VAIP process is estimated to cost $108 million and to produce market-level service delivery plans and facility master plans. However, the VAIP master plans incorrectly assume that all future growth in services will be provided directly through VA facilities without considering alternatives, such as purchasing care from the community. GAO recommended that VA consider discontinuing the VAIP facility master plans pending an assessment of their value as a facility-planning tool. VA agreed with the recommendation and is implementing it while pursuing a national realignment strategy.. Key elements of the Department of Defense's (DOD) 2005 Base Realignment and Closure (BRAC) process could benefit VA's asset and infrastructure review. The key elements included: (1) establishing goals for the process, (2) developing criteria for evaluating closures and realignments, and (3) establishing an organizational structure to develop closure and realignment options. GAO identified key challenges that affected DOD's implementation of BRAC 2005 and the results achieved; these challenges would need to be addressed if VA is to successfully apply the process. These challenges included: (1) large, complex recommendations required sustained senior leadership's attention and a high level of coordination among many stakeholders, and (2) the large number of actions that depend on each other for successful implementation. What GAO Recommends In the April 2017 report, GAO made recommendations related to capital planning and stakeholder involvement. VA concurred with the recommendations to the extent that they were within its control and has started making improvements.

gao_GAO-18-661T

gao_GAO-18-661T_0

Background VA’s Current Disability Compensation Appeals Process VA’s process for deciding veterans’ eligibility for disability compensation begins when a veteran submits a claim to VA. Staff in one of VBA’s 56 regional offices assist the veteran by gathering additional evidence, such as military and medical records, that is needed to evaluate the claim. Based on this evidence, VBA decides whether the veteran is entitled to compensation and, if so, how much. A veteran dissatisfied with the initial claim decision can generally appeal within 1 year from the date of the notification letter sent by VBA. Under the current appeals process (now referred to by VA as the legacy process), an appeal begins with the veteran filing a Notice of Disagreement. VBA then re-examines the case and generally issues a Statement of the Case that represents its decision. A veteran dissatisfied with VBA’s decision can file an appeal with the Board. In filing that appeal, the veteran can indicate whether a Board hearing is desired. Before the Board reviews the appeal, VBA prepares the file and certifies it as ready for Board review. If the veteran requests a hearing to present new evidence or arguments, the Board will hold a hearing by videoconference or at a local VBA regional office. The Board reviews the evidence and either issues a decision to grant or deny the veteran’s appeal or refers (or remands) the appeal back to VBA for further work. VA’s New Appeals Process The Act made changes to VA’s appeals process that will generally take effect no earlier than February 2019, which is approximately 18 months after enactment. According to its appeals plan, VA intends to implement the Act by February 2019, by replacing the current appeals process with a process offering veterans who are dissatisfied with VBA’s decision on their claim five options: two of those options afford the veteran an opportunity for an additional review of VBA’s decision within VBA, and the other three options afford them the opportunity to bypass additional VBA review and appeal directly to the Board. Under the new appeals process, the two VBA options will be: 1. Request higher-level review: The veteran asks VBA to review its initial decision based on the same evidence but with a higher-level official reviewing and issuing a new decision. 2. File supplemental claim: The veteran provides additional evidence and files a supplemental claim with VBA for a new decision on the claim. The veteran could also request a VBA hearing. The three Board options will be: 3. Request Board review of existing record: The veteran appeals to the Board and asks it to review only the existing record without a hearing. 4. Request Board review of additional evidence, without a hearing. 5. Request Board review of additional evidence, with a hearing. In November 2017, VA initiated a pilot test of the new VBA higher-level review and supplemental claim options. According to VA’s appeals plan, a purpose of this pilot—the Rapid Appeals Modernization Program (RAMP)—is to reduce legacy appeals by providing veterans with a chance for early resolution of their claims within VBA’s new process. Participation in RAMP is voluntary, but veterans must withdraw their pending legacy appeal to participate, according to VA’s appeals plan. VA Has Addressed Some Aspects of GAO’s Recommendations on Appeals Reform Planning In our March 2018 report, we found that VA could help ensure successful implementation of appeals reform by addressing gaps in its planning. We recommended four actions that VA should take: (1) address all legally required elements required by the Act; (2) articulate how it will monitor and assess the performance of the new appeals process compared to the legacy process, (3) augment its master schedule to manage the project, and (4) address risk more fully. VA has taken steps in response to all four, but has not fully addressed our recommendations. VA Has Yet to Provide Complete Information on GAO’s Recommendation to Address the Act’s Required Elements In our March 2018 report, we found that VA’s November 2017 plan for implementing a new disability appeals process while attending to appeals under way in the current (legacy) process, addressed most, but not all, elements required by the Veterans Appeals Improvement and Modernization Act of 2017. Specifically, we found that VA’s appeals plan addressed 17 of 22 elements required by the Act. For the five remaining elements, it partially addressed 4 elements related to monitoring implementation, projecting productivity, and workforce planning, and did not address 1 element related to identifying total resources. This element called for delineating the resources needed by VBA and the Board to implement the new appeals process and address legacy appeals. We recommended in March 2018 that the Secretary of Veterans Affairs address all 22 required elements in the Act in VA’s appeals plan to Congress. This included delineating resources required for all VBA and Board appeals options using sensitivity analyses and results from the RAMP test where appropriate and needed. Since our 2018 report, VA has taken some action on the five elements that were not fully addressed. For example, VA’s updated plan added details related to projecting staff productivity, identifying total resources, as well as about personnel requirements and projections for processing legacy appeals. For identifying total resources, VA added FTE information for other offices that help implement the appeals process and prepared a model to project resource needs. VA’s updated plan, however, continues to only partially address 3 elements related to monitoring implementation and workforce planning, and now addresses the 1 element related to projecting productivity and partially addresses the 1 element related to delineating the total resources. For total resources, VA’s updated plan does not delineate the total resources required by VBA and the Board. Until VA’s appeals plan provides complete information on all required elements, Congress may not have the information needed to conduct oversight of the agency’s efforts to implement and administer the new process while addressing legacy appeals. VA Has Partially Addressed GAO’s Recommendation to Measure, Monitor, and Assess Performance In our 2018 report, we found that VA could improve its planning practices related to monitoring and assessing performance on a range of key dimensions of success. Specifically, the plan had not (1) established timeliness goals for two of the three Board options (i.e., Board review of additional evidence without a hearing and Board review of additional evidence with a hearing); (2) articulated additional aspects of performance important for managing appeals, such as accuracy of decisions, veteran satisfaction with the process, or cost; (3) provided important details about what aspects of the new appeals’ performance would be compared to what aspects of the legacy process’ performance; or (4) explained how the agency would monitor whether resources are being appropriately devoted to both the new and legacy appeals processes and how it will track both sets of workloads. To address these gaps, we recommended that the Secretary of Veterans Affairs clearly articulate in VA’s appeals plan how VA will monitor and assess the new appeals process compared to the legacy process. These include specifying a balanced set of goals and measures with related baseline data, such as timeliness goals for all VBA appeals options and Board dockets, and measures of accuracy, veteran satisfaction, and cost. In its May 2018 updated plan, VA addressed some but not all aspects of this recommendation. Specifically: Timeliness goals and balanced measures. VA’s updated plan states that the agency is collecting data to inform its development of a complete and balanced set of measures for all new appeals options (e.g., timely and accurate processing of appeals while ensuring veteran satisfaction). VA’s original plan had outlined timeliness goals for the two VBA options and for the Board option that does not include new evidence or a hearing. However, VA does not intend to establish timeliness goals or balanced measures for all options until after fully implementing the new appeals process. Further, VA officials told us they are working to produce metrics required under the Act, but have yet to fully articulate a plan for monitoring. For example, there is not a specific plan to monitor the accuracy of decisions under or veteran satisfaction with the new process. Until VA identifies a complete set of timeliness goals and balanced measures, the agency will not have a way to determine how well the new process is performing. Comparison of new and legacy processes. VA’s updated plan states that VA is working toward capturing the metrics listed in section 5 of the Act, which could help VA measure relative performance of the new and legacy processes. However, VA’s updated plan does not state how VA will assess whether the new process addresses problems in the legacy process. For example, according to VA’s updated plan and agency officials we interviewed, VA believes it cannot measure the timeliness of legacy appeals processing from when an appeal is filed to its resolution. According to VA, developing this measure is not feasible because the legacy process has no defined endpoint. Submission of additional evidence by veterans can, at any point, cause additional cycles of re- adjudication. However, VA has not articulated other options for comparing the timeliness of the new and legacy processes in its May 2018 update to its plan. Without this assessment, VA cannot determine the extent to which the new process, which also allows for multiple appeal opportunities, will achieve final resolution of veterans’ appeals sooner, on average, than the legacy process. Moreover, VA’s updated plan does not fully explain how the agency will use the Act’s metrics to assess relative performance of the new and legacy appeals processes on issues like accuracy, veteran satisfaction, or cost. Monitoring processing of legacy versus new appeals. VA’s updated plan articulates VA’s intention to use sensitivity and other analyses to monitor and address workload changes in its legacy and new appeals processes. These analyses could better position VA to manage the two parallel processes. Nevertheless, VA has not established complete and balanced goals and measures or developed a plan for comparing the new and legacy processes. In recent communications on the status of implementing our recommendations, VA officials indicated they plan to address some of these monitoring and performance issues in the next update. Until VA does so, the agency risks not fully understanding whether the new process is an improvement, or whether veterans with appeals in the legacy process are experiencing poor results. VA Has Made Little Progress in Addressing GAO’s Recommendation to Augment Its Master Schedule for Implementation Our March 2018 report also identified elements of a high-quality and reliable implementation schedule that were missing from VA’s master schedule for appeals reform. Specifically, we reported that VA’s master schedule—which the agency included with its November 2017 plan—did not (1) include all key activities; (2) show which activities must finish prior to the start of other activities, or the amount of time an activity could be delayed before the delay affects VA’s estimated implementation date; (3) reflect interim goals and milestones for monitoring implementation; or (4) assign resources for activities. We recommended that the Secretary of Veterans Affairs augment the master schedule for VA’s appeals plan to reflect all activities—such as modifications to IT systems—as well as assigned responsibilities, interdependencies, start and end dates for key activities for each workgroup, and resources. These steps establish accountability and reduce overall risk of implementation failures. In its updated plans, VA took steps to develop interim goals and milestones for monitoring implementation, among other positive actions, but the master schedule still included gaps in sound practices for project management. Specifically: Key activities and their duration. The updated master schedule VA provided in its May 2018 plan added activities, but VA continues to exclude some major activities—including those beyond the planned February 2019 implementation date—and their duration. For example: The updated master schedule does not include a small-scale pilot of the new Board options, even though this pilot is occurring at the same time VA is preparing for full implementation. In response to our questions about this issue, as of July 2018, VA officials said they are adding related pilot test activities to the master schedule. Many activities in the master schedule have the same label or description, such as “communications,” “change management,” “implementation,” “training,” and “hosting,” that do not clearly identify their associated end product without the need to review high-level summary or predecessor activity names. The updated master schedule lacks details and transparency regarding Caseflow, the new information technology system for VA’s appeals process. While VA identified the overall functionality and general timing needed for Caseflow, the steps to accomplish them lack specificity. Further, VA’s updated plan indicates Caseflow will be “minimally ready” by the end of calendar year 2018. At a June 2018 meeting with VA, we asked officials to define the term “minimally ready” and what additional activities or functionality, if any, they planned after reaching this milestone. In response, VA officials pointed us to another source that they said outlined the remaining functionality to complete Caseflow. However, when we consulted this source, we could not determine what functionality listed was to be implemented before or after October 2018. The updated master schedule also lacks start and finish dates as well as status information (e.g., not started, in planning, in progress, complete, etc.) for many of the activities. Sequencing and linkages among activities. VA’s updated plan provided new details about some sub-activities related to processing legacy appeals, monitoring implementation, drafting Board policies, and training. Moreover, the May 2018 updated master schedule was reorganized to improve its flow and alignment, according to VA officials. However, the overall updated master schedule generally does not indicate logical relationships regarding the sequence in which activities should occur, and whether any delays in one activity will dynamically affect other activities linked to it. This type of logic is necessary to define both when an activity may start and finish and when an activity must start and finish for meeting a specified program completion date. These are known as early and late dates, respectively. For example, the plan does not indicate the latest date regulations can fall behind schedule before the planned February 2019 implementation date is impacted, or related activities such as training. This sound planning practice is especially important because VA officials said the agency is concurrently executing many of the activities. Without logical relationships, the master schedule is less effective for modeling the impact of delayed or accelerated activities on related activities, and ultimately for estimating the final implementation date. Interim goals and milestones for monitoring implementation. VA has taken steps to address this aspect of the recommendation. In addition to reiterating the use of an agency-wide governance structure to coordinate implementation of its new appeals process, VA in its updated May 2018 plan added indicators to monitor and assess its readiness for full implementation. Indicators include monitoring the status of implementing regulations and information technology as well as considering any lessons learned through its piloting of the new process. These “readiness indicators” could help VA better identify potential issues related to implementation of the new appeals process. However, the master schedule does not show sequencing and linkages for these indicators. Establishing resources. VA’s updated plan states the agency will use existing resources to implement the new appeals process. Moreover, the master schedule identifies the “owners” or parts of the organization that are playing a role in appeals reform, such as the Veterans Health Administration (VHA). However, other than identifying the “owners” for the activities, resources needed are not identified for the groups of related activities identified in the master schedule or for processing legacy and new appeals processes once implemented in February 2019. By not estimating these resources, VA’s plan does not illuminate resource constraints and indicate whether other parts of the organization or workgroups are dedicated full-time to the tasks or activities for which they are responsible, or whether other constraints exist on funding or time. In general, neither the plan nor the master schedule refers to underlying budget or cost documents or information. In recent discussions on the status of implementing our recommendations, VA officials indicated they plan to address some of these issues in the August 2018 update. Until all necessary activities are accounted for, VA cannot be certain whether key activities are scheduled in the correct order, resources are properly allocated, and key risks have been identified, among other sound practices for guiding implementation and accountability. Furthermore, to the extent that the master schedule is used for internal coordination, the absence of necessary elements could hinder coordination, increasing the likelihood of disruption or delay. VA Has Addressed Many, but Not All Aspects of GAO’s Recommendation to More Fully Assess Risk In our 2018 report, we found that VA’s November 2017 appeals plan could more fully assess key risks related to implementing the new appeals process. In particular, we found that VA’s plan did not include testing of new Board options or clearly define how it would assess the RAMP pilot test of the VBA-only options before implementing the process more broadly. Further, we reported that VA’s plan had not comprehensively reflected key risks because the agency had not established a complete and balanced set of goals and measures, which are a necessary pre-condition to effectively assessing risk. We recommended the Secretary of Veterans Affairs ensure that the appeals plan more fully addresses risk associated with appeals reform by, for example, assessing risks against a balanced set of goals and measures, articulating an assessment plan for RAMP, and testing or conducting sensitivity analyses of all appeals options—prior to fully implementing the new appeals process. In its updated May 2018 plan, VA took many steps to address our recommendation, although opportunities exist to better assess risks associated with implementing appeals reform and managing appeals workloads in the legacy process. Specifically: Testing all aspects of the new appeals process. Since our March 2018 report, VA has taken steps to pilot test the three new Board appeals options. In its May 2018 updated plan, VA describes a small-scale test program—the Board’s Early Applicability of Appeals Modernization (BEAAM)—to collect information about what options veterans choose and their experiences using the new appeals options. For BEAAM, the Board is partnering with veterans service organizations to identify 50 veterans who are dissatisfied with a recent claim decision, and allowing these veterans to appeal directly to the Board. Participating veterans have begun opting in, and VA plans to collect information on adjudication of these appeals. In addition, for veterans dissatisfied with their RAMP decisions, as of October 2018 the Board will begin adjudicating their appeals to further test new Board processes and technology. VA officials also reported progress with developing new sensitivity analyses that will allow the agency to change assumptions related to key variables—both individually and in conjunction with one another. VA anticipates these analyses will allow the agency to project potential budget needs and staffing requirements and more accurately predict resolution of legacy appeals given certain assumptions. Further, VA anticipates using the analyses to determine distribution of resources, and quickly react to changes in its pending legacy and new appeals processes, and other trends. By taking these steps, VA may be better positioned to estimate future disability appeals inventories, timeliness, and resource needs as well as assess risks associated with implementing a new appeals process. Defining success criteria and articulating how to assess RAMP and BEAAM. In its updated plan, VA broadly defines what it hopes to achieve with the RAMP and BEAAM pilots, such as providing information on veterans’ choices in the new process, testing new technology and procedures, and estimating workloads. It also states that VA will use the results to inform the assumptions in its sensitivity analyses. In addition, the updated plan states that VBA is refining the methods to evaluate RAMP. The applicability of BEAAM results to a fully implemented appeals process may be limited. For example, the BEAAM pilot and the Board’s implementation of RAMP provide limited time in which to conduct and assess the results. Moreover, because VA’s test is very small in scale (up to 50 veterans), it will be important for VA to consider, for example, whether these appeals reflect the complexity of cases and the range of circumstances expected in a fully implemented new appeals process. In a mid-May 2018 meeting with VA officials, we raised these and similar concerns. VA officials said they would consider these concerns. Finally, although VA’s updated plan includes a timeline for testing and assessing the new processes, VA’s updated schedule indicates that VA is planning to assess RAMP results between February 15, 2019 and May 10, 2019. These dates occur after VA intends to fully implement its new process. Our recommendation specifies that testing and assessment of pilot results should occur prior to full implementation. Comprehensively assess risks. Within VA’s updated plan, VA has added to its “risk register,” which describes risks associated with many elements of its plan and related mitigation strategies. However, VA’s updated plan has not established a complete and balanced set of goals and measures as discussed above, which are a necessary pre-condition to effectively assessing risk. Having a complete set of goals and measures would allow VA to better identify and target risks associated with reaching these goals while concurrently managing two processes. Thus, VA may not have comprehensively reflected key risks in its updated plan. In conclusion, although VA intends to fully implement the new disability appeals process in about 6 months (February 2019), VA still has an opportunity to create a stronger foundation of sound planning practices. To its credit, VA has taken a number of positive steps toward implementing our prior recommendations to improve its planning for disability appeals reform while it attends to legacy appeals. Efforts such as testing Board appeals options and resuming sensitivity analysis will provide useful information to guide VA through the uncertainty often associated with process change. However, VA needs to fully address our four recommendations to reasonably assure smooth implementation of appeals reform. As we noted in our prior work, VA is undertaking a complex endeavor that involves updating and creating new processes while on-boarding hundreds of new staff and implementing new technology—an endeavor that will affect the lives of hundreds of thousands of veterans with disabilities. Such an undertaking requires an appropriate level of planning to improve VA’s chance of success. VA’s continued efforts to address our recommendations will better position the agency in its implementation of new appeals processes. Chairman Roe, Ranking Member Walz, and Members of the Committee, this concludes my prepared statement. I would be pleased to respond to any questions you may have at this time. GAO Contact and Staff Acknowledgments For further information about this testimony, please contact Elizabeth H. Curda at (202) 512-7215 or curdae@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this testimony. Other key contributors to this testimony include James Whitcomb (Assistant Director), Daniel Concepcion (Analyst in Charge), and Michele Grgich. In addition, key support was provided by Susan Aschoff, Mark Bird, Grace Cho, Alex Galuten, Joel Green, Sheila R. McCoy, Karen Richey, Almeta Spencer, and Walter Vance. This is a work of the U.S. government and is not subject to copyright protection in the United States. The published product may be reproduced and distributed in its entirety without further permission from GAO. However, because this work may contain copyrighted images or other material, permission from the copyright holder may be necessary if you wish to reproduce this material separately.

Why GAO Did This Study VA's disability compensation program pays cash benefits to veterans with disabilities connected to their military service. In recent years, the time needed to complete appeals of VA's decisions on claims has increased. For appeals resolved in fiscal year 2017, veterans waited an average of 3 years. The subset of appeals resolved by the Board of Veterans Appeals—a separate VA agency that provides a higher level of appeals review—took on average 7 years to resolve. The Veterans Appeals Improvement and Modernization Act of 2017 makes changes to VA's current (legacy) process, giving veterans options to have their claims further reviewed by VA or appeal directly to the Board. The Act requires VA to submit a plan to Congress and GAO for implementing a new appeals process (which VA submitted in November 2017) and periodic updates (which VA submitted in February and May 2018). The Act also includes a provision for GAO to assess VA's original plan. In March 2018, GAO found that VA could help ensure successful implementation of appeals reform by addressing gaps in planning and made four recommendations, with which VA agreed. This testimony focuses on the steps VA has taken to address GAO's recommendations and what aspects remain unaddressed. For this statement, GAO reviewed VA's May 2018 updated plan, and interviewed VA officials and reviewed information they provided about steps taken to implement GAO's recommendations. What GAO Found In a March 2018 report, GAO assessed the Department of Veterans Affairs' (VA) November 2017 plan for changing how veterans appeal disability claim decisions and found that VA could do more to successfully implement these reforms. The March 2018 report made four recommendations to address planning gaps. Since then, VA has updated its plan and taken some steps to address aspects of these recommendations, but further steps are needed: Address all legally required elements . GAO reported that VA's plan did not address one and partially addressed four of 22 elements required by the Veterans Appeals Improvement and Modernization Act of 2017 (Act), and recommended VA fully address them all. In a May 2018 update to its plan, VA took steps to address the five elements, such as developing productivity projections and a model to forecast resource needs for processing appeals. These steps address one element related to projecting productivity, and partially address the four remaining elements. Articulate performance measurement . GAO also recommended VA clearly articulate how it will monitor and assess the new appeals process relative to the legacy process. This recommendation includes specifying timeliness goals for five new appeals options to be made available to veterans, and additional goals or measures of performance, such as accuracy in processing appeals. VA's updated plan states that the agency will develop goals and measures for all appeals options after fully implementing appeals reform. Contrary to sound planning practices, it does not articulate these performance goals and measures now, which would provide a vision for what successful implementation would look like. Lacking this vision, VA does not have an “end state” to guide its implementation and help establish accountability. Augment project plan . GAO recommended VA augment its master schedule for implementing appeals reform to include all key activities and reflect other sound practices for guiding implementation and establishing accountability. Although VA's May 2018 updated master schedule added activities, it omitted a pilot test of the new Board of Veterans' Appeals (Board) options. More generally, the plan does not reflect interdependencies among activities. Until all key activities are accounted for and the master schedule reflects sound practices, VA cannot provide reasonable assurance that it has the essential information needed to manage its appeals reform implementation. Address risk fully . GAO recommended that VA's appeals plan more fully address risks in implementing a new process by, for example, testing all appeals options prior to full implementation. In its updated plan, VA stated it will pilot all five new appeals options. By taking these steps, VA should be better positioned to assess implementation risks. However, the updated plan does not have well-defined, measurable criteria for assessing lessons learned from these pilots and does not articulate how well these lessons translate to a broader context. Taking these steps would improve VA's ability to assess and mitigate risks as it implements its reforms.

gao_GAO-18-454

gao_GAO-18-454_0

Background Overview of Coast Guard’s Major Acquisition Programs As of March 2018, the Coast Guard’s portfolio of major acquisitions has 10 programs, 2 more than during our June 2014 review, when the Coast Guard had 8 major acquisition programs (see figure 1). DHS defines major acquisition programs as those with life-cycle cost estimates of at least $300 million. Appendix II provides information on the programs included in the Coast Guard’s major acquisition portfolio in 2018. Programs in the acquisition portfolio progress through a series of four acquisition phases, accompanied by a series of acquisition decision events (ADE), outlined in the DHS’s acquisition life-cycle framework (acquisition process). Figure 2 depicts the acquisition process. Oversight of the Coast Guard’s Acquisition Portfolio The Coast Guard currently has three cross-directorate groups that include members from the acquisitions, resources, and requirements directorates and are responsible for addressing and overseeing issues across the Coast Guard. Since 2011, these three groups—the Executive Oversight Council, the Systems Integration Team, and the Resource Councils— have helped oversee the Coast Guard’s acquisition portfolio. Table 1 provides information on the roles and responsibilities of these three groups. Each of these groups has a charter to identify its purpose and scope of responsibilities, which involve providing cross-directorate representation and information on all of the acquisition programs to help manage its portfolio. The Coast Guard updated the roles and responsibilities for two of its cross-directorate groups in its Major Systems Acquisition Manual, including how the groups are to interact and work together within the established acquisition governance framework. For example, the Resource Councils are to serve as advisors to the senior-level Executive Oversight Council. Each of the Resource Councils is to report directly to the Executive Oversight Council for issues within its own domain and report to the Systems Integration Team for issues that cross domains. The Executive Oversight Council oversees the acquisition governance framework and is positioned to delegate tasks to the other two cross- directorate groups or obtain information from them to assist in the management of acquisitions to address problems related to acquisitions. Asset Service Life and Maintenance of Legacy Assets Coast Guard assets are developed with a specific design service life. According to Coast Guard officials, the design service life for aircraft is established as a maximum number of flight hours; while for cutters, the design service life is the number of years the cutter is expected to operate based on contractual design requirements. An asset’s design service life can be extended through major maintenance events, such as Service Life Extension Projects (SLEP). SLEPs are funded with the Coast Guard’s acquisition, construction, and improvements appropriation account whereas routine depot-level maintenance is funded with the Coast Guard’s operating expenses appropriation account. SLEPs address specific systems and major maintenance to extend the service life of an asset beyond the original plan. A SLEP is not designed to increase an asset’s capability; it extends the service life by replacing obsolete, unsupportable, or maintenance- intensive equipment. Table 2 provides more details about the design service life and maintenance history of select legacy assets. Prior GAO Reviews We have issued several reports since 2012 on the Coast Guard’s management of its acquisition portfolio and the oversight of its depot-level maintenance resources. We have made several recommendations in these reports. For example, in September 2012, we found a mismatch between resources needed to support all approved major acquisition program baselines and expected funding levels. This resulted in the Coast Guard requesting funding for programs as a part of its annual budget process below the levels identified in programs’ life-cycle cost estimates, resulting in a bow wave of future funding requirements. At the time, DHS and the Coast Guard acknowledged this resource challenge, but we found they had not developed a clear strategy for moving forward. At that time, DHS officials stated that funding variability results in inevitable trade-off decisions being made on an annual basis. We recommended that the Coast Guard conduct a comprehensive portfolio review to develop revised baselines that reflect acquisition priorities as well as realistic funding scenarios. DHS concurred with our recommendation. Since 2014, the Coast Guard has undertaken efforts to address this issue, but, as of October 2017, we found these efforts have not led to the significant trade-off decisions needed to improve the affordability of the Coast Guard’s portfolio. Additionally, in September 2012, we found that the Coast Guard had established an acquisition governance framework and that the Executive Oversight Council was well positioned to receive information from other cross-directorate groups in order to manage the acquisition portfolio. However, while the Executive Oversight Council had been active in overseeing individual programs, it had not met to oversee the portfolio collectively. Officials told us at that time that the portfolio oversight was done through the annual budget process. We found this approach to managing portfolio affordability was ineffective and facilitated immediate trade-offs, and did not provide the best environment to make decisions in developing a balanced long-term portfolio. We recommended that the Coast Guard identify the Executive Oversight Council as the governing body to oversee the Coast Guard’s acquisition enterprise with a portfolio management approach. In addition, this council should supplement individual program reviews with acquisition portfolio-wide reviews to make performance and affordability trade-off decisions that will help ensure the Coast Guard is acquiring a balanced portfolio to meet mission needs. DHS concurred and the Coast Guard updated the Executive Oversight Council’s charter in 2014 to require the group to annually oversee the acquisitions collectively as a balanced long-term and affordable portfolio. Similarly, in June 2014, we found that the Coast Guard had repeatedly delayed and reduced its capabilities through its annual budget process and did not know the extent to which it would meet mission needs and achieve desired results. We reported that this was because the Coast Guard did not have a long-term fleet modernization plan that identified all acquisitions needed to meet mission needs over the next 20 years within available resources. We recommended that the Coast Guard develop a 20-year fleet modernization plan that identifies all acquisitions needed to maintain the current level of service and the fiscal resources necessary to build the identified assets. We recommended this plan consider trade-offs in cases where the fiscal resources needed to execute the plan are not consistent with annual budgets. DHS concurred, but, according to Coast Guard officials, the plan has yet to be approved. In addition to reporting on the Coast Guard’s management of its acquisition portfolio, we have also issued several reports on how it oversees depot-level maintenance funding. For example, in July 2012, we found that the Coast Guard’s depot-level maintenance cost-estimating process did not fully reflect best practices. We recommended that the Coast Guard conform its estimated depot-level maintenance expenditures with cost-estimating best practices. DHS concurred; however, it raised several points that we found could limit the implementation of the recommendation. DHS stated, for example, that cost-estimating best practices are most applicable for new acquisitions. As our report noted, our cost-estimating best practices guide is intended to be applicable to programs and assets in all stages of their life cycles, including maintenance and support. Additionally, in March 2017, we found that the cost estimates were not adjusted or updated over the course of an asset’s service life, leading to a large discrepancy between expected and actual annual depot-level maintenance expenditures. We recommended that the Coast Guard periodically update standard support levels, which are annual estimates for depot-level maintenance over the course of an asset’s life cycle, to account for actual expenditures. DHS concurred with our recommendation and plans to complete actions to implement it by December 2018. Effectiveness of the 5-Year Capital Investment Plan Is Limited and Coast Guard Does Not Review Its Acquisition Portfolio Collectively The Coast Guard’s 5-year CIP, a congressionally mandated report, does not fully reflect cost realities or acquisition needs. For example, the most recent CIP—from fiscal years 2018 through 2022—projects funding for its portfolio of major acquisitions that, over the 5-year period, exceeds average budget requests in the last several years. As such, we found that the Coast Guard continues to face the same programmatic risks that annual CIP-based planning perpetuates, similar to what we have been reporting since 2011. To address funding constraints, the Coast Guard has been in a reactive mode by making prioritization decisions through the annual budget process without identifying how trade-off decisions made in the current budget cycle will affect the future of the acquisition portfolio. As a result of this planning process, and as we found in 2012 and in this current review, the Coast Guard has continued to defer planned acquisitions to future years and left a number of operational capability gaps unaddressed that could affect future operations. Moreover, the Coast Guard has not conducted portfolio-wide oversight through its cross-directorate groups. Coast Guard CIP-based Planning Has Shortcomings That Limit Its Effectiveness The 5-year CIP is the Coast Guard’s key acquisition portfolio planning tool. However, since 2011, we have reported on shortcomings that limit its effectiveness. As required by statute, the Coast Guard prepares a 5-year CIP that is required to be updated and submitted annually with the administration’s budget request. This 5-year CIP provides information on the proposed budget for the upcoming fiscal year and the following 4 fiscal years. Coast Guard officials told us the 5-year CIP is the starting point for developing the acquisition, construction, and improvements budget for a given year, which funds asset acquisitions as well as major sustainment projects and infrastructure investments. Officials also indicated that operational commanders provide input for the budget, as do senior Coast Guard officials for operations, resources, and others who have a role within its resource governance construct. As we have previously found, the Coast Guard’s 5-year CIPs continue to demonstrate a pattern of certain planning practices, to include: not identifying priorities or trade-offs between acquisition programs and not showing the effect of current decisions on the overall affordability of the acquisition portfolio; projecting funding levels for the current budget year that do not reflect the full extent of the Coast Guard’s projected acquisition needs; and projecting funding levels for future years that frequently surpass the average funding amounts requested by the Coast Guard in recent years. These shortcomings limit the Coast Guard’s ability to manage the affordability of its acquisition portfolio. Coast Guard officials said the CIP reflects the highest priorities of the department within the given top funding level and that prioritization and trade-off decisions are made as part of the annual budget cycle. However, these decisions, and the resulting impacts on affected programs, are not articulated in the CIPs. While the Coast Guard is not required under statute to identify the effects of trade-off decisions in the CIP, failing to show which acquisitions would take on more risk so other acquisitions can be prioritized and adequately funded within budget parameters also makes it difficult for Congress and other stakeholders, such as DHS and the Office of Management and Budget (OMB), to understand other options the Coast Guard considered. GAO’s Cost Estimating and Assessment Guide states that comparative analyses showing facts and supporting details among competing alternatives, such as budget priorities, should consider trade-offs needed to identify solutions and manage risk. Our past work has also highlighted other best practices for portfolio management, such as demonstrating comprehensive knowledge of the portfolio, including needs, gaps, and how to address those gaps; prioritizing investments through alignment of requirements, acquisition, and budget processes; and use of long-term planning. As we found in September 2012, the Coast Guard’s approach of relying on the annual budget process to manage portfolio affordability does not provide the best basis for making decisions to develop a more balanced and affordable portfolio in the long-term. In June 2014, we also found that there is no evidence that short-term budget decisions will result in a good long-term strategy, and the Coast Guard’s annual budget-driven trade-off approach creates constant churn as program baselines must continually re-align with budget realities instead of budgets being formulated to support program baselines. This results in trade-off decisions between capability and cost being pushed into the future. For example, the Coast Guard has a stated requirement for three medium icebreakers and three heavy icebreakers, and has initiated an acquisition program for heavy icebreakers. Assets acquired under this program will replace the Coast Guard’s only operating heavy icebreaker—the Polar Star—which is well past the end of its original design service life. The Coast Guard currently plans to have the three heavy icebreakers delivered in 2023, 2025, and 2026. Additionally, the Coast Guard operates one medium icebreaker, the Healy, which has an expected end of service life in 2029. Despite the requirement for three medium icebreakers, Coast Guard officials said they are not currently assessing acquisition of the medium polar icebreakers because they are focusing on the heavy icebreaker acquisition and plan to assess the costs and benefits of acquiring medium polar icebreakers at a later time. This planning approach can also lead to delayed capabilities and program risks, as the 5-year CIP does not prioritize acquisition programs in its projections for the 5-year period. We found in 2017 that both acquisition needs—as articulated in program baselines—as well as the 5-year CIP’s funding projections frequently surpass the average requested funding amounts in recent years. Similarly, in this review, we found this to be the case in the 5-year CIP covering fiscal years 2018 through 2022. Congressional appropriations for Coast Guard acquisition, construction, and improvements in fiscal years 2013 through 2018 exceeded the Coast Guard’s requests. Explanatory materials on the annual appropriations acts for these fiscal years indicated, among other things, that funding was provided above requested amounts for procurement of one HC-130J aircraft each year. The House Committee on Appropriations report accompanying the Homeland Security appropriations bill for 2018 noted that the Secretary is expected to include adequate funding in the fiscal year 2019 budget request to normalize the recapitalization of the HC-130 fleet. Recent annual appropriations acts also direct the use of funds to contract for three National Security Cutters that were not a part of the program of record for the fleet of these cutters. Absent any additional funding appropriated by Congress above Coast Guard requests, the acquisition portfolio put forth in the fiscal year 2018 CIP will not be affordable by fiscal year 2019 based on average recent budget requests. Figure 3 shows aggregate projected funding for various major Coast Guard acquisitions over the fiscal year 2018 through 2022 CIP, along with average budget requests and appropriations from fiscal years 2014 through 2018. Further, the previous Commandant of the Coast Guard testified in November 2017 that an annual acquisition budget of $2 billion is needed to modernize the fleet and address other critical priorities, such as the recapitalization of the Coast Guard’s icebreaker fleet. However, the fiscal year 2018 through 2022 CIP, dated October 2017, does not reflect this need for any year in its 5-year budget window. By not providing comprehensive information in the CIP on the acquisitions needed to perform its missions as well as the trade-offs necessary at different funding levels, the Coast Guard is not providing decision makers, including those in Congress, information to help decide which programs are the highest priority and which funding increases may or may not be consistent with the Coast Guard’s programs of record, as approved by DHS. For example, even though recent annual appropriations acts direct the use of funds to contract for three additional National Security Cutters, the Coast Guard had not identified a need for these cutters as they were not part of the original program of record. Coast Guard Is Developing a 20-Year Long-term Major Acquisitions Plan The Coast Guard has initiated the development of a 20-year Long-term Major Acquisitions Plan, but it is incomplete as of March 2018. According to Coast Guard officials, the Coast Guard’s efforts were in response to Congressional direction in 2016. In February 2016, Congress directed that the Coast Guard develop a Long-term Major Acquisitions Plan to cover the upcoming 2017 fiscal year, and for each of the 20 fiscal years thereafter, and stated that it should be updated every 2 years. Specifically, each plan is to include the following: (1) the number and types of cutters and aircraft to be decommissioned; (2) the number and types of cutters and aircraft to be acquired to replace the cutters and aircraft or address an identified capability gap; and (3) the estimated level of funding in each fiscal year required to acquire the cutters, aircraft, and command and control systems as well as acquire, construct, or renovate shore-side infrastructure. As of November 2017, officials told us that the Coast Guard was developing a 20-year Long-term Major Acquisitions Plan that specifically focused on its highest priority recapitalization and sustainment efforts for its assets and will focus on meeting the intent of the 2016 congressional mandate. These officials said that the plan will also be based on the Coast Guard’s 5-year CIP and will contain the necessary sustainment activities for current assets, according to service life limitations and recapitalization efforts for assets that reach the end of their service lives. Coast Guard officials stated that the plan will not be a budget document, but rather an overall planning document for future budgets. As of March 2018, the Coast Guard had not completed this long-term plan. Coast Guard Made Progress in Reducing Cost of the Portfolio but Reactive Planning Has Created Impending Surge of Unfunded Acquisitions Since our 2014 review, the Coast Guard has generally demonstrated improved fiscal management of the major programs in its acquisition portfolio and made progress in acquiring the assets in the portfolio. At that time, we found that program cost increases were consuming significant amounts of funding, and the Coast Guard was further from fielding its planned fleet than it was in 2009, in terms of the fiscal resources needed to finish those programs—or the remaining investment required. Since 2014, program costs have generally been stable and, from 2014 to 2018, the Coast Guard reduced the remaining investment required to complete those acquisitions by $4.9 billion or 24 percent (see table 3). However, while the Coast Guard has reduced the remaining investment required to complete its acquisition portfolio, there is little room for additional major acquisitions based on recent budget requests. For example, the National Security Cutter and Fast Response Cutter—two of the Coast Guard’s most expensive acquisitions programs—both experienced delays and were not delivered as originally scheduled. As a result, these delays stretched its acquisition budget longer than intended. Going forward, our analysis indicated that once the Coast Guard begins funding construction of Offshore Patrol Cutters—another major acquisition program critical in replacing vessels well past their service lives—that program is expected to consume a significant portion of the Coast Guard’s planned acquisition, construction, and improvements budget between 2018 and 2032, also raising uncertainties in how the Coast Guard will be able to fund other priorities. According to the previous Commandant of the Coast Guard, the Offshore Patrol Cutter is the Coast Guard’s top priority and, as such, the Coast Guard will prioritize its budget requests for the Offshore Patrol Cutter before other assets, potentially limiting funds requested for other acquisition programs. This approach will limit the portfolio for the foreseeable future and affect other new programs, such as the Heavy Polar Icebreaker and Waterways Commerce Cutter. These two programs represent critical needs for the Coast Guard, as the legacy assets they are intended to replace are well past their designed service lives, but there are limited resources for them if acquisition of the current portfolio is to be completed as scheduled. The polar icebreaker program has an estimated total acquisition cost of more than $3 billion and, according to the Coast Guard, is needed to alleviate a potential icebreaking mission capability gap. Heavy icebreakers are needed, as Coast Guard officials also indicated, to provide year-round access to the Polar Regions, including the clearance of a navigable channel for access to the National Science Foundation’s McMurdo Research Station on Ross Island, Antarctica as well as to facilitate other national security interests in polar waters. DHS approved the icebreaker program for entry into the obtain phase of the acquisition process in March 2018. The Coast Guard—in partnership with the Navy—is expected to award a contract for design and construction of up to 3 heavy polar icebreakers by June 2019, and plans for the first icebreaker to be delivered by the end of fiscal year 2023. We recently reported on this program in April 2018 and have an ongoing review that is expected to be completed by summer 2018. DHS recently approved a new program, known as the Waterways Commerce Cutter program, to recapitalize aging vessels such as its fleet of 35 Inland Tenders (river, buoy, and construction tenders). The assets in the current fleet continue to age beyond their expected service lives and the Waterways Commerce Cutter program is currently in the analyze/select phase of the acquisition process. Coast Guard officials said they are still determining how many new vessels are needed to provide capabilities similar to the current fleet of vessels that replace or relocate river buoys and builds fixed aids to navigational marine structures. A life-cycle cost estimate has not yet been developed for this program, but, according to Coast Guard officials, the preliminary rough order of magnitude estimate for total acquisition cost is $1.1 billion. As we reported in July 2017, the Coast Guard has no method in place to capture the effects of deferred acquisitions on its future portfolio. The lack of a long-term plan, as discussed earlier, and determining priorities and making trade-off decisions based on the annual budget have put the Coast Guard in a reactive planning mode each year. We found that this type of reactive planning and the Coast Guard’s constrained budget environment have created a bow wave of near-term unfunded acquisitions, negatively affecting future acquisition efforts and potentially affecting future operations. This bow wave consists of new acquisition programs and recapitalization efforts, as well as high-cost maintenance projects that use the acquisition construction and improvements account, which continue to put pressure on available resources. These projects include some that are not currently identified in the acquisition portfolio. For instance, the Coast Guard’s 87-foot patrol boats are forecast to require recapitalization beginning in 2023. Additionally, the ocean-going 175-foot coastal buoy tenders—not included in the Waterways Commerce Cutter program—are past the point in their service lives when a midlife maintenance availability would normally have been conducted. However, we found that the Coast Guard has historically operated vessels well past their expected end of service life, and it will likely need to do so with these assets given limited available acquisition funding. Furthermore, the Coast Guard has identified more than $1.5 billion in shore infrastructure projects, which are paid for with funding from the acquisition, construction, and improvements account that it has not been able to address, primarily due to lack of funding, among other reasons. Some of these projects are detailed in an unfunded priorities list the Coast Guard submitted to congressional committees in July 2017 pursuant to statutory requirements. Among the projects identified are recapitalization for waterfront facilities damaged in hurricanes; major acquisition systems infrastructure associated with homeporting the ninth National Security Cutter; and a number of pier replacements, building construction, and navigational aid realignment projects in several locations. The Explanatory Statement regarding the Consolidated Appropriations Act, 2018, reflected approximately $135 million in acquisition, construction, and improvements funding for shore infrastructure/construction projects, including for some previously unfunded priorities. We currently have an ongoing review to assess Coast Guard shore infrastructure projects and expect to issue a report in early 2019. Figure 4 shows the current and future acquisitions that, based on current Coast Guard programs and requirements, need to be addressed in order for the Coast Guard to meet its statutory missions, along with the backlog of shore infrastructure projects noted above. For more information about the Coast Guard’s 11 mission areas, including which assets perform each mission, see appendix III. Federal standards for internal control state that quality information that is appropriate, current, complete, accurate, accessible, and timely is necessary for an organization to achieve its objectives. The Coast Guard has not communicated quality information to Congress or demonstrated how deferred acquisitions will affect the future acquisition portfolio. Including information in the CIP, such as how trade-off decisions will affect other programs in the portfolio, would allow decision makers, including Congress, to better understand Coast Guard priorities and how changes to one program might potentially affect other programs. Coast Guard Does Not Conduct Oversight of Its Acquisitions Portfolio Collectively The Coast Guard has a management body in place to conduct oversight of its major acquisition programs; however, this management body has not conducted oversight across the entire acquisition portfolio from a collective approach. Among the Coast Guard’s three cross-directorate groups, the Executive Oversight Council is positioned to oversee the portfolio collectively and has the potential to implement key portfolio-wide management practices, including conducting formal reviews and issuing reports. This council has cross-directorate senior-level management representation, access to information on acquisition programs, and support from the other two cross-directorate groups (the Systems Integration Team and the Resource Councils). However, this council has not carried out these portfolio-wide practices. Since 2012, the responsibilities of the Executive Oversight Council regarding portfolio- wide management have been changed multiple times (see figure 5). In 2014, the Coast Guard updated the Executive Oversight Council’s charter, in response to our September 2012 recommendation, adding the responsibility for portfolio-wide oversight to include conducting an annual review to assess and oversee acquisitions collectively. However, during our current review, we found that the Coast Guard revised the council’s charter in June 2017, removing this responsibility. According to Executive Oversight Council officials, this responsibility was removed from the 2017 charter because the council did not conduct these annual reviews. Instead, Executive Oversight Council officials indicated that the council facilitates a balanced and affordable portfolio of acquisition programs through the individual program-level reviews. GAO’s best practices work states that successful organizations assess product investments in aggregate, rather than as independent products or programs. For example, by considering the requirements, acquisition, and budget processes collectively, it helps organizations prioritize their product investments. In addition, Coast Guard officials said that a portfolio-wide affordability review or assessment is undertaken by the Systems Integration Team—a cross-directorate, cross-enterprise group below the flag/Senior Executive Service-level—to help inform the annual budget process. The Systems Integration Team’s responsibilities outlined in its current charter include addressing issues tasked by the Executive Oversight Council chair, reporting to the council on cross-programmatic issues, and providing recommendations to the council. For example, officials with the Systems Integration Team said they met with, and gathered information from, each of the Resource Councils and briefed the Executive Oversight Council in February 2018 with proposals for looking at investments collectively across the Coast Guard enterprise to include potential priorities and trade-offs. They said the briefing included a review of the upcoming annual budget, a look at the overall portfolio of major acquisition programs over the next 10 years, and prospective new start initiatives at low, medium, and high funding levels. It is unclear what actions the Executive Oversight Council has taken as a result of the Systems Integration Team briefing. However, we found that the Executive Oversight Council did not review the portfolio from a collective perspective. Further, the members of the Systems Integration Team, who inform and report to the senior-level Executive Oversight Council, are not at the appropriate senior position to oversee or make decisions for the acquisition portfolio. Specifically, the Executive Oversight Council’s revised 2017 charter states that the Systems Integration Team is to support the council in its role to facilitate a balanced and affordable portfolio as a whole. However, as the higher-level cross-directorate group, the Executive Oversight Council has not engaged in overseeing or reporting on the acquisition portfolio collectively and annually. OMB’s 2017 Capital Programming Guide outlines a capital programming process, including how agencies should effectively and collectively manage a portfolio of capital assets. This OMB guidance states that a senior-level executive review committee should be responsible for reviewing the agency’s entire capital asset portfolio on a periodic basis and for making decisions or priorities on the proper composition of agency assets needed to achieve strategic goals and objectives within the budget limits. In the case of the Coast Guard, only the Executive Oversight Council has representation at the senior-level executive level and has the responsibility for oversight of its major acquisition programs. Without collective portfolio reviews at the senior management level, the Coast Guard does not have sufficient cross-directorate information to determine needed trade-offs in the major acquisitions realm, considering budget realities. Coast Guard Plans to Extend Service Lives of Certain Legacy Assets as Sustainment Costs and Other Risks Are Increasing Given the Coast Guard’s limited acquisition budget in recent years, it is unclear how the Coast Guard will be able to fund planned Service Life Extension Projects (SLEP) on several aging assets in order to sustain them—that is, keep them operating at acceptable levels—until replacement assets are available. We found that each of these sustainment efforts involves a certain amount of risk. For example, according to Coast Guard officials, they plan to operate H-65 and H-60 helicopters to flight hours beyond what has been flown for those aircraft. In addition, several of the Coast Guard’s aging cutters have spent more on depot-level maintenance than was planned. Combined, these cutters—the 210-foot and 270-foot Medium Endurance Cutters, the icebreaker Polar Star, and Inland Tenders—expended in excess of $460 million more than what was originally estimated (standard support levels) from 2010 to 2017. When combined with the challenges facing the acquisition portfolio noted above, the Coast Guard will likely struggle to pay for the maintenance of older assets, a situation that could lead to deferred maintenance and lost operational capability, as we found in our July 2012 review and in our current review. As discussed earlier, the 20- year long-term plan, if completed as directed by our June 2014 recommendation and subsequent congressional direction, will begin to lay out the prioritization of all efforts, trade-offs, and impacts. Coast Guard Intends to Extend the Service Lives of Certain Legacy Assets The Coast Guard currently operates several assets that have passed, or will soon pass, the end of their design service lives—the total period for which they were designed to operate. We found that these legacy assets are generally meeting metrics for availability to conduct operations; however, they are in need of major maintenance overhauls—or SLEPs— in order to continue providing capabilities to operators. According to Coast Guard officials, SLEPs are necessary because the Coast Guard does not have the funds available to initiate a new major acquisition program to recapitalize these assets in the short term, or because a significant amount of maintenance work is required to keep these assets operational until replacements are fielded. Table 4 provides details about the Coast Guard’s plans for SLEPs for selected assets. These planned SLEPs involve several risks including technical, scheduling, and funding. While SLEPs will extend these assets’ expected service lives, they will also add cost to an already constrained Coast Guard acquisition, construction, and improvements account. Since these projects use these funds, we would expect them to be included in the Coast Guard’s forthcoming 20-year long-term plan so that decision makers and stakeholders can see their effects on the broader acquisition portfolio. Additional detail on these planned SLEPs follows. H-65 and H-60 Aircraft The Coast Guard is planning to conduct a SLEP that will add an additional 10,000 hours to the H-65 rotary-wing aircraft, taking the service life of each aircraft in the fleet to 30,000 hours. The Coast Guard is evaluating alternatives to extend the service lives of the H-60 fleet. According to DHS, two options the Coast Guard is considering include utilizing newer H-60 aircraft from the Navy and conducting a SLEP on those aircraft to extend their service lives to 20,000 hours or extending the life of the current fleet to 30,000 hours. Coast Guard officials said that this will allow both the H-65 and H-60 aircraft to operate into the mid- 2030s so that the Coast Guard can focus funds from the acquisition, construction, and improvements account on the Offshore Patrol Cutter procurement and align its next helicopter acquisition effort with the Department of Defense’s future vertical lift acquisition plans. However, there are risks associated with these SLEP plans. According to Coast Guard officials, they plan to operate H-65 and H-60 helicopters to flight hours beyond what has been flown for those aircraft. The Coast Guard is working with the original manufacturers to identify structural components that would need to be replaced to accomplish the service life extension. From fiscal years 2012 to 2017, the H-65 operational availability—time available to conduct missions—averaged 70.9 percent and the H-60 averaged 73.5 percent, compared to their target of 71 percent. Both aircraft generally met their target but are approaching their end of service lives, with the H-65 expected to reach its 20,000 flight hour limit starting in 2020 and the H-60 in 2023. The Coast Guard expects the H-65 SLEP to cost about $61.6 million, but the H-60 SLEP cost is unknown because the effort has not progressed to the acquisition decision event at which a cost estimate is required to be approved. The H-60 SLEP was recently approved for entry into the analyze and select phase, where it was designated as a level 1 program, which DHS defines as programs with estimated life-cycle costs greater than or equal to $1 billion. Heavy Icebreaker Polar Star The Coast Guard conducted reactivation work on the Polar Star from 2010 to 2013, and the icebreaker resumed its missions for the annual breakout of the National Science Foundation’s McMurdo Research Facility in Antarctica in 2014. The Coast Guard is planning a SLEP on the Polar Star to keep it operational until the first and second new heavy polar icebreakers are delivered (planned for 2023 and 2025, according to current acquisition plans) in order to bridge a potential operational gap. This approach would allow the Coast Guard to operate a minimum of two heavy icebreakers once the first polar icebreaker is delivered. The approach would also provide the Coast Guard with a self-rescue capability—the ability for one icebreaker to rescue the other if it became incapacitated while performing icebreaking operations. The Coast Guard’s plan to conduct the Polar Star SLEP during its existing annual depot-level maintenance periods may not be feasible given the amount of maintenance already required on the cutter. The Polar Star’s mission capable rating has been decreasing in recent years and reached a low point of 29 percent—well below the target of 41 percent—from October 2016 to September 2017. Based on mission capable data, we found this is mostly due to additional time spent in depot-level maintenance, which has increased in recent years from about 6 months in 2015 to more than 8 months in 2017. Additionally, the Polar Star has required extensions of about 3 months for its annual dry dock periods—the period of time when a cutter is removed from the water so that maintenance can be conducted—in 2016 and 2017 to complete required maintenance activities. These dry docks were originally planned to last between 2-1/2 months and 4 months. These extensions also compressed the amount of time that the crew had to prepare for its annual mission to Antarctica, which, according to members of the Polar Star crew, placed a large stress on the crew, risked the quality of work, and reduced or eliminated the crews’ planned rest and personal preparation for their roughly 4-month deployment. Based on our analysis, these delays and extensions are likely to continue as the cutter ages. According to Coast Guard officials, the Polar Star’s SLEP work will be conducted during the annual dry dock periods by adding an additional 1 or 2 months to the annual dry docks. However, if the work is unable to be completed during this time frame, it could force the Coast Guard to miss its commitment to conduct the annual Antarctica mission. Coast Guard maintenance officials stated that until the Polar Star completes the SLEP, its repairs will likely continue to get more expensive and time consuming. We will continue to monitor the Polar Star’s SLEP through our annual review of DHS programs. As we found in July 2017, the Polar Star SLEP effort has a rough order cost estimate of $75 million, which is based on the reactivation work completed in 2013. However, this estimate may be unrealistic based on assumptions the Coast Guard used, such as that it would continue to use parts from the Coast Guard’s other heavy polar icebreaker, the Polar Sea, which has been inactive since 2010. The Coast Guard’s recent assessment of the Polar Star’s material condition—the physical condition of the cutter, which includes the hull structure, habitability, major equipment systems, and spare parts availability—was completed in January 2018. The material assessment stated that many of the available parts from the Polar Sea have already been removed and installed on the Polar Star. As a result of the finite parts available from the Polar Sea, the Coast Guard may have to acquire new parts for the Polar Star that could increase the $75 million SLEP estimate. The Polar Star’s recent material assessment will form the basis to determine which systems will be overhauled during the SLEP and for a more detailed cost estimate. The Coast Guard expects the program to reach the obtain phase of the acquisition life cycle by December 2019, at which time the Polar Star could reach the end of its current useful service life (currently projected to be between 2020 to 2023). This timeline contains risk that the Polar Star could be rendered inoperable before the cutter is able to undergo a SLEP. Medium Endurance Cutters The Coast Guard operates two fleets of Medium Endurance Cutters (270- foot and 210-foot cutters) and both are either approaching or have exceeded their design service lives. According to Coast Guard maintenance officials, the primary problem facing the 270-foot Medium Endurance Cutters is obsolescence given the age of these cutters. The cutters have several systems that are no longer manufactured, and in many cases the original manufacturer no longer makes parts for the systems, such as the generators, fire pumps, and main diesel engines. In order to sustain the 270-foot Medium Endurance Cutters until the Offshore Patrol Cutters—replacements for the Medium Endurance Cutters—are delivered, the Coast Guard is planning to conduct a SLEP. Officials stated they are evaluating how many of the 13 cutters will undergo the SLEP. The Coast Guard does not have a cost estimate for the SLEP, but officials said that the project should enter the obtain phase and complete its first cost estimate by June 2019. Despite the age and condition of the cutters, the mission capable rate for the 270-foot Medium Endurance Cutters has been increasing since the fleet first started using the metric in August 2014 and has met its minimum target of 49 percent. Specifically, the 270-foot Medium Endurance Cutters’ mission capable rate increased from 47.6 percent in 2015 to 69.4 percent in 2017. This indicates that the Coast Guard has been increasing the amount of time that the cutters are available to conduct operations. However, the mission capable rating of 69.4 percent in 2017 is above the maximum target—61 percent—which means the Coast Guard is operating the cutters more than planned. This could be troublesome since the percentage of time above the 61 percent target is time that is allocated to depot-level maintenance, meaning these cutters are not spending as much time as planned in maintenance. In May 2016, we found that deferring maintenance can lead to declining ship conditions and longer maintenance periods that can reduce a ship’s operational availability. The Coast Guard is also evaluating how long the 270-foot Medium Endurance Cutters should remain in service. According to Coast Guard officials, this decision is at least partially dependent on the delivery of the Offshore Patrol Cutters—specifically the shipbuilder’s ability to deliver 2 cutters per year, which is expected to start in fiscal year 2024 with the 4th and 5th cutters. Officials stated that the Coast Guard does not plan to operate any Medium Endurance Cutters once all 25 Offshore Patrol Cutters are operational, yet the fiscal year 2018 through 2022 CIP report indicates that 7 of the 270-foot Medium Endurance Cutters will still be in service when all 25 Offshore Patrol Cutters are delivered and operational. Officials said this is a contingency plan in case not all Offshore Patrol Cutters are delivered on time. As we found in June 2017, the Coast Guard completed refurbishment work on the 210-foot and 270-foot Medium Endurance Cutters in 2014, but this was not intended to extend the cutters’ service lives. Figure 9 shows the delivery dates for the Offshore Patrol Cutters and the decommissioning dates for the legacy Medium Endurance Cutters. The fiscal year 2018 through 2022 CIP shows that there is little, if any, gap between when the 210-foot and 270-foot Medium Endurance Cutters will be removed from service and when the Offshore Patrol Cutters will be operational. However, both Medium Endurance Cutter classes will be well past their end of service lives by the time they are decommissioned. For instance, in our July 2012 report, we reported that the 210-foot Medium Endurance Cutter Dependable reached its end of service life in 2006. In addition, based on the fiscal year 2018 through 2022 CIP, we found that the Coast Guard plans for the cutter to operate for an additional 23 years (until 2029) without any major sustainment work to extend its service life. While it is not unusual for the Coast Guard to operate cutters for longer than originally planned, the acquisition schedule for fielding the Offshore Patrol Cutters will result in some of the Medium Endurance Cutters being expected to operate up to 30 years beyond their original design service lives when they are removed from service. In the February 2017 Sustainability Assessment of the 210-foot Medium Endurance Cutters, the Coast Guard rated 5 of the 14 cutters as a high risk for sustainability, which reflects either a poor material condition or high maintenance costs. Additionally, the most recent material condition assessments for the Medium Endurance Cutters, which were completed in 2015, found that the: 210-foot Medium Endurance Cutters cannot be expected to meet operational requirements using the normal depot-level maintenance funding levels due to the time required to complete maintenance and the increased maintenance costs in recent years; and mission effectiveness of the 270-foot Medium Endurance Cutters will continue to degrade without a near-continuous recapitalization of older sub-systems. Further, according to the fiscal year 2018 through 2022 CIP, the Coast Guard is planning to operate some of the Medium Endurance Cutters for about the same period of time as other Medium Endurance Cutters that will undergo the SLEP project. This raises questions as to how those cutters that do not go through the SLEP will continue operating until their planned decommissioned date, which in some cases is the same time period as those cutters undergoing the SLEP. As shown in figure 9, the 210-foot Medium Endurance Cutter Alert will be decommissioned in 2030 and will not undergo a SLEP, while the 270-foot Medium Endurance Cutter Bear will also be decommissioned in 2030 and could undergo the SLEP. In July 2012, we found that as assets age beyond their design service lives, they can negatively affect the Coast Guard’s operational capacity to meet mission requirements as the cutters require more maintenance. As discussed earlier, in response to Congressional direction, as the Coast Guard continues its development of a 20-year Long-term Major Acquisitions Plan, it is important to include more details about the 270-foot Medium Endurance Cutter SLEP, including when the SLEP should begin and how much service life the SLEP should add to the cutters. Depot-Level Maintenance Costs for Aging Assets Are Increasing As legacy assets operate longer than originally planned, they are becoming costlier to maintain, which introduces risk to an already constrained Coast Guard budget. For example, depot-level maintenance expenditures from fiscal years 2010 to 2017 for the 210-foot and 270-foot Medium Endurance Cutters, Polar Star, and Inland Tenders exceeded by $460 million the assets’ estimated costs for depot-level maintenance (standard support levels—the Coast Guard’s annual estimates for depot- level maintenance) since these assets are near the end of or have exceeded their expected service lives. Specifically, over the 8-year period the: 210-foot Medium Endurance Cutters’ expenditures were about $151 million (219 percent) more, 270-foot Medium Endurance Cutters’ expenditures were $192 million (265 percent) more, Polar Star’s expenditures were about $15 million (31 percent) more, Inland Tenders expenditures were about $102 million (151 percent) more than standard support levels. The most recent material assessments for the 210-foot and 270-foot Medium Endurance Cutters, completed in September 2015 and November 2015 respectively, stated that the cutters’ current standard support level funding is not sufficient to continue funding the necessary maintenance activities. The assessments noted that there is the likelihood that maintenance will be deferred, postponed, or modified to accommodate this funding shortfall and that the cutters could degrade at an increasing rate if additional funding is not identified. According to Coast Guard guidance, once the cost to maintain or repair equipment is in excess of 50 percent of a cutter’s annual standard support level, it is considered to have zero years of remaining service life. The 210-foot and 270-foot Medium Endurance Cutters and Inland Tenders exceeded this threshold each year from 2010 to 2017 and the Polar Star exceeded this threshold in 2016. This indicates that, although the legacy cutters we reviewed continue to perform missions, the Coast Guard is accepting a significant level of risk based on the cutters’ increased depot-level maintenance expenditures, and that these cutters could experience catastrophic failures. Such an event could result in assets being removed from service without available replacement assets. Our March 2017 recommendation that the Coast Guard periodically update standard support levels to account for actual expenditures would, if implemented, begin to address this problem so that standard support levels would better align with depot-level maintenance expenditures. Conclusions The Coast Guard continues to use the annual budget process to address the affordability of its portfolio of major acquisition programs by making trade-off decisions that result in delayed acquisitions and reduced capabilities. This approach places decision makers, including those in Congress in a position of committing fiscal resources to individual programs without knowing whether they are affordable or achievable within the context of the Coast Guard’s overall portfolio. While the 5-year CIP shows the Coast Guard’s immediate trade-off decisions, it does not show how these decisions could affect other programs in the portfolio or future acquisition efforts. Not providing comprehensive information in the CIP on the acquisitions needed to perform its missions, the trade-offs necessary at different funding levels, and the impact of the trade-off decisions made, the CIP limits the information available to decision makers, including those in Congress. In addition, the Coast Guard currently is not conducting key oversight that could facilitate a balanced, affordable portfolio. While the Coast Guard has a group in place to conduct portfolio reviews as a part of the annual budget cycle in the Systems Integration Team, it does not have senior- level executive representation or responsibilities necessary for the oversight and management of the portfolio as a whole. The Executive Oversight Council is a flag/Senior Executive Service-level group that monitors major risks and provides direction to other cross-directorate teams. In the past, this council had a documented role to annually review and oversee the Coast Guard’s overall acquisition portfolio, but it never conducted these reviews. Without collective portfolio reviews at the senior management level, the Coast Guard does not have sufficient information to determine needed trade-offs between the major acquisition programs while also considering the affordability of the portfolio and budget realities. Recommendations for Executive Action We are making the following two recommendations to the Coast Guard: The Commandant of the Coast Guard should work with Congress to include in the Coast Guard’s annual 5-year CIP a discussion of the acquisition programs it prioritized that describes how trade-off decisions made could affect other acquisition programs, such as by delaying recapitalization efforts or needing to conduct Service Life Extension Projects for legacy assets. (Recommendation 1) The Commandant of the Coast Guard should require the Executive Oversight Council, in its role to facilitate a balanced and affordable acquisition portfolio, to annually review the acquisition portfolio collectively, specifically for long-term affordability. (Recommendation 2) Agency Comments and Our Evaluation We provided a draft of this report to DHS for review and comment. DHS’s written comments are reprinted in appendix IV. The Coast Guard also provided technical comments that we incorporated into the report as appropriate. In responding to a draft of our report, DHS concurred with our first recommendation and non-concurred with our second recommendation. In its response, with respect to our second recommendation DHS noted that several existing organizations within the Coast Guard—such as its Investment Board, Deputies Council, and Investment Review Board—are responsible for making decisions regarding out-year funding. Further, DHS noted that the Executive Oversight Council works outside the Planning, Programming, Budgeting, and Execution process and that the phrase “long-term affordability” is subject to interpretation in the context of our recommendation. DHS also stated that, to meet the spirit of our recommendation, the Coast Guard will update the Executive Oversight Council’s charter to require a review of the collective acquisition portfolio, specifically evaluating long-term planning. We believe that updating the Executive Oversight Council’s charter to include long-term planning is a positive step. However, long-term affordability, as discussed throughout this report, should include the budget realities faced by the Coast Guard in its major acquisition portfolio. If the planning accounts for long-term funding considerations to achieve the Coast Guard’s acquisition goals and objectives, we believe the intent of our recommendation would be met. We are sending copies of this report to the Secretary of Homeland Security and the Commandant of the Coast Guard. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-4841 or makm@gao.gov. Contact points for our Office of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix V. Appendix I: Objectives, Scope, and Methodology The objectives of this report are to assess (1) the extent to which the Coast Guard has made changes to how it manages its acquisition portfolio and (2) how the Coast Guard is sustaining existing assets until new assets become operational. To examine the extent to which the Coast Guard has changed how it manages its acquisition portfolio, we assessed Coast Guard practices for managing the portfolio’s affordability and long-term planning. We looked at the portfolio to determine how its composition changed since our last Coast Guard acquisition portfolio review in 2014. We selected a range of acquisition programs based on if they were already major acquisition programs (programs with a life-cycle cost estimate greater than or equal to $300 million or a total acquisition cost greater than or equal to $100 million) by definition, the programs were part of our 2014 review, or they are likely to be major acquisition programs that will require significant funding in the near future. Using acquisition program baselines, we also identified changes in the expected total acquisition costs of these programs using the threshold costs and compared them with what we reported in 2014. We used the threshold acquisition costs—the maximum amount the program should cost as approved by DHS—when referring to the total acquisition cost of a program. We calculated the remaining investment required for each program by taking the total acquisition cost, as reported in the program’s acquisition program baseline, and subtracting the funding reflected for the program in Explanatory Statements regarding annual appropriations acts through fiscal year 2018 appropriations. We reviewed program documentation and interviewed officials from program offices and the Coast Guard’s capabilities and engineering directorates. These discussions helped identify program achievements as well as any risks associated with realizing planned cost, schedule, and capability targets. We analyzed Coast Guard 5-year Capital Investment Plans (CIP) that supported the budget requests for fiscal years 2014 through 2018 to determine how the Coast Guard has managed the affordability of its acquisition portfolio. We also compared annual appropriations acts and accompanying explanatory materials since fiscal year 2014 with acquisition needs and capability gaps identified in the CIPs. We compared Coast Guard practices for managing the affordability of its acquisition portfolio and long-term planning with best practices outlined in GAO’s Cost Estimating and Assessment Guide and prior GAO reports. In addition, we reviewed Atlantic Area Command annual area planning assessments for fiscal years 2011 through 2016 and other Coast Guard documents highlighting shore-side infrastructure and vessel recapitalization needs. We reviewed charters for Coast Guard cross directorate groups—the Executive Oversight Council and Systems Integration Team—that help oversee Coast Guard acquisitions—to identify responsibilities and membership for these organizations, and conducted interviews with officials from these bodies to better understand their portfolio oversight activities. We also reviewed surface and aviation fleet mix studies and other strategy and planning documents. We interviewed Coast Guard officials about the anticipated content of the 20- year Long-term Major Acquisitions Plan. Additionally, we interviewed officials from the Coast Guard resources directorate; the Coast Guard’s two operational commands (Pacific Area Command and Atlantic Area Command); the Department of Homeland Security (DHS) offices for Program Accountability and Risk Management, and Program Analysis and Evaluation; and the Office of Management and Budget (OMB) to discuss Coast Guard planning and budget preparation. We visited Eastern Shipbuilding Group and the Offshore Patrol Cutter Project Resident Office in Panama City Beach, Florida, to discuss Offshore Patrol Cutter production. To examine how the Coast Guard is sustaining existing assets until new assets become operational, we selected assets to review that were at or approaching their end of design service lives—an estimated period before the asset reaches obsolescence—and if the Coast Guard was planning to conduct a Service Life Extension Project (SLEP). We collected and analyzed program documentation on asset operational availability and mission capability, sustainment needs and maintenance history, and plans for extending the service lives of selected assets. We assessed Coast Guard expenditures on depot-level maintenance—which, according to the Coast Guard, is maintenance that is beyond the capability of the crew—for fiscal years 2010 to 2017 for legacy assets, and compared them with standard support levels—annual funding estimates for depot- level maintenance—for those assets over that same time period. We interviewed Coast Guard officials from the Long Range Enforcer Product Line Office, which is responsible for sustainment of the Polar Star, the Coast Guard’s only active heavy icebreaker. We conducted site visits— based on Coast Guard’s availability of assets—to the Coast Guard’s Medium Endurance Cutter Product Line Office in Portsmouth, Virginia, and the Aviation Logistics Center in Elizabeth City, North Carolina. We also toured a 270-foot Medium Endurance Cutter in Portsmouth, Virginia, and interviewed the officers serving on the cutter at the time. Based on the nature of the information we collected, we are not making any generalizable statements from these site visits. We conducted this performance audit from March 2017 to July 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: The Coast Guard’s Major Acquisition Portfolio The Coast Guard’s major acquisition portfolio comprises 10 surface, aviation, and command and control programs. Major acquisition programs are those with life-cycle cost estimates of at least $300 million. Table 5 provides quantities and descriptions of each major acquisition program in the Coast Guard’s 2018 portfolio. Appendix III: Coast Guard Missions The Coast Guard performs 11 statutory missions, some of which align with DHS missions (such as undocumented migrant interdiction; defense readiness; and ports, waterways, and coastal security) and some of which are broader (such as search and rescue, and living marine resources). Table 6 shows select Coast Guard assets we reviewed and which of the 11 statutory missions they perform. Appendix IV: Comments from the U.S. Department of Homeland Security Appendix V: GAO Contact Staff and Acknowledgments GAO Contact Marie A. Mak, (202) 512-4841 or makm@gao.gov. Staff Acknowledgments In addition to the contact above, Rick Cederholm, Assistant Director, Peter W. Anderson, John Crawford, Kristine Hassinger, Andrew Redd, Suzanne Sterling, and Roxanna Sun all made key contributions to this report. Related GAO Products Homeland Security Acquisitions: Leveraging Programs’ Results Could Further DHS’s Progress to Improve Portfolio Management. GAO-18- 339SP. Washington, D.C.: May 17, 2018. Coast Guard Acquisitions: Status of Coast Guard’s Heavy Polar Icebreaker Acquisition. GAO-18-385R. Washington, D.C.: April 13, 2018. Coast Guard Acquisitions: Limited Strategic Planning Efforts Pose Risk for Future Acquisitions. GAO-17-747T. Washington, D.C.: July 25, 2017. Coast Guard Recapitalization: Matching Needs and Resources Continue to Strain Acquisition Efforts. GAO-17-654T. Washington, D.C.: June 7, 2017. Coast Guard Cutters: Depot Maintenance Is Affecting Operational Availability and Cost Estimates Should Reflect Actual Expenditures. GAO-17-218. Washington, D.C.: March 2, 2017. Coast Guard Aircraft: Transfer of C-27J Aircraft Is Complex and Further Fleet Purchases Should Coincide with Study Results. GAO-15-325. Washington, D.C.: March 26, 2015. Coast Guard Acquisitions: Better Information on Performance and Funding Needed to Address Shortfalls. GAO-14-450. Washington, D.C.: June 5, 2014. Coast Guard: Portfolio Management Approach Needed to Improve Major Acquisition Outcomes. GAO-12-918. Washington, D.C.: September 20, 2012. Coast Guard: Legacy Vessels’ Declining Conditions Reinforce Need for More Realistic Operational Targets. GAO-12-741. Washington, D.C.: July 31, 2012. Coast Guard: Action Needed as Approved Deepwater Program Remains Unachievable. GAO-11-743. Washington, D.C.: July 28, 2011. Coast Guard: Progress Being Made on Deepwater Project, but Risks Remain. GAO-01-564. Washington, D.C.: May 2, 2001.

Why GAO Did This Study The Coast Guard spends billions of dollars on its major acquisition programs to meet its missions. GAO's prior work has identified the Coast Guard's reliance on its annual budget process to manage its acquisition portfolio as a challenge. GAO was asked to review the recapitalization of the Coast Guard's acquisition portfolio. This report assesses, among other topics, the extent to which the Coast Guard has made changes to how it manages its acquisition portfolio. GAO assessed Coast Guard's major acquisition programs to determine changes since GAO's 2014 portfolio review. GAO analyzed program baselines and interviewed Coast Guard officials. GAO analyzed the CIP for fiscal years 2014 through 2018, and reviewed the EOC's documentation. What GAO Found The Coast Guard, a component within the Department of Homeland Security (DHS), continues to manage its acquisitions through its annual budget process and the 5-year Capital Investment Plan (CIP)—congressionally mandated and used for oversight. This management approach creates constant churn as program baselines must continually re-align with budget realities instead of budgets being formulated to support program baselines. Further, Coast Guard officials said the CIP reflects the highest priorities of the department—such as the Offshore Patrol Cutter, which is the Coast Guard's highest priority—and that trade-off decisions are made as part of the annual budget process. However, the effects of these decisions, such as which acquisitions would take on more risk so others can be prioritized and adequately funded, are not communicated in the CIP to key decision makers, because including such information is not statutorily required. Over the years, this approach has left the Coast Guard with a build up—or bow wave—of near-term unfunded acquisitions, negatively affecting recapitalization efforts and limiting the effectiveness of long-term planning. Including the effects of these trade-offs in the CIP would align with GAO's cost estimating best practices. Until it does so, the Coast Guard limits its ability to manage its acquisition portfolio in the long-term, beyond the time covered in the 5-year CIP. In response to a September 2012 GAO recommendation, the Coast Guard updated the Executive Oversight Council's (EOC)—a cross-directorate group that oversees major acquisition programs—charter in 2014 to require annual reviews of the acquisition portfolio collectively. However, EOC officials said that these annual reviews never occurred, and GAO found that the annual review requirement was removed from the charter in 2017. Thus, the Coast Guard is without a senior-level group charged to collectively review and ensure affordability of its acquisition portfolio. The Office of Management and Budget's Capital Programming Guide states that a senior-level executive committee should be responsible for reviewing the agency's entire asset portfolio and for making decisions on the proper composition of assets needed to achieve strategic goals within budget constraints. What GAO Recommends GAO recommends that the annual CIPs reflect acquisition trade-off decisions and their effects, and that the EOC review the overall acquisition portfolio and its affordability annually. DHS concurred with the CIP recommendation. DHS did not concur with the EOC recommendation. It noted that other existing Coast Guard bodies are responsible for evaluating and prioritizing funding. However, DHS stated that the EOC charter will be updated to require it to review the overall acquisition portfolio, including long-term planning. If this long-term planning accounts for budget realities for the acquisition portfolio, GAO believes the intent of the recommendation will be met.

gao_GAO-19-209

gao_GAO-19-209_0

Background DOD acquires new weapons for its warfighters through a management process known as the defense acquisition process. This process has multiple phases, including: (1) technology maturation and risk reduction, (2) engineering and manufacturing development, and (3) production and deployment. In this report we refer to these three phases as concept development, system development, and production. Programs typically complete a series of milestone reviews and other key decision points that authorize entry into a new acquisition phase. DOD Instruction 5000.02 delegates responsibility for developing and procuring weapon systems to the military departments and other defense agencies. This policy does not specify a standard organizational structure—or program structure—to manage acquisition programs, but rather states that programs are to be tailored as much as possible to the characteristics of the product being acquired, and to the totality of circumstances associated with the program including operational urgency and risk factors. In addition, DOD’s guidance for managing its workforce states that the approach should be flexible, adaptive to program changes, and responsive to new management strategies. DOD decides how many personnel and how much program funding to request for each military department through the Planning, Programming, Budgeting, and Execution (PPBE) process. DOD programming policy requires the military departments and defense agencies to develop a program objective memorandum that identifies and prioritizes requirements and total funding needs for the current budget year and 4 additional years into the future. As a part of this process, the departments also estimate the personnel requirements and program funding needed to execute their mission, including support for the commands and PEOs that are responsible for managing acquisition programs. The results of the PPBE process, including proposed funding levels for programs, are captured in the President’s annual budget request to Congress. For example, in its budget request, DOD identifies and requests the total number of civilian full-time equivalent personnel, among other things. Congress then authorizes and appropriates the funding to pay for civilian personnel for each military department. When budgeting for contracted services, DOD estimates the cost of the tasks to be performed but not the number of individuals that may perform those tasks. The military departments, commands and PEOs then distribute approved funding (which, in part, is used to pay for civilian personnel and contractor support) to the various organizations including the programs that are responsible for managing and supporting defense acquisitions. Each military department has a different approach to developing its budget request, and program budgets may be spread across multiple types of appropriations that are organized into various categories based on their purpose such as research, development, testing and evaluation, or procurement. Similarly, the military departments fund their personnel through several different types of appropriations, including (1) operation and maintenance; (2) military personnel; and (3) research, development, test, and evaluation. Requests for funding are included in different documents and often presented in multiple volumes that can be hundreds of pages long. DOD’s Financial Management Regulation provides instructions for the formulation and presentation of the budget request to Congress, including general categories of costs that might be included in program specific budgets. In addition, the regulation requires DOD components to include specific budget exhibits for certain acquisition programs to provide more insight into those programs’ funding needs. Several Factors Affect the Workforce Size, Composition, and Mix, As Well As the Organizational Structure of Selected DOD Acquisition Programs Several interrelated factors influenced the workforce size, composition, and mix, as well as the organizational structure of the 11 major defense acquisition programs we reviewed. We found the following: Program workforce size and composition were influenced by the degree to which the program assumed responsibility for technical development and integration, as well as the program’s stage within the acquisition life cycle. Program workforce mix varied depending on the use of contractor personnel, which was based on the workload requirements and the availability of government personnel to provide the skills needed. Programs were generally structured as either stand-alone—new, high priority, complex weapon system platforms with dedicated personnel—or as part of a portfolio of related programs to share personnel across programs. Factors Affecting Selected Programs’ Workforce Size and Composition The number and composition of personnel that supported the selected major defense acquisition programs varied considerably. As shown in figure 1, the total number of personnel supporting the 11 selected programs ranged from 30 to 397, and the composition of those personnel varied based on the needs of the program. While program officials cited a number of factors that influenced the selected programs’ workforce size and composition, including department priority and complexity, we identified two overarching factors—(1) the level of program responsibility for technical development and integration, and (2) the stage of the acquisition life cycle. First, we found programs that assume more responsibility for technical development and integration have more personnel—primarily those that perform engineering as well as test and evaluation functions. The two largest of the selected programs we reviewed, the Navy’s Next- Generation Jammer Mid-Band (NGJ Mid-Band) and Columbia Class Ballistic Missile Submarine (Columbia), assumed significant responsibility for system development and integration, activities a prime contractor often undertook for the other programs we reviewed. For example, NGJ Mid-Band officials explained that the program is responsible for overseeing software integration and other efforts directly. In this case, in addition to personnel assigned to the program office, the Navy relies on personnel from other organizations such as the Naval Air Warfare Center Aircraft Division instead of a prime contractor to develop the software needed to operate the system, conduct system testing, and manage integration into the platform. Similarly, the Columbia program maintains responsibility for many aspects of development and integration of the submarine including most hull, mechanical, and electrical components. As a result, about two-thirds of the 309 personnel supporting the program are performing engineering and technical tasks. In contrast, two programs with fewer personnel, the Air Force’s B-2 Defensive Management System Modernization program (DMS-M) and Navy’s John Lewis Class Fleet Replenishment Oiler (T-AO), assigned significant responsibility for development and integration to their respective prime contractors. The Defensive Management System Modernization program reported to us that it has a total of 11 engineering and technical personnel, and T-AO reported that it has 35 engineering and technical personnel. Secondly, we found that program workforce size and composition changed in response to the amount and nature of the work programs perform at different stages of the acquisition life cycle. For example, officials from our selected programs stated they generally planned to increase in size as they progressed from concept development to system development and also planned to concurrently increase the proportion of engineering and technical personnel. Program officials stated that as the program progresses into the logistics support stage, the number of personnel supporting the program generally decreases as programs release some personnel to other assignments while retaining enough personnel to manage the logistics support stage. Figure 2 shows how the size and composition of Army’s Joint-Air-to-Ground Missile (JAGM) program changed from concept development into production. A program’s total development and procurement cost was not necessarily related to the number of personnel supporting the program for the 11 programs we reviewed. All 11 selected programs are classified as major defense acquisition programs and ranged in total acquisitions cost from $1.5 billion to $103.2 billion. Our analysis, shown in table 1 below, indicates that total cost did not significantly influence the number of personnel supporting these programs. Selected Programs Used Contractor Support to Help Meet Workload Requirements All 11 selected programs used contractors to help meet workload requirements, but the level of contractor support varied from approximately 5 percent to 72 percent of total program personnel, as shown in figure 3. Program officials told us that while they generally try to use civilian or military personnel to meet workload requirements, they use contractor support when the number of government personnel allocated to the program is not sufficient to meet their needs, the technical skills are not available or are limited within the government, or to fulfill short-term tasks that are too brief to justify hiring government personnel. Program officials stated the extent to which their programs use contractor support often depends on the number civilians allocated to the program by the command or PEO. In the case of the three selected programs with the fewest personnel, the officials stated that the number of personnel authorizations allocated to the program by their respective command or PEO did not meet their estimated workload requirements. For example, the B-2 Defensive Management System Modernization program estimated it needed 82 personnel in fiscal year 2018, but was only allocated 13 personnel. As a result, program officials stated that they used program funds to pay for contractor support personnel to partially offset the government civilian staffing shortfalls. Officials at the Air Force Life Cycle Management Center, the organization that allocated personnel to the B-2 program office, told us that civilian personnel are allocated based on the risk associated with each program. Program officials told us that contractor support personnel are used to augment civilian and military personnel by providing skills or technical expertise that are limited or not available in the government. We found that over two-thirds of the contractors that supported the 11 selected programs we reviewed were performing engineering and technical functions. For example, the John Lewis Class Fleet Replenishment Oiler (T-AO) is a commercially-derived ship design. As such, program officials stated that the required engineering expertise resides in the commercial sector, which resulted in contracted engineers comprising about 77 percent of the program’s total engineering personnel. Program officials also stated that it is more effective to use contractor support personnel to perform tasks that are relatively short in duration than to go through the lengthy process of hiring government personnel. Contracting for support allows the program to grow and shrink to meet personnel requirements as they change. For example, Joint Air-to-Ground Missile program officials stated they contracted for support to execute tasks that are not recurring, such as developing the required documents to get approval to start production. Among the 11 programs we reviewed, the Air Force’s Military Global Positioning System User Equipment (MGUE) program has a unique workforce mix. Twenty-four percent of MGUE’s program personnel were military, and MGUE was the only one of the 11 selected programs that had FFRDC personnel. Program officials stated that the challenge of obtaining civilian personnel with the required technical skills in a high cost-of-living area around Los Angeles, California required the program to rely more heavily on military personnel and contractors to support the program. Program officials stated this is in part because it is easier to assign military personnel in high cost-of-living areas than it is to hire civilian personnel. In addition, programs in the Air Force’s Space and Missile Systems Center often rely on FFRDC personnel from Aerospace Corporation, which is located in the Los Angeles area and provides technical expertise that is specific to space systems. Program officials from the other 10 programs we reviewed reported that they did not have FFRDC personnel. Military Departments Structured Selected Acquisition Programs to Leverage Available Personnel with the Necessary Skills While differences existed in the organizational structure of the 11 programs we reviewed, we identified factors that affected which of the two common approaches the military departments used to leverage available personnel with the necessary skills: New, high priority, complex weapon system platforms that require a significant amount of development and integration, such as the Navy’s Columbia and the Army’s Armored Multi-Purpose Vehicle, are structured as distinct standalone program offices with dedicated program personnel. Nine of the 11 selected programs were managed in a portfolio-based program structure which included multiple related acquisition programs. For these portfolio-based programs, personnel were shared across the related programs to help meet fluctuating workload requirements and maximize personnel resources. Figure 4 compares the structure of a standalone program to the structure of a portfolio-based program with multiple acquisition programs managed under it. The figure also illustrates how the Air Force’s MGUE program was situated within the Air Force’s Global Positioning Systems portfolio of programs. In both types of organizational structures illustrated above, the PEO and the program office have personnel that oversee and support the programs. These personnel may be dedicated to one program or may split time between multiple portfolio-based programs. For example, the Air Force PEO for Space has more than 5,000 military, civilian, and contractor personnel and is responsible for managing 41 programs, the responsibility for which is distributed among multiple program offices. One of these program offices, the Global Positioning Systems program office, has 628 personnel. This program office is responsible for overseeing and supplementing the staff of several programs, including the Military Global Positioning System User Equipment Program, which has about 70 personnel. According to PEO and program officials, acquisition programs may be managed within portfolios for several different reasons: Programs are part of the same weapon system platform. The B-2 Defensive Management System Modernization program and the F-15 Eagle Passive Active Warning Survivability System program are examples of upgrades to existing systems on mature aircraft and are managed within a portfolio of programs within the B-2 and F-15 system program offices, respectively. Programs have interrelated technologies. The Air Force’s MGUE program is managed within the GPS program office, which also manages other GPS satellite and ground system programs. Programs have related acquisition strategies. The Navy’s John Lewis Class Fleet Replenishment Oiler (T-AO) program is managed within a portfolio of commercially designed and developed ships. This program is managed within a program office that oversees approximately 85 types of commercially derived auxiliary ships, boats, service craft, and special mission ships. Regardless of how the acquisition program is structured, other DOD organizations also provide personnel to support a program’s workload requirements. There are various specialized DOD organizations that support programs and provide specific acquisition functions or skill sets, such as contracting, cost estimating, and engineering. For the 11 selected programs we reviewed, these organizations supported multiple programs and were either structured (1) within the PEO that was responsible for the programs we reviewed or (2) external to the PEO. These external support organizations include contracting commands, warfare centers, and engineering organizations that are intended to provide the program specialized technical expertise from across the military department. Program officials stated that these organizations may share personnel with a program on a full or part-time basis, and the shared personnel may or may not be co-located with the program. Figure 5 is a notional representation of the way that programs are supported by different organizations. The major defense acquisition programs we reviewed used different approaches to organizing and leveraging support organizations. For example: The Navy programs we reviewed relied on naval warfare centers to provide the engineering expertise necessary to design, build, maintain, and repair the Navy’s aircraft, ships, and submarines. For example, the Navy’s NGJ Mid-Band relies heavily on warfare centers, including the Naval Air Warfare Center Weapons Division and the Naval Air Warfare Center Aircraft Division, to support the program. We found that about 60 percent of the total number of personnel supporting the program office were from these organizations. The Army programs we reviewed relied on support organizations such as the Army Contracting Command for contracting functions, the Aviation and Missile Research Development and Engineering Center for engineering expertise, and others to provide life cycle management support. The Air Force programs we reviewed relied on support organizations established within their command. For example, Air Force’s Life Cycle Management Center has organizations dedicated to supporting all of its programs. These organizations provide support, such as contracting and cost estimating expertise, to programs managed under the Air Force’s Life Cycle Management Center. Personnel within these organizations are not staffed to one particular program, but share their time among many of the programs the Center is responsible for managing. Personnel Costs for Selected DOD Acquisition Programs Are Included in Multiple Parts of the Budget Justification Documents and Are Not Specifically Identified for Individual Programs The personnel costs for each major defense acquisition program we reviewed are included in different parts of the President’s annual budget request, including budget justification documents, but are not always clearly identifiable due to different approaches used to report such costs. The DOD Financial Management Regulation gives the military departments flexibility in how they submit program personnel costs. For example, it suggests the use of “typical” personnel cost categories for research, development, test, and evaluation programs to include in their individual program budget exhibits, but it also allows the departments to use the personnel cost categories they deem to be the most appropriate when formulating the budget request. In reviewing DOD’s budget requests for fiscal years 2018 and 2019 associated with the 11 selected programs, we found that personnel costs are budgeted for in two main wayscentrally by the military department, or by an individual programdepending on whether the requests are for military, civilian, or contractor support services. Personnel costs that are program-funded are included in individual program budget justification requests, whereas personnel costs that are centrally funded by the military departments are aggregated into one or more line items in the military department’s specific appropriation request. Table 2 shows how each military department funds military and civilian personnel and contractor support services for major defense acquisition programs. Each military department centrally budgets for military personnel through its respective Military Personnel appropriation requests, which aggregate personnel funding. These requests include funding for pay, travel, and other personnel-related costs. As these costs are combined and not associated with a specific program, we could not determine the costs of the military personnel supporting the 11 selected programs by reviewing DOD’s budget justification documentation. In contrast, support contractor costs were included in each program’s individual budget request. The military departments also centrally budget for some civilian personnel, but there are differences between the departments regarding which appropriations categories they use to request these funds. Regardless of the appropriation, we found that the budget requests do not identify civilian personnel costs by specific program; therefore, we could not determine the costs of the centrally funded civilian personnel supporting the 11 programs we selected. For example, in fiscal year 2019, the Air Force requested funding for the civilian personnel supporting its acquisition programs in development through the Research, Development, Test, and Evaluation appropriation. It grouped the costs into eight categories that represent various missions such as Cyber, Network, and Business Systems; Global Battle Management; and Nuclear Systems. The Air Force budget request indicates the total amount of funds requested, but does not identify the estimated number of personnel that these funds will support. Figure 6 illustrates how the Air Force requested funds for its civilian acquisition workforce in fiscal year 2019. The Navy and Army request funds for civilian personnel primarily through their respective operation and maintenance appropriations. This appropriation is used to fund a wide range of costs necessary to manage, operate and maintain worldwide facilities and military operations. These operation and maintenance budgets are divided into numerous categories related to various missions, functions, or activities. For example, the Navy’s Operation and Maintenance budget requests funding for civilian personnel in several categories, such as “Ship Operational Support and Training” and “Administration.” The Army Operation and Maintenance budget requests funding for civilian acquisition personnel in one combined category labeled as “Other Service Support.” Apart from the portions of the budget described above, certain DOD programs have specific budget exhibits that identify its funding requirements. In reviewing the exhibits for the 11 selected programs, we found that individual program requests include personnel costs that are not funded centrally such as contractor support services costs, but these costs are generally not specifically identified as personnel costs. For example, according to program officials, the Air Force’s B-2 Defensive Management Modernization program requested funds in its exhibit accompanying the fiscal year 2019 Research Development, Test, and Evaluation budget request labeled “PMA,” which stands for Program Management Administration. According to program officials, PMA includes costs for contractor support services, government travel, and other costs but does not include civilian personnel costs (see figure 7). In reviewing and discussing the budget exhibits for the 11 selected programs with program officials, we found that personnel costs, including civilian, contractor, and FFRDC, were generally spread across multiple budget request lines that were associated with various tasks but were not specifically identified as personnel costs. These include the following: Development Test & Evaluation For example, the Navy’s Joint Precision Approach and Landing System’s fiscal year 2019 Research Development, Test and Evaluation budget exhibit included personnel costs across seven lines that represented various efforts including ship integration, test and evaluation, systems engineering, and program management support, as shown in figure 8. Of the 11 program’s fiscal year 2019 budgets we reviewed, one identified personnel costs on a single line, and the remaining 10 programs included personnel costs in two or more budget lines. Agency Comments We provided a draft of this report to DOD for comment. DOD provided technical comments that we incorporated into this report as appropriate. We are sending copies of this report to the appropriate congressional committees; the Acting Secretary of Defense and the Secretaries of the Army, Navy, and Air Force, as well as the Under Secretary of Defense for Personnel and Readiness. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-4841 or dinapolit@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix II. Appendix I: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Justin Jaynes (Assistant Director); Bradley Terry (Analyst-in-Charge); Matthew T. Crosby; Stephanie Gustafson; Heather B. Miller; Karen Richey; Miranda Riemer; Robin Wilson; and Chris Zakroff made significant contributions to this review.

Why GAO Did This Study In 2018, DOD estimated that its 82 major defense acquisition programs would cost over $1.69 trillion in total to acquire. DOD relies on program offices—composed of civilian, military and contractor support personnel—to manage and oversee these technically complex programs. GAO was asked to review factors affecting DOD's personnel needs for its acquisition programs and how DOD budgets for the costs associated with these personnel. This report describes (1) factors affecting the workforce size, composition, and mix of contractor and government personnel, as well as organizational structure for selected programs; and (2) how personnel costs associated with those selected programs are included in DOD's budget justification documents. GAO reviewed DOD acquisition, workforce, and financial management policies and regulations and identified a non-generalizable sample of 11 major defense acquisition programs, including programs from each military department that were recently approved to enter into system development. GAO requested information from each of these programs to identify the number and type of personnel supporting the program, reviewed program documentation, and interviewed program officials. GAO also reviewed DOD's budget justification documents for fiscal years 2018 and 2019. What GAO Found The workforce size, composition, and mix, as well as the organizational structure of the 11 Department of Defense (DOD) major defense acquisition programs GAO reviewed were influenced by several interrelated factors. These factors include the government's role in developing and integrating key technologies, the availability of government personnel to provide the skills needed, and whether the program was managed as part of a portfolio of related programs or as a stand-alone program. For example, programs that assumed more responsibility for developing and integrating key technologies generally had a larger workforce, which was primarily composed of engineering and technical personnel. Program officials GAO met with stated that they generally prefer to use government personnel, but use contractor support when the number of government personnel allocated to the program is not sufficient to meet their needs, the technical skills are not available or are limited within the government, or to fulfill short-term tasks that are too brief to justify hiring government personnel. GAO also found that DOD structured the 11 programs to allow them to leverage available personnel with the necessary skills. Two programs were structured as standalone programs because they were new, high priority, and complex. The other nine programs were managed as a part of a portfolio of related programs. For example, the Air Force's F-15 program office manages a number of programs that add capabilities to the existing system. DOD's Financial Management Regulation, which governs the formulation and presentation of DOD's budget request, gives DOD flexibility in how it submits program personnel costs. Consequently, the personnel costs for the 11 programs GAO reviewed were not separately and distinctly identified from other costs. For example, costs for civilian and military personnel are often centrally funded through appropriations categories that support many DOD activities and do not provide information on specific program personnel costs. GAO also found that costs for contractor support are often combined with other costs in individual program budget exhibits.

gao_GAO-18-682

gao_GAO-18-682_0

Background The federal government has recognized 573 Indian tribes as distinct, independent political communities with inherent powers of a limited sovereignty, which has never been extinguished. These tribes can vary significantly in regard to tribal size, population, and ownership status of land. For instance, some tribal lands include reservations—land set aside by treaty, federal law, or executive order for the residence or use of an Indian tribe. Some tribal lands include parcels with different ownership; parcels may be held in trust by the federal government for the benefit of a tribe or an individual tribal citizen. Trust and restricted lands can affect a tribe’s ability to use their land as collateral to obtain a loan. In addition, the size of a tribe’s land base can range from less than one square mile to more than 24,000 square miles (the size of West Virginia). Some tribes are located in extremely remote, rural locations and others are located in urban areas. The term “broadband” commonly refers to Internet access that is high speed and provides an “always-on” connection, so users do not have to reestablish a connection each time they access the Internet. Broadband providers deploy and maintain infrastructure to connect consumers to the Internet and provide Internet service through a number of technologies. Broadband infrastructure may include burying fiber-optic or copper cables, stringing cable on existing poles, or erecting towers for wireless microwave links, which relay wireless Internet connections from tower to tower. Figure 1 illustrates some of the options for broadband deployment infrastructure. To install this infrastructure, providers must obtain permits from government entities with jurisdiction over the land or permission from public utilities to deploy infrastructure on existing utility poles. The federal government has emphasized the importance of ensuring Americans have access to broadband, and a number of agencies, including FCC and RUS, currently provide funding to subsidize broadband deployment in areas in which the return on investment has not attracted private investment. FCC funds a number of programs through the Universal Service Fund, which may increase broadband deployment on tribal lands. One program is the high-cost program (renamed the Connect America Fund (CAF) in 2011), which provides financial support to both wireline and wireless telecommunications carriers that provide telecommunications services (referred to as providers in this report) to supplement their operating costs to serve consumers in rural or remote areas, where the cost of providing service is high. From 2010 to 2017, a total of $34.5 billion in annual and standalone Universal Service Fund high-cost support was disbursed to providers, as follows: In total, the high-cost program and Connect America Fund provided $34.1 billion from 2010 to 2017 in financial support to providers, consisting of annual disbursements between $3.7 and $5 billion. The Mobility Fund Phase I provided $300 million in 2012 in one-time support to providers to expand broadband service in areas where service was not available, including tribal lands. The Tribal Mobility Fund Phase I provided $49.8 million in 2014 in one-time support to providers to deploy broadband service to unserved tribal lands. To be eligible to receive for Universal Service Fund program support, a provider must be designated an eligible telecommunications carrier (ETC) by the appropriate state or by FCC. Under FCC rules, which many state programs mirror, ETCs must meet certain service obligations as described below: provide a 5-year plan showing how program support will be used to improve its coverage, service quality, or capacity in each service area where it seeks designation; demonstrate its ability to remain functional in emergency situations; demonstrate that it will satisfy consumer protection and service quality offer local usage plans comparable to those offered by the incumbent carrier in the areas for which it seeks designation; and acknowledge that it may be required to provide equal access to other providers within the service area if all other ETCs in the designated service area relinquish their designations. In addition to FCC’s funding, RUS has a current program and had a prior program and NTIA had a prior program that provided funding to improve broadband service in unserved or underserved areas. The RUS and NTIA prior programs were authorized by the American Recovery and Reinvestment Act of 2009 (Recovery Act) to expand high-speed Internet service in unserved areas, and there is no current funding for these programs. RUS’s Community Connect program currently provides grants to rural communities to provide high-speed Internet service to unserved areas. The Community Connect program is significantly smaller than FCC’s programs, with $95.2 million awarded to 36 recipients from 2010 to 2017. The purpose of the RUS Community Connect program is to provide financial assistance to eligible applicants that will provide broadband service that fosters economic growth and delivers enhanced educational, healthcare, and public-safety benefits. In addition, RUS previously administered the Broadband Initiatives Program (BIP), authorized by the Recovery Act to expand high-speed Internet service in unserved areas. BIP funding included $2.2 billion dedicated to deploy broadband infrastructure. Through the program, RUS funded a total of 247 infrastructure projects with the requirement that all projects be fully completed by June 30, 2015. In addition to the infrastructure awards, 12 technical assistance grants went to tribal communities to develop regional plans to provide broadband service in rural areas that remain critically unserved. NTIA administered a prior program also authorized by the Recovery Act called the Broadband Technology Opportunities Program (BTOP). NTIA made available $3.1 billion in BTOP funding to deploy broadband infrastructure. Through the program, NTIA awarded a total of 116 infrastructure grants with the requirement that all projects be fully complete within 5 years of the award date. Selected Tribes Partnered with Various Entities to Increase Broadband Deployment and Outcomes Varied Although we identified some partnership arrangements between tribes and other entities to increase broadband deployment on tribal lands using prior authorized funding, based on our review, these arrangements are not being used under currently available programs. As previously noted, there are greater costs associated with deploying broadband on unserved tribal lands because the unserved areas are generally rural, with possibly rugged terrain, and have low population densities. Because of these greater costs, there may be little to no private sector incentive to deploy broadband or enter into a partnership arrangement to do so. During our review, we did not find any partnership arrangements that leveraged currently available federal funding from FCC’s CAF or RUS’s Community Connect Program. The seven partnership examples we identified were ones that obtained federal funding under past programs, namely BIP and BTOP that were funded by the Recovery Act. Among these examples, tribes partnered with several different types of entities that were eligible to receive federal grants, including (1) private providers; (2) a community access network provider; (3) an electric cooperative; (4) a regional consortium; and (5) tribally owned telecommunications companies (which we will refer to as tribally owned providers). These types of arrangements are explained below. Outcomes of these partnership arrangements varied, as reported by tribal officials and other stakeholders we interviewed, but these stakeholders did not always agree on the outcomes. Private Providers Private providers can partner with a tribe to deploy broadband infrastructure on tribal lands. We found two instances in which a tribe partnered with private providers to improve broadband service. Pine Telephone Company and Choctaw Nation. With the land the tribe has jurisdiction over covering over 10 counties across 12,000 square miles in Oklahoma, the Choctaw Nation’s lands encompasses about 15 percent of the State of Oklahoma’s total area—an area larger than the entire state of Maryland. According to the Oklahoma Department of Transportation, the Choctaw Nation is the largest employer in the southeastern Oklahoma region and its businesses are key contributors to the state’s economy. However, a tribal official told us that the tribal government has struggled to meet the tribe’s broadband needs. According to the Choctaw Nation official, Pine Telephone Company (Pine), a privately owned company, has a history of partnering with the Choctaw Nation. In 2010, Pine received a BTOP grant of $9.5 million to deploy broadband infrastructure to underserved areas of Southeastern Oklahoma, including Choctaw Nation lands. According to tribal officials and representatives from Pine, the partnership enabled tribal government agencies and buildings— including public schools, public safety agencies, fire and police departments, and a health clinic—to get broadband service. The partnership also improved broadband service for the Choctaw Nation. Additionally, Pine had been proactive in partnering with the Choctaw Nation to secure federal grants and assist with land use and rights-of- way issues, according to a tribal official. Pine representatives told us that partnering with the Choctaw Nation had been beneficial based on their common interest to increase broadband service to the area. Inland Cellular, First Step, and Nez Perce. The Nez Perce Tribe’s reservation consists of 750,000 acres located in north central Idaho. Tribal officials told us that the terrain on the reservation makes broadband deployment challenging because it has very large hills and deep valleys; additionally the reservation is sparsely populated. Tribal officials told us that prior to 2010, there was no broadband service available on the Nez Perce reservation. In 2010, the tribe received a BTOP grant of $1.6 million for the Nez Perce Broadband Enhancement Project; the project was completed in 2013. The tribe used that federal grant to deploy 216 miles of broadband (wireless) infrastructure across its reservation to provide broadband service in four northern Idaho counties. As part of the project, the tribe partnered with two private providers, Inland Cellular and First Step, to expand broadband service on the reservation. The tribe used BTOP funding for infrastructure buildout in areas in need of connectivity, while Inland Cellular and First Step focused their efforts on infrastructure buildout in more populated areas. According to Nez Perce and Inland Cellular officials, the partnership resulted in broadband service being provided to previously underserved rural communities and 17 community institutions, including schools and public safety organizations. Because the partners each own towers on the reservation, the officials told us they could collocate equipment on each other’s towers, an approach that resulted in more reliable service. Further, Nez Perce officials and Inland Cellular representatives told us that their partnership was complementary, in that Inland Cellular offered voice services and the tribe’s enterprise offered data services. Community Access Network Provider Community access network providers are typically owned and operated by public entities rather than by a private corporation. All profits are reinvested to operate, maintain, and expand the community network. Community access networks focus on building broadband infrastructure that allows multiple Internet service providers to offer their services to customers. For example, rather than having one choice for Internet service, community access network providers will allow several service providers to compete for customers. Northwest Open Access Network (NoaNet). NoaNet, a utility network that offers communities access to broadband infrastructure, has deployed infrastructure in rural areas of Washington State, including on tribal lands. NoaNet received two BTOP grants in 2010—one grant for $84 million and the other for $54 million—to enhance existing infrastructure and improve broadband service in unserved areas. NoaNet representatives told us that over the course of several years, NoaNet deployed 2,300 miles of fiber-optic cable across tribal lands in Washington State and partnered with several Indian tribes and nations, including the Kalispel Indian Community of the Kalispel Reservation, Lower Elwha Tribal Community, and Yakama Nation, to deploy broadband infrastructure. For example, NoaNet representatives told us they partnered with Yakama Nation and exchanged a NoaNet-owned asset for access to a power source and the right to install fiber-optic lines on Yakama tribal land. According to NoaNet representatives, NoaNet’s infrastructure buildout improved broadband services and created new economic development opportunities for several tribes in Washington State. For example, they said NoaNet collaborated with Yakama Nation Networks—a wireless network and tribal enterprise serving the tribe— to provide faster broadband service to the reservation. Further, the NoaNet representatives said the availability of broadband service created new technical jobs with professional growth opportunities on the reservation. NoaNet representatives added that NoaNet enabled high-speed Internet service to the Makah Tribe’s health clinic, government offices, school, and library, where they previously had no Internet service at all. Moreover, they told us that partnerships are beneficial in helping tribes gain telecommunications experience. Similarly, according to a tribal representative from Jamestown S’Klallam Tribe, NoaNet’s infrastructure buildout helped the tribe obtain broadband services for its library and also helped create economic opportunities for the tribe. Electric Cooperative Rural electric cooperative networks typically serve areas that have low population density where traditional providers do not want to serve because of limited opportunities for financial return on investment. Kit Carson Electric Cooperative and Taos Pueblo. Kit Carson Electric Cooperative (KCEC) is a member-owned, nonprofit electric distribution cooperative that operates a fiber-optic broadband network. In 2010, KCEC received $64 million in grant funding from RUS’s BIP to create a 2,400-mile broadband network in northern New Mexico and provide broadband service to businesses and homes, including those on the Taos Pueblo and Picuris Pueblo. In an August 2016 presentation to the New Mexico state legislature, KCEC stated that that the project connected tribal members and community institutions, created job opportunities, and improved public safety by improving emergency communications services. According to Taos Pueblo officials, the impetus to work with KCEC was to improve broadband service to meet immediate economic, education, health service, and public safety needs of the tribe. However, based on our meetings with both KCEC representatives and Taos Pueblo officials they have different perspectives about the success of this partnership at delivering broadband service to the tribe. For example, KCEC representatives told us that the cooperative constructed the fiber-optic network and connected the government buildings and homes of Taos Pueblo and Picuris Pueblo members as promised, and that KCEC has responded to service interruptions when they occurred on tribal lands. On the other hand Taos Pueblo officials told us, that KCEC did not deploy broadband infrastructure to enable service to all homes and buildings on tribal lands as the tribe had expected. Similarly, KCEC representatives told us that they worked regularly with the Taos and Picuris tribal governments and had good relationships with them; they noted that they meet with tribal leadership every quarter to maintain effective communications and address any issues. In contrast, according to Taos Pueblo officials, KCEC did not solicit tribal input when building out the fiber-optic network, and only met with Taos Pueblo officials about once a year and did not follow up on the issues the tribe raised. Further, according to KCEC representatives, in its federal funding application, KCEC made a commitment that the Taos and Picuris tribal lands would be the first areas targeted for building out the network, and the representatives said that KCEC completed 100 percent of the construction and connected the tribal governments as promised. Taos Pueblo officials, however, said that their tribe was the last to receive service and that KCEC did not complete the broadband construction, including service to the homes of some of its members, because KCEC exhausted its BIP funding. Regional Consortium We have previously reported that regional consortium, which are typically formed by groups to undertake an enterprise beyond the resources of any one member, can sponsor regional networks that focus on building broadband networks and providing broadband services to schools, medical providers, public safety agencies, and other community institutions. North Central New Mexico Economic Development District. Located in northern New Mexico, the Pueblo of Pojoaque, Santa Clara Pueblo, Tesuque Pueblo, and Ohkay Owingeh partnered with local governments to establish the North Central New Mexico Economic Development District (the District), a regional consortium, to address the socio-economic needs of its members. In 2008, regional planners and government officials identified broadband as the region’s number- one infrastructure priority because rural north central New Mexico relied significantly on dial-up Internet service and lacked affordable service to small businesses, libraries, schools, and other community institutions. In 2010, the District received a BTOP grant of $10.6 million to build a community-owned broadband network, known as REDI Net. The District sought the BTOP grant to improve rural healthcare services, make public and higher education more accessible, and improve local government services, like public safety. REDI Net’s construction included upgrading existing infrastructure and deploying 136 miles of new fiber-optic cable across the region and on pueblo lands to replace low-performing dial-up service with faster, more affordable broadband service. According to the project’s progress report submitted to NTIA, the partnership enabled broadband infrastructure to be deployed across the four participating lands and connect 110 community institutions. The project’s description stated that REDI Net was being used to deliver telemedicine services, distance-learning applications, and critical communications for emergency first-responders. According to REDI Net representatives, in 2017, REDI Net became a standalone organization, separate from the District, and currently charges a monthly fee for the pueblos to use the broadband network. A REDI Net representative told us that the biggest outcome of the partnership has been the improved relationships and collaboration among the Pueblo of Pojoaque, Santa Clara Pueblo, Tesuque Pueblo, and Ohkay Owingeh and other local municipalities. Tribally Owned Providers Some tribes have created their own telecommunications companies to provide broadband access to their communities. Based on the examples we identified, a tribe may create its own telecommunications or broadband company or a tribe may partner with an existing tribal enterprise such as an electrical utility to provide broadband services. Navajo Nation and Navajo Tribal Utility Authority. The Navajo Nation—which spans across Arizona, New Mexico, and Utah— partners with a tribally owned entity, the Navajo Tribal Utility Authority (NTUA), to provide broadband service to residents and households. According to a NTUA representative, the Navajo Nation has diverse, challenging terrain—which includes canyons, valleys, timber forest, desert, and mountains—making it difficult to provide broadband service to tribal residents. In 2010, NTUA received a BTOP grant of $32 million to deploy broadband infrastructure covering 15,000 square miles across the three states. According to the project’s progress report submitted to NTIA, by 2013, NTUA leveraged BTOP funding to deploy 570 miles of fiber-optic cable and 775 miles of wireless infrastructure resulting in a total of 1,345 new network miles. According to NTUA representatives, the partnership between the Navajo Nation and NTUA increased broadband deployment on the nation and created new opportunities for NTUA to partner with other private providers to further expand broadband services. For example, NTUA representatives said NTUA partnered with a private broadband provider, Commnet, to deploy wireless broadband infrastructure that enabled tribal citizens to receive 4G LTE service. NTUA and Commnet representatives told us NTUA’s relationship with Navajo Nation represented an attractive business opportunity for Commnet because of NTUA’s established rights-of-ways on the Navajo Nation’s tribal lands. Saint Regis Mohawk Tribe and Mohawk Networks. The Saint Regis Mohawk Tribe, located in the northern region of New York, received a $10.5 million BIP grant in 2010 to complete a large broadband project expanding access to unserved areas. According to tribal officials and Mohawk Networks representatives, the tribe completed a $15 million broadband infrastructure project laying 68 miles of fiber and connecting 1,500 tribal households and community institutions. Upon completion of the BIP broadband project, the tribal officials said the tribe launched its tribally owned broadband provider, Mohawk Networks, LLC in 2015, to respond to tribal residents’ need for reliable, cost-effective broadband service. Tribal officials said Mohawk Networks currently provides high-speed Internet to tribal homes and businesses. According to tribal officials, in addition to providing broadband service to tribal residents for the first time, the partnership between Saint Regis Mohawk Tribe and its tribally owned broadband provider created new jobs and opportunities to expand broadband services. For example, the officials said the partnership resulted in the creation of a tribal subsidiary, North Country Broadband Services, Inc., to deploy wireless infrastructure to neighboring counties, thus generating new revenue for Mohawk Networks. Few Federal Funds Were Provided to Tribal Entities to Increase Broadband Deployment from 2010 to 2017 FCC and RUS are the primary sources of federal funding to deploy broadband infrastructure in rural and remote areas where the cost of providing service is high, including tribal lands. Based on our review of the funding provided by four federal programs targeted to increase deployment in unserved areas, very little has gone directly to tribes or to tribally owned broadband providers. Specifically, from 2010 to 2017, we found that less than 1 percent of FCC funding and about 14 percent of RUS funding went directly to tribes and tribally owned providers. Combined, FCC and RUS funding totaled $34.6 billion during that time period and tribes and tribally owned providers received $235 million, or about 0.7 percent. While the majority of the funding from the four programs we reviewed from both agencies is provided to deploy broadband to rural, unserved, or underserved areas, only one source of funding, FCC’s Tribal Mobility Fund Phase I, is dedicated specifically to deploying broadband on tribal lands. The National Broadband Plan stated in 2010 that tribes needed substantially greater financial support than was available to them at the time and that accelerating tribal broadband deployment would require increased funding. Furthermore, the National Congress of American Indians expressed concerns that the needs for federally funded broadband projects are greater on tribal lands but tribes do not receive the appropriate share of federal funding aimed at increasing broadband deployment. Through our analysis we found that 14 tribal entities received federal funding from FCC and RUS to increase broadband deployment from 2010-2017 (see fig. 2). Of the four main programs we reviewed, tribes and tribally owned providers received the following funds: Connect America Fund: Nine tribally owned providers received high- cost support funding totaling $218.1 million. Mobility Fund Phase I: One tribally owned provider received support totaling $3.3 million. Tribal Mobility Fund Phase I: No tribal providers received funding. RUS Community Connect Grants: Four tribal entities received $13.5 million. Stakeholders Cited Barriers for Tribes to Obtain Federal Funding and Federal Agencies Have Taken Few Actions to Address the Barriers The tribal officials, tribal associations, and tribally owned broadband providers we interviewed cited several barriers that tribes may face when seeking federal funding for broadband deployment. The two primary barriers these interviewees cited were (1) the statutory requirement for ETC designation and (2) grant application requirements. Statutory Requirement for ETC Designation FCC’s Connect America Fund (CAF) is the largest source of federal funding for broadband deployment in unserved and underserved areas; however, very few tribes are currently eligible for this source of funding. At the time of our review, FCC officials told us there were 11 tribes that have providers that are designated as ETCs and therefore would be eligible to receive CAF funding. Although FCC adopted rules in 2011 to create CAF and modernize the program so that it could support broadband capable networks, FCC officials told us that most ETCs are the telephone companies that were in existence when Congress passed the Telecommunications Act of 1996. According to FCC officials, FCC has explored whether it has authority to allow non-ETC providers to receive CAF support payments but determined that the statute is clear that only ETCs can receive program support. Between 2012 and 2017, FCC officials said FCC received nine ETC applications, four of which were from tribally owned providers. Of those four, only one tribally owned provider was designated an ETC. Three tribes we contacted said they would like the opportunity to receive CAF support to deploy broadband on tribal lands, but they realize they are not eligible to receive funding unless they have the ETC designation. Moreover, officials from two tribes and a tribal association stated that while they want to provide broadband services in their communities, they did not seek the ETC designation because of the ETC service obligations described above. The Leech Lake Band of Ojibwe applied for ETC status in 2013. We met with tribal officials who told us that the tribe was providing broadband service in its community through its own, tribally chartered telecommunications company and at the time of our visit, they had been waiting several years for a decision from FCC on their ETC application. The tribal officials told us that if FCC did not make a decision soon, the tribal government would need to shut down the broadband network, as the tribe’s original decision to fund the network assumed there would be a CAF subsidy to help defray the costs. The Leech Lake reservation is rural with low population density and is surrounded by the Chippewa National Forest. Subsequent to our meeting with the tribe in November 2017, the tribe withdrew its application in March 2018, noting that it was ceasing its attempt to run its telecommunications company specifically “due to inaction” by FCC. According to representatives from a tribal association we contacted, FCC has provided ETCs with billions of dollars to deploy service to unserved areas through the Universal Service Fund programs, but FCC’s efforts have not always been successful in the hardest to reach areas, particularly tribal lands. The representatives noted that FCC’s competitive market approach does not work where competition cannot be supported and that there needs to be a different approach. Similarly, tribal officials from Idaho told us that rural service providers are able to operate due to CAF support, but the tribe is not eligible to receive those subsidies. Officials said although the provider in their area has received millions of dollars in CAF subsidies, it has not deployed broadband on the tribal lands. Other tribal officials from Washington State told us that although private providers received CAF subsidies to deploy broadband service to their reservation, the private providers told the tribe it would be years before they offer service on tribal lands. In 2014, FCC conducted its Rural Broadband Experiment to open up eligibility for CAF funding to non-ETC providers. FCC made $100 million available for the experiment and applicants included a diverse group of entities, including competitive providers, electric utilities, wireless Internet service providers, and others. However, while this experiment opened the application process to non-ETC providers, it did not remove the ETC requirement. CAF support awarded through this experiment was provisional pending the broadband providers’ obtaining ETC status. According to FCC documentation, there were 181 applicants for the experiment, but only 16 ended up meeting all the requirements to receive funding. None of those 16 entities was tribal. Grant Application Requirements Stakeholders we interviewed said tribes may face barriers completing federal grant applications to obtain funding for broadband deployment. In particular, two community access providers, five tribally owned providers, and one regional consortium we contacted said that meeting the application requirements was difficult. Representatives from eight of the tribes we contacted told us that in general, the language included in the federal grant applications is difficult to understand or the administrative requirements of federal grants are burdensome. Another tribal representative told us he would only recommend applying for RUS’s Community Connect program if the tribe has an entire team of dedicated people to manage the grant process. Some of the tribal officials we contacted cited difficulties preparing required application materials between the time a grant announcement was made and the submission deadline. For example, tribal officials we contacted from New Mexico and Oklahoma stated that the constrained time frames prevented them from effectively preparing a comprehensive application package. In some cases, the narrow application windows prevented the tribes from applying at all. Furthermore, tribal officials, tribal associations, and tribally owned broadband providers told us that complying with the following regulatory requirements for RUS Community Connect grants could be challenging for tribes: Preparing existing and proposed network design: RUS’s Community Connect program requires applicants to submit information on the network’s design that contains all the technical information on the applicant’s existing (if applicable) and proposed network. The network design is typically completed by a licensed engineer. Tribal officials in Washington State told us that conducting analyses of existing infrastructure and what improvements are needed can be cost- prohibitive for some tribes because it requires financial resources that the tribe may not have before applying for the grant. Many of these costs are related to the expense of bringing in outside experts or consultants who are needed to perform the technical studies. Another tribal representative told us since the tribe has no way of knowing if the grant will be approved, spending money to complete the application is a large risk. According to RUS officials, the Community Connect program is not authorized to fund pre-planning activities. Demonstrating financial sustainability within 5 years: The RUS Community Connect grant application requires a “financial forecast” that includes the applicant’s existing operations and the proposed project and must be supported by a detailed narrative that explains the methodology and assumptions used to develop the projections, including the number of subscribers projected to take the applicant’s service. The financial forecast must cover at least 5 years, and it is used by RUS to determine whether the proposed project is financially sustainable. However, tribal officials from Idaho told us that it is not feasible for tribes to show financial sustainability (a return on investment) in 5 years in high cost areas. They noted that a period of 15 years may be needed to produce a return on investment in those areas, and this requirement prevents tribes from qualifying for Community Connect grants. Obtaining matching funds required to apply for federal grants: RUS’s Community Connect program requires grant applicants to provide matching funds of at least 15 percent from non-federal sources and does not accept in-kind contributions of goods or services. The matching fund requirement can be difficult for some tribes to obtain. For example, officials from RUS and the tribal entities we contacted told us that tribes often times do not have the upfront cash to meet the matching requirement. According to a tribal association we contacted, obtaining credit is a serious problem for some tribes. In general, tribes cannot collateralize tribal property, and therefore often times are unable to get bank loans for infrastructure projects. The National Broadband Plan recommended that federal agencies facilitate tribal access to broadband funding opportunities. Furthermore, recognizing the need to reduce barriers to expand broadband deployment, the Broadband Opportunity Council, established in March 2015, issued a report stating that federal agencies should use all available and appropriate authorities to identify and address regulatory barriers that may unduly impede either broadband deployment or the infrastructure to augment broadband deployment. RUS officials said they have held a number of external training and outreach events, such as workshops and seminars, with tribes over the past 5 years to provide information about RUS’s broadband programs. For example, in April 2018, before the 2018 Community Connect grant’s application deadline, RUS hosted a webinar on various requirements for grant applications. RUS officials told us that RUS’s outreach efforts generally focus on specific programs and instructing potential applicants on program requirements and how to complete application packages. However, beyond these outreach efforts, RUS officials said they have not undertaken a formal assessment to identify and address the regulatory barriers that tribes may face in obtaining RUS funding for broadband deployment. When we asked RUS officials about the feasibility of doing so, they said that they have limited resources and multiple competing priorities for those resources. RUS officials also noted that BIP authorized and provided funding for technical assistance for applicants, funding that enabled RUS to address some of the barriers tribes face. Nevertheless, lacking such an assessment, tribes may continue to face the regulatory barriers described above in obtaining RUS funding for broadband deployment on their lands. According to the National Broadband Plan, local entities (including tribal, state, regional, and local governments) decide to offer broadband services when no providers exist that meet local needs, and local entities do so after trying to work with established carriers to meet local needs. Several of the tribes we visited told us they were trying to deploy broadband infrastructure or offer service because the private providers were not building out on their lands. For example, one tribe stressed that unlike private providers, they would prioritize tribal areas needing broadband service, but they need federal funding to do so. Conclusions An estimated 35 percent of Americans living on tribal lands lack broadband service, which could hinder tribal efforts to promote self- governance, economic opportunity, education, public safety, and cultural preservation. However, little federal funding aimed at increasing broadband service actually goes to tribal entities, even though the National Broadband Plan stressed that tribes needed substantially greater financial support and recommended that federal agencies facilitate tribal access to broadband funding opportunities. Tribes may face barriers in obtaining federal funds to deploy broadband, and the Broadband Opportunity Council recognized the need for federal agencies to reduce the barriers that are impeding broadband deployment. However, RUS has not taken steps to identify or address the barriers tribes face when applying for RUS grant funding. By identifying and addressing any regulatory barriers that impede tribal entities’ access to RUS funding, RUS could help tribes obtain funding to expand broadband deployment on tribal lands. Recommendation for Executive Action The Secretary of Agriculture should direct the Administrator of RUS to undertake an assessment to identify any regulatory barriers that may unduly impede efforts by tribes to obtain RUS federal grant funds for broadband deployment on tribal lands and implement any steps necessary to address the identified barriers. (Recommendation 1) Agency Comments We provided a draft of this report to FCC, RUS, and NTIA for comment. FCC and RUS provided technical comments, which we incorporated as appropriate; NTIA did not have any comments. A Department of Agriculture official indicated in an e-mail message that RUS neither agreed nor disagreed with the recommendation. RUS’s technical comments noted that RUS has and will continue to work with tribes to facilitate broadband deployment, whether tribes have the desire and capacity to provide the service or whether another provider is able to bring that service to tribal areas. We are sending copies of this report to the appropriate congressional committees, the Chairman of FCC, the Secretary of Agriculture, the Secretary of Commerce, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-2834 or goldsteinm@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix II. Appendix I: Objectives, Scope, and Methodology This report discusses (1) examples of partnership arrangements that tribal entities have used to increase broadband deployment on tribal lands and the outcomes of those partnerships, (2) the amount of funding provided to tribal entities for broadband deployment from key federal programs, and (3) stakeholder-identified barriers that tribal entities face in obtaining federal funding for broadband deployment and the extent to which federal agencies have taken action to address those barriers. To address these objectives, we reviewed relevant federal statutes, including the Communications Act of 1934, as amended, and Federal Communications Commission’s (FCC) regulations, orders, and policy statements including FCC’s Statement of Policy on Establishing a Government-to-Government Relationship with Indian Tribes. In addition, we reviewed documentation and interviewed officials from FCC, including officials from the Office of Native Affairs and Policy; U.S. Department of Agriculture’s Rural Utilities Service (RUS); U.S. Department of Commerce’s National Telecommunications and Information Administration (NTIA); and U.S. Department of Housing and Urban Development’s Office of Native American Programs. To gather information on partnership arrangements that tribes have entered to increase broadband deployment on tribal lands and their outcomes, we conducted a review of relevant published literature that included government reports, industry articles, and publications from associations, non-profits, and public policy research organizations. Although we were not able to identify an industry-accepted definition of partnerships, we used the term partnerships to refer to instances in which a tribal nation or tribal government works with another entity to design, build, or operate infrastructure assets, or other capital assets to improve or enhance broadband service. This also included partnerships between a tribe and its tribally owned broadband provider. To identify examples of tribal broadband partnerships for our review, we first interviewed agency officials, tribes, private providers, and other stakeholders such as tribal associations. We also identified broadband projects with a tribal partnership component by reviewing reports from 2010 to 2017 for the following federal programs: (1) FCC’s Universal Service Fund high-cost program and Connect America Fund (including the Mobility Fund Phase I (Auction 901) and Tribal Mobility Fund Phase I (Auction 902)); (2) RUS’s Community Connect Program; (3) RUS’s Broadband Initiatives Program; and (4) NTIA’s Broadband Technology Opportunities Program. While there may be other tribal partnership examples that exist, through these efforts we identified seven broadband projects with a tribal partnership component completed within the last 5 years (2013 to 2017). We interviewed tribal leaders and officials from the seven tribes that were involved in the selected partnerships, and visited six tribes in Idaho, New Mexico, and Washington State. When meeting with tribal leaders and officials, we used the same semi-structured interview questions for all tribes; however, tribal officials may not have answered all questions. Because we limited our review to these seven selected partnership examples, our findings are not generalizable. To determine the amount of funding from key federal programs provided to tribal entities for broadband deployment, we first identified the federal programs that provide broadband funding from NTIA’s Guide to Federal Funding of Broadband Projects. The guide lists 17 federal programs that fund broadband infrastructure. Of those federal programs, we focused our review on four programs, three in FCC and one grant program in RUS, selected because they provide the most directly relevant funding for broadband deployment in unserved areas, which includes tribal lands. We first identified federal agencies and programs that provide grants or loans to tribal and non-tribal entities to buildout broadband infrastructure on tribal lands including: FCC, RUS, U.S. Department of Commerce Economic Development Administration (EDA), and NTIA. We interviewed federal agency officials to identify any additional federal programs that provided funding in the last 7 years. We excluded federal loan programs because they may require letters of credit and or assets as collateral, which is often not a feasible option for tribes given land ownership issues. We also considered but excluded those federal programs that are not directly related to broadband expansion and deployment, such as the Department of Housing and Urban Development’s Choice Neighborhoods Program, whose primary purpose is housing related. We compiled total funding data for these four federal programs and the amount of funding provided to tribes and tribal entities for broadband deployment projects for 2010 to 2017. We took steps to assess the reliability of the data—such as cross-checking the data, following up with agency officials, and reviewing documentation—and found the data were sufficiently reliable for the purposes of summarizing total funding and the amount provided to tribes and tribal entities. Because we relied upon titles or names of grant recipients to identify those grants awarded to tribes and tribally owned providers, our analysis may not include some grants awarded to broadband providers that deploy infrastructure to larger service areas that may also include tribal lands. To determine stakeholder-identified barriers that tribal entities face in obtaining federal funding for broadband deployment and federal government efforts to address those barriers, we interviewed FCC and RUS officials and the tribal government officials, tribally owned broadband providers, and tribal associations listed in table 1. We interviewed representatives from 17 tribes in different locations with varying population sizes and levels of broadband deployment. Additionally, we interviewed officials from 9 tribally owned and 7 private broadband providers operating on tribal lands. We selected these broadband providers to interview because they received federal support to serve on tribal lands or because they were a designated eligible telecommunications carrier (ETC) serving tribal interests. Furthermore, we identified and interviewed industry stakeholders such as research groups and telecommunications associations on their views regarding barriers to obtaining federal program assistance for broadband deployment on tribal lands. These stakeholders were selected based on their exposure to broadband issues on tribal lands such as representing tribally owned broadband providers. The views obtained from these interviews are not generalizable to all tribes, all broadband providers, or all industry stakeholders. We also reviewed a report from the Broadband Opportunity Council directing agencies to identify and address regulatory barriers that may unduly impede broadband deployment and assessed RUS’s efforts to address the regulatory barriers tribes may face in attempting to obtain RUS funding for broadband deployment. For a complete list of entities we interviewed see table 1. We conducted this performance audit from September 2017 to September 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: GAO Contact and Staff Acknowledgments GAO Contact Mark L. Goldstein, (202) 512-2834 or goldsteinm@gao.gov. Staff Acknowledgments In addition to the contact named above, Sally Moino (Assistant Director); Tina Paek (Analyst in Charge); Rose Almoguera; Sharon Dyer; Hannah Laufe; Serena Lo; Cheryl Peterson; Malika Rice; Amy Rosewarne; Jay Spaan; James Sweetman, Jr.; Hai Tran; and Jade Winfree made key contributions to this report.

Why GAO Did This Study In 2018, FCC estimated that 35 percent of Americans living on tribal lands lack broadband service compared to 8 percent of Americans overall. Various federal programs support increasing broadband deployment in unserved areas, including tribal lands. Tribes can form partnerships with private sector companies and others to deploy broadband infrastructure on tribal lands. GAO was asked to provide information on these partnerships. This report discusses (1) examples and outcomes of tribal partnership arrangements, (2) the amount of federal funding provided to tribal entities for broadband deployment, and (3) stakeholder-identified barriers that tribes face in obtaining federal funding and the extent to which federal agencies have addressed those barriers. GAO identified partnerships by reviewing federally funded broadband projects that included a partnership component; analyzed federal funding dedicated to broadband deployment; interviewed agency and tribal government officials, tribal associations, tribally owned broadband providers, and industry stakeholders; and assessed RUS's efforts to address the regulatory funding barriers tribes may face. The information presented is illustrative and is not generalizable to all tribes or all partnerships. What GAO Found GAO identified some partnership arrangements between tribes and other entities to increase broadband access on tribal lands. Among the seven examples GAO identified, tribes partnered with different types of entities including private broadband providers, a community access network provider, an electric cooperative, a regional consortium, and tribally owned broadband providers. According to the tribes participating in the partnerships, almost all of the partnerships improved broadband service on tribal lands. The Federal Communications Commission (FCC) and the Rural Utilities Service (RUS) are the primary sources of federal funding to deploy broadband infrastructure where the cost of providing service is high, including on tribal lands. GAO reviewed funding for four programs, three in FCC and one grant program in RUS, and found that in total, less than 1 percent has gone directly to tribes or to tribally owned broadband providers to expand broadband service. GAO found that 14 tribal entities received federal funding from FCC and RUS to increase broadband deployment for 2010–2017. The tribal officials, tribal associations, and tribally owned broadband providers GAO contacted cited several barriers tribes face in obtaining federal funding to deploy broadband service on tribal lands. For example, they said tribes face regulatory barriers in applying for RUS's grant funding, including (1) preparing existing and proposed network design, (2) demonstrating financial sustainability of the broadband project within 5 years, and (3) obtaining matching funds. An interagency council recently recommended that federal agencies identify and address regulatory barriers that may unduly impede broadband deployment. Although RUS conducts some outreach with tribes, it has not undertaken a formal assessment to identify and address the regulatory barriers that tribes may face in obtaining RUS's funding for broadband deployment. RUS officials told GAO that they do not have the resources to do so. Nevertheless, lacking such an assessment, tribes may continue to face regulatory barriers in obtaining RUS funding for broadband deployment on their lands. By identifying and addressing any regulatory barriers that impede tribal entities' access to RUS funding, RUS could help tribes obtain funding to expand broadband deployment on tribal lands. What GAO Recommends GAO recommends that RUS identify and address regulatory barriers that impede tribal entities from obtaining RUS funding for broadband deployment. RUS neither agreed nor disagreed with this recommendation.

gao_GAO-19-99

gao_GAO-19-99_0

Background Headquartered in Washington, D.C., the Corps has eight regional divisions and 38 districts that carry out its domestic civil works’ program (see fig. 1). Corps headquarters primarily develops policies based on administration guidance; plans the direction of the organization; and approves projects to recommend for inclusion in the President’s annual budget request to Congress. The divisions approve projects for submission to headquarters and coordinate projects within their districts, while the districts plan and implement the projects. The Corps’ construction appropriations account has three main business lines— aquatic ecosystem restoration, flood risk management, and navigation— that correspond to the three primary missions of its civil works program. Some projects may be multipurpose and fit into more than one business line within the program. For fiscal years 2008 through 2017, the President’s budgets requested about $4.78 billion per year, on average, in discretionary funding for the Corps’ civil works program to plan, construct, operate, and maintain a wide range of water resources projects (see fig. 2). Of this total, the budget requested an average of about $1.39 billion total per year for construction projects in the aquatic ecosystem restoration, flood risk management, and navigation business lines. The total construction amounts in Corps budget documents include funding for the hydropower business line; hydropower funding is represented in the other discretionary funding in this figure. We excluded those funding amounts from our business line totals, in part because these projects are now mainly funded through the operations and maintenance account. They represented less than 4 percent of the construction account requests from fiscal years 2008 through 2017. Discretionary funding refers to the level of budget authority, outlays, or other budgetary resources (other than those for mandatory programs) that is provided in, and controlled by, appropriation acts. Construction Projects Included in the President’s Budget Requests for the Corps Located In Over Half of the States For fiscal years 2008 through 2017, the President’s budget requests for the Corps included 164 construction projects located in 31 states, the District of Columbia, and Puerto Rico. The five states for which the most funds were requested in this period were: Florida, with $2.4 billion requested for 15 construction projects, Illinois, with $2.3 billion requested for 15 construction projects, California, with $1.6 billion requested for 24 construction projects, Washington, with $924 million requested for 9 construction projects, Kentucky, with $646 million requested for 6 construction projects. The projects in these five states accounted for 61 percent of the $12.9 billion that the President requested for Corps construction projects in these years. Figure 3 shows the total number of construction projects and funds included in the President’s budget requests for fiscal years 2008 through 2017, by state, within the Corps’ three main business lines— aquatic ecosystem restoration, flood risk management, and navigation. See appendix I for the number of construction projects and funds requested for each fiscal year during this period, by state. See appendix II for a list of the names of construction projects, locations, business lines, and funds requested per year during this period. For fiscal years 2015 through 2017, the Corps provided us with detailed data for the construction projects included in the President’s budget requests. This information included latitudinal and longitudinal coordinates to locate projects within states and divisions, as well as business-line-specific data. A total of 71 projects were included in the budget requests for these business lines during this period. Construction projects included in the President’s budget requests for the Corps’ three main business lines for fiscal years 2015 through 2017 were geographically distributed in 26 states and Puerto Rico. According to our analysis of Corps budget data, most of the projects were near either water sources or Corps-constructed water infrastructure. Figure 4, which is an interactive map, identifies the locations and describes budget data for each construction project (see interactive instructions). See appendix III for a list of these construction projects by Corps division, state, business line, and funds requested for each fiscal year. Corps headquarters officials said that the Corps does not specifically use geographic locations to select construction projects to recommend for inclusion in the President’s budget requests. However, Corps officials explained that geographic characteristics, such as population, might have affected how they considered including construction projects within specific business lines. For example, Corps officials within the flood risk management business line may have considered a construction project located in a population center that could be severely impacted by a flooding event to be a higher priority over other projects in less populated areas. For fiscal years 2015 through 2017, the Corps requested more than $3 billion for the 71 construction projects that fell within the three main business lines: Aquatic ecosystem restoration. The President’s budget requested $618 million for 15 Corps construction projects in the aquatic ecosystem restoration business line, which were located in California, Florida, Georgia, Illinois, Louisiana, Maryland, Minnesota, Oregon, and Washington. According to the Corps budget guidance, these projects were located in areas of federal significance that have some degree of habitat scarcity, connectivity, and special-status species, among other characteristics. Moreover, according to the Corps budget guidance, construction projects in this business line emphasize the restoration of nationally or regionally significant habitats where the solution primarily involves modifying the hydrology and geomorphology. For example, the goals of the South Florida Ecosystem Restoration program—a collection of several projects— includes improving the health of over 2.4 million acres of the south Florida ecosystem (including Everglades National Park), enhancing water supply, and maintaining flood mitigation, according to a Corps document and Corps officials. According to a Corps document, since 2000, the Corps has invested a total of $2.4 billion in the program including other initiatives, such as the Comprehensive Everglades Restoration Plan and Central and Southern Florida. Flood risk management. The President’s budget requested $1.33 billion for 33 Corps construction projects in the flood risk management business line, which were located in California, Florida, Illinois, Kansas, Kentucky, Massachusetts, Missouri, New Jersey, Ohio, Oklahoma, Pennsylvania, Tennessee, Texas, Virginia, and West Virginia. According to the Corps’ budget guidance and strategic plan, these projects are located in areas that may experience riverine and coastal flooding, and they are to provide water supply storage. For example, the Bluestone Lake project, in West Virginia, is to address deficiencies that could lead to a breach of a dam built by the Corps in the 1940s. According to the Corps, the dam’s spillway cannot discharge enough water without substantially increasing the potential for a breach of the dam. According to the Corps, a breach could cause catastrophic flooding along the largest river valleys in West Virginia, including locations of major manufacturing and chemical industries, and put 165,000 lives at risk. The Corps started the project in 1998, and plans to award the next phase of the project in 2022. A draft supplementary study has been completed to identify a plan to address this additional deficiency, according to the Corps. The Corps is planning for a 10-year construction period, with an estimated cost of $575 million, according to a Corps document and a headquarters official. Navigation. The President’s budget requested $908 million for 20 Corps construction projects in the navigation business line, which were located in coastal, inland, and intracoastal navigation systems in California, Georgia, Illinois, Louisiana, Missouri, New Jersey, New York, Ohio, Oregon, South Carolina, Texas, Virginia, Washington, and Wisconsin. According to Corps’ budget guidance, these projects are intended to provide safe, reliable, cost-effective and environmentally sustainable waterborne transportation systems for the movement of commercial goods. For example, the Corps’ Olmsted Locks and Dam project is located on the Ohio River, which connects to the Tennessee, Cumberland, and Mississippi rivers and is considered critical for commercial navigation. According to the Corps, the project consists of two 110-foot by 1,200-foot locks, which are located adjacent to the Illinois bank, and a dam comprising of five tainter gates, which control the amount of water that flows downstream. According to a Corps document, over the last several years, approximately 80 million tons of bulk commodities (for example, coal, grain, rock, and sand) per year, on average, have passed through navigation structures that are part of the project. The Corps estimates that this project has been under construction for nearly 30 years. According to Corps documents and headquarters officials, the project became operational as of September 2018, with a total estimated cost of $3 billion by the time of project completion. Corps Used a Multi- level Process to Prioritize Construction Projects To prioritize construction projects to recommend for inclusion in the President’s budget requests for fiscal years 2008 through 2017, the Corps used a process involving the three levels of its organization—districts, divisions, and headquarters. Districts Prepared Work Packages for Ranking To begin the process, Corps district officials divided projects into work packages—increments of work that can be considered for inclusion in the budget. According to Corps policy guidance for budget development (budget guidance), these work packages should contribute to the overall project and be executed without being dependent on the funding of additional work packages. According to a district official, district officials then assigned one of six priority levels to indicate the order in which work packages for the same project should be completed for that fiscal year. Priority levels are categories used to differentiate work packages within the same project. Corps budget guidance instructed district officials to assign priority levels based on criteria including whether a project is new or continuing and where a work package falls within a project’s overall work plan. Corps budget guidance also instructed officials to group work packages either by business line or appropriations account, depending on the fiscal year, based on the budget guidance for that fiscal year. Districts, Divisions, and Headquarters Ranked Work Packages Corps districts, divisions, and headquarters consecutively ranked the work packages, as shown in figure 5. In doing so, they established criteria specific to the business line for each project. Corps budget guidance provided instructions on which criteria to use for each business line to determine rankings in a particular year. The ranking criteria in the guidance—such as the rate of economic return, populations at risk, or the environmental impact—varied nearly every year from fiscal years 2008 to 2017 for two of the three main business lines: flood risk management and navigation (see appendix IV for the criteria used, by business line and fiscal year). Corps officials said they routinely revised the criteria while developing the annual budget guidance, for reasons such as addressing changes in the policy guidance from the Assistant Secretary of the Army for Civil Works or OMB, or improving the ranking process. Corps officials we interviewed noted that although each level used the same criteria to rank work packages, the districts, divisions and headquarters had different focuses and increasing numbers of work packages to rank and compare. Specifically, officials at each level considered the overall needs of their respective jurisdictions when making ranking decisions: districts had a local focus; divisions had a regional or watershed focus; and headquarters had a national focus. The number of work packages to be ranked increased at each level according to Corps officials: districts ranked local work packages; divisions re-ranked work packages from four to seven districts; and headquarters re-ranked work packages from all of the divisions nationwide. After ranking all work packages within their respective jurisdictions by business line or appropriations account, officials from all three levels entered the rankings into the database for use in the budget review process. According to information from the Civil Works Integrated Funding Database, the Corps ranked more than 25,000 work packages for the fiscal year 2017 budget recommendation. Headquarters Developed Final Recommendations for Budget Requests According to one headquarters official, Corps officials in the Program Development Branch at headquarters facilitated discussions among business line and account managers to develop the final rankings of all work packages. As part of this process, headquarters officials noted that business line managers compared work packages with different characteristics across business lines or accounts. According to Corps budget guidance and headquarters officials, business line managers and account managers are instructed to consider two key factors when determining their final rankings each fiscal year. Specifically, those managers are instructed to give top priority to work packages that significantly impact the risk to human life posed by potential disasters. In addition, the managers are to prioritize work packages that address a legal requirement to mitigate potential negative effects caused by construction, such as adverse environmental effects. Using the final rankings, Corps headquarters officials said they developed the final budget recommendations for each fiscal year, including a recommended funding amount for each project, with input from various levels of the organization. More specifically, to determine the budget recommendations, Corps headquarters officials obtained feedback from district commanders, generals, directors, and the Chief of Engineers. In fiscal year 2017, the Corps used its final rankings to determine recommended funding amounts for 89 construction projects, each of which included one or more work packages; ultimately, these projects comprised about 298 work packages. Once the Corps headquarters officials developed these recommendations, they briefed the Assistant Secretary of the Army for Civil Works on their recommendations. An official from the Assistant Secretary’s office said they reviewed the Corps recommendations and compared them with the Assistant Secretary’s priorities, after which they developed the final recommendations to send to OMB for review and potential inclusion in the President’s budget requests. According to a Corps official, 34 percent of construction projects included in the fiscal year 2017 President’s budget request received funding. Agency Comments and Our Evaluation We provided a draft of this report for review and comment to the Department of Defense. The department told us they had no comments on the draft report. We are sending copies of this report to the appropriate congressional committees, the Secretary of Defense, the Chief of Engineers and Commanding General of the U.S. Army Corps of Engineers, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-3841 or fennella@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix V. Appendix I: Corps Construction Projects Included in the President’s Budget Requests, by State, Fiscal Years 2008 through 2017 Table 1 lists the 164 construction projects and funds requested, by state, within the U.S. Army Corps of Engineers’ (Corps) three main business lines—aquatic ecosystem restoration, flood risk management, and navigation—for fiscal years 2008 through 2017. Appendix II: Corps Construction Projects Included in the President’s Budget Requests, Fiscal Years 2008 through 2017 Table 2 lists the names and locations of the 164 construction projects the U.S. Army Corps of Engineers (Corps) identified as included in the President’s budget requests for its three main business lines for fiscal years 2008 through 2017. The Corps’ three main business lines are aquatic ecosystem restoration, flood risk management, and navigation. Appendix III: Corps Construction Projects Included in the President’s Budget Requests, by Division, Fiscal Years 2015 through 2017 Table 3 lists the names of the 71 construction projects, their locations, and the agency divisions that manage them, as shown in interactive figure 4 of this report, and includes the figure’s rollover information. The U.S. Army Corps of Engineers (Corps) identified these projects as included in the President’s budget requests for its three main business lines for fiscal years 2015 through 2017. The Corps’ three main business lines are aquatic ecosystem restoration, flood risk management, and navigation. Appendix IV: U.S. Army Corps of Engineers’ Criteria Used to Rank Construction Projects in Main Business Lines Table 4 lists the criteria included in the U.S. Army Corps of Engineers’ (Corps) annual budget guidance for its three main business lines for fiscal years 2008 through 2017. The Corps’ three main business lines are aquatic ecosystem restoration, flood risk management, and navigation. Appendix V: GAO Contact and Staff Acknowledgments GAO Contact Anne-Marie Fennell, (202) 512-3841 or fennella@gao.gov. Staff Acknowledgments In addition to the contact named above, Vondalee R. Hunt (Assistant Director), Leah E. English, Kerstin Hudon, Susan Malone, and Cynthia Norris made significant contributions to this report. Important contributions were also made by Melinda Cordero, Justin Fisher, Juan Garay, Patricia Moye, and Danny Royer. Related GAO Products Army Corps of Engineers: Factors Contributing to Cost Increases and Schedule Delays in the Olmsted Locks and Dam Project. GAO-17-147. Washington, D.C.: February 16, 2017. Standards for Internal Control in the Federal Government. GAO-14-704G. Washington, D.C.: September 2014. Army Corps of Engineers: The Corps Needs to Take Steps to Identify All Projects and Studies Eligible for Deauthorization. GAO-14-699. Washington, D.C.: August 21, 2014. Army Corps of Engineers: Cost Increases in Flood Control Projects and Improving Communication with Nonfederal Sponsors. GAO-14-35. Washington, D.C.: December 20, 2013. Army Corps of Engineers: Budget Formulation Process Emphasizes Agencywide Priorities, but Transparency of Budget Presentation Could be Improved. GAO-10-453. Washington, D.C.: April 2, 2010. A Glossary of Terms Used in the Federal Budget Process. GAO-05- 734SP. Washington, D.C.: September 2005.

Why GAO Did This Study Through its civil works program, the Corps plans, designs, constructs, operates, and maintains a range of water resources projects for purposes such as aquatic ecosystem restoration, flood risk management, and navigation. To support these projects, the Corps requests funding through the annual budget and appropriation process. For fiscal year 2017, the President's budget requested $4.6 billion for Corps' water resources projects, of which about $1 billion was for construction projects. GAO was asked to review budget requests for construction projects under the Corps' civil works program, including the geographic distribution of those projects. This report examines for fiscal years 2008 through 2017 (1) the geographic distribution of the construction projects included in the President's budget requests for the Corps, and (2) how the Corps prioritized such projects for inclusion in the President's budget requests. GAO summarized available budget data for fiscal years 2008 through 2017; reviewed the Corps' budget guidance and documents; mapped locations for construction projects in budget requests for years when sufficient information was available; and interviewed Corps headquarters and division officials. What GAO Found For fiscal years 2008 through 2017, construction projects included in the President's budget requests to Congress for the U.S. Army Corps of Engineers (Corps) were geographically distributed in 31 states, the District of Columbia, and Puerto Rico. During this 10-year period, the President requested over $12.9 billion for 164 construction projects included in the Corps' three main missions—aquatic ecosystem restoration, flood risk management, and navigation. The Corps provided GAO with detailed information on the location of construction projects included in the budget requests for the 3 most recent fiscal years at the time of its review—2015 through 2017. These projects, shown in the figure below, spanned 26 states and Puerto Rico. They were typically located near sources of water or Corps-constructed water infrastructure. To prioritize construction projects to include in the President's budget requests for fiscal years 2008 through 2017, the Corps used a process involving each of its three organizational levels—districts, divisions, and headquarters. Districts divided projects into work packages and assigned 1 of 6 priority levels to indicate the order in which work packages from the same project should be completed. Districts grouped these work packages by business line or appropriations account based on the Corps' budget guidance for the fiscal year and then ranked them. Then Corps divisions and headquarters ranked the work packages. To assign rankings, Corps officials applied criteria specific to the business line of each project. These criteria often varied by fiscal year to address changes to policy guidance. Across the organization, Corps officials ranked more than 25,000 packages for fiscal year 2017. After assigning rankings, headquarters developed final budget recommendations to submit to the Assistant Secretary of the Army for Civil Works, who in turn provided the recommendations to the Office of Management and Budget for review and potential inclusion in the President's budget requests. What GAO Recommends GAO is not making recommendations in this report. The Department of Defense stated that they had no comments on the draft report.

gao_GAO-19-49

gao_GAO-19-49_0

Background In addition to improving the government-wide acquisition of IT, FITARA was intended to assist Congress in holding covered agencies accountable for their progress towards reducing duplication and achieving cost savings. The act also enhanced the CIO’s authority in covered agencies for the formulation and approval of their agency’s IT budgets. In this regard, the act requires CIOs to have a significant role in the decision processes for all annual and multi-year planning, and to approve the IT budget requests of the agency. In June 2015, OMB released guidance that describes how agencies are to implement the requirements of FITARA. The guidance is intended to, among other things: assist agencies in aligning their IT resources with statutory requirements; establish government-wide IT management controls that will meet the law’s requirements, while providing agencies with flexibility to adapt to unique agency processes and requirements; strengthen the relationship between agency CIOs and component CIOs; and strengthen the CIO’s accountability for IT costs, schedules, performance, and security. The guidance identifies a number of actions that agencies are to take to establish a basic set of roles and responsibilities (the common baseline) for CIOs and other senior agency officials. One such action is that agencies are to conduct a self-assessment to determine whether their current policies and procedures meet or do not meet the common baseline requirements. If the agencies do not meet the requirements, they are to submit an implementation plan describing the changes they intend to make to their policies and procedures in order to ensure that the common baseline requirements are met. Further, the guidance notes that senior agency officials—including Chief Financial Officers and Chief Acquisition Officers—are to work in partnership to facilitate successful implementation of the common baseline and to ensure the CIO is a strategic partner in agency strategies, budgets, and operations. In its guidance, OMB states that agency CIOs are allowed to delegate certain responsibilities from the common baseline to other agency officials, such as component agency CIOs. For example, CIOs can delegate to these officials, inclusion in the planning, programming, and budgeting stages for programs with IT resources. However, according to the guidance, agency CIOs cannot delegate their responsibility for reviewing and approving the major IT investments portion of the budget request. The guidance further states that, for delegated responsibilities, agency CIOs are to establish plans that demonstrate how they will retain accountability. These delegation plans should include procedures for ensuring that the delegated official will execute the responsibility with the appropriate level of rigor. In addition to the FITARA implementation guidance, OMB Circular A-130 establishes general requirements for the planning; budgeting; governance; acquisition; and management of federal information, personnel, equipment, funds, IT resources, and supporting infrastructure and services. The circular identifies responsibilities for planning, programming, and budgeting that reinforce requirements in OMB’s FITARA implementation guidance. Moreover, in May 2018 the President issued an executive order that reinforces requirements in OMB’s FITARA implementation guidance. The order noted that its purpose was to further enhance the effectiveness of CIOs by, among other things, requiring agency heads to ensure that the CIO has a significant role in all IT-related annual and multi-year planning, programming, budgeting, and execution decisions. In addition, the executive order noted that agency heads are to direct the CIO to be a voting member of and to chair agency governance boards, including investment review boards, that have purview over IT or that set agency-wide IT standards. We have previously testified that, while agencies have made progress in implementing FITARA, its further implementation is critical to improving IT management. We have also noted that, continued congressional oversight of agencies’ implementation of this law is essential to help ensure that these efforts succeed. In addition, in an August 2018 report, we noted that 23 federal agencies had reported wide variations in the authority over component-level IT spending. For example, 8 agencies reported that the CIO had 100 percent authority over the agencies’ IT spending (including for components), while 10 agencies reported that these officials had authority for less than 50 percent of such spending. These widely varying levels of authority over agency-wide IT spending existed, in part, because OMB’s guidance did not completely define the authority that CIOs should have over this spending. Accordingly, we recommended that OMB define the authority that the CIOs are to have when agencies report on their authority over IT spending. OMB subsequently agreed with our recommendation. Departments Took Steps to Establish Policies and Procedures for IT Budgeting, but Lacked Plans to Fully Implement FITARA Requirements As previously mentioned, OMB’s guidance on implementing FITARA requires departments to develop policies and procedures to address a number of requirements identified in the basic set of roles and responsibilities (the common baseline) for CIOs. These include the eight selected common baseline requirements related to the CIO’s responsibility for IT budgeting. As identified in table 1, these requirements can be categorized into three areas: (1) CIO visibility into IT resources, (2) CIO input into IT resource plans, and (3) CIO review and approval of IT budgets. OMB required federal agencies and departments to establish FITARA implementation plans that articulated policies and procedures for addressing each of the common baseline requirements and that described changes the departments intended to make to address any gaps in their policies and procedures. Further, for delegated responsibilities, OMB required agency CIOs to establish delegation plans that demonstrate how they intend to retain accountability for the requirement and ensure that the delegated official will execute the responsibility with the appropriate level of rigor. Toward this end, the selected departments had taken steps to establish policies and procedures that addressed the common baseline requirements established by OMB; however, most of the departments, or their component agencies, lacked comprehensive policies and procedures that fully addressed all of the requirements. Specifically, of the eight common baseline requirements that we reviewed, all four departments and their respective component agencies had fully documented one requirement in their policies and procedures, and either had partially documented or had not documented the other seven requirements. While shortfalls existed for each department, DOJ had the most comprehensive IT budgeting policies and procedures, followed by Treasury, HHS, and DOE, respectively. In addition, department CIOs at Treasury, HHS, and DOE had delegated many of the responsibilities for addressing the IT budgeting requirements to component CIOs—thus, these component agencies were to supplement their departments’ policies and procedures with their own IT budgeting policies and procedures for the responsibilities they were delegated. Among the three respective component agencies to which Treasury, HHS, and DOE had delegated IT budgeting responsibilities, the extent to which the components had documented requirements in their policies and procedures varied. For example, IRS had documented policies and procedures for all four of its delegated requirements. In addition, CMS had documented policies and procedures that satisfied two, partially satisfied two, and did not satisfy one of the five requirements delegated to that component. For its part, NNSA had not documented any of its five delegated requirements in the component’s policies and procedures. Figure 1 illustrates the extent to which all four departments’ policies and procedures had addressed the selected OMB common baseline requirements, and is followed by a discussion of each category of requirement. In addition, the figure highlights areas where component agencies addressed their delegated responsibilities or did not address their delegated responsibilities. Appendices II through V (for DOE, HHS, DOJ, and Treasury, respectively) provide additional details about our assessment of the extent to which the departments’ policies and procedures had addressed the selected OMB common baseline requirements, as well as the extent to which component agencies fulfilled their delegated responsibilities. CIO Visibility into IT Resources Establish the Level of Detail with Which IT Resources Are to Be Described in Order to Inform the CIO during the Planning and Budgeting Processes All four departments had policies and procedures that addressed the level of detail with which IT resources are to be described in order to inform the CIO during the planning and budgeting processes. For example, three of the four departments (DOE, HHS, and Treasury) had policies that required the IT budget to include a description of IT resource categories that are required by OMB’s IT capital planning guidance, such as government labor and certain infrastructure resources. DOJ’s policy required the IT budget to include a description of 49 different IT resource categories. As a result, the departments have increased the likelihood that IT resources will be consistently described with the appropriate level of detail for the CIO. Establish Agency-Wide Policy for the Level of Detail with Which Planned Expenditures for All Transactions That Include IT Resources Are to Be Reported to the CIO Each of the four departments had documented within their IT budgeting procedures the level of detail that was required for reporting planned IT expenditures to the CIO. However, the procedures did not explicitly require that every transaction that related to IT resources be included in the planned expenditure reporting to the CIO. Without explicitly requiring that all transactions that have IT resources be included in the reporting of planned expenditures, there is increased risk that the CIO cannot ensure that all budget requests contain complete and accurate resource estimates, in a consistent manner, to inform the department’s annual IT budget. CIO Input into IT Resource Planning Include the CIO in the Planning and Budgeting Stages for Programs That Are Fully or Partially Supported with IT Resources The departments varied in the extent to which their policies and procedures included a requirement for the department-level CIO to be included in programs supported with IT resources. For example, DOE and DOJ took steps to ensure that their CIOs are included in the planning and budgeting of programs with IT resources by requiring that each IT acquisition request include information about the investment in the CIO’s IT portfolio that is to support the acquisition. Adding this investment information to each acquisition request is intended to allow the CIO to ensure that the requests are factored into resource planning for the IT budget. However, DOE’s, HHS’s, and Treasury’s policies and procedures did not always require that the CIO be included in the planning and budgeting stages for every program with IT resources. For example, DOE’s policies called for the CIO to be included in the budget development process by requiring the program offices to submit their IT budget to the CIO for review and approval annually. On the other hand, this policy did not apply to NNSA’s IT programs because the responsibility to meet this requirement was delegated to that component. NNSA drafted procedures to carry out the delegated responsibility, but the procedures did not call for the DOE CIO—in addition to the component-level CIO—to have input into the IT budget, as required by the Secretary of Energy’s October 2016 FITARA implementation memorandum. According to officials in NNSA’s Office of the CIO, the component expects to finalize the procedures by the end of August 2018; however, the officials did not say whether the finalized procedures would include a requirement for NNSA to obtain input from the DOE CIO on its IT budget. By not requiring that the department-level CIO be included in the planning and budgeting stages for programs that are fully or partially supported with IT resources, DOE, HHS, and Treasury are at increased risk that the CIO is not providing input into key IT resource planning decisions. Include the CIO as a Member of Governance Boards That Inform Decisions Regarding All IT Resources, Including Component-Level Governance Boards The charters for all four department-level investment review boards that inform decisions regarding IT resources indicated that their respective CIOs were included as members. In addition, the charters for component-level investment review boards that inform decisions regarding IT resources at CMS and IRS included their respective component-level CIOs. However, a similar review board at FBI did not include the component-level CIO as a member and NNSA had not yet finalized its charter. Further, none of the charters for the selected components’ investment review boards indicated that the department-level CIOs were members. Among the three CIOs at DOE, HHS, and Treasury that had delegated the responsibility of component-level board membership to component CIOs, the department-level CIOs at these agencies had not established procedures for ensuring that the components had implemented this responsibility, as required by OMB. As previously mentioned, while department CIOs were allowed to delegate this responsibility, OMB requires department CIOs to establish delegation plans that describe each requirement being delegated, demonstrate how the department CIOs will retain accountability for the requirement, and ensure that the delegated official executed the responsibility with the appropriate level of rigor. By not requiring that the department-level CIO be included in key governance board decisions regarding IT investments or establishing delegation plans that outline such activities for component CIOs, the selected departments are at increased risk that the CIO is not providing input into key IT resource planning decisions. Document the Processes by Which Program Leadership Works with the CIO to Plan an Overall Portfolio of IT Resources Departments varied in the extent to which they had documented the process by which program leadership is to work with the CIO to plan an overall portfolio of IT resources. For example, DOJ and Treasury documented a detailed process with roles and responsibilities for how program leadership is to work with their CIOs to plan resources for the overall portfolio through their IT governance process. However, DOE and HHS had partially documented the process they were to follow to meet this requirement. Specifically, these departments documented that they were to utilize department-level governance boards to plan IT resources with program leadership for investments subject to the governance board reviews. However, they did not always document how CIOs were to work with program leadership in planning IT resources for other investments that were not subject to department-level governance board reviews, such as existing HHS investments that are greater than or equal to $20 million annually and DOE investments initiated by NNSA. As a result of shortfalls in documenting policies and procedures that require the CIO to work with program leadership to plan the IT portfolio, DOE and HHS are at an increased risk that the CIO’s role in the formulation of IT budgets is limited. CIO Review and Approval of the IT Budget Request Ensure the CIO Has Reviewed and Approved the Major IT Investments Portion of the Budget Request The four selected departments varied in the extent to which they had documented in their policies and procedures how their CIOs are to review and approve the major IT investments portion of the budget request. For example, DOJ had a documented process for how the CIO is to review and approve all major IT investments through the department’s annual budget planning and IT portfolio review processes. In contrast, the other three departments partially addressed the requirement by documenting the requirement to review and approve certain major investments, but not all major investments. To illustrate, DOE documented policies and procedures that required the CIO to review major IT investments, but the policies and procedures did not apply to major IT investments within NNSA and the national laboratories, including those related to high-performance computing. Further, NNSA had draft policies and procedures requiring its component-level CIO to review major IT investments. However, these policies and procedures had not yet been finalized and approved, and they did not include a requirement for the DOE CIO’s review of these investments. Moreover, the department had not developed policies and procedures stipulating this requirement for the national laboratories. While officials in the department’s Office of the CIO stated that they plan to revise policies and procedures for the national laboratories to include the CIO in their annual planning processes, the officials did not identify a time frame for completing those revisions. As another example, HHS had documented a process that required the department-level CIO to review and approve new major IT investments greater than or equal to $20 million. In addition, the process required that the review and approval of new and existing major investments between $10 million and $20 million annually be delegated to the department’s component CIOs. Accordingly, CMS met this requirement at the component level by documenting IT investment review board policies and procedures that require the component CIO to review and approve major, high-risk, and mission critical IT investments with estimated costs of less than $20 million. However, HHS did not fully address the department-level requirement in that its process did not document how the department-level CIO would review and approve existing (as opposed to new) major investments greater than or equal to $20 million annually. As a result of not fully documenting the process for how the departments are to meet this requirement, DOE, HHS, and Treasury are at increased risk that major investments will be submitted for the budget without being reviewed and approved by the CIO. Ensure the CIO Has Reviewed IT Resources That Are to Support Major Program Objectives and Significant Increases and Decreases in IT Resources The departments we reviewed varied in the extent to which they had documented in their policies and procedures how they are to ensure that the CIO has reviewed IT resources that are to support major program objectives and significant increases and decreases in resources. For example, DOJ and Treasury had documented in their policies and procedures their CIOs’ role in reviewing IT resources that support major program objectives and significant increases and decreases in their resources. However, the other two departments—HHS and DOE—had not documented this role for their CIOs. Specifically, HHS policies and procedures did not include a requirement for the CIO to review significant increases and decreases in IT resources. In addition, the HHS CIO delegated to component-level CIOs the responsibility to review IT resources that support major program objectives for investments of less than $20 million annually. However, HHS had not established procedures for ensuring its components carried out the responsibility, and the component agency we selected—CMS—did not include this requirement in its procedures. Similarly, DOE had not documented procedures for the department-level CIO’s role in reviewing IT resources that support major program objectives and significant increases and decreases in IT resources. For NNSA programs, DOE delegated the responsibility to the NNSA CIO. However, NNSA had not documented the NNSA CIO’s role in reviewing planned IT support for major program objectives, as well as significant increases and decreases in IT resources. Until DOE and HHS develop policies and procedures that include how the CIO is to review whether each investment’s IT resources support major program objectives and have increased or decreased significantly, they will have less assurance that the IT budget request consistently supports the departments’ goals and objectives and that the CIOs have approved significant changes in the budget. Ensure the CIO Has Reviewed Whether the IT Portfolio Includes Appropriate Estimates of All IT Resources Included in the Budget Request None of the four departments had documented in their policies and procedures how their CIOs are to ensure, as part of the IT budget review and approval process, that the IT portfolio includes appropriate estimates of all resources. Specifically, DOE, HHS, Treasury, and DOJ had not documented in their policies and procedures the necessary steps that their CIOs would need to take in order to ensure that the portfolios included the appropriate estimates of all IT resources in the budget requests. In addition, Treasury delegated this responsibility to its component CIOs for component-level investments, and IRS had documented procedures for validating the estimates of all IT resources for the IRS budget request. However, Treasury did not document the necessary steps to ensure that its delegated authorities were being carried out, as required by OMB. Without documented policies and procedures for the steps the CIO is to take to review whether the IT portfolio includes appropriate estimates of all IT resources included in the budget request or delegation plans that outline such activities for component CIOs, the selected departments may be limited in their ability to assure that their CIOs are effectively positioned to consistently and adequately review and approve the IT budget request. Gaps in Departments’ FITARA Implementation and Delegation Plans The shortcomings in the four departments’ policies and procedures related to CIO visibility into IT resources, CIO input into IT resource planning, and CIO review and approval of the IT budget request were due, in part, to having not addressed in their FITARA implementation and delegation plans how they intended to implement the OMB common baseline requirements. For example, none of the four departments’ FITARA implementation plans addressed how they intended to implement the requirement that all transactions related to IT resources be included in planned expenditure reporting to the CIO. These departments’ implementation plans also did not address the requirement that the CIO review whether the IT portfolio includes appropriate estimates of all IT resources identified in the budget request. Officials in DOE’s Office of the CIO stated that the department is in the process of determining ways to add specific review criteria to its capital planning policies and procedures to identify how the department is to review the appropriateness of IT resources in the portfolio. Had such procedures been documented and identified in the department’s FITARA implementation plan, it would have been better positioned to demonstrate how this common baseline requirement is being addressed. In addition, the HHS and Treasury FITARA delegation plans did not address how their CIOs would ensure components carried out their responsibilities for reviewing and approving the IT budget request. Officials in HHS’s Office of the CIO stated that delegation memorandums issued to their components included procedures for ensuring components carried out their responsibilities. However, the delegation plans they provided to us did not include such procedures. Officials in Treasury’s Office of the CIO stated that they did not believe that it was their responsibility to have procedures for verifying that components are carrying out their delegated responsibilities because they viewed it as an audit function. However, having such procedures is called for by OMB’s FITARA implementation guidance. Without FITARA implementation plans that address the shortfalls in policies and procedures for ensuring the implementation of OMB’s common baseline requirements, departments have limited assurance that their CIOs will implement the requirements as intended by OMB and FITARA. In addition, without identifying the steps within the FITARA delegation plans that departments intend to take to ensure the responsibilities delegated to components are appropriately carried out, the departments may have limited assurance that these actions have been taken. Selected Departments Demonstrated Mixed Implementation of IT Budgeting Requirements in Developing Fiscal Year 2017 Budgets for a Sample of Investments While it is important for federal agencies to establish policies and procedures that describe how they are to carry out IT budgeting requirements identified in FITARA and OMB guidance, it is equally important for them to implement the requirements when planning and budgeting for individual IT investments and to retain supporting documentation that would demonstrate that they have done so. Among the eight selected OMB common baseline requirements related to IT budgeting, five of the requirements are applicable at the individual investment level. Table 2 shows how these five common baseline requirements would be implemented at the individual investment level, as well as the related categories. The selected departments and their respective component agencies varied in the extent to which they could demonstrate that they had implemented the five common baseline requirements when developing their fiscal year 2017 funding requests for 16 sampled investments. Figure 2 and the subsequent discussion summarize the extent to which the departments and their component agencies could demonstrate that they had implemented the five requirements in developing fiscal year 2017 budgets for the sample of investments that we reviewed. As described earlier, we reviewed the largest major and non-major investment for each of the four departments and four component agencies. In addition, appendices II through V provide further details about our assessments of the extent to which the departments and component agencies demonstrated that they had implemented the five requirements. CIO Visibility into IT Resources IT Resources for Each Investment Are Described in Order to Inform the CIO during the Planning and Budgeting Processes For the investments that we reviewed, the departments and their components varied in the extent to which they could demonstrate that they had described their investments’ IT resources. For example, DOJ and Treasury described specific IT resources, such as costs for personnel and software, in spreadsheets or databases for processing annual requests for resources for each proposed IT investment. Conversely, HHS and DOE did not fully describe in supporting documentation their respective IT resources for their investments included in our review. For example, HHS could not demonstrate that it had described the department-level non-major investment’s IT resources. In addition, although HHS described non-labor resources that were allocated for a portion of the sampled department-level major investment, the department did not describe labor resources for the investment. HHS also could not account for the investment’s entire funding request— leaving nearly $17 million in resources that were not described. Officials in HHS’s Office of the CIO were unable to explain why supporting documentation for the investment only accounted for a portion of the investment’s total funding request, and not the entire request. HHS and DOE officials provided various reasons as to why their departments did not describe in supporting documentation all of the IT resources associated with the investments we sampled. For example, HHS could not demonstrate that it had described IT resources for the non-major investment that we reviewed because officials in the Office of the CIO did not have the supporting documentation associated with its funding request. In addition, according to officials in HHS’s Office of the CIO, the department’s omission of required labor resources from program office artifacts supporting the funding request for the department-level major investment was an oversight. According to the officials, during the budget formulation cycle, the department did not consistently maintain documentation for its investments that would describe the IT resources and lacked a mature governance process for reviewing the IT resources associated with the investment. Moreover, officials in DOE’s Office of the CIO stated that the department’s budgeting procedures did not call for clearly identifying specific IT resources. However, at a minimum, DOE’s budgeting procedures required that the budget estimate for investments include planned government labor expenditures. Until HHS and DOE describe IT resources within their investments, the CIO may have limited visibility into what the resources are that are being requested in the annual IT budget. CIO Input into IT Resource Planning The CIO Is Included in the Planning and Budgeting Stages for Investments with IT Resources The extent to which each of the four departments’ could demonstrate that their CIOs were included in the planning and budgeting stages for the sampled investments with IT resources varied. Specifically, of the four investments we reviewed for each agency, DOJ and its component included the CIO in the planning and budgeting stages via an annual IT portfolio review that included the four sampled investments. On the other hand, HHS, Treasury, and DOE—along with their components—could not always demonstrate that the department-level CIO was included for their investments. For example, within HHS, its component agency—CMS—partially implemented the requirement for both of the sampled investments. Specifically, CMS documented the CIO’s review and approval of each investment’s detailed IT resource estimates during governance board reviews. However, HHS’s supporting documentation did not demonstrate that the department CIO was involved in the planning process for these investments even though its capital planning and investment control policy required this official to review, validate, and approve these IT investments through the department-level review board. Further, Treasury could not demonstrate that the department’s CIO was included in the planning and budgeting stages for the two department-level investments that we reviewed. According to officials in Treasury’s Office of the CIO, the relevant documentation was not retained for the selected department-level investments because procedures were not in place to document reviews by the CIO and certain artifacts that may have documented such reviews were no longer available in part due to employee turnover within the program offices responsible for the investments. Until DOE, HHS, and Treasury include the CIO in the planning and budgeting stages for investments with IT resources, they may be at risk of duplicating resources or funding investments without the CIO’s knowledge or approval. Program Leadership Works with the CIO to Plan the Investment’s IT Resources The selected departments varied in their ability to demonstrate that their CIOs worked with program leadership across the investments we sampled for the fiscal year 2017 funding request—both within and across the departments. For example, DOJ demonstrated that the CIO worked with program leaders in planning IT resources for both the major and non-major investment at the department level by jointly developing a plan for how business units were to utilize funds for IT services. At the component-level, FBI demonstrated that IT officials assisted program leadership in the planning of the major investment, but could not demonstrate that the CIO worked with program leadership on both the major and non-major investment. In addition, HHS fully demonstrated that the CIO worked with program leaders in planning IT resources for its major investment. For example, the CIO reviewed detailed IT resource narratives and line item estimates for the investment at a department-level governance board meeting with program leadership. However, HHS could not demonstrate that the CIO worked with program leadership to plan the non-major investment. At the component-level, CMS partially demonstrated that the CIO took such actions to plan the component-level investments. Specifically, the CIO at CMS worked with program officials to review and approve detailed IT resource requests for the investments. However, HHS could not demonstrate that its CIO was also involved in planning IT resources with program leadership for the same investments, as required by the department’s policy. Officials in the Office of the CMS CIO stated that they believed that the CMS CIO was an authorized delegate for this responsibility. However, the officials could not provide documentation of the delegation as required by OMB. Further, Treasury could not demonstrate that the CIO had worked with program leadership in planning IT resources for the department-level investments. According to officials in the Office of the CIO, they could not demonstrate the actions the CIO took to work with program leadership because documentation that would show the interaction was not retained. The officials stated that documentation was not retained due to turnover within the program offices responsible for the investments. At the component-level, IRS partially demonstrated that the CIO took action to work with program leadership for a portion of the component-level investments’ budget through IT budget reviews. However, IRS could not demonstrate coordination with program leadership for the full amount of the investments’ budget because the agency did not maintain a document trail for lower-level budgeting activities that included all relevant resource planning for the investments. Lastly, DOE could not fully demonstrate that the CIO had worked with program leadership in planning IT resources across all four investments at the department and component. DOE could not demonstrate this, in part, because the Office of the CIO’s internal process, during the formulation of the fiscal year 2017 budget, did not require input from all relevant stakeholders, including senior leadership, directors, and program managers. Officials in DOE’s Office of the CIO acknowledged the gap in its process and stated that the department and its component agency— NNSA—are working to establish processes that include senior management and program officials in the planning process. As of May 2018, DOE did not have a time frame for establishing these processes. The lack of consistent partnership of program leaders and the CIO to plan an investment’s IT resources at the department and component levels limits the ability of the CIO to have a significant role in the formulation of the department’s IT budget. CIO Review and Approval of IT Budgets The CIO Reviews Whether the Investment’s IT Resources Support Major Program Objectives and Have Increased or Decreased Significantly The selected departments varied in the extent to which they could demonstrate that the CIO had appropriately reviewed all the investments we sampled. For example, DOJ demonstrated that the CIO reviewed whether the IT resources for the department- and component-level investments supported major program objectives and whether there were increases and decreases in IT resources for the investments. In addition, HHS partially addressed the requirement for its component-level investments. Specifically, while the component-level CIO at CMS reviewed changes in the investments’ resources, supporting documentation did not show that alignment with major program objectives was reviewed. Further, Treasury and DOE could not demonstrate that their CIOs reviewed whether the investment’s IT resources support major program objectives and any significant increases or decreases in resources for their department-level investments. According to officials in the offices of the CIO at Treasury and DOE, relevant documentation that would have demonstrated review activities had not been maintained for the investments. Until DOE, HHS, and Treasury can consistently demonstrate that the CIO has reviewed whether each investment’s IT resources support major program objectives and have increased or decreased significantly, the departments will have less assurance that the IT budget request supports their goals and objectives and that significant changes in the budget are appropriate. The CIO Reviews Whether the Investment’s Estimates of IT Resources in the Portfolio and Budget Request Are Appropriate The selected departments varied in the extent to which they could demonstrate that the CIO took steps to review whether the investment’s estimates of IT resources in the portfolio and budget request were appropriate. For example, the CIO for DOE’s component agency—NNSA— demonstrated the review and approval of the non-major investment’s estimates of IT resources. However, NNSA could not demonstrate that the CIO reviewed the estimates for the major investment because it did not retain documentation that would provide details on the investment’s budget formulation and approval. In addition, HHS’s component agency—CMS—partially demonstrated implementation of the requirement on the major investment. Specifically, the CIO for CMS reviewed and approved supporting documentation for the investment’s detailed resource estimates totaling more than $500 million in developing the fiscal year 2017 budget request. However, the fiscal year 2017 budget request for this investment was $399 million, and according to officials in the CMS Office of the CIO, the CMS CIO did not review and approve the lowered estimate—ensuring the IT portfolio reflected an appropriate estimate. According to CMS officials in the Office of the CIO, the lowered estimate was the result of the user fees portion of the investment being removed from the request before it was submitted to OMB because it was not funded by annual appropriations. However, OMB’s fiscal year 2017 IT capital planning guidance required departments to report all budgetary sources of funding for each investment, including amounts available for obligation through collection of fees, as well as annual appropriations. Further, Treasury could not demonstrate that the CIO had reviewed the resource estimates for the department-level investments. At the component level, IRS demonstrated that officials in the Office of the CIO reviewed supporting documentation for detailed cost estimates for the component-level investments. However, these cost estimates only accounted for a portion, and not the full amount, of the investment. Finally, DOJ could not demonstrate that the component agency CIO ensured that the IT portfolio included appropriate estimates of all IT resources for the non-major investment at the component level. While officials in the FBI’s Office of the CIO stated that the component’s CIO was involved in reviewing detailed resource estimates for the investment prior to its submission to the department-level CIO, they could not provide supporting documentation because the FBI had not established procedures that explicitly required documenting the performance of this activity. Until the CIOs at DOE, HHS, Treasury, and DOJ consistently review IT resource estimates for each investment, departments will have less assurance that the estimates in the budget request are appropriate. Departments Lacked Processes for Ensuring That Their IT Budgets Are Informed by Reliable Costs; the Administration Has Introduced an Initiative to Improve Cost Visibility GAO and international standards recommend certain quality assurance practices that can assist departments in developing an IT budget that is informed by reliable cost information. These practices include, among others: (1) ensuring government labor costs have been accurately reported for all investments, (2) aligning contract costs with the investments, and (3) utilizing budget object class data to capture all IT programs. Further, having documented IT capital planning processes to implement these practices is important because OMB requires department CIOs to fully account for and report on planned expenditures in their annual IT budget requests. All of the four selected departments—DOE, HHS, DOJ, and Treasury— lacked quality assurance processes to ensure government labor costs have been accurately reported, align contract costs with IT investments, and utilize budget data to capture all IT programs. However, OMB’s fiscal year 2019 IT capital planning guidance introduced several major changes to the budgeting process which, if effectively implemented, should provide departments and CIOs with enhanced visibility into IT costs across the portfolio and additional assurance that the budget is being informed by all relevant IT costs. The Administration’s Efforts to Revise IT Budget Reporting Could Provide Department CIOs Additional Visibility into IT Spending OMB’s fiscal year 2019 IT capital planning guidance, released in August 2017, introduced several major changes to the federal IT budgeting process, including the practice of using a set of low-level cost categories to group spending. Subsequently, the President’s Management Agenda, released in March 2018, identified “improving outcomes through federal IT spending transparency” as one of the Administration’s 14 cross-agency priority goals. According to the President’s Management Agenda, the Administration intends to accomplish the cross-agency priority goal related to improving federal IT spending transparency by, among other things, increasing the granularity in IT budget reporting by utilizing a set of cost categories from OMB’s capital planning guidance. Figure 3 identifies the standard cost categories that OMB plans to implement in IT budget reporting. In its fiscal year 2019 IT capital planning guidance, OMB has recognized potential value in utilizing budget object classes, or similar financial data, to provide CIOs and Chief Financial Officers additional visibility into costs that inform the budget. As noted in the guidance, OMB expects that this new approach for utilizing financial data to inform the IT investment portfolio will enable the reconciliation of this portfolio with the department’s budget submitted by the Chief Financial Officer. Further, this effort is to help CIOs work more closely and in partnership with the Chief Financial Officers by using budget object classes and sub-object classes as a way to reconcile different presentations of estimated costs. In addition, OMB’s guidance stated that departments should begin to identify where they lack capabilities or resources to deliver financial data for the new low-level IT cost categories (shown in figure 3), consider what changes are necessary to achieve the new reporting requirements, and take steps to align reporting with the categories. Moreover, the President’s Management Agenda stated that the changes to how IT spending is to be categorized were made, in part, because federal executives have long known that they could better manage the more than $90 billion in federal government IT spending with increased visibility and more accurate data. The President’s Management Agenda action plan identified several milestones and due dates for accomplishing the goal of improving federal IT spending transparency, such as determining data sources necessary for departments to report within the low-level cost categories and establishing the common tools and services for the required reporting by June 2019. Moreover, the action plan stated that federal departments are expected to report all of the spending within their IT portfolio against the cost categories by September 2019. Given that improving federal IT spending transparency has been identified as one of the President’s top 14 management priorities and is critical to enabling department CIOs in carrying out their IT budgeting authorities from FITARA, it is important that OMB and departments take action now in order to meet the 2019 reporting requirements. The Administration’s approach for obtaining additional granularity on department IT investment spending, when implemented, should provide departments and CIOs enhanced visibility into IT costs across the portfolio. If implemented effectively, this approach could also provide departments additional assurance that their budgets are being informed by relevant IT costs. Conclusions Recognizing the importance of CIOs’ ability to be responsible for IT budgeting, OMB’s common baseline includes eight requirements that departments’ policies and procedures should address to implement FITARA. While the four selected departments in our review either fully or partially addressed the majority of the requirements, none fully addressed all of them. The lack of policies and procedures was due, in part, to the fact that departments had not adequately addressed all of the required common baseline requirements in their FITARA implementation and delegation plans, as directed by OMB. Until the departments establish policies and procedures that address all requirements, they risk inconsistently applying requirements that are key to providing their CIOs visibility into resources, input to resource plans, and meaningful review and approval of IT budgets. In addition, the lack of policies and procedures has hampered the departments’ ability to demonstrate their implementation of the common baseline requirements for their investments. While DOJ fully demonstrated implementation for the selected requirements for the majority of the investments we sampled, HHS and Treasury partially demonstrated implementation for a majority of their investments, and DOE had not demonstrated implementation for the majority of its investments. As a result, departments were not always able to show that these CIOs had adequate input to resource plans and review of their IT budgets. Without retaining supporting documentation to show how common baseline requirements have been addressed on individual investments, the departments will be challenged in consistently demonstrating that CIOs are sufficiently involved in planning and budgeting annual IT expenditures. Finally, the four selected departments lacked quality assurance processes for ensuring their IT budgets are informed by reliable cost information. This resulted in billions of dollars that were requested without departments having comprehensive information to support those requests. Among other things, this was due to a lack of processes for periodically reviewing data quality and estimation methods for government labor estimates, as well as a lack of processes to cross-walk IT spending data in their procurement and accounting systems with investment data in their IT portfolio management systems. The Administration’s new approach of using a standard set of low-level cost categories to group IT spending could help departments address their lack of processes if properly implemented. It is important that OMB and departments meet the 2019 milestone dates associated with this approach so that department CIOs have additional transparency into IT spending and can make informed budget decisions. Nonetheless, departments will continue to have limited insight into IT budgeting until they capture all relevant IT costs in their budgets. Recommendations for Executive Action We are making a total of 43 recommendations, including 9 to DOE, 6 to NNSA, 10 to HHS, 4 to CMS, 4 to DOJ, 1 to FBI, 8 to Treasury, and 1 to IRS. The Secretary of Energy should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that establish department-wide policy for the level of detail of planned expenditure reporting to the CIO for all transactions that include IT resources. (Recommendation 1) The Secretary of Energy should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that include the CIO in the planning and budgeting stages for all programs that are fully or partially supported with IT resources. (Recommendation 2) The Secretary of Energy should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that include the CIO as a member of governance boards that inform decisions regarding all IT resources, including component-level boards. (Recommendation 3) The Secretary of Energy should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that document the processes by which program leadership works with the CIO to plan an overall portfolio of IT resources. (Recommendation 4) The Secretary of Energy should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that document the process for the CIO’s review and approval of the major IT investments portion of the budget request. (Recommendation 5) The Secretary of Energy should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that document the CIO’s role in reviewing IT resources that are to support major program objectives and significant increases and decreases in IT resources. (Recommendation 6) The Secretary of Energy should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that document the steps the CIO is to take to ensure whether the IT portfolio includes appropriate estimates of all IT resources included in the budget request. (Recommendation 7) The Secretary of Energy should direct the Office of the CIO and other offices, as appropriate, to take steps to ensure that the actions taken to comply with OMB’s common baseline for implementing FITARA on individual investments are adequately documented. (Recommendation 8) The Secretary of Energy should ensure that the Office of the CIO and other offices, as appropriate, establish quality assurance processes— such as data quality checks, reviews of estimation methods, linkages between the IT portfolio and procurement system data, and linkages between the IT portfolio and financial system data—for ensuring the annual IT budget is informed by complete and reliable information on anticipated government labor, contract, and other relevant IT expenditures. (Recommendation 9) The Administrator of NNSA should ensure that the Office of the CIO and other offices, as appropriate, develop and implement policies and procedures that establish agency-wide policy for the level of detail with which planned expenditures for all transactions that include IT resources are to be reported to the CIO. (Recommendation 10) The Administrator of NNSA should ensure that the Office of the CIO and other offices, as appropriate, develop and implement policies and procedures that include the CIO in the planning and budgeting stages for all programs that are fully or partially supported with IT resources. (Recommendation 11) The Administrator of NNSA should ensure that the Office of the CIO and other offices, as appropriate, develop and implement policies and procedures that include the CIO as a member of governance boards that inform decisions regarding all IT resources. (Recommendation 12) The Administrator of NNSA should ensure that the Office of the CIO and other offices, as appropriate, develop and implement policies and procedures that document the process for the CIO’s review and approval of the major IT investments portion of the budget request. (Recommendation 13) The Administrator of NNSA should ensure that the Office of the CIO and other offices, as appropriate, develop and implement policies and procedures that document the CIO’s role in reviewing IT resources that are to support major program objectives and significant increases and decreases in IT resources. (Recommendation 14) The Administrator of NNSA should direct the Office of the CIO and other offices, as appropriate, to take steps to ensure that the actions taken to comply with OMB’s common baseline for implementing FITARA on individual investments are adequately documented. (Recommendation 15) The Secretary of Health and Human Services should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that establish department-wide policy for the level of detail of planned expenditure reporting to the CIO for all transactions that include IT resources. (Recommendation 16) The Secretary of Health and Human Services should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that include the CIO in the planning and budgeting stages for all programs that are fully or partially supported with IT resources. (Recommendation 17) The Secretary of Health and Human Services should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that include the CIO as a member of governance boards that inform decisions regarding all IT resources, including component-level boards. (Recommendation 18) The Secretary of Health and Human Services should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that document the processes by which program leadership works with the CIO to plan an overall portfolio of IT resources. (Recommendation 19) The Secretary of Health and Human Services should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that document the process for the CIO’s review and approval of the major IT investments portion of the budget request. (Recommendation 20) The Secretary of Health and Human Services should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that document the CIO’s role in reviewing IT resources that are to support major program objectives and significant increases and decreases in IT resources. (Recommendation 21) The Secretary of Health and Human Services should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that document the steps the CIO is to take to ensure whether the IT portfolio includes appropriate estimates of all IT resources included in the budget request. (Recommendation 22) The Secretary of Health and Human Services should direct the department CIO to establish, for any OMB common baseline requirements that are related to IT budgeting that have been delegated, a plan that specifies the requirement being delegated, demonstrates how the CIO intends to retain accountability for the requirement, and ensures through quality assurance processes that the delegated official will execute such responsibilities with the appropriate level of rigor. (Recommendation 23) The Secretary of Health and Human Services should direct the Office of the CIO and other offices, as appropriate, to take steps to ensure that the actions taken to comply with OMB’s common baseline for implementing FITARA on individual investments are adequately documented. (Recommendation 24) The Secretary of Health and Human Services should ensure that the Office of the CIO and other offices, as appropriate, establish quality assurance processes—such as data quality checks, reviews of estimation methods, linkages between the IT portfolio and procurement system data, and linkages between the IT portfolio and financial system data—for ensuring the annual IT budget is informed by complete and reliable information on anticipated government labor, contract, and other relevant IT expenditures. (Recommendation 25) The Administrator of CMS should ensure that the Office of the CIO and other offices, as appropriate, develop and implement policies and procedures that include the CIO in the planning and budgeting stages for all programs that are fully or partially supported with IT resources. (Recommendation 26) The Administrator of CMS should ensure that the Office of the CIO and other offices, as appropriate, develop and implement policies and procedures that document the processes by which program leadership works with the CIO to plan an overall portfolio of IT resources. (Recommendation 27) The Administrator of CMS should ensure that the Office of the CIO and other offices, as appropriate, develop and implement policies and procedures that document the CIO’s role in reviewing IT resources that are to support major program objectives and significant increases and decreases in IT resources. (Recommendation 28) The Administrator of CMS should direct the Office of the CIO and other offices, as appropriate, to take steps to ensure that the actions taken to comply with OMB’s common baseline for implementing FITARA on individual investments are adequately documented. (Recommendation 29) The Attorney General should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that establish department-wide policy for the level of detail of planned expenditure reporting to the CIO for all transactions that include IT resources. (Recommendation 30) The Attorney General should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that include the CIO as a member of governance boards that inform decisions regarding all IT resources, including component-level boards. (Recommendation 31) The Attorney General should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that document the steps the CIO is to take to ensure whether the IT portfolio includes appropriate estimates of all IT resources included in the budget request. (Recommendation 32) The Attorney General should ensure that the Office of the CIO and other offices, as appropriate, establish quality assurance processes—such as data quality checks, reviews of estimation methods, linkages between the IT portfolio and procurement system data, and linkages between the IT portfolio and financial system data—for ensuring the annual IT budget is informed by complete and reliable information on anticipated government labor, contract, and other relevant IT expenditures. (Recommendation 33) The FBI Director should direct the Office of the CIO and other offices, as appropriate, to take steps to ensure that the actions taken to comply with OMB’s common baseline for implementing FITARA on individual investments are adequately documented. (Recommendation 34) The Secretary of the Treasury should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that establish department-wide policy for the level of detail of planned expenditure reporting to the CIO for all transactions that include IT resources. (Recommendation 35) The Secretary of the Treasury should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that include the CIO in the planning and budgeting stages for all programs that are fully or partially supported with IT resources. (Recommendation 36) The Secretary of the Treasury should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that include the CIO as a member of governance boards that inform decisions regarding all IT resources, including component-level boards. (Recommendation 37) The Secretary of the Treasury should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that document the process for the CIO’s review and approval of the major IT investments portion of the budget request. (Recommendation 38) The Secretary of the Treasury should ensure that the Office of the CIO and other offices, as appropriate, address gaps in the department’s FITARA plans by developing and implementing policies and procedures that document the steps the CIO is to take to ensure whether the IT portfolio includes appropriate estimates of all IT resources included in the budget request. (Recommendation 39) The Secretary of the Treasury should direct the department CIO to establish, for any OMB common baseline requirements that are related to IT budgeting that have been delegated, a plan that specifies the requirement being delegated, demonstrates how the CIO intends to retain accountability for the requirement, and ensures through quality assurance processes that the delegated official will execute such responsibilities with the appropriate level of rigor. (Recommendation 40) The Secretary of the Treasury should direct the Office of the CIO and other offices, as appropriate, to take steps to ensure that the actions taken to comply with OMB’s common baseline for implementing FITARA on individual investments are adequately documented. (Recommendation 41) The Secretary of the Treasury should ensure that the Office of the CIO and other offices, as appropriate, establish quality assurance processes—such as data quality checks, reviews of estimation methods, linkages between the IT portfolio and procurement system data, and linkages between the IT portfolio and financial system data—for ensuring the annual IT budget is informed by complete and reliable information on anticipated government labor, contract, and other relevant IT expenditures. (Recommendation 42) The IRS Commissioner should direct the Office of the CIO and other offices, as appropriate, to take steps to ensure that the actions taken to comply with OMB’s common baseline for implementing FITARA on individual investments are adequately documented. (Recommendation 43) Agency Comments and Our Evaluation We provided a draft of this report to the four departments and four component agencies included in our review, as well as to OMB. In response, we received comments from two departments and three component agencies (HHS, CMS, DOJ, FBI, and IRS) which agreed with our recommendations. One department (DOE) partially agreed with one recommendation and agreed with the other recommendations made to it, as well as with the recommendations made to its component agency (NNSA). In addition, one department (Treasury) neither agreed nor disagreed with the recommendations. Further, OMB provided technical comments, which we incorporated in the report, as appropriate. The following departments and component agencies agreed with all of the recommendations that we directed to them: HHS provided written comments, reprinted in appendix VI, stating that it concurred with the 10 recommendations made to the department, and with the 4 recommendations made to CMS. Of the recommendations made to the department, HHS stated that the processes it currently has in place address the various gaps in the department’s FITARA plans, documentation, and quality assurance processes. However, HHS did not provide additional evidence to demonstrate that the weaknesses we identified have been mitigated. Thus, we maintain that the department needs to take further actions to address our recommendations. Until it takes the appropriate actions to address gaps in its FITARA plans, document the actions taken to comply with OMB’s guidance, and implement key quality assurance processes, the department will be at increased risk that its CIO is not effectively engaged in IT budgeting decisions. With regard to its component agency, HHS stated that CMS would take action to implement the recommendations made to it by updating the relevant policies and procedures to more explicitly identify the role of the CIO in developing the IT budget. In comments provided via email on September 27, 2018, an audit liaison in the Internal Review and Evaluation Office of the Justice Management Division stated that DOJ concurred with the four recommendations made to the department, and with the one recommendation made to FBI. IRS provided written comments, reprinted in appendix VII, stating that it concurred with our recommendation, has taken steps to begin implementing our recommendation, and is committed to making further progress toward fully implementing all OMB requirements when planning and budgeting for its individual investments. DOE provided written comments, reprinted in appendix VIII, in which it concurred with eight of the nine recommendations made to the department and partially concurred with one recommendation. The department also concurred with all six recommendations made to NNSA. Of the nine recommendations made to DOE, the department stated that it already had processes in place, or had taken action to address six of the recommendations, including the recommendation with which it partially concurred. However, the department did not provide sufficient evidence to demonstrate that the weaknesses we identified had been mitigated. Thus, we maintain that the recommendations warrant further actions. Until DOE takes the appropriate actions to address gaps in its FITARA plans, document the actions taken to comply with OMB’s guidance, and implement key quality assurance processes, the department will be at increased risk that the CIO is not effectively engaged in IT budgeting decisions. In addition, DOE stated that NNSA’s Office of the CIO plans to develop policies and procedures—in collaboration with the component agency’s Office of Acquisition and Project Management and the agency’s Office of Management and Budget—that should address the findings and six recommendations made to NNSA. The department anticipates that the policies and procedures will be finalized by March 31, 2019. Lastly, Treasury responded via email on September 28, 2018, but did not state whether it agreed or disagreed with our eight recommendations. Specifically, an audit liaison in Treasury’s Office of the CIO stated that the department believes it is implementing most of the OMB common baseline requirements in practice, but agreed that gaps exist in its policies and documentation. The official added that the department had started work on strengthening existing policies and procedures or developing new ones to close the gaps uncovered by our review. We are sending copies of this report to the appropriate congressional requesters; OMB; the Secretaries of the Departments of Energy, Health and Human Services, and the Treasury; the Attorney General; the Administrator of NNSA, the Administrator of CMS, the FBI Director, and the IRS Commissioner. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-4456 or at harriscc@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix IX. Appendix I: Objectives, Scope, and Methodology Our objectives were to determine the extent to which selected federal agencies (1) established policies and procedures in place that address the information technology (IT) budgeting requirements of the Federal Information Technology Acquisition Reform Act (FITARA) and related Office of Management and Budget (OMB) IT budget guidance, (2) could demonstrate that they had developed fiscal year 2017 IT budgets for sampled investments consistent with FITARA and OMB guidance, and (3) implemented processes to ensure that annual IT budgets are informed by reliable cost information. To address our objectives, we first identified the subset of the 24 agencies covered by the Chief Financial Officers Act that had a fiscal year 2017 IT budget request of at least $1 billion. From this subset, we then identified the four agencies having the two highest and the two lowest average initial FITARA self-assessment scores, which included an assessment of OMB’s common baseline for IT management (common baseline). In the event that one or more agencies had the same average self-assessment scores, we selected the agency with the largest fiscal year 2017 IT budget. Based on these criteria, we selected four departments for our review: (1) the Department of Energy (DOE), (2) the Department of Health and Human Services (HHS), (3) the Department of Justice (DOJ), and (4) the Department of the Treasury (Treasury). In addition, for each of these four departments, we selected their component agencies that had the largest fiscal year 2017 IT budget request. The components within the four selected departments were: (1) the National Nuclear Security Administration (NNSA) within DOE, (2) the Centers for Medicare and Medicaid Services (CMS) within HHS, (3) the Federal Bureau of Investigation (FBI) within DOJ, and (4) the Internal Revenue Service (IRS) within Treasury. For the first objective, we compared the selected departments’ policies and procedures to requirements selected from OMB’s FITARA guidance (referred to as the common baseline) that related to developing departments’ IT budgets. In selecting the requirements, we reviewed 10 areas related to budget formulation and execution within OMB’s common baseline, and used professional judgment to select 8 requirements that we believed would significantly impact the development and approval of departments’ annual IT budgets. In doing so, we excluded one requirement that affected the development of annual IT budgets to a lesser extent and combined one requirement that was similar to another. Specifically, we excluded the requirement from the area related to the chief information officer’s (CIO) role in program management because the CIO’s review of program management artifacts could not be directly related to the IT budget review and approval process. In addition, two common baseline areas had a similar requirement that the CIO be involved in the internal planning of IT resources prior to the budget submission. We combined these into one requirement for our review. We consulted with OMB officials in the Office of the Federal CIO on the requirements that we selected and how we planned to evaluate them and the officials agreed with our approach. The eight OMB common baseline requirements within budget formulation and execution that we identified and selected are: establish the level of detail with which IT resources are to be described in order to inform the CIO during the planning and budgeting processes; establish agency-wide policy for the level of detail with which planned expenditures for all transactions that include IT resources are to be reported to the CIO; include the CIO in the planning and budgeting stages for programs that are fully or partially supported with IT resources; include the CIO as a member of governance boards that inform decisions regarding all IT resources, including component-level governance boards; document the processes by which program leadership works with the CIO to plan an overall portfolio of IT resources; ensure the CIO has reviewed and approved the major IT investments portion of the budget request; ensure the CIO has reviewed IT resources that are to support major program objectives and significant increases and decreases in IT resources; and ensure the CIO has reviewed whether the IT portfolio includes appropriate estimates of all IT resources included in the budget request. After determining the eight requirements that we would review, we categorized them into three areas: CIO visibility into IT resources, CIO input to IT resource plans, and CIO review and approval of IT budgets. We then reviewed the current policies and procedures that each department had documented for its IT budgeting process to determine whether the department documented a process for how they would address each of the eight common baseline requirements we selected for review. In addition to policies and procedures, we also reviewed each department’s FITARA implementation plan, which included a description of the steps the department must take to ensure that all FITARA and OMB requirements would be implemented; and the delegation memorandums from department CIOs, in which formal assignments of responsibilities to other department officials are documented, where applicable. In cases where the department CIO fully or partially delegated responsibilities to component officials, we requested relevant documentation from the agency component with the largest fiscal year 2017 IT budget request. In such cases, we based our determination of departments’ implementation of the requirement on (1) the extent to which the component agency had documented policies and procedures that carried out the delegated requirement and (2) the extent to which the department CIO had procedures for ensuring the delegation was being carried out by the components. With regard to our second objective, we determined whether the selected departments had implemented key IT budgeting requirements for a non-generalizable sample of investments in their fiscal year 2017 budget formulation. In doing so, we chose 16 investments—the largest major and non-major investments at the department level and the largest major and non-major investments at the component level—based on the selected departments’ fiscal year 2017 IT budget request. Although the information obtained is not generalizable to all of the departments’ investments, the sample provided a range of examples and conditions under which the departments were implementing requirements found in OMB’s common baseline. We then identified a subset of requirements from the eight department requirements for developing IT budgets found in OMB’s common baseline for which implementation could be observed at the investment level. In doing so, we used professional judgment to select the subset of requirements where actions taken to implement these requirements could be observed for individual investments. Specifically, we did not select the requirement to establish agency-wide policy for the level of detail with which planned expenditures for all transactions that include IT resources are to be reported to the CIO, because the requirement primarily applied to enterprise-wide policymaking and reporting. Also, we did not select the requirement to include the CIO as a member of governance boards that inform decisions regarding all IT resources for review on individual investments because certain investments may not have been subject to governance board reviews during fiscal year 2017. In addition, we did not select the requirement to ensure the CIO has reviewed and approved the major IT investments portion of the budget request for review on individual investments because half of the investments we selected were not classified as major investments. The five requirements for which we selected and reviewed implementation at the investment level were: IT resources for each investment are described in order to inform the CIO during the planning and budgeting processes; the CIO is included in the planning and budgeting stages for investments with IT resources; program leadership works with the CIO to plan the investment’s IT the CIO reviews whether the investment’s IT resources support major program objectives and have increased or decreased significantly; and the CIO reviews whether the investment’s estimates of IT resources in the portfolio and budget request are appropriate. For each investment, if available, we obtained artifacts for the fiscal year 2017 budget submission—such as briefings, reports, meeting minutes, memorandums, and other relevant documentation showing the CIO’s involvement in relevant reviews or decisions. We compared this documentation to relevant OMB requirements for developing the IT budget at the individual investment level. To address our third objective, we assessed the selected departments’ efforts to develop their fiscal year 2017 budget with reliable cost information by comparing the department’s IT capital planning and budgeting processes against best practices identified by us and the International Organization for Standardization—such as capturing government labor costs, aligning contract costs with investments, and utilizing budget object class data. We selected these three practices (from among others) because of their potential to inform the development of a complete and accurate IT budget for a federal department. Capturing government labor costs. We reviewed each selected departments’ IT capital planning policies and procedures and government labor estimates within the 2017 IT budget submission reported on the Federal IT Dashboard. For each selected department, we identified the processes by which forecasted government labor costs are to be captured within investment proposals submitted with the annual IT budget. We then analyzed each department’s 2017 IT investment proposals to determine whether the department was capturing government labor for each investment. In doing so, we analyzed each selected department’s IT portfolio submitted with its fiscal year 2017 budget to determine whether the identified investments had included planned government labor costs, as required by OMB. Aligning contract costs with investments. We reviewed each selected departments’ IT capital planning policies and procedures, contract-related information within the 2017 IT budget submission reported on the Federal IT Dashboard, and contract obligation data reported within the Federal Procurement Data System-Next Generation. From these document reviews, we identified the processes by which contract-related costs are to be captured within investment proposals submitted with the annual IT budget. We then determined whether departments were able to align current contracts with 2017 IT investment proposals. In doing so, we identified fiscal year 2016 contracts that departments reported in the Federal Procurement Data System-Next Generation that had an IT-related product or service code and an expected completion date that extended into fiscal year 2017 or beyond. We then attempted to match, using the unique procurement identification number for each contract, a corresponding IT investment for those contracts in departments’ fiscal year 2017 IT budget data. For contracts that we could not find a match, or alignment, with investments in departments’ fiscal year 2017 IT budget data, we identified dollars obligated on those contracts from October 2016 through September 2017. Utilizing budget object class data. We reviewed each selected departments’ IT capital planning policies and procedures, budget object classes that are to help track IT financial transactions, and OMB’s fiscal year 2019 IT capital planning guidance that calls for greater use of IT financial data. We then assessed whether departments’ IT capital planning processes utilized budget object class information to ensure that relevant IT costs are being captured as investments for the annual IT budget. We assessed the reliability of reported government labor costs by identifying instances in which investments had not included planned government labor costs and by corroborating those instances with officials in the departments’ offices of the CIO. We determined that the data were sufficiently reliable for our purposes. Where we identified data quality issues in capturing government labor costs for department investments, we included those in the findings of this report. We also assessed the reliability of Federal Procurement Data System-Next Generation data by performing electronic testing of selected data elements and reviewing existing information about the system and the data it produces. Specifically, we reviewed the data dictionary, data validation rules, and the fiscal year 2016 Federal Government Procurement Data Quality Summary for agency data in the Federal Procurement Data System-Next Generation. We determined that the data were sufficiently reliable for our purposes. We supplemented our review with interviews with officials in the departments’ offices of the CIO, Chief Financial Officer, and program offices to include discussions of our observations of any shortfalls in their processes. We conducted this performance audit from January 2017 to November 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Department of Energy Detailed Analysis DOE established department-wide IT capital planning and budgeting procedures that document the level of detail with which IT resources are to be described during the planning and budgeting process. The documented level of detail included OMB’s required reporting on government labor and certain resources for infrastructure investments. This requirement was assigned at the department level. DOE documented IT capital planning procedures for the level of detail of planned expenditure reporting. However, the procedures did not explicitly require that all transactions with an IT component be included in the expenditure reporting. DOE’s FITARA implementation plan stated that the NNSA CIO was to work with the chief financial officer and chief acquisition officer to update NNSA’s policies to ensure they documented the level of detail for planned expenditure reporting for all IT transactions. However, NNSA had not established such policies and procedures. DOE documented IT capital planning and annual budgeting procedures that included the CIO in the planning and budgeting stages for some, but not all programs identified as having IT resources. Specifically, at the time of our review, the department had not yet documented procedures for ensuring the CIO was included in budget decisions for all programs with IT resources, including those within NNSA and national laboratories. According to DOE’s FITARA implementation plan and a memorandum from the Secretary providing further instructions on FITARA, NNSA was to provide an opportunity for the department CIO to provide input in its planning and budgeting stages for programs with IT resources and to document related review processes. However, at the time of our review, NNSA had not yet established the procedures that were to detail how this process was to be carried out. DOE developed charters that included the CIO as a member of department-level IT governance boards, but had not included the CIO as a member of component-level IT investment review boards. According to DOE’s FITARA implementation plan, NNSA was to allow the DOE CIO to provide input into NNSA acquisition decisions through its IT investment review board. However, at the time of our review, NNSA had not yet finalized its investment review board charter and related procedures to include the DOE CIO. Department Rating GAO’s assessment DOE DOE documented IT governance board procedures by which the CIO is to work with program leadership in planning IT resources for some, but not all of the programs subject to department-level governance board reviews. Specifically, at the time of our review, the department had not yet documented procedures by which the CIO was to work with program leadership in planning IT resources within NNSA and national laboratories, including high-performance computing efforts. This requirement was assigned at the department level. DOE documented IT capital planning and governance board procedures for its CIO to review and approve some, but not all of its major IT investments. Specifically, at the time of our review, the department had not yet documented procedures for the CIO to review major investments within NNSA and national laboratories, including high-performance computing efforts. According to DOE’s FITARA implementation plan, the NNSA CIO was to review and approve NNSA major IT investments and provide the DOE CIO an opportunity to review and provide input prior to the final decision. However, at the time of our review, procedures to do so had not yet been established. At the time of our review, DOE had not yet documented procedures for reviewing IT resources that are to support major program objectives and significant increases and decreases in IT resources for other department and component agency budget requests. DOE delegated the responsibility to meet this requirement to NNSA for their programs. According to DOE’s FITARA implementation plan, the NNSA CIO was to review and approve NNSA major IT investments and provide the DOE CIO an opportunity to review and provide input prior to the final decision. However, at the time of our review, procedures to do so had not yet been established. In addition, NNSA had not documented procedures by which the NNSA CIO was to review IT resources that are to support major program objectives as well as significant increases and decreases in IT resources. DOE had not documented procedures for this requirement. This requirement was assigned at the department level. ● = The department provided documentation that demonstrated that the department or delegated component agency satisfied the OMB common ◑ = The department provided documentation that demonstrated that the department or delegated component agency satisfied some but not all of the ○ = The department could not provide documentation that demonstrated that the department or delegated component agency satisfied any of the OMB common baseline requirement. N/A = The component agency was not officially delegated the identified responsibilities for the OMB common baseline requirement. GAO’s assessment NNSA could not demonstrate that the CIO reviewed the investment’s alignment with major program objectives or changes in resources. The CIO reviewed IT resources that are to support major program objectives through business case materials that described the goals that the investment supported. In addition, the CIO reviewed decreases in the annual resource requirements by reviewing acquisition planning artifacts that included the investment’s cost estimate details. DOE could not demonstrate that the CIO reviewed the appropriateness of IT resource estimates underlying the investment’s budget request. DOE could not demonstrate that the CIO reviewed the appropriateness of IT resource estimates underlying the investment’s budget request. NNSA could not demonstrate that the CIO reviewed the appropriateness of IT resource estimates underlying the investment’s budget request. The NNSA CIO took steps to ensure the appropriateness of IT resource estimates included in the investment’s budget request by reviewing acquisition planning artifacts that included details on the fiscal year 2017 cost estimates. ● = The department or component demonstrated that it had implemented the OMB common baseline requirement on the investment. ◑ = The department or component demonstrated that it had implemented some but not all of the OMB common baseline requirement on the ○ = The department or component could not demonstrate that it had implemented the OMB common baseline requirement on the investment. This requirement was assigned at the department level. HHS documented IT capital planning procedures for the level of detail of planned expenditure reporting. However, the procedures had not explicitly required that all transactions with an IT component are included in the expenditure reporting. This requirement was assigned at the department level. HHS documented IT capital planning procedures for including the CIO in the planning and budgeting stages for new programs with IT resources that are greater than or equal to $20 million annually. However, at the time of our review, the department had not documented procedures for ensuring the CIO was included in the budget decisions for existing programs with IT resources that are greater than or equal to $20 million annually or for other programs that may have IT resources, such as those that are not primarily IT-oriented. The HHS CIO delegated the responsibility for carrying out this requirement to component CIOs for investments less than $20 million annually. However, HHS had not established procedures for ensuring its components were carrying out this responsibility. CMS documented annual IT budget instructions and governance board procedures for the component CIO’s involvement in the planning and budgeting stages for major investments less than $20 million annually. However, CMS had not documented procedures for how the CIO was to be involved in budgeting decisions for non-major investments. HHS developed charters that included the CIO on department-level governance boards that inform decisions regarding IT resources, such as the HHS Domain IT Steering Committee and the Chief Technology Officer Council. However, the HHS CIO was not a member of the Service and Supply Fund board—which reviews and approves operations and common service spending across the department—and other component-level IT investment review boards at CMS. The HHS CIO delegated the responsibility for carrying out this requirement to CMS’s CIO for investments less than $20 million annually. However, HHS had not established procedures for ensuring components were carrying out this responsibility. Selected OMB common baseline requirements CMS included its CIO as a member of the IT investment review board to oversee investments that are less than $20 million annually, consistent with the delegation from the HHS CIO. HHS documented IT capital planning and governance board procedures by which the CIO is to work with program leadership to plan IT resources for new investments greater than or equal to $20 million annually. However, the department had not established procedures by which the CIO is to work with program leadership in planning resources for existing investments greater than or equal to $20 million annually. HHS’s CIO delegated the responsibility of reviewing and approving IT investments to components for investments less than $20 million annually. However, HHS had not established procedures for ensuring its components were carrying out this responsibility. CMS documented the procedures by which program leadership was to work with the CMS CIO to plan IT resources for selected major and non- major investments through its IT investment review board. However, CMS had not established procedures for how the CIO was to work directly with program leadership on non-major IT investments that are not subject to the IT investment review board. HHS documented IT capital planning and governance board procedures by which the CIO is to review and approve new major IT investments greater than or equal to $20 million annually. However, the department had not established procedures by which the CIO was to review and approve other major IT investments, including major investments greater than or equal to $20 million annually that are not new investments. The HHS CIO delegated the responsibility of the requirement to review and approve major investments between $10 million and $20 million annually to its component CIOs. However, at the time of our review, HHS had not established procedures for ensuring its components carried out the responsibility. CMS documented procedures for its CIO to review and approve major IT investments that are between $10 million and $20 million annually through its IT capital planning and governance board procedures, consistent with its delegation from the HHS CIO. HHS had not documented procedures for the CIO’s review of significant increases and decreases in IT resources. In addition, the HHS CIO delegated the responsibility for the requirement to review IT resources that are to support major program objectives to component-level CIOs for investments less than $20 million annually. However, HHS had not established procedures for ensuring its components carried out the responsibility. CMS had not documented procedures for how the component was to review IT resources that are to support major program objectives, consistent with its delegated responsibility by the HHS CIO. Department Rating GAO’s assessment HHS HHS has not documented procedures for this requirement. This requirement was assigned at the department level. ● = The department provided documentation that demonstrated that the department or delegated component agency satisfied the OMB common ◑ = The department provided documentation that demonstrated that the department or delegated component agency satisfied some but not all of the ○ = The department could not provide documentation that demonstrated that the department or delegated component agency satisfied any of the OMB N/A = The component agency was not officially delegated the identified responsibilities for the OMB common baseline requirement. The HHS CIO was included in the planning and budgeting stages of the sampled investment by reviewing the IT resources through budget analysis meetings with the program office and department-level governance board. However, the review board only discussed nearly $15.8 million of the $33 million in total IT resources for the investment that was reported to OMB, and HHS officials could not demonstrate that the CIO was involved in the planning and budgeting stages for the remaining portion of the budget request. HHS could not demonstrate that the CIO was involved in the planning and budgeting stages for the sampled investment’s IT resources. The CMS CIO was included in the planning and budgeting stages for the sampled investment by reviewing and approving the investment’s budget request. However, HHS could not demonstrate that the department-level CIO was involved in the budgeting process for the investment’s IT resources through governance board reviews, as required by HHS policy. In addition, HHS could not demonstrate that the responsibility for this requirement had been delegated to the CMS CIO for the fiscal year 2017 budget request. The CMS CIO was included in the planning and budgeting stages for the sampled investment by reviewing the investment’s budget. However, HHS could not demonstrate that the department-level CIO was involved in the budgeting process for the investment’s IT resources through governance board reviews, as required by HHS policy. In addition, HHS could not demonstrate that the responsibility for this requirement had been delegated to the CMS CIO for the fiscal year 2017 budget request. HHS demonstrated that the CIO worked with program leadership to plan the investment’s IT resources through a department-level governance board meeting with program leadership, a briefing with the program office, and direction to investment managers to plan for a different funding scenario when determining the investment’s IT resource estimate. HHS could not demonstrate that the CIO worked with program leadership to plan the sampled investment’s IT resources. The CMS CIO worked with program leadership to plan the investment’s IT resources by chairing the CMS IT investment review board that approved the investment’s funding proposal. However, HHS could not demonstrate that the department-level CIO was involved in the planning process for the investment’s IT resources as required by HHS policy. In addition, HHS could not demonstrate that the responsibility for this requirement had been delegated to the CMS CIO for the fiscal year 2017 budget request. The CMS CIO worked with program leadership to plan the investment’s IT resources by chairing the CMS IT investment review board that reviewed the investment’s funding proposal. However, HHS could not demonstrate that the department-level CIO was involved in the planning process for the investment’s IT resources as required by HHS policy. In addition, HHS could not demonstrate that the responsibility for this requirement had been delegated to the CMS CIO for the fiscal year 2017 budget request. Selected OMB common baseline requirements CIO review and approval of IT budgets The HHS CIO reviewed the investment’s alignment with major program objectives during an annual operational analysis review in December 2015. However, HHS did not demonstrate that the CIO reviewed the increase in IT resources for the investment totaling $33 million, more than double its initial estimate of nearly $15.8 million. HHS could not demonstrate that the CIO reviewed the investment’s alignment with major program objectives or changes in resources. The CMS CIO reviewed changes in resources identified within individual activity funding requests related to the investment. However, CMS could not demonstrate that the CIO reviewed the investment’s alignment with major program objectives. In addition, HHS could not demonstrate that the department-level CIO was involved in reviewing changes in IT resources for the investment. Moreover, HHS could not demonstrate that the responsibility for reviewing changes in IT resources had been delegated to the CMS CIO for the fiscal year 2017 budget request. The CMS CIO reviewed changes in resources identified within individual activity funding requests related to the investment. However, CMS could not demonstrate that the CIO reviewed the investment’s alignment with major program objectives. In addition, HHS could not demonstrate that the department-level CIO was involved in reviewing changes in IT resources for the investment. Moreover, HHS could not demonstrate that the responsibility for reviewing changes in IT resources had been delegated to the CMS CIO for the fiscal year 2017 budget request. The HHS CIO took steps to determine the appropriateness of nearly $15.8 million of the IT resource estimates for this investment by reviewing a line item budget estimate prepared for a department-level governance board. However, HHS could not demonstrate that the CIO took steps to determine the appropriateness of the IT resources for the remaining portion of the investment’s total 2017 budget request of $33 million as reported to OMB. HHS could not demonstrate that the CIO or a designee reviewed the appropriateness of the IT resource estimates underlying the investment’s 2017 budget request. The CMS CIO took steps to determine the appropriateness of the investment’s IT resources totaling approximately $500 million by reviewing the detailed budget request at an IT investment review board meeting. However, according to the Federal IT Dashboard, the fiscal year 2017 budget request for this investment totaled $399 million, and CMS could not demonstrate that the CIO took steps to determine the appropriateness of the revised budget total. In addition, HHS could not demonstrate that the department-level CIO reviewed the appropriateness of the investment’s IT resources. Moreover, HHS could not demonstrate that the responsibility for this requirement had been delegated to the CMS CIO for the fiscal year 2017 budget request. Selected OMB common baseline requirements The CMS CIO took steps to determine the appropriateness of the investment’s IT budget request by reviewing the IT resource request through the investment review board. However, HHS could not demonstrate that the department-level CIO reviewed the appropriateness of the investment’s IT resources. In addition, HHS could not demonstrate that the responsibility for this requirement had been delegated to the CMS CIO for the fiscal year 2017 budget request. ● = The department or component demonstrated that it had implemented the OMB common baseline requirement on the investment. ◑ = The department or component demonstrated that it had implemented some but not all of the OMB common baseline requirement on the ○ = The department or component could not demonstrate that it had implemented the OMB common baseline requirement on the investment. DOJ established department-wide IT capital planning and budgeting procedures that document the level of detail with which IT resources are to be described during the planning and budgeting process. The documented level of detail went beyond OMB’s minimum required reporting to include 49 IT resources across four business areas. This requirement was assigned at the department level. DOJ documented IT capital planning procedures for the level of detail of planned expenditure reporting. However, the procedures did not explicitly require that all transactions with an IT component are included in the expenditure reporting. This requirement was assigned at the department level. DOJ documented procedures for including the CIO in the planning and budgeting stages for programs with IT resources. This requirement was assigned at the department level. DOJ documented governance board charters that included the CIO as a member of department-level IT governance boards. However, the CIO was not included a member of key component-level IT investment review boards, including those at FBI. This requirement was assigned at the department level. DOJ documented procedures in its IT Governance Guide by which the CIO is to work with program and component leadership in planning the overall portfolio of IT resources. This requirement was assigned at the department-level. DOJ documented procedures in its IT Governance Guide and Department Investment Review Council for the CIO to review and approve major IT investments. This requirement was assigned at the department level. DOJ documented procedures in its IT Governance Guide and IT capital planning guidance for the CIO to review IT resources that are to support major program objectives and significant increases and decreases in resources. This requirement was assigned at the department level. DOJ did not document procedures for this requirement. This requirement was assigned at the department level. ● = The department provided documentation that demonstrated that the department or delegated component agency satisfied the OMB common ◑ = The department provided documentation that demonstrated that the department or delegated component agency satisfied some but not all of the ○ = The department could not provide documentation that demonstrated that the department or delegated component agency satisfied any of the OMB N/A = The component agency was not officially delegated the identified responsibilities for the OMB common baseline requirement. FBI included the DOJ CIO in the planning and budgeting stages for the sampled investment through a review of the funding request during an annual IT portfolio review in October 2015. FBI included the DOJ CIO in the planning and budgeting stages for the sampled investment through a review of the funding request during an annual IT portfolio review in October 2015. DOJ’s CIO collaborated with program leaders in planning IT resources for the investment through development of the annual operating plan for DOJ’s working capital fund and through meetings with component business leadership and meetings with component CIOs. DOJ’s CIO collaborated with program leaders in planning IT resources for the investment through development of the annual operating plan for DOJ’s working capital fund. While FBI program leadership collaborated with IT representatives from the Criminal Justice and Information Services Division in planning IT resources, FBI could not demonstrate that the CIO was involved in the planning or that the responsibility had been delegated to the division. FBI could not demonstrate that the CIO worked with program leadership to plan the sampled investment’s IT resources. DOJ’s CIO reviewed IT resources that are to support major program objectives and changes in IT resources through development of the annual operating plan for DOJ’s working capital fund and in an annual IT portfolio review in October 2015. DOJ’s CIO reviewed IT resources that are to support major program objectives and changes in IT resources through development of the annual operating plan for DOJ’s working capital fund and in an annual IT portfolio review in October 2015. FBI obtained a review from the DOJ CIO regarding IT resources that are to support major program objectives and significant changes in IT resources through an annual IT portfolio review in October 2015. FBI obtained a review from the DOJ CIO regarding IT resources that are to support major program objectives and significant changes in IT resources through an annual IT portfolio review in October 2015. DOJ’s CIO took steps to ensure the investment included appropriate estimates of IT resources in its budget request by reviewing spreadsheets with additional and more detailed cost information during an annual IT portfolio review in October 2015. DOJ’s CIO took steps to ensure the investment included appropriate estimates of IT resources in its budget request by reviewing spreadsheets with additional and more detailed cost information during an annual IT portfolio review in October 2015. While IT representatives within the Criminal Justice and Information Services Division were involved in validating underlying IT resource estimates, FBI could not demonstrate that the CIO was involved in the planning or that the responsibility had been delegated to the division. Selected OMB common baseline requirements FBI could not demonstrate that the CIO or designee reviewed the appropriateness of IT resource estimates underlying the investment’s budget request. ● = The department or component demonstrated that it had implemented the OMB common baseline requirement on the investment. ◑ = The department or component demonstrated that it had implemented some but not all of the OMB common baseline requirement on the ○ = The department or component could not demonstrate that it had implemented the OMB common baseline requirement on the investment. This requirement was assigned at the department level. Treasury documented IT capital planning procedures for reporting investments’ planned IT expenditures. However, the procedures did not explicitly require that all transactions with an IT component are included in the expenditure reporting. This requirement was assigned at the department level. While the department documented procedures for including the CIO in the planning and budgeting stages for department-level programs that are identified as having IT resources, it did not document procedures for ensuring the CIO is included in all department-level programs that may have IT resources, including those that are not primarily IT-oriented. In addition, Treasury’s CIO delegated this requirement to component CIOs for component-level investments. However, the agency had not established procedures for verifying that components were carrying out this delegated responsibility. IRS documented annual IT budgeting procedures for including the IRS CIO in the component’s planning and budgeting stages for all programs that have IT resources. Treasury developed department-level IT governance board charters that included the CIO as a member. In addition, the Treasury CIO delegated the responsibility for this requirement to component CIOs for component- level IT governance boards. However, the department had not established procedures for verifying that components were carrying out this delegated responsibility. IRS included its CIO as a member of its IT investment review board through its documented IT budgeting procedures. Treasury has documented procedures in its IT capital planning guidance by which the CIO works with program and component leadership in planning IT resources. This requirement was assigned at the department level. Selected OMB common baseline requirements CIO review and approval of IT budgets While Treasury documented procedures in its IT capital planning guidance for the CIO to review major IT investments with each component, the department had not documented procedures for how the CIO was to approve those investments. In addition, the Treasury CIO delegated to component CIOs the responsibility to develop proposed IT planning and budgeting artifacts while the Treasury CIO would retain the authority to approve them. However, the department had not established procedures for verifying that components were carrying out this delegated responsibility. IRS documented annual IT budgeting procedures that described how the component CIO was to review and approve major IT investments. Treasury established procedures through its Quarterly Performance Reviews and its IT capital planning guidance to review IT resources that are to support major program objectives and significant changes in IT resources. For example, during the annual Spring portfolio review with each component, the CIO was to discuss strategic IT changes for the component and any significant resource changes that have occurred on individual investments. This requirement was assigned at the department level. Treasury had not documented procedures for ensuring the appropriateness of IT resource estimates for department-level investments. In addition, Treasury delegated the responsibility for this requirement to component CIOs for component-level investments. However, the department had not established procedures for verifying that components were carrying out this delegated responsibility. IRS documented annual IT budgeting procedures for ensuring the appropriateness of IT resources within the component-level IT portfolio. The procedures included validating annual IT demand requests that form the basis of the budget request. ● = The department provided documentation that demonstrated that the department or delegated component agency satisfied the OMB common ◑ = The department provided documentation that demonstrated that the department or delegated component agency satisfied some but not all of the ○ = The department could not provide documentation that demonstrated that the department or delegated component agency satisfied any of the OMB N/A = The component agency was not officially delegated the identified responsibilities for the OMB common baseline requirement. IRS demonstrated that managers from the Office of the CIO worked with program leadership from IRS business units in developing a portion of the sampled investment’s budget of $286 million through IT budget reviews conducted by an integrated review team in July 2015. However, the component agency could not demonstrate that Office of the CIO officials worked with program leadership to develop the budget for the full investment totaling $468 million. IRS demonstrated that managers from the Office of the CIO worked with program leadership from IRS business units in developing a portion of the sampled investment’s budget of $40 million through IT budget reviews conducted by an integrated review team in July 2015. However, the component agency could not demonstrate that Office of the CIO officials worked with program leadership to develop the budget for the full investment totaling $343 million. Treasury could not demonstrate that the CIO reviewed the investment’s alignment with major program objectives or changes in resources. Treasury could not demonstrate that the CIO reviewed the investment’s alignment with major program objectives or changes in resources. IRS demonstrated that managers from the Office of the CIO reviewed alignment with major program objectives and changes in underlying resources for a portion of the sampled investment’s budget of $286 million through IT budget reviews conducted by an integrated review team in July 2015. However, the component agency could not demonstrate that Office of the CIO officials reviewed alignment with major program objectives and changes in resources for the full investment totaling $468 million. IRS demonstrated that managers from the Office of the CIO reviewed alignment with major program objectives and changes in underlying resources for a portion of the sampled investment’s budget of $40 million through IT budget reviews conducted by an integrated review team in July 2015. However, the component agency could not demonstrate that Office of the CIO officials reviewed alignment with major program objectives and changes in resources for the full investment totaling $343 million. Treasury could not demonstrate that the CIO or designee reviewed the appropriateness of IT resource estimates underlying the investment’s budget request. Treasury could not demonstrate that the CIO or designee reviewed the appropriateness of IT resource estimates underlying the investment’s budget request. Selected OMB common baseline requirement IRS demonstrated that managers from the Office of the CIO reviewed detailed cost estimates for a portion of the sampled investment’s budget of $286 million through IT budget reviews conducted by an integrated review team in July 2015. However, the component agency could not demonstrate that Office of the CIO officials reviewed detailed cost estimates for the full investment totaling $468 million. IRS demonstrated that managers from the Office of the CIO reviewed detailed cost estimates for a portion of the sampled investment’s budget of $40 million through IT budget reviews conducted by an integrated review team in July 2015. However, the component agency could not demonstrate that Office of the CIO officials reviewed detailed cost estimates for the full investment totaling $343 million. ● = The department or component demonstrated that it had implemented the OMB common baseline requirement on the investment. ◑ = The department or component demonstrated that it had implemented some but not all of the OMB common baseline requirement on the ○ = The department or component could not demonstrate that it had implemented the OMB common baseline requirement on the investment. Appendix VII: Comments from the Internal Revenue Service Appendix VIII: Comments from the Department of Energy Appendix IX: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Dave Powner (Director), Nicole Jarvis (Assistant Director), Joshua Leiling (Analyst-in-Charge), Chris Businsky, Kara Epperson, Rebecca Eyler, Suellen Foth, Torrey Hardee, Tarunkant Mithani, Monica Perez-Nelson, and Andrew Stavisky made key contributions to this report.

Why GAO Did This Study In December 2014, Congress enacted FITARA, which was intended to improve covered agencies' acquisitions of IT. FITARA also provided an opportunity to strengthen the authority of CIOs to provide needed direction and oversight of agencies' IT budgets. GAO was asked to review whether CIOs' IT budgeting practices are consistent with FITARA and OMB's implementing guidance. This report addresses the extent to which selected federal agencies (1) established policies and procedures that address IT budgeting requirements, (2) could demonstrate that they had developed fiscal year 2017 IT budgets for sampled investments consistent with FITARA and OMB guidance, and (3) implemented processes to ensure that annual IT budgets are informed by reliable cost information. GAO selected four departments to review. These departments had the two highest and the two lowest average initial selfassessments scores of compliance with OMB's FITARA guidance, as well as a fiscal year 2017 IT budget of at least $1 billion. Within each of the departments, GAO also selected the component agencies with the largest fiscal year 2017 IT budget. For each selected department and component agency, GAO reviewed relevant IT budget policies and procedures, analyzed a sample of major and non-major investment proposals against key OMB requirements, and determined whether selected departments captured government labor costs, among other things. What GAO Found The departments GAO reviewed—the Departments of Energy (DOE), Health and Human Services (HHS), Justice (DOJ), and the Treasury (Treasury)—took steps to establish policies and procedures that align with eight selected Office of Management and Budget (OMB) requirements intended to implement information technology (IT) acquisition reform legislation (commonly referred to as the Federal Information Technology Acquisition Reform Act, or FITARA) and to provide the chief information officer (CIO) visibility into and oversight over the IT budget. For example, of the eight OMB requirements, all four departments had established policies and procedures related to the level of detail with which IT resources are to be described in order to inform the CIO during the planning and budgeting processes. Agencies varied, however, as to how fully they had established policies and procedures related to some other OMB requirements, and none of the four departments had yet established procedures for ensuring that the CIO had reviewed whether the IT portfolio includes appropriate estimates of all IT resources included in the budget request. (See table.) Where the departments had not fully established policies and procedures, it was due, in part, to having not addressed in their FITARA implementation and delegation plans how they intended to implement the OMB requirements. Until departments develop comprehensive policies and procedures that address IT budgeting requirements established by OMB, they risk inconsistently applying requirements that are intended to facilitate the CIO's oversight and approval of the IT budget. Departments varied in the extent to which they could demonstrate implementation of key IT budgeting requirements when developing fiscal year 2017 funding requests for sampled investments. Specifically, while DOJ demonstrated that it had fully implemented the selected requirements for the majority of the investments GAO sampled, HHS and Treasury partially demonstrated implementation for a majority of the sampled investments, and DOE could not demonstrate implementation for the majority of the sampled investments. For example, DOE, HHS, and Treasury were not able to fully show that their CIOs had reviewed whether estimates of IT resources included in the budget request were appropriate for two of their respective departments' largest fiscal year 2017 IT investments. Departments often could not demonstrate that they had implemented selected IT budgeting requirements at the investment level because they had not established comprehensive policies and procedures that required them to do so. As a result, departments could not show that CIOs were sufficiently involved in planning fiscal year 2017 IT expenditures at the individual investment level. All four selected departments lacked quality assurance processes for ensuring their IT budgets were informed by reliable cost information. Specifically, the selected departments did not have IT capital planning processes for (1) ensuring government labor costs have been accurately reported, (2) aligning contract costs with IT investments, and (3) utilizing budget object class data to capture all IT programs. This resulted in billions of dollars in requested IT expenditures without departments having comprehensive information to support those requests, and nearly $4.6 billion in IT contract spending that was not explicitly aligned with investments in selected departments' IT portfolios. This was due to a lack of processes for periodically reviewing data quality and estimation methods for government labor estimates, as well as a lack of mechanisms to cross-walk IT spending data in their procurement and accounting systems with investment data in their IT portfolio management systems. In August 2017, OMB developed a new approach of using a standard set of categories to group IT spending that, if properly implemented, has the potential to provide departments and CIOs enhanced visibility into IT costs across the portfolio. Nevertheless, until departments establish processes for assessing or otherwise ensuring the quality of relevant IT cost data used to inform their IT budgets, department CIOs will have less assurance that their budget includes appropriate and comprehensive estimates of IT resources. What GAO Recommends GAO is making 43 recommendations to the eight selected departments and component agencies to address gaps in their IT budgeting policies and procedures, demonstrate implementation of OMB requirements, and establish procedures to ensure IT budgets are informed by reliable cost information. HHS, the Centers for Medicare and Medicaid Services, DOJ, the Federal Bureau of Investigation, and the Internal Revenue Service agreed with our recommendations. DOE partially agreed with one recommendation and agreed with the other recommendations made to it, as well as with the recommendations made to its component agency—the National Nuclear Security Administration. Treasury neither agreed nor disagreed with the recommendations.

gao_GAO-18-425

gao_GAO-18-425_0

Background Research has found that girls’ participation in sports has increased dramatically since the passage of Title IX. However, research has also found that progress toward equal sports participation between boys and girls has slowed since 2000, and a participation gap remains between the sexes. We previously reported that federal data from school year 2013- 14 showed that national girls’ participation rates in public high school interscholastic sports remained nearly 10 percentage points lower than boys’ rates. The same data showed that at nearly half of schools, girls’ share of sports participation was less than their share of enrollment by 5 percentage points or more. Education’s Role Within Education, OCR enforces and implements Title IX, which applies at all educational levels, including colleges, universities, and public school districts, with limited exceptions. OCR’s most recent annual report describes its mission as ensuring equal access to education and promoting educational excellence throughout the nation through vigorous enforcement of civil rights laws. OCR’s core activities include responding to civil rights complaints filed by the public and conducting agency-initiated investigations to enforce federal civil rights laws; providing technical assistance to help institutions achieve compliance with the civil rights laws that OCR enforces; and issuing regulations and policy guidance to ensure equal access to educational opportunity. OCR also conducts the Civil Rights Data Collection (CRDC), which collects key information related to civil rights from public elementary and secondary schools and school districts, including information on interscholastic sports and teams offered for boys and girls and their participation. With respect to athletics, Education’s Title IX regulations require schools that offer sports teams to provide equal opportunities for members of both sexes. The regulations, along with OCR guidance, specify key elements OCR considers, among other things, in determining whether schools are offering equal opportunities (see fig. 1). OCR uses the number of participants on a school’s sports teams as a proxy for participation opportunities when determining whether those opportunities are proportionate for boys and girls. Recipients of federal education funds, such as public school districts, bear the responsibility for complying with Title IX. Districts are required to designate an employee to coordinate efforts under Title IX, and to make this Title IX coordinator visible. In 2014, we recommended OCR clarify and disseminate information on the roles and responsibilities of these Title IX coordinators. In response, during fiscal year 2015, OCR issued several pieces of Title IX guidance, including a Dear Colleague letter delineating the specific requirements and duties of coordinators, in addition to a letter to coordinators and a Title IX resource guide, which includes guidance on monitoring compliance in athletics. This guide states that the Title IX coordinator should work closely with many different members of the school community, including athletics administrators. Regarding athletics, it recommends tools that Title IX coordinators can use to encourage equal opportunities in athletics, which include evaluating whether there is unmet interest in a particular sport and comparing expenditures on boys’ and girls’ sports teams as an indicator of benefits provided to those teams. Athletics Administrators Reported Schools Took Some Steps to Encourage Equal Opportunities, but About Half Did Not Receive Title IX Coordinator Support To Encourage Equal Opportunities, the Majority of Athletics Administrators Reported That Their Schools Assessed Resources Provided to Boys’ and Girls’ Teams The majority of public high schools assessed some aspects of their sports programs over the past 2 years to encourage equal opportunities for boys’ and girls’ sports teams, according to our nationally generalizable survey of athletics administrators. Specifically, the estimated percentage of schools assessing key athletic resources provided to these teams ranged from 63 percent of schools assessing travel opportunities to 76 percent assessing uniforms (see fig. 2). In our interviews with eight athletics administrators, we heard a variety of approaches to assessing these resources. For example, when scheduling practice times and competitions, five athletics administrators said that they scheduled a boys’ competition only if they could also schedule a girls’ competition. Four athletics administrators described watching practices, inspecting equipment to identify when it needed replacement, or replacing equipment as their coaches requested it. Four athletics administrators said that coaches can sometimes influence the distribution of resources. For instance, one athletics administrator noted that in the past, his school had unequal facilities for boys’ baseball and girls’ softball, stemming in part from the boys’ baseball coach being a stronger advocate for his team. However, these athletics administrators generally described working with the coaches to ensure that resource allocation did not create inequalities. Most schools reported using a mix of public and private funds to support their athletic programs. An estimated 75 percent of public high schools received public funding (state or local) for their sports programs; for some individual sports or school athletics programs, public funding may be the primary funding source. We estimate that about 52 percent of schools that received public funding monitored or directed its use to help encourage equal resources for boys’ and girls’ teams. OCR’s Title IX Resource Guide encourages Title IX coordinators to periodically review expenditures on male and female athletic teams as part of their review of resources. At four schools, athletics administrators told us that they paid attention to the actual resources girls and boys received rather than focusing on expenditures, and three of these administrators explained there could be valid reasons for spending differences. For example, one athletics director said that both boys’ and girls’ hockey teams at his school participated in annual tournaments, but the girls preferred a tournament that did not require a hotel stay, so it was less expensive. In addition, we estimate that about 81 percent of public high schools had at least one booster club and, according to our survey, an estimated 51 percent of these schools monitored or directed the club to encourage equal opportunities. Among the eight athletics administrators we interviewed, relationships with booster clubs varied. For instance, some issued booster club guidelines and approved their purchases in advance, while others had no oversight of booster club expenditures. For example, one athletics administrator told us that he provides booster club presidents with written guidelines and approves purchases to make sure they do not create a Title IX compliance issue. Another athletics administrator’s school had recently undergone negotiations to obtain access to booster club expenditure records for the first time so that they could regularly review those expenditures. OCR has stated in compliance decisions, and OCR officials confirmed to us in interviews, that it considers resources provided through the use of private funds, including booster funding, in assessing whether schools are providing equivalent resources to teams of each sex. An official from a national association representing athletics administrators stated that administrators who take the association’s Title IX trainings are often surprised to learn they should monitor or direct booster club spending to help ensure equal opportunities. Some Athletics Administrators Reported That Their Schools Used Surveys and Other Tools to Gauge Student Sports Interests, Which Can Help Encourage Equal Opportunities In addition to assessing the various school and booster club resources provided to boys’ and girls’ teams, some schools recently took steps to gauge student interest in specific sports as a means of encouraging equal opportunities, according to our survey. For example, we estimate that 40 percent of schools surveyed students about their sports interests over the last 2 school years and 25 percent added or changed their sports offerings based on requests from the underrepresented sex in their school’s sports program (see fig. 3). An estimated 31 percent of schools had not recently used any of these tools, or did not know if they had used the tools, to gauge student interest. And, according to our analysis of Education’s data, 60 percent of schools had one sex underrepresented by more than 5 percent in their sports programs in school year 2013-14. OCR guidance states that where one sex is underrepresented in sports, schools can demonstrate they are providing equal participation opportunities by using multiple indicators to identify, among other things, whether the sports currently offered meet student interest. OCR guidance also states that in its investigations the agency determines on a case-by-case basis whether sports participation numbers at a school are disproportionate, and whether the school is taking sufficient steps to accommodate the athletic interests and abilities of both girls and boys. In addition, OCR guidance describes tools that schools and school districts can use to assess for themselves whether action is needed to address any underrepresentation, or to otherwise encourage equal athletic opportunities. According to the guidance, these efforts should be led by the school district’s Title IX coordinator. About Half of Athletics Administrators Were Either Unaware of or Unsupported by Their District Title IX Coordinator About 51 percent of athletics administrators were either not aware of or not supported by their Title IX coordinator, according to our survey. Specifically, we estimate that 40 percent of athletics administrators– serving about 6,110 schools and 5 million students–were unaware of a Title IX coordinator in their school district and that an additional 12 percent were aware of their Title IX coordinator but received little to no support from them (see fig. 4). We also found that almost all of the athletics administrators who were not aware of having a Title IX coordinator were in a district that had, in fact, designated one. Specifically, when we matched athletic administrators’ survey responses with OCR’s data and extrapolated to the population overall, we estimated that 99 percent of the athletics administrators who were not aware of a Title IX coordinator in their district were in a school district that had listed a coordinator in school year 2013-14. Further, an estimated 26 percent of athletics administrators wanted additional guidance or assistance related to encouraging equal opportunities for boys and girls, according to our survey. Given the significant number of athletics administrators who reported being unaware of or unsupported by their Title IX coordinators, our survey results raise questions as to whether Title IX coordinators—whom school districts must designate and make visible in accordance with Title IX regulations—are familiar with and using OCR’s guidance on their role and responsibilities. This guidance states that the Title IX coordinator should support and work closely with members of the school community, including athletics administrators, to ensure compliance with Title IX. When asked about these survey results, officials from an association for Title IX coordinators and for other related administrators told us that they were not surprised that a number of athletics administrators were not aware of or supported by their Title IX coordinator, because the results are consistent with what they hear when interacting with their members across the country. Based on these interactions, these association officials said they have observed that there is often a separation between athletics and other school departments, and that Title IX coordinators without an athletics background may be reluctant to engage in oversight of that department. Based on their experiences providing training to Title IX coordinators, these association officials also said that Title IX coordinators’ familiarity with Title IX requirements has improved somewhat since the release of OCR’s 2015 guidance delineating their role and responsibilities, but their familiarity with these requirements is still generally low, particularly with respect to athletics. In these officials’ opinion, this lack of understanding is due in part to the complex and wide-ranging nature of Title IX and to the lack of resources for training in many school districts. These and other subject matter specialists we interviewed said that other potential factors contributing to athletics administrators’ lack of awareness of their Title IX coordinator included high turnover among athletics administrators and myriad responsibilities of staff in both roles. When Title IX coordinators do not work closely with athletics administrators, as OCR guidance suggests they do, they may miss opportunities to make those administrators aware of tools the guidance recommends that could help advance equal opportunities. In addition, OCR guidance recognizes that the most serious Title IX violations tend to occur in districts without a supportive Title IX coordinator. OCR officials said that they had learned from their complaint investigations and compliance reviews that some athletics administrators were not working with their districts’ Title IX coordinators. However, these officials said they did not know the extent to which Title IX coordinators themselves were aware of and using the tools recommended in their guidance because, outside of these enforcement activities, OCR generally does not collect information on Title IX coordinators’ knowledge of or activities related to the guidance. Standards for internal control in the federal government state that agencies should both obtain quality information from and communicate quality information to external parties to help achieve the agency’s objectives and address risks. In OCR’s case, its objectives include ensuring schools actively encourage equal opportunities for boys and girls as articulated in Education’s Title IX regulations and OCR guidance, and risks include violations of Title IX that have not resulted in formal complaints. Absent better information on Title IX coordinators’ awareness and use of Title IX guidance, OCR may not have a complete picture of school districts’ ongoing efforts to encourage equal opportunities, challenges they encounter in doing so, and successful strategies that might be shared with a broader audience. Collecting and analyzing this information could enable OCR to target its communication to Title IX coordinators, and further encourage them to work with athletics administrators on ensuring equal athletic opportunities. Available Opportunities Helped Drive Public High School Sports Participation Levels, but Family Resources and Other Factors Could Limit Participation The Number of and Interest in Opportunities Offered by Schools Encouraged Higher Participation Levels The number of participation opportunities schools offered, as well as student interest in those opportunities and in working with specific coaches at the school, were top factors that encouraged interscholastic sports participation among public high school students, according to our survey of public high school athletics administrators. We estimate that over 70 percent of athletics administrators viewed the number of interscholastic athletic participation opportunities at their school as encouraging boys and girls to participate in high school sports (see fig. 5). Our 2017 report on high school sports access and participation found that in school year 2013-14, public high schools overall offered the same number of sports and teams for boys and girls. Of the nine subject matter specialists we interviewed for this report, six described specific knowledge of factors that encourage or discourage participation in high school sports. All six of these subject matter specialists agreed that opportunity is an important factor affecting student participation, especially for girls; several specialists also said that the continued participation gap shows that girls do not have access to an equal number of roster spots on teams as boys. For example, one researcher, as well as a representative of a national association of athletics administrators, suggested that one reason the gap between boys and girls persists is that schools do not offer girls’ sports with roster sizes equivalent to popular boys’ sports, such as football. One of the eight athletics administrators said that this was the case at her school, noting that none of her girls’ teams came close to the size of the football teams. The gap may be particularly acute for minority girls, according to one subject matter specialist. Our 2017 report on public high school sports access and participation found that, for both boys and girls, fewer students attended high minority and high poverty schools that offered sports, compared to students at other schools, and these schools had lower participation rates when they did offer sports. We estimate that over 75 percent of athletics administrators viewed the level of student interest in the sports offered by their school as encouraging participation in their school’s teams. Several subject matter specialists agreed that offering sports that align with students’ specific interests is an important aspect of providing meaningful opportunities, but a few also noted that some schools fail to consider which sports most interest their female students. We estimate that 70 percent or more of athletics administrators viewed student interest in working with certain coaches as a factor that encouraged participation at their school. As explained by one researcher, coaching quality plays a large role in encouraging high school sports participation and a good coach can pull students into a sport and keep them participating. Alternatively, another researcher noted that less qualified or inexperienced coaches depress participation. These views are consistent with our work, in which we reported that the quality of coaching is a key factor in maximizing the positive effects of sports participation on students’ personal development. In addition, research shows that the state of athletic facilities can also affect a student’s choice to participate in high school sports, and a few athletics administrators and subject matter specialists we interviewed also cited this as a factor. For example, one study found that proximity to sports facilities was a factor predicting children’s participation in team sports. Another study found that student participation in interscholastic sports is higher at schools with more sports facilities compared with schools that have few sports facilities. A few of the subject matter specialists and one athletics administrator made similar observations about the relationship between facilities, participation, and inequity. For example, the athletics administrator said that at his high school, baseball and softball participation has decreased because their athletic facilities are located off campus, requiring additional travel for both students and parents for practices and games. Additionally, representatives from two advocacy groups noted that parents may have concerns related to school sports facilities, particularly for the safety of their daughters. For example, one said that some schools have girls’ teams practice in off-campus facilities, sometimes in unsafe neighborhoods, without offering transportation. The other said parents may be concerned when fields are insufficiently lit or their daughters come home late from practices. Factors Related to Family and Community Resources and Cultural Expectations May Discourage Participation in Sports We found no clear consensus in our survey of athletics administrators regarding factors that tend to discourage students from participating in sports, and the eight athletic administrators we interviewed had mixed views on the subject. That said, the most frequently mentioned factors that were perceived to discourage participation (representing an estimated 15-35 percent of athletic administrators) were (1) competing responsibilities, (2) lack of access to athletic feeder programs, (3) the perceived benefits of joining club teams, and (4) participation costs to the student. Competing responsibilities. Over one-quarter of athletics administrators cited students’ competing responsibilities as discouraging participation in public high school sports. This could include a range of responsibilities, including schoolwork, other school activities, and family obligations. Among the athletic administrators we interviewed, one noted that many students at his magnet school were more focused on academics than athletics. A few cited examples of competing responsibilities that were tied to family resources. For example, two said that many of their students have jobs and family responsibilities that prevent them from participating in sports. One of these administrators said that his school’s student population largely comes from lower-income families, and many are juggling jobs; in response, the school changed practice schedules to better match students’ availability, which has made it easier for more students to participate. Lack of access to athletic feeder programs. Some athletics administrators also mentioned a lack of access to athletic feeder programs—club or community-based youth sports programs that train younger children before they enter high school—as discouraging participation in sports at their public high schools. In addition, a few of the eight athletics administrators we interviewed saw this lack of access as being closely related to community or family resources. Two of these administrators, who worked in lower-income schools, reported that younger children in their area have very little access to community or club sports and that students who do not have previous exposure to sports may lack the skills to participate at the high school level. One of these administrators also said that having more community youth sports might increase student interest in playing at the high school level. A third athletics administrator said that feeder programs help drive participation in high school sports. In his school’s competitive environment, students trying out for sports for the first time when they get to high school are, in most cases, likely to be cut from the team. He added that he has found that family income is a major contributing factor to children’s ability to begin training early, which puts lower income students at a disadvantage. His point was echoed in one research study that found that as family income increases, boys and girls tend to enter organized sports at a younger age. Perceived benefits of joining club teams. Athletics administrators also mentioned the presence of club teams that students may choose over school teams as discouraging participation in public high school teams. This may be particularly true for higher-income students, as competitive travel and club teams—which parents and students may see as offering higher-caliber coaching, more specialized training, and greater opportunities to compete against elite athletes—can be quite expensive. Several subject matter specialists we interviewed cited this as an issue that affects high school sports participation. Further, a few of the high school athletics administrators we interviewed observed decreased student participation at their schools due to the presence of club teams. One athletics administrator from the Southwest explained that participation is weaker for his school’s Olympic sports, such as swimming, due to competition from club sports. He noted that at his high school, this phenomenon makes it more difficult to recruit other students because the school teams become less competitive. Another athletics administrator from the Midwest explained that at his high school, the presence of club sports disproportionally depressed girls’ participation in high school sports. In particular, he said the popularity of club girls’ volleyball in the winter reduced participation in his girls’ basketball teams. Participation costs. The cost to students of participating in athletics was also mentioned by some athletics administrators as discouraging participation in public high school teams. The subject matter specialists and athletics administrators we interviewed had mixed views on the effect of costs on student participation. Among the subject matter specialists, two said that the increasing prevalence of fees in high school sports programs is threatening participation by lower-income students. One athletics administrator agreed, saying that in the past he has dissuaded his school district from charging participation fees for this reason. Another said that his school does not charge fees, but students could still be discouraged by the fundraising required for “extras” such as team t-shirts. In contrast, one subject matter specialist said that it is typically higher-income schools that charge students fees to participate in sports, and therefore fees do not generally affect students in lower-income schools. In addition to the four most commonly cited barriers from our survey, several research studies noted that cultural expectations around family responsibilities and gender roles may also discourage some student groups more than others. For example, one study found that Hispanic girls quit sports to take care of younger siblings at higher rates than their white peers. This and another study noted that students from recent immigrant families may also be discouraged from participating in sports because of different cultural expectations around prioritizing sports, and girls may be additionally affected by expectations around gender roles. For example, it found that immigrant parents are more likely than non- immigrant parents to believe that boys are more interested in sports than girls, and that 75 percent of immigrant sons were involved with organized or team sports compared with 43 percent of immigrant daughters. Similarly, a study of sports involvement among East African immigrant girls found that those the researchers interviewed face social barriers to participation, such as peer criticism, parents’ fears of interactions with male athletes, and lack of parental support. Several of the subject matter specialists and athletics administrators with whom we spoke made similar observations around cultural expectations. One suggested that differences in sports participation among immigrant communities may stem from the opportunities to play sports in the family’s country of origin, noting that the United States is unique in tying sports teams to its academic institutions. Officials from two advocacy organizations, one of which advocates for the Hispanic community, noted that some Hispanic families expect daughters to come home after school to help care for their siblings. This can interfere with participating in after-school activities. In addition, one athletics administrator we interviewed, whose school serves a predominantly Hispanic community, commented that his coaches have seen girls from this community quit sports teams on several occasions due to family responsibilities. One of the advocacy organization officials added that schools wanting to improve participation among Hispanic girls should, for example, consider more creative scheduling to allow these students to attend practices. Conclusions While sports participation for girls has risen dramatically over the last 45 years, a significant gap still remains between boys and girls in public high school sports. Our findings suggest that the reasons for this gap are varied and complex, and according to our survey, at many schools, athletics administrators are not aware of or do not receive support from their Title IX coordinator. OCR’s guidance suggests that uninvolved Title IX coordinators are associated with serious Title IX violations, but OCR does not collect information about coordinators’ level of involvement with districts and schools outside of its complaint investigations and compliance reviews. Better information about Title IX coordinators’ awareness and use of OCR’s guidance could help OCR support schools’ and districts’ efforts to provide equal opportunities in their sports programs. Recommendation for Executive Action The Department of Education’s Assistant Secretary for Civil Rights should determine the extent to which Title IX coordinators at the K-12 level are aware of and using the tools recommended in OCR’s existing guidance and any barriers preventing their use of this guidance, and use this information in OCR’s efforts to encourage them to work with athletics administrators on ensuring equal athletic opportunities. (Recommendation 1) Agency Comments and Our Evaluation We provided a draft of this report to Education for review and comment. Education provided written comments that are reproduced in appendix III, as well as technical comments that we incorporated, as appropriate. In its written comments, Education stated that it partially concurs with our recommendation that OCR determine the extent of K-12 Title IX coordinators’ knowledge and use of tools in its existing guidance and use this information in its efforts to encourage them to work with athletics administrators to help ensure equal athletic opportunities. Specifically, Education stated that when OCR conducts investigations in response to complaints it would look for opportunities to examine whether K-12 Title IX coordinators were aware of, and using, the tools in OCR’s guidance. Education also said that when OCR engages in technical assistance activities, it will encourage Title IX coordinators to work with athletics administrators to encourage equal opportunities. Education also said that it will consider our recommendation during its frequent reviews of the agency’s communications practices. We agree that these are important first steps in helping ensure that Title IX coordinators are working with athletics administrators and otherwise fulfilling their responsibilities to encourage equal opportunities. However, given our finding that about half of public high school athletics administrators were unaware of or unsupported by their Title IX coordinator, we continue to believe the systemic approach we recommend is necessary. The activities that OCR described in its response are predicated on a complaint being filed or technical assistance being requested. This narrow approach means that OCR will likely not learn the full extent to which K-12 Title IX coordinators are unaware of or not using the tools in OCR’s guidance. It also means that its reviews of the agency’s communication practices may be hampered by incomplete information on how best to encourage Title IX coordinators to use these tools and work with athletics administrators to ensure equal opportunities. As agreed with your offices, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies to the appropriate congressional committees, the Secretary of Education, and other interested parties. In addition, this report will be available at no charge on GAO’s website at http://www.gao.gov. If you or your staff should have any questions about this report, please contact me at 617-788-0580 or nowickij@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix IV. Appendix I: Scope and Methodology The objectives of this study were to examine: (1) what measures public high schools and athletics administrators have taken to encourage equal athletic opportunities for boys and girls, and (2) what factors affect boys’ and girls’ participation levels in public high school sports programs. To address these objectives, we used a variety of methods, including a web- based survey of public high school athletics administrators; follow-up interviews with eight survey respondents; reviews of federal law, regulations, and guidance; and interviews with federal officials at the Department of Education (Education) and with subject matter specialists. Web-based Survey of Public High School Athletics Administrators To obtain school-level perspectives on factors that affect boys’ and girls’ participation levels in public high school sports programs and approaches schools and athletics administrators have used to encourage equal athletic opportunities, we designed and administered a survey to athletics administrators at a generalizable, stratified random sample of public high schools in the United States. The survey included questions about what sports and levels of competition the school offered for each sex, how many boys and girls participated in sports in school year 2015-16, and factors that encourage and discourage girls’ and boys’ sports participation at the school. The survey also included a variety of questions related to the school’s and athletics administrator’s activities to encourage equal opportunities in the prior 2 years, challenges they faced in encouraging equal opportunities, sources of guidance on Title IX, booster club structures and oversight, and data they maintained on funding and expenditures. In addition, it included a question on whether, to the athletics administrator’s knowledge, their school district had a Title IX coordinator. Our population of interest for the survey was athletics administrators at public high schools. In terms of the schools, we defined our target population as public schools offering at least one high school level grade (9, 10, 11, or 12) that appeared in both Education’s Common Core of Data (CCD) and Civil Rights Data Collection (CRDC) for the 2013-14 school year, were located in the 50 states and the District of Columbia, and indicated in the CRDC that they offered interscholastic sports. We excluded schools that were listed as closed or not operational according to the school year 2015-16 CCD, as well as single-sex schools and schools located in U.S. territories. We also obtained the most current school contact information from the school year 2015-16 CCD. We originally selected a stratified random sample of 813 from a population of 15,330 schools in our sampling frame. However, we ultimately excluded 26 schools from our original population and sample because they had closed, did not serve high school grades, or did not offer interscholastic sports, and thus were not considered eligible for our survey. In addition, we found schools in the population and sample that shared sports programs and athletics administrators, effectively reducing the population by 10 schools and the sample by 3 schools for purposes of our survey. This resulted in a sample of 784 schools from the eligible population of 15,294. We stratified this sample based on school type (charter or traditional), concentration of minority students (low = 0-25 percent, mid = 26-74 percent, high = 75-100 percent), locale type (urban, suburban, or rural), and participation rates of male and female students in school year 2013- 14. This created 24 strata as noted in table 1. For the participation rate strata, we calculated each school’s male and female students’ participation rates using data from the school year 2013-14 CRDC. Participation rates were defined as the number of sports participants of that gender divided by the number of enrolled students of that gender, and these rates were then compared to determine which gender had higher participation: females or males. We placed schools with equal participation rates for males and females into the “Females” participation category because, given the overall higher participation rates for boys, schools with both equal participation rates and higher rates for girls are rarer. We chose these strata to ensure schools with the stratum characteristics were included in the sample. The total sample size of n=813 was inflated for an expected 60 percent response rate, and we distributed the sample across the strata for workload and analysis considerations. The sample size in table 1 optimizes for some groups, while controlling the distribution across the 24 strata. Specifically, we calculated the Neyman optimal sample size that resulted in an overall 5 percent margin of error for an attribute estimate. We allocated samples across strata to achieve precision goals at two levels: overall population percentage estimates with margins of errors within plus or minus 5 percentage points, and subpopulation percentage estimates (i.e. school type, minority level, locale, or participation group) with margins of errors within plus or minus 10 percentage points, both at the 95 percent confidence level. Additionally, we ensured a minimum sample of 10 schools in every stratum. Based solely on the constraint of an overall margin of error within plus or minus 5 percentage points, some reporting groups were expected to have margins of errors that were less than 10 percentage points without the need of additional explicit constraints. For other reporting groups, we implemented constraints so that the designed margin of error was within plus or minus 10 percentage points. Specifically, we included the following constraints for margins of error of attribute estimates with 95 percent confidence intervals within each reporting group, for a realized response rate of 60 percent: margins of error within plus or minus 5 percentage points overall, within plus or minus 10 percentage points for urban schools, within plus or minus 10 percentage points for high minority schools, within plus or minus 10 percentage points for charter schools. Because we followed a probability procedure based on random selections, our sample is only one of a large number of samples that we might have drawn. Since each sample could have provided different estimates, we expressed our confidence in the precision of our particular sample’s results as a 95 percent confidence interval (e.g., plus or minus 7 percentage points). This is the interval that would contain the actual population value for 95 percent of the samples we could have drawn. Unless otherwise noted, all percentage estimates in this report have confidence intervals within plus or minus 7.7 percentage points. For other estimates, the confidence intervals or margins of error are presented along with the estimates themselves. We took several steps to minimize non-sampling error. We used several methods to identify the names and email addresses of the athletics administrators for our selected sample of schools. In some states, the high school associations had directories we used to obtain this information. For those that did not, we searched the school’s website or called the school or district. We administered the survey from June through early September 2017. To obtain the maximum number of responses to our survey, we sent e-mails to the principals of the schools in the sample prior to the survey’s launch, asking them to support and encourage their athletics administrator to complete the survey. We also worked with the National Federation of State High School Associations to have the state associations e-mail their members and encourage them to participate in the survey. Finally, we sent direct reminder emails to nonrespondents and contacted nonrespondents over the telephone. We took additional steps to minimize non-sampling errors, including pretesting draft instruments and using a web-based administration system. During survey development, we met with officials from national groups representing high school activities associations and athletics administrators and held discussion groups with nine athletics administrators to explore the feasibility of responding to the survey questions. We then pretested the draft instrument from April through May 2017 with five athletics administrators in public high schools that were diverse across a range of characteristics, such as region, school type and locale, and minority enrollment. In the pretests, we asked about the clarity of the questions and the flow and layout of the survey. A survey specialist independent of the project team within GAO also reviewed a draft of the questionnaire prior to its administration. Based on feedback from the pretests and the independent review, we made revisions to the survey instrument. To further minimize non-sampling errors, we used a web- based survey, which allowed respondents to enter their responses directly into an electronic instrument. Using this method automatically created a record for each respondent in a data file and eliminated the errors associated with a manual data entry process. Despite these efforts, like most surveys, our survey had nonresponse. Specifically, the weighted response rate was 42 percent. Survey nonresponse raises the possibility that those athletics administrators who did respond to the survey may not be representative of the intended population, due to nonresponse bias. We carried out a nonresponse bias analysis and identified three potential factors that may have been related to athletics administrators’ propensity to respond: school concentration of minority students, school size, and region. In order to adjust for the potential nonresponse bias, we adjust the sampling weight with a nonresponse adjustment to form a final weight. Data analyzed using the final, nonresponse-adjusted sampling weight is assumed to be missing at random, given the nonresponse adjustments, and therefore unbiased for the intended population. We used response propensity weighting class adjustments based on a model that included the variables identified in the nonresponse bias analysis. We conducted our analysis using survey software that accounted for the sample design and weighting. Survey Follow-Up Interviews To gain further insights into factors that encourage or discourage participation in sports, schools’ efforts to encourage equal opportunities, and the role of the Title IX coordinator, we conducted follow-up interviews with 8 athletics administrators, chosen from the 105 who had responded to our survey as of late August and indicated that they were willing to participate in a follow-up discussion on their responses. Specifically, we selected respondents to obtain diversity in their responses to a few key survey questions, as well as certain characteristics of their schools. In making our selections, we considered their responses to survey questions on: their awareness of their Title IX Coordinator, activities their schools conducted within the last two years to encourage equal opportunities for boys and girls in sports, and whether the school maintains expenditure data on sports and their willingness to share these data. We identified school characteristics with the data sources used to create our survey sampling frame. The characteristics we considered to further narrow our selection were: school type (charter or traditional), school locale (urban, suburban, or rural), concentration of minority students (low-, mid-, or high-minority). Additionally, we reviewed open-ended responses in the survey to determine if there were answers that necessitated additional discussion or clarification (see table 2). In our interviews with the athletics administrator at each school, which we conducted by phone, we asked officials to describe their relationship with their Title IX coordinator, familiarity with Title IX requirements overall, and their familiarity with state and local guidance, specifically. We also asked them to describe their efforts to encourage equal opportunities in sports and the nature of challenges they have faced in doing so. In addition, we asked them about funding sources and their use of expenditure data, the role of booster clubs, and the role of outside funding. For each school where the athletics administrator reported that they had expenditure data, we requested a copy of these data. We obtained expenditure data from three schools. In some cases we obtained additional documentation such as booster club guidelines, processes for adding school sports, and participation data. Because we selected the schools for follow-up interviews judgmentally and only conducted eight interviews we cannot generalize our findings about their policies, practices, and challenges. Review of Law, Regulations, and Guidance and Interviews with Education Officials To understand the requirements for providing equal athletic opportunities in public high schools and how Education’s Office for Civil Rights (OCR) monitors and supports public school districts in meeting these requirements, we reviewed Title IX of the 1972 Education Amendments (Title IX), Education’s Title IX regulations, and related guidance documents. We also interviewed OCR and other Education officials. In addition, we reviewed selected research studies that provided context and insight into factors affecting high school sports participation. Interviews with Subject Matter Specialists To obtain additional context and insights, we selected and interviewed subject matter specialists, including researchers and officials from advocacy groups and associations. We selected these subject specialists so that, together with the athletics administrators we surveyed and interviewed, they would provide a variety of perspectives on factors that affect boys’ and girls’ participation in high school sports and approaches schools use to encourage equal athletic opportunities. The researchers and officials we interviewed were located at: the Institute for Research on Women and Gender and the Sport, Health, and Activity Research and Policy Center at the University of Michigan, the Tucker Center for Research on Girls and Women in Sport at the University of Minnesota, the Department of Recreation, Sport and Tourism at the University of Illinois, National Women’s Law Center, Women’s Sports Foundation, National Interscholastic Athletics Administrators Association National Federation of State High School Associations, and Association of Title IX Administrators. We conducted this performance audit from February 2017 to May 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Survey of Public High School Athletics Administrators The questions we asked in our survey of public high school athletics administers, as well as definitions used, are shown below. The blanks in the survey were filled in with the name of each athletics administrator’s school. Some questions were only asked if the athletics administrator responded a certain way to a prior question. Appendix IV: GAO Contact and Staff Acknowledgements GAO Contact Staff Acknowledgements In addition to the individual named above, Bill MacBlane (Assistant Director), Lauren Gilbertson and Jamila Jones Kennedy (Analysts-in- Charge), Christina S. Cantor, MacKenzie Cooper, Jill Lacey, Benjamin Sinoff, Andrew Stavisky, Sonya Vartivarian, and Khristi Wilkins made key contributions to this report. Also contributing to this report were James Bennett, Deborah Bland, Barbara Bovbjerg, Randy De Leon, Holly Dye, David Forgosh, Amy MacDonald, and Sheila R. McCoy. Related GAO Products K-12 Education: High School Sports Access and Participation. GAO-17-754R. Washington, D.C.: September 14, 2017. Child Welfare: Federal Agencies Can Better Support State Efforts to Prevent and Respond to Sexual Abuse by School Personnel. GAO-14-42. Washington, D.C.: January 27, 2014. K-12 Education: School-Based Physical Education and Sports Programs. GAO-12-350. Washington, D.C.: February 29, 2012.

Why GAO Did This Study Research has found that sports participation yields many benefits for youth. Girls' participation in sports has increased dramatically since the passage of Title IX in 1972, but is still lower than for boys. Further, investigations by OCR, which enforces and implements Title IX, have highlighted instances of disparities in the resources provided to girls' and boys' teams. GAO was asked to review how public high schools encourage equal athletic opportunities. This report examines (1) measures public high schools and athletics administrators have taken to encourage equal athletic opportunities for boys and girls, and (2) factors that affect boys' and girls' participation levels in public high school sports programs. GAO conducted a nationally generalizable probability survey of athletics administrators at 784 public high schools. GAO interviewed nine subject matter specialists selected to provide a range of perspectives. GAO also reviewed relevant federal laws, regulations, and guidance and interviewed OCR officials. What GAO Found According to GAO's nationally generalizable survey of athletics administrators, public high schools recently took various measures to encourage equal opportunities for boys and girls in sports. For example, a majority assessed resources such as equipment, travel opportunities, and facilities that they provided to girls' and boys' teams and some schools took steps to gauge student interest in specific sports as a means of encouraging equal opportunities, according to GAO's survey. Education's Office for Civil Rights (OCR) guidance indicates that Title IX coordinators—which school districts are required to designate and make visible per regulations for Title IX of the 1972 Education Amendments (Title IX)—should work closely with athletics administrators to determine whether action is needed to address any underrepresentation, or to otherwise encourage equal athletic opportunities. However, GAO estimates that 51 percent of athletics administrators either were unaware of or unsupported by their Title IX coordinator, according to the survey (see figure). These findings raise questions as to whether Title IX coordinators are familiar with and using Education's guidance. Officials from an association for Title IX coordinators said this lack of communication with athletics administrators may be related to some Title IX coordinators' limited understanding of Title IX and athletics. OCR officials said that they did not know the extent to which Title IX coordinators are working with their athletics administrators to encourage equal athletic opportunities because Education generally does not collect this information. Better information on Title IX coordinators could help Education support school districts' efforts to encourage equal sports opportunities for girls and boys. The factors that most affect boys' and girls' participation in public high school sports are the number of, and interest in, participation opportunities offered, according to GAO's survey and interviews with nine subject matter specialists. Though the survey provided no clear consensus on factors that discourage students from participating in sports, athletics administrators most often perceived students' competing responsibilities as discouraging participation. What GAO Recommends GAO is recommending that OCR determine the extent of K-12 Title IX coordinators' knowledge and use of tools in its existing guidance and use this information in its efforts to encourage them to work with athletics administrators to help ensure equal athletic opportunities. Education partially concurred, stating it would consider GAO's recommendation in its complaint investigations, technical assistance activities, and communication practice reviews.

gao_GAO-18-32

gao_GAO-18-32_0

Background NAS and Prenatal Opioid Use NAS is a withdrawal condition within infants that can result from the prenatal use of opioids by pregnant women. Prenatal opioid use occurs when a woman, during the course of her pregnancy, uses an opioid- based medication or substance. Prenatal opioid use can take various forms, including (1) the use of prescriptions for pain management, such as fentanyl and oxycodone; (2) medication-assisted treatment for opioid addiction, such as methadone and buprenorphine; (3) prescription drug misuse or use disorder (such as using an opioid without a prescription, using a different dosage than prescribed, or continuing to use an opioid when it is no longer needed for pain); and (4) illicit opioid use, such as heroin use. These types of prenatal opioid use are not mutually exclusive. A 2014 study found that almost 22 percent of pregnant Medicaid beneficiaries filled a prescription for an opioid during their pregnancy. Medication-assisted treatment—an approach that combines the use of certain medications and behavioral therapy—is generally considered by HHS and medical specialty societies to be the standard of care for treating pregnant women with opioid use disorders, depending on the individual and her circumstances. SAMHSA and several medical specialty societies, including the American College of Obstetricians and Gynecologists and the American Society of Addiction Medicine, have noted that providing medication-assisted treatment during pregnancy prevents complications associated with illicit opioid use, encourages prenatal care, and reduces the risk of obstetric complications. Further, women may use multiple substances in addition to opioids during pregnancy—known as maternal polysubstance use—such as tobacco, alcohol, or anti-depressants, among others. GAO reported in 2015 that the gaps in efforts to address prenatal opioid use and NAS most commonly cited by federal agency officials and experts were related to the treatment of prenatal opioid use and NAS. Agency officials and experts said that there has not been adequate research comparing different types of treatment approaches and that research is needed on how best to treat a pregnant woman with an opioid use disorder so that the treatment is most effective for the woman while offering minimal risk to the fetus. See GAO-15-203. For more information on factors that can affect access to medication- assisted treatment, see GAO, Opioid Addiction: Laws, Regulations, and Other Factors Can Affect Medication-Assisted Treatment Access, GAO-16-833 (Washington, D.C.: Sept. 2016). respiratory distress. There is currently no national standard of care for screening or treating NAS. There have been a few scoring tools developed to screen the infant to determine the appropriate course of treatment. Health care providers predominantly diagnose NAS using the Finnegan Neonatal Abstinence Scoring Tool, which calculates a score based on a variety of central nervous, metabolic, respiratory, and gastro-intestinal symptoms that might be observed. The American Academy of Pediatrics and the American College of Obstetricians and Gynecologists recommend that infants with NAS should not be initially treated with medication, known as pharmacologic treatment. Instead, these organizations recommend starting with non-pharmacologic treatment, which includes placing the infant in a dark and quiet environment, swaddling, breastfeeding, rooming-in with the mother, and providing high-calorie nutrition, among other things. For example, rooming-in—allowing the mother to reside with the infant during the infant’s treatment—may have benefits, such as helping to develop a bond between the mother and infant and to reduce the severity of the infant’s NAS symptoms. Pharmacologic treatment, such as using methadone or morphine, may be necessary only for the relief of moderate to severe signs of NAS. See figure 1 for more information on non-pharmacologic and pharmacologic treatment. Federal Resources on NAS and Prenatal Opioid Use HHS has published several guidance and educational resources related to NAS and prenatal opioid use. These documents serve as tools to help stakeholders, including state entities and health care providers, who work with this population. For example, SAMHSA published a clinical report for health care providers in 2016 that provides recommendations to help them when making decisions regarding the evaluation, care, and treatment of women with opioid use disorders and infants with NAS. As of July 2017, SAMHSA is in the process of producing a clinical guide based on this report and expects to publish an updated report later this year. In another example, SAMHSA published a guidance document in 2016 that aims to support the efforts of states, tribes, and local communities in addressing the needs of pregnant women with opioid use disorders and their infants. Among other things, the document includes strategies and guidance to promote care coordination among stakeholders, including child welfare agencies and medical professionals, when treating infants with NAS. (See app. II for a list of federal educational resources related to NAS and prenatal opioid use published by HHS.) Medicaid Program Overview As we previously noted, more than 80 percent of NAS cases are paid for by Medicaid, which is a federal-state health care program that finances health care coverage for low-income and medically needy populations, including children and aged or disabled adults. States administer their Medicaid programs within broad federal requirements and according to a state plan approved by CMS, the federal agency within HHS that oversees Medicaid. The Medicaid program allows states to design and implement their programs within certain federal parameters, resulting in more than 50 distinct state-based programs. For example, states generally determine the type and scope of services to cover, set payment rates that different health care providers will receive for various covered services, and pay these providers for claims submitted for services rendered. In addition, states vary in the extent to which they enroll beneficiaries in managed care versus delivering care through the more traditional fee-for-service model. Under a managed care delivery model, states typically contract with managed care plans to provide a specific set of Medicaid-covered services to beneficiaries and pay them a set amount per beneficiary—referred to as capitation payments—to provide those services. Under fee-for-service, Medicaid pays health care providers a fee for each service provided to a Medicaid beneficiary. Medicaid’s Early and Periodic Screening, Diagnostic, and Treatment benefit, which states are required to provide, covers comprehensive health screenings, preventive health services, and all medically necessary treatment and services—for Medicaid eligible-children under the age of 21—to correct or ameliorate health conditions discovered through screenings. Most Infants with NAS Complete Treatment in Various Hospital Settings, and in Selected States Medicaid Pays for Services Using Bundled Payments Hospital Settings for Treating Infants with NAS According to the literature we reviewed, most infants with NAS in the United States are treated in a hospital setting, often in the neonatal intensive care unit (NICU), which has a relatively high daily cost of care. Stakeholders we interviewed told us that these infants may also be treated in other hospital settings. According to state and perinatal collaborative officials in the four selected states we reviewed, infants diagnosed with NAS begin—and most complete—treatment for the condition in various hospital settings which provide different levels of care: a well newborn nursery (level I), special care nursery (level II), or NICU (level III or IV). For example, according to these officials, most infants with NAS in Vermont are treated in well newborn nurseries, while most infants with NAS in Kentucky and Wisconsin are treated in NICUs. West Virginia perinatal collaborative officials told us that about a third of infants with NAS are treated in well newborn nurseries, while two-thirds receive treatment in either a special care nursery or NICU. According to perinatal collaborative officials and hospital providers in the four selected states, the severity of the infant’s NAS symptoms or the hospital’s capability to treat NAS can determine whether the infant receives care in a nursery or NICU. Health care providers in the four selected states described the general clinical approach for treating infants with NAS. According to these providers, they generally start with non-pharmacologic treatment—for example, swaddling or placing the infant in a quiet, dark room. Health care providers may continue to monitor and assess the severity of the infant’s NAS symptoms using one of the available scoring tools for NAS. If the infant’s symptoms meet or exceed a certain threshold, these providers may initiate pharmacologic treatment by administering morphine or methadone, for example. Some perinatal collaborative officials that we interviewed in the four selected states told us that not all hospitals may have the capability to provide pharmacologic treatment. For example, these officials told us that level I hospitals—hospitals with only well newborn nurseries—in Kentucky and Wisconsin may not provide pharmacologic treatment to infants with NAS because these hospitals may not have the staff expertise to administer the needed medication and monitor the infants who receive it. Instead, these hospitals may transfer infants with NAS who require pharmacologic treatment to hospitals with higher levels of care, such as those with a NICU. Table 1 provides information on the eight selected hospitals in our review that provide NAS services. According to a 2015 study we reviewed, nationwide, infants with NAS who require pharmacologic treatment generally have longer average hospital stays (23 days) compared with infants with NAS who do not require such medication (17 days). Health care providers from our selected hospitals also indicated a similar trend—the average length of hospital stay in calendar year 2016 for infants with NAS who received pharmacologic treatment ranged from 7 to 30 days, while the stays for infants who did not require such medication ranged from 3 to 7 days. Medicaid Payments for NAS Treatment Services Provided in Hospital Settings Medicaid generally pays for NAS treatment services in our four selected states using a diagnosis-related group (DRG) based payment system, in which hospitals receive a fixed amount for a bundle of services. In general, the DRG-based system used in Medicaid pays for the medical services necessary for treating infants with NAS, such as medication, bed space, and nursing staff, according to CMS officials. CMS officials said that the DRG-based system generally does not pay for professional services, such as physician visits; instead, these services are typically paid under a fee-for-service payment schedule, in which states or contracted managed care plans pay health care providers directly for their services. Officials in our selected states said information on total Medicaid payments for hospital-based NAS services was not readily available. Several DRGs are typically used to bill Medicaid for services provided to infants. However, these codes alone cannot provide an accurate estimate of Medicaid payments for NAS treatment services because the codes are not used exclusively for NAS. For example, according to some health care providers we interviewed, two DRG codes that may be used to bill Medicaid and other payers for NAS treatment services are 791 (prematurity with major problems) and 793 (full term neonate with major problems). However, these codes could be used to bill for over 2,000 diagnoses—for example, pneumonia or measles. One state official said that while they could provide us with information on Medicaid payments for these infants, they could not parse out the costs by diagnosis codes, such as those related to NAS. Thus, estimates of total Medicaid payments based only on DRG codes likely overstate the amount paid for NAS hospital-based services. Officials from two of the four selected states told us that their state has a public health surveillance system that tracks the incidence of infants diagnosed with NAS; however, the surveillance systems do not capture financial information, including Medicaid payments for NAS. At our request, one of the states cross-referenced their surveillance and Medicaid data and estimated that in 2016, their state Medicaid program spent over $22 million to treat 1,565 infants with NAS. While selected states generally could not provide information on total Medicaid payments for infants with NAS, some hospitals in our selected states were able to generate this information at our request using diagnosis codes that they identified as related to NAS from hospital claims data. Six of our selected eight hospitals reported that in calendar year 2016, the average Medicaid payment for treating infants with NAS ranged from about $1,500 to about $20,200 per infant per stay. The wide range in Medicaid payment averages may be because the averages included both infants who did and did not require pharmacologic treatment and because these hospitals treated infants in various settings, such as a nursery or a NICU. The literature we reviewed also had limited information on Medicaid payments for NAS treatment services provided in hospitals. A recent study reported that from 2009 through 2012—the most recent data available at the time of the study—Medicaid payments to hospitals for NAS treatment services increased from about $564 million to $1.2 billion nationwide. Some Infants with NAS May Complete Treatment in Non- Hospital Settings Available in Certain States, and in These States Medicaid Pays for a Subset of Services Non-Hospital Settings for Treating Infants with NAS While most infants with NAS typically complete treatment in a hospital setting, stakeholders told us that some of these infants may be transferred to a non-hospital setting to complete pharmacologic treatment and continue non-pharmacologic treatment. HHS officials told us that there is not a comprehensive list of facilities that may treat infants with NAS outside of the hospital. Based on information from the stakeholders we interviewed and the literature we reviewed, we identified two types of non-hospital settings available in certain states that treat infants with NAS: (1) outpatient clinics and programs and (2) neonatal withdrawal centers. For the purposes of this report, we defined neonatal withdrawal centers as facilities that can treat infants who are prenatally exposed to drugs, including infants with NAS, within the facility. Outpatient clinics and programs to treat NAS Through stakeholder interviews and the literature we reviewed, we identified examples of outpatient clinics and programs in certain states where infants with NAS can continue pharmacologic treatment after their discharge from the hospital. For example, some stakeholders we interviewed told us about a Neonatal Medical Follow-Up Clinic in Vermont used to follow-up with infants with NAS who have been discharged from the hospital and are being weaned off methadone on an outpatient basis. Hospital providers train the infant’s family on how to administer the infant’s medication at home and provide a referral to the clinic. After hospital discharge, the infant and family have follow-up visits in the clinic every 1 to 2 weeks, during which the family discusses with health care providers the weaning schedule and demonstrate how they administer the infant’s medication. Health care providers told us that they also encourage the family to continue providing non-pharmacologic treatment to the infant. After weaning is complete, the infant continues to follow-up at the clinic every 1 to 2 months until the infant reaches 12 to 18 months of age. Literature we reviewed indicated that other outpatient treatment clinics or programs such as the one in Vermont have been established or considered in other states. Specifically, four studies we reviewed described instances in which infants began their treatment in the hospital but completed their treatment through a dedicated outpatient program in Florida, Ohio, and Pennsylvania. Each study noted that the inpatient-to- outpatient approach can result in a shorter hospital length of stay. For example, one 2015 study found that infants who began treatment in a hospital and completed their treatment in an outpatient setting stayed in the hospital an average of 11 days, compared to infants who completed treatment in the hospital, where the stays averaged about 25 days. However, the studies also noted that the inpatient-to-outpatient approach resulted in a longer overall treatment duration across the two settings. Neonatal withdrawal centers to treat NAS Some stakeholders we interviewed, including health care providers, described examples of neonatal withdrawal centers in two states, where infants with NAS can continue pharmacologic treatment after their discharge from the hospital. Health care providers in these facilities told us that in Washington and West Virginia, some infants with NAS who began treatment in a hospital may be referred to these facilities, where they reside until they complete treatment and are discharged from the facility. These providers explained that in these facilities, the infants are placed in nursery rooms, where health care providers can monitor them and administer and adjust their medication as needed. In addition, nursing staff or other caregivers are responsible for providing continuous non-pharmacologic treatment, and mothers are encouraged to visit and continue this care. For example, health care providers told us that in Washington, two to three infants may share a nursery room where trained caregivers provide them with non-pharmacologic treatment. In West Virginia, health care providers said infants are typically placed in individual nursery rooms where nurses provide them with non- pharmacologic treatment. The rooms in the West Virginia facility are also equipped with a rocking chair to encourage mothers to visit and provide this care as well. Health care providers told us that the facility currently offers one nursery room equipped with a bed to help prepare mothers on what to expect after discharge; they also said that they encourage mothers to spend the night prior to the infant’s discharge from the facility. (See text box below). Treating infants with neonatal abstinence syndrome (NAS) in a neonatal withdrawal center One health care provider from a neonatal withdrawal center told us that the practice of rooming-in helps to facilitate the bond between the mother and infant. He also said that rooming-in allows health care providers to model care for the mothers and for mothers to learn how to care for their infants with NAS. Health care providers told us that the facility currently offers one nursery room equipped with a bed to help prepare mothers on what to expect after discharge and that they encourage mothers to spend the night prior to the infant’s discharge from the facility. One health care provider told us that one mother, after staying overnight with her infant, realized that she was not prepared to take care of her infant and consequently gave up custody of the infant. Because of rooming-in, health care providers were able to ensure that the infant was safe because the mother came to this realization at the facility, rather than alone at home. Although the lack of physical space at the facility currently makes it difficult to accommodate rooming-in for the entire course of the infant’s treatment, these providers noted the importance of this practice and that they are committed to parental involvement when treating infants with NAS at their facility. Figure 2 depicts nursery rooms in the neonatal withdrawal center in West Virginia. Efforts are also underway to open a neonatal withdrawal center in Arizona and Ohio, according to stakeholders we interviewed. Stakeholders we interviewed and the literature we reviewed suggest some limitations as well as benefits of treating infants with NAS in non- hospital settings, including factors to consider in these settings. Health care providers from one of the hospitals we visited in Vermont told us that their hospital is the only one in the state that allows infants with NAS to complete pharmacologic treatment through the Vermont outpatient clinic because they have established the necessary infrastructure to ensure families’ compliance and safe practices at home. These providers said that they worked with one local pharmacy to ensure proper dispensing of the medication. Additionally, these providers measured the amount of medication left over at each follow- up visit with the families. Some state and perinatal collaborative officials told us that neonatal withdrawal centers may not be the best environment to treat infants with NAS because these settings may limit a mother’s access to her infant, since she may not always be allowed to reside with the infant. Such limits, according to officials, do not facilitate bonding between mother and infant. Another state perinatal collaborative official, as well as health care providers and staff, told us that neonatal withdrawal centers may be better for treating infants with NAS because the environment is quieter and less stimulating than hospital settings, such as NICUs. Several studies we reviewed also emphasized that the inpatient-to- outpatient approach requires ongoing coordination, communication, and commitment from multidisciplinary providers, as well as the families. These studies highlighted instances in which these approaches reduced the length of the infants’ stay in a hospital, though the studies emphasized that more work needs to be done to determine whether these are the optimal approaches for infants with NAS, as well as the potential long-term benefits of such approaches. Medicaid Payments for NAS Treatment Services Provided in Non-Hospital Settings Medicaid pays for NAS treatment services provided in the non-hospital settings we identified in certain states, according to CMS officials and other stakeholders we spoke with, but generally pays for these services separately, in contrast with the single bundled payment paid to hospitals. State officials and health care providers in the non-hospital settings we examined described various ways in which Medicaid covered services they provided to treat infants with NAS. For example: Outpatient follow-up clinic in Vermont. State officials and staff at this facility told us that the Vermont Medicaid program pays for an infant’s outpatient physician visits using a fee-for-service payment schedule. They added that the Vermont Medicaid program also pays for the infant’s medication used in pharmacologic treatment and explained that the pharmacy that dispenses the medication bills Medicaid for these services. Neonatal withdrawal center in Washington. Health care providers at this facility told us that the Washington Medicaid program or their contracted managed care plans pay for physician visits using a fee- for-service payment schedule, noting that the facility decided to stop billing Medicaid for medical supplies because of the low reimbursement. Additionally, these providers suggested that because the facility does not meet the Medicaid standards required for receiving payment for hospital inpatient, nursing, or other covered facility services, the facility is ineligible to receive Medicaid payment for the costs of room and board. These providers said that they receive funding for the cost of these services through state appropriations, foster care payments, city contracts, grants, and private donations. Neonatal withdrawal center in West Virginia. State officials and health care providers at this facility told us that West Virginia pays for NAS services through two mechanisms, depending on whether the infant is in foster care. Specifically, if the infant is in foster care, the facility receives a bundled payment from the state Medicaid program and the Bureau of Children and Families. However, if the infant is not in foster care, the state Medicaid program pays for physician visits using a fee-for-service schedule. Additionally, the facility can receive payment under a per diem rate that is negotiated with state Medicaid managed care plans. The health care providers said that they also receive funding through grants and private donations to help cover the costs of NAS services. Stakeholders we interviewed and literature we reviewed suggest that the costs of treating infants with NAS in non-hospital settings were lower than treating them in hospital settings. However, supporting data and research of the costs in different settings are anecdotal or otherwise limited. For example: Health care providers from the neonatal withdrawal center in Washington told us that their facility could treat infants at a lower average cost per day than could hospitals—at about $700 per infant per day compared to an average cost of about $1,500-2,500 per infant per day in a hospital. These providers said that this cost savings is in part due to their limited staffing of nurses and their ability to leverage specially trained caregivers to provide infants with non-pharmacologic treatment and hands-on care, such as feeding and bathing. These providers also said they use volunteers to help with household duties, such as laundry and replenishing supplies. A health care provider from the neonatal withdrawal center in West Virginia conducted a study that found that the average daily charges per infant were about $400 in their facility, compared to about $2,600 in a special care nursery and $4,000 in a NICU. Two studies we reviewed found that inpatient-to-outpatient treatment approaches reduced hospital costs for NAS treatment; however, these studies were not generalizable and did not account for the duration of treatment across the two settings. Specifically, one study found that an inpatient-to-outpatient treatment approach reduced hospital length of stay by 55 percent—estimated to save hospitals $396 million annually—compared with treatment provided solely in a hospital. The second study found that infants who received care for NAS through an inpatient-to-outpatient treatment approach had an average length of stay of 13 days and cost about $14,000, while an inpatient- only approach had an average length of stay of 25 days and cost about $28,000. Recommended Practices for Addressing NAS Include Prioritizing Non-Pharmacologic Treatment, While Challenges Include Maternal Use of Multiple Substances Recommended Practices for Addressing NAS Include Prioritizing Non- Pharmacologic Treatment, Educating Mothers, and Addressing Stigma The 32 stakeholders we interviewed and the literature we reviewed identified several recommended practices for addressing NAS—that is, treating women with opioid use disorders during pregnancy or treating infants diagnosed with NAS after birth. The most frequently recommended practices were (1) prioritizing non-pharmacologic treatment, such as allowing the mother to reside with the infant during treatment, to facilitate the mother-infant bond; (2) educating mothers on prenatal care, treatment for NAS, and available resources for after an infant’s discharge; (3) educating health care providers on the stigma faced by women who use opioids during pregnancy and on how to screen for and treat NAS; and (4) using a protocol in a hospital or non-hospital setting for screening and treating infants with NAS. These recommended practices are described in more detail below. Volunteer programs to provide non- pharmacologic treatment for neonatal abstinence syndrome (NAS) Some stakeholders told us that some hospitals have established volunteer cuddler programs that train volunteers to help provide some of these non-pharmacologic treatments—-namely, swaddling, feeding, soothing, and coddling infants. However, health care providers at some facilities noted that volunteers are not necessarily available during late shifts. indicated that non-pharmacologic treatment may (1) facilitate the mother- infant bond, (2) reduce the severity of NAS symptoms, (3) reduce the need for pharmacologic treatment, and (4) reduce the length of an infant’s hospital stay. For example, two of the articles we reviewed noted that rooming-in has been shown to help decrease the need for pharmacologic treatment, the number of admissions to the NICU, and the length of an infant’s hospital stay. Additionally, 17 of the stakeholders we interviewed and nine articles we reviewed recommended that mothers be allowed to breastfeed while their infants are treated for NAS, as it helps to build a bond between the mother and infant. Most of these articles also noted that breastfeeding has been shown to reduce the severity of NAS. Educating mothers on prenatal care, treatment for NAS, and resources for after an infant’s hospital discharge. Most stakeholders we interviewed and several of the literature articles we reviewed recommended providing comprehensive, ongoing education to mothers on prenatal care and treatment for NAS and on the resources that are available after an infant’s discharge. (See text box below). The stakeholders and literature indicated that this education may (1) facilitate a non-combative relationship between the mother and health care providers; (2) help to reassure and support the mother, who may feel responsible for the infant’s suffering, in addition to facilitating treatment of NAS; and (3) help the mother understand her infant’s behavior and develop greater confidence in her parenting skills. For example, one article noted that an infant’s withdrawal behavior, such as fisting, back arching, and jaw clenching, may be misinterpreted by the mother as dislike of touch, and that educating mothers on these behaviors can help alleviate feelings of guilt. Education for mothers on prenatal care, treatment for neonatal abstinence syndrome (NAS), and resources for after an infant’s hospital discharge Explaining to the mother during the prenatal period what she can expect when the infant is born to help ensure she understands the effects of and treatment for NAS; Informing the mother about non-pharmacologic treatment techniques that can help reduce the severity of the infant’s NAS symptoms; Modeling good parenting skills, such as demonstrating how to comfort an infant who may be crying inconsolably for hours because of withdrawal; and Informing the mother about contraception for preventing future pregnancies. Educating health care providers on the stigma faced by women who use opioids during pregnancy, and how to screen for and treat NAS. Most stakeholders we interviewed and several of the literature articles we reviewed recommended educating health care providers, including providers who are not addiction specialists, on both the stigma faced by women who use opioids during pregnancy as well as on how to screen for and treat infants with NAS. The stakeholders and literature indicated that this education may: (1) improve care so that mothers with opioid use disorders feel more comfortable seeking and obtaining prenatal care, (2) help health care providers know how to recognize NAS symptoms to help ensure infants receive appropriate treatment, and (3) allow for more consistency among these providers in NAS screening and treatment. For example, 26 stakeholders told us that educating health care providers about stigma is important because provider attitudes affect how and if pregnant women obtain prenatal care and treatment for their opioid use disorders, which can affect the severity of NAS. Additionally, several articles we reviewed noted the importance of educating and training clinicians on how to administer the screening tools used to identify infants with NAS, which helps ensure infants are identified and receive optimal care. Using a protocol for screening and treating infants with NAS. While there is no single national standard of care for screening and treating NAS, most stakeholders we interviewed and several of the literature articles we reviewed recommended that hospital and non-hospital settings use a protocol to screen for and treat infants with NAS. The stakeholders and literature indicated that having a protocol can help: (1) identify infants at risk for NAS, (2) ensure that care is provided consistently, and (3) reduce the length of stay for infants receiving pharmacologic treatment. For example, the stakeholders we interviewed explained that a standard protocol also helps health care providers understand the tools used to assess the severity of NAS; know the types of medication used in treatment, including amounts and duration; and learn how to wean the infant off these medications. Similarly, one article we reviewed noted that infants who were treated at facilities that adopted standard treatment protocols experienced shorter durations of pharmacologic treatment compared with infants who were treated at facilities that did not use a standard protocol. Challenges Faced by Health Care Providers in Addressing NAS Include Maternal Use of Multiple Substances and Stigma Faced by Women Who Use Opioids Stakeholders we interviewed and literature we reviewed identified several challenges health care providers face in their efforts to address NAS. The most frequently cited challenges included (1) the use of multiple substances by pregnant women, which can exacerbate NAS; (2) the stigma faced by women who use opioids during pregnancy, which may affect whether they seek prenatal care to address NAS, among other things; (3) hospital staff burden and limited physical capacity to care for infants with NAS; (4) limited coordination of care for mothers and infants with NAS; and (5) gaps in research and data on NAS. These challenges are described in more detail below. The use of multiple substances by pregnant women, which can exacerbate NAS. Most stakeholders we interviewed and some of the literature we reviewed noted that the use of multiple substances by pregnant women, including opioids—referred to as maternal polysubstance use—can be a challenge, and some stated that the use of these substances can exacerbate NAS symptoms. According to the stakeholders, the substances can include methamphetamines, nicotine, alcohol, cocaine, marijuana, benzodiazepines, and Gabapentin. The stakeholders and literature indicated that maternal polysubstance use can lead to multiple conditions in the infant—such as prematurity or Hepatitis C—that can exacerbate NAS symptoms and prolong the length of an infant’s hospital stay. For example, one expert noted that many women with opioid use disorders are also heavy cigarette smokers, and the nicotine typically exacerbates NAS withdrawal symptoms. Additionally, officials from a hospital and non-hospital setting we visited told us that they had developed a separate protocol for treating infants exposed to multiple substances that includes the use of several medications to address the more severe NAS withdrawal symptoms. Stigma faced by women who use opioids which may affect whether they seek prenatal care to mitigate the severity of NAS, among other things. Most stakeholders we interviewed and several of the literature articles we reviewed noted that the stigma faced by pregnant women with opioid use disorders is a challenge in addressing NAS. The stakeholders and literature indicated that stigma may: (1) prevent pregnant women from seeking substance use treatment or prenatal care; (2) prevent them from disclosing their drug use to health care providers during pregnancy; or (3) cause the women to fear punitive effects, such as losing custody of their children, being detained, or losing their jobs. For example, officials from one perinatal quality collaborative told us that these women may fail to seek care because of stigma, which can ultimately make it more difficult for health care providers to build relationships with these women and identify infants at risk for NAS. Hospital staff burden and limited physical capacity to care for infants with NAS. According to most stakeholders we interviewed and some literature we reviewed, staff burden and a limited physical capacity at facilities can pose challenges for addressing NAS. The stakeholders and literature indicated that there is increased burden on staff to care for these infants because they require frequent, personal attention. For example, the stakeholders explained that a hospital may have to increase the number of nurses on duty in order to provide the care the infants need. Health care providers at one hospital said that nurses still struggle to care for infants with NAS, even with additional staff, because these infants are overstimulated, cry, and do not eat or sleep well. As a result, they require much time and one-on-one attention—including cuddling— from nurses. With respect to physical capacity, some stakeholders told us that limited physical capacity can make it difficult to (1) find space in the facility where the infants can be protected from high levels of stimulation and (2) facilitate the mother-infant bond. For example, some stakeholders told us that hospitals may not have a dedicated space for rooming-in, making it more difficult to facilitate bonding between mothers and infants. Limited coordination of care for mothers and infants with NAS. Most stakeholders we interviewed explained that the lack of coordination among health care providers and others for the mother and infant with NAS during the prenatal period, after the infant is born, and following the infant’s discharge can be a challenge. This coordination includes organizing patient care activities and sharing information among health care providers, social workers, and all other participants concerned with the mother and infant’s care. The stakeholders indicated that this lack of coordination can make it difficult for families to get the resources or support they need. (See text box below). For example, some stakeholders told us that women may miss health care visits because of a lack of access to enabling services such as transportation or child care. Limited coordination of care for mothers and infants with neonatal abstinence syndrome (NAS) One expert told us that there is a disproportionate number of infants with NAS born in rural areas. Infants in these areas may be discharged from the hospital without many follow-up services, such as transportation and care coordination. Gaps in research and data on NAS. Some stakeholders we interviewed noted that gaps in research and data on NAS make it challenging to conduct research on the affected population and fully understand the magnitude of the problem. The stakeholders indicated that there are gaps in adequate research and data on (1) the different types of treatment approaches for NAS; (2) the extent and effects of maternal polysubstance use among pregnant women; (3) the long-term effects of prenatal drug exposure, including the effects seen in childhood and adolescence; and (4) the efforts to ensure more consistent provider diagnosis and screening, such as through an improved screening tool. For example, the stakeholders told us that gaps in research and data may contribute to a lack of a national standard of care for screening and treating infants with NAS. According to some stakeholders, this may result in missed opportunities for identifying and treating infants with NAS. Some stakeholders also told us that because of gaps in research on the long- term effects of prenatal drug exposure, there is limited information on the types of services that infants with NAS may need in early childhood. Additionally, some stakeholders noted they found that because NAS was not consistently diagnosed and coded in medical records using diagnosis codes, the condition may be under-reported, and researchers may be limited in their ability to track these infants. HHS’s Strategy Includes Recommendations Related to Addressing NAS but Lacks Priorities, Timeframes, and Responsibilities for Implementing the Recommendations In May 2017, HHS published the Protecting Our Infants Act: Report to Congress, which—among other things—presents a strategy that identifies key recommendations related to addressing NAS. Specifically, HHS’s strategy—known as the Protecting Our Infants Act: Final Strategy—made 39 recommendations related to the prevention, treatment, and related services for NAS and prenatal opioid use. Of the 39 recommendations HHS made in its report, we found that 28 of them directly relate to the recommended practices or challenges that we describe above. For example, the Strategy recommends the following: promoting non-pharmacologic treatment, such as rooming-in; providing continuing medical education to health care providers for managing and treating infants with NAS, such as on NAS treatment protocols; conducting research on the long-term effects of prenatal drug exposure so that appropriate services can be developed for infants with NAS; and establishing clear definitions of NAS and standardizing the use of diagnosis codes to collect more meaningful and actionable data on NAS. According to the Strategy, the recommendations will be used to inform planning and policy across HHS. However, HHS does not include any information in the Strategy on how the department and other stakeholders will implement the recommendations. Specifically, HHS does not include in its Strategy the following: the explicit priorities among the numerous recommendations and associated efforts the department has initiated related to NAS; timeframes for partial or full implementation of these recommendations; clear roles and responsibilities for the recommendations, such as the extent to which HHS will need to rely on the medical community and federal and public stakeholders for implementation; and the methods that will be used to assess the department’s progress in implementing any of these recommendations. HHS officials told us that they expect to develop a separate plan to guide implementation of the recommendations and that efforts to develop this plan were likely to begin in July 2017. However, as of September 2017, HHS could not provide any documentation that it had started to develop this implementation plan or establish a timeline for completing the plan; nor was HHS able to provide any information on what the plan may include. Having such a plan in place is important to ensure priorities are known and responsibilities are clear so that agencies and stakeholders can take appropriate action. Federal internal control standards call for agencies to have defined objectives clearly as part of their objective- setting process and to assign roles and responsibilities for achieving these objectives. Objectives defined in specific and measurable terms allow for the assessment of performance toward achieving objectives. Furthermore, leading principles on sound planning we have identified in our prior work call for developing robust plans to achieve agency goals. Until HHS finalizes an implementation plan that includes specific priorities, timeframes, responsibilities, and methods for evaluating progress, it is at risk of not being able to provide reasonable assurance that it can successfully implement these recommendations in a timely manner and assess the effectiveness of its efforts. Conclusions The rising opioid crisis has caused a significant increase in the number of infants born and diagnosed with NAS, a condition that affects infants and their families, hospitals, and other health care providers who are treating them. The increase in infants born with NAS also increases medical and other treatment costs experienced by the federal government and states. HHS recently published a strategy with key recommendations that have the potential to address some of the challenges related to treating NAS. However, HHS lacks a sound plan for implementing these recommendations. The absence of such planning raises questions about whether and when HHS will be able to implement these recommendations in a timely manner and be able to assess its progress. Recommendation for Executive Action The Secretary of HHS should expeditiously develop a plan—that includes priorities, timeframes, clear roles and responsibilities, and methods for assessing progress—to effectively implement the NAS-related recommendations identified in the Protecting Our Infants Act: Final Strategy. (Recommendation 1) Agency Comments and Our Evaluation We provided a draft of this report to HHS for review, and HHS provided written comments, which are reprinted in appendix III. HHS also provided technical comments, which we incorporated as appropriate. In its written comments, HHS concurred with our recommendation to expeditiously take steps to address NAS and re-stated that its Strategy will be used to inform planning and policy across HHS. Specifically, HHS said that as part of its broader initiative to address the opioid crisis, the department will develop and implement a plan—that will include priorities, timeframes, roles and responsibilities, and methods for assessing progress—to address as appropriate and possible, the NAS-related recommendations in its Strategy. HHS also stated that full implementation would be contingent on funding, though it provided no information on how much funding was needed or how the funding would be used. Developing a plan to guide implementation can help the department determine what resources, if any, are needed to implement the recommendations in its Strategy. We are sending copies of this report to the appropriate congressional addressees, the Secretary of Health and Human Services, and other interested parties. In addition, the report will be available at no charge on GAO’s website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-7114 or at iritanik@gao.gov. Contact points for our Office of Congressional Relations and Office of Public Affairs can be found on the last page of this report. GAO staff who made major contributions to this report are listed in appendix IV. Appendix I: Detailed Scope and Methodology To address our first three audit objectives to describe care settings for treating infants with neonatal abstinence syndrome (NAS), Medicaid payment for NAS treatment, and the recommended practices and challenges for addressing NAS, we selected 32 stakeholders based on their relevant experience to cover a range of perspectives on NAS. Specifically, these stakeholders included those from site visits we conducted to four states—Kentucky, Vermont, West Virginia, and Wisconsin. We selected these states because they met the following criteria: 1. the state had one of the top 10 highest incidence rates of NAS, according to data from the Centers for Disease Control and Prevention (CDC) for 2013, the most recent year of publicly available data; 2. the state provided variation in United States geographic regions with high rates of NAS, as of 2012; 3. more than 40 percent of births in the state were financed by Medicaid, according to a 2016 Kaiser Family Foundation Medicaid Budget Survey; and 4. the state has a perinatal quality collaborative—a state or multi-state network of teams working to improve health outcomes for mothers and infants—with work related to NAS, which we identified through the American College of Obstetricians and Gynecologists. As part of these site visits, we interviewed (1) officials from each of the four states, including Medicaid officials, Maternal and Child Health Directors, and Women’s Services Coordinators; (2) representatives from the four state perinatal collaboratives; (3) health care providers (including physicians or nurses) from eight hospitals of varying levels of care (two hospitals in each state), which were selected based on recommendations from the state perinatal collaboratives because of the hospitals’ experience treating NAS; and (4) officials from a residential treatment facility in each of the four states that provide prenatal and postpartum care to mothers, which were also selected based on recommendations from the state perinatal collaboratives regarding the facilities’ experience with pregnant women with opioid use disorders and their infants with NAS. In addition to our site visits, we selected 12 additional stakeholders that included health care providers or administrators in non-hospital settings across the United States; officials from medical specialty societies; and experts. Specifically, we spoke with (1) health care providers (including physicians or nurses) or administrators from four non-hospital settings in Arizona, Ohio, Washington, and West Virginia, which were selected based on recommendations from stakeholders we interviewed and on the availability of such settings and their experience treating NAS; (2) health care providers from five medical specialty societies, including the American Academy of Pediatrics, the American College of Obstetricians and Gynecologists, the American Society of Addiction Medicine, MedNAX (a network of physicians that specialize in neonatal care, including NAS treatment), and the National Association of Neonatal Nurses; and (3) three experts, including the authors of published literature we reviewed. We interviewed each of these 32 stakeholders and requested information from stakeholders about treating infants with NAS, including the utilization of available hospital and non-hospital care settings and associated costs of treatment services. For example, we requested protocols for screening and treating infants with NAS from hospital and non-hospital care settings. We reviewed available protocols provided by hospitals and a non-hospital care setting. We also reviewed available information reported by state officials, hospital and non-hospital providers, and state perinatal collaboratives on the utilization of hospital and non-hospital care settings, the facilities’ cost of treating NAS in hospital and non-hospital settings, the lengths of stay for treating infants with NAS, or the amount of Medicaid payments for treating infants with NAS. We discussed the information provided by stakeholders and examined the information for obvious errors. The information obtained from these stakeholders is not generalizable to other states or other hospital and non-hospital settings. In addition, in some cases, stakeholders used different methods to collect the information they reported, including information on Medicaid payments; as a result, the information reported by stakeholders is not directly comparable. Additionally, we interviewed officials from HHS, including those from the Centers for Medicare & Medicaid Services (CMS) and HHS’s Behavioral Health Coordinating Council—which includes officials from the Substance Abuse and Mental Health Services Administration (SAMHSA), the Indian Health Service, the Centers for Disease Control and Prevention (CDC), and the Food and Drug Administration, among others—concerning NAS treatment services, settings of care, Medicaid payment, and recommended practices and challenges related to addressing NAS. We also conducted a comprehensive literature review to identify relevant studies on NAS published in peer-reviewed journals from January 2013 to December 2016. We searched more than 40 databases for research published in relevant peer-reviewed journals, including BIOSIS Previews®, Embase®, Gale Group Health Periodicals Database, MEDLINE®, and New England Journal of Medicine. Key search terms included “neonatal abstinence syndrome,” “neonatal opioid withdrawal syndrome,” and “newborn infants.” After excluding duplicates, we identified and reviewed 325 abstracts. For those abstracts we found relevant, we obtained and reviewed the full study and selected 40 that were relevant to (1) hospital and non-hospital settings and related treatment services for infants with NAS; (2) the costs associated with treating infants with NAS, including Medicaid payments for services in these care settings; or (3) recommended practices and challenges for addressing NAS. We examined the methodologies for each of these studies and interviewed some of their authors. We determined that the studies were sufficiently reliable for our audit objectives. For a complete list of the studies we reviewed, see below. To examine our last audit objective on HHS’s strategy related to addressing NAS, we interviewed agency officials and reviewed agency documents on the agency’s efforts to develop a strategy. Specifically, we interviewed relevant officials from CMS and HHS’s Behavioral Health Coordinating Council concerning their efforts to develop a strategy related to addressing NAS. In reviewing relevant HHS documents, we focused on HHS’s Protecting Our Infants Act Report to Congress, which includes a strategy to address identified gaps, challenges, and recommendations related to NAS and prenatal opioid use. In addition, we reviewed the relevant standards for internal control in the federal government and the relevant criteria from GAO’s body of work on effectively managing performance under the Government Performance and Results Act (GPRA) of 1993 and the GPRA Modernization Act of 2010. Studies GAO Reviewed Allocco, E., M. Melker, F. Rojas-Miguez, C. Bradley, K. A. Hahn, and E. M. Wachman. “Comparison of Neonatal Abstinence Syndrome Manifestations in Preterm Versus Term Opioid-Exposed Infants.” Advances in Neonatal Care, vol. 16, no.5 (2016). Artigas, V. “Management of Neonatal Abstinence Syndrome in the Newborn Nursery.” Nursing for Women’s Health, vol. 18, issue 6 (Dec. 2014/Jan. 2015). Asti, L., J. S. Magers, E. Keels, J. Wispe, and R. E. McLead. “A Quality Improvement Project to Reduce Length of Stay for Neonatal Abstinence Syndrome.” Pediatrics, vol. 135, no. 6 (2015). Busch, D. W. “Clinical management of the Breast-Feeding Mother-Infant Dyad in Recovery from Opioid Dependence.” Journal of Addictions Nursery, vol. 27, no. 2 (2016). Casper, T. and M. Arbour. “Evidence-Based Nurse-Driven Interventions for the Care of Newborns with Neonatal Abstinence Syndrome.” Advances in Neonatal Care, vol. 14, no. 6 (2014). Chau, K. T., J. Nguyen, B. Miladinovic, C. M. Lilly, T. L. Ashmeade, and M. Balakrishnan. “Outpatient Management of Neonatal Abstinence Syndrome: A Quality Improvement Project.” The Joint Commission Journal on Quality and Patient Safety, vol. 42, no. 11 (2016). Cirillo, C. and K. Francis. “Does Breast Milk Affect Neonatal Abstinence Syndrome Severity, the Need for Pharmacologic Therapy, and Length of Stay for Infants of Mothers on Opioid Maintenance Therapy During Pregnancy?” Advances in Neonatal Care, vol. 16, no.5 (2016). Clark, L. and A. Rohan. “Identifying and Assessing the Substance- Exposed Infant.” MCN in Advance (2015). Demirci, J. R., D. L. Bogen, and Y. Klionsky. “Breastfeeding and Methadone Therapy: The Maternal Experience.” Substance Abuse, vol. 36, no. 2 (2015). Edwards, L. and L. F. Brown. “Nonpharmacologic Management of Neonatal Abstinence Syndrome: An Integrative Review.” Neonatal Network, vol. 35, no. 5 (2016). Gregory, K. E. “Caring for the Infant with neonatal Abstinence Syndrome in a Community-Based Setting.” The Journal of Perinatal & Neonatal Nursing, (2014). Grim, K., T. E. Harrison, and R. T. Wilder. “Management of Neonatal Abstinence Syndrome from Opioids.” Clinics in Perinatology, (2013). Hahn, J., A. Lengerich, R. Byrd, R. Stoltz, J. Hench, S. Byrd, and C. Ford. “Neonatal Abstinence Syndrome: The Experience of Infant Massage.” Creative Nursing, vol. 22, issue 1 (2016). Hall, E. S., S. L. Wexelblatt, M. Crowley, J. L. Grow, L. R. Jasin, M. A. Klebanoff, R. E. McClead, J. Meinzen-Derr, V. K. Mohan, H. Stein, and M. C. Walsh. “A Multicenter Cohort Study of Treatments and Hospital Outcomes in Neonatal Abstinence Syndrome.” Pediatrics, vol. 134, no. 2 (2014). Hall, E. S., S. L. Wexelblatt, M. Crowley, J. L. Grow, L. R. Jasin, M. A. Klebanoff, R. E. McClead, J. Meinzen-Derr, V. k. Mohan, H. Stein, and M. C. Walsh. “Implementation of a Neonatal Abstinence Syndrome Weaning Protocol: A Multicenter Cohort Study.” Pediatrics, vol. 136, no. 4 (2015). Holmes, A. V., E. C. Atwood, B. Whalen, J. Beliveau, J. D. Jarvis, J. C. Matulis, and S. L. Ralston. “Rooming-In to Treat Neonatal Abstinence Syndrome: Improved Family-Centered Care at Lower Cost.” Pediatrics, vol. 137, no. 6 (2016). Jones, H. E., K. Deppen, M. L. Hudak, L. Leffert, C. McClelland, L. Sahin, J. Starer, M. Terplan, J. M. Throrp Jr., J. Walsh, and A. A. Creanga. “Clinical Care for Opioid-Using Pregnant and Postpartum Women: The Role of Obstetric Providers.” American Journal of Obstetrics & Gynecology, vol. 210, issue 4 (2014). Jones, H.E., C. Seashore, E. Johnson, E. Horton, K.E. O’Grady, K. Andringa, M. R. Grossman, B. Whalen, and A.V. Holmes. “Brief Report: Psychometric Assessment of the Neonatal Abstinence Scoring System and the MOTHER NAS Scale.” American Journal on Addictions, (2016). Kraft, W.K., M. W. Stover, and J. M. Davis. “Neonatal Abstinence Syndrome: Pharmacologic Strategies for the Mother and Infant.” Seminars in Perinatology, vol. 40, issue 3 (2016). Krans, E. E., G. Cochran, and D. L. Bogen. “Caring for Opioid Dependent Pregnant Women: Prenatal and Postpartum Care Considerations.” Clinical Obstetrics and Gynecology, vol. 58, no. 2 (2015). Lee, J., S. Hulman, M. Musci Jr., and E. Stang. “Neonatal Abstinence Syndrome: Influence of a Combined Inpatient/Outpatient Methadone Treatment Regimen on the Average Length of Stay of a Medicaid NICU Population.” Population Health Management, vol. 18, no.5 (2015). MacMullen, N. J., L. A. Dulski, and P. Blobaum. “Evidence-Based Interventions for Neonatal Abstinence Syndrome.” Pediatric Nursing, vol. 40, no. 4 (2014). Maguire, D. J., “Mothers on Methadone: Care in the NICU.” Neonatal Network, vol. 32, no. 6 (2013). Marcellus, L. “Supporting Women with Substance Use Issues: Trauma- Informed Care as a Foundation for Practice in the NICU.” Neonatal Network, vol. 33, no.6 (2014). McKeever, A. E., S. Spaeth-Brayton, and S. Sheerin. “The Role of Nurses in Comprehensive Care Management of Pregnant Women with Drug Addiction.” Nursing for Women’s Health, vol. 18, no.4 (2014). Meyer, M. and J. Phillips. “Caring for Pregnant Opioid Abusers in Vermont: A Potential Model for Non-Urban Areas.” Preventive Medicine, vol. 80 (2015). Newnam, K. M. “The Right Tool at the Right Time: Examining the Evidence Surrounding Measurement of Neonatal Abstinence Syndrome.” Advances in Neonatal Care, vol. 14, no.3 (2014). Orlando, S. “An Overview of Clinical Tools Used to Assess Neonatal Abstinence Syndrome.” Journal of Perinatal and Neonatal Nursing, vol. 28, no.3 (2014). Patrick, S.W., M.M. Davis, C.U. Lehman, and W.O. Cooper. “Increasing Incidence and Geographic Distribution of Neonatal Abstinence Syndrome: United States 2009 to 2012.” Journal of Perinatology, vol. 35 (2015). Patrick, S.W., H.C. Kaplan, M. Passarella, M.M. Davis, and S.A. Lorch. “Variation in Treatment of Neonatal Abstinence Syndrome in U.S. Children’s Hospitals, 2004-2011.” Journal of Perinatology, vol. 34, no. 11 (2014). Patrick, S.W., J. Dudley, P.R. Martin, F.E. Harrell, M.D. Warren, K.E. Hartmann, E.W. Ely, C.G. Grijalva, and W.O. Cooper. “Prescription Opioid Epidemic and Infant Outcomes.” Pediatrics, vol. 135, no.5 (2015). Patrick, S.W., R. E. Schumacher, J. D. Horbar, M. E. Buus-Frank, E. M. Edwards, K. A. Morrow, K. R. Ferrelli, A. P. Picarillo, M. Gupta, and R. F. Soll. “Improving Care for Neonatal Abstinence Syndrome.” Pediatrics, vol. 137, no. 5 (2016). Reece-Stremtan, S., and K. A. Marinelli. “ABM Clinical Protocol #21: Guidelines for Breastfeeding and the Drug-Dependent Woman, Revised 2015.” Breastfeeding Medicine, vol. 10, no.3 (2015). Shaw, M. R., C. Lederhos, M. Haberman, D. Howell, S. Fleming, and J. Roll. “Nurses Perceptions of Caring for Childbearing Women Who Misuse Opioids.” The American Journal of Maternal-Child Nursing, vol. 41, no. 1 (2016). Sublett, J. “Neonatal Abstinence Syndrome: Therapeutic Interventions.” The American Journal of Maternal-Child Nursing, vol. 38, no. 2 (2013). Sutter, M. B., L. Leeman, and A. Hsi. “Neonatal Opioid Withdrawal Syndrome.” Obstetrics and Gynecology Clinics of North America, vol. 41, issue 2 (2014). Teague, A. H., A. J. Jnah, and D. Newberry. “Intraprofessional Excellence in Nursing: Collaborative Strategies for Neonatal Abstinence Syndrome.” Neonatal Network, vol. 34, no.6 (2015). Terplan, M., A. Kennedy-Hendricks, and M. S. Chisolm. “Prenatal Substance Use: Exploring Assumptions of Maternal Unfitness.” Substance Abuse: Research and Treatment (2015). Tolia, V. N., S. W. Patrick, M. M. Bennett, K. Murthy, J. Sousa, P. B. Smith, R. H. Clark, and A. R. Spitzer. “Increasing Incidence of the Neonatal Abstinence Syndrome in U.S. Neonatal ICUs.” The New England Journal of Medicine, vol. 372, no. 22 (2015). Wiles, J. R., B. Isemann, L. P. Ward, A. A. Vinks, and H. Akinbi. “Current Management of Neonatal Abstinence Syndrome Secondary to Intrauterine Opioid Exposure.” Journal of Pediatrics, vol. 165, no. 3 (2014). Appendix II: Department of Health and Human Services’ Resources Related to Neonatal Abstinence Syndrome and Prenatal Opioid Use The Department of Health and Human Services (HHS) has published several guidance and educational resources related to neonatal abstinence syndrome and prenatal opioid use. According to HHS, these documents serve as tools to help stakeholders, including state entities and health care providers, and policymakers. Examples of these resources are listed below. Centers for Disease Control and Prevention (CDC), Pregnancy and Opioid Medications Factsheet, accessed June 8, 2017, https://www.cdc.gov/drugoverdose/pdf/pregnancy_opioid_pain_factsheet- a.pdf. CDC Public Health Grand Rounds, Primary Prevention and Public Strategies to Prevent Neonatal Abstinence Syndrome (Atlanta, GA: CDC, last updated August 18, 2016), accessed July 19, 2017, https://www.cdc.gov/cdcgrandrounds/archives/2016/August2016.htm. CDC, Treating for Two: Safer Medication Use in Pregnancy Initiative (Atlanta, GA: CDC, last updated May 5, 2016), accessed June 8, 2017, https://www.cdc.gov/pregnancy/meds/treatingfortwo. Department of Health and Human Services, Opioids: The Prescription Drug & Heroin Overdose Epidemic (Washington, D.C., last reviewed March 24, 2016), accessed June 8, 2017, https://www.hhs.gov/opioids/index.html. Department of Health and Human Services, National Center for Substance Abuse and Child Welfare, Resources & Topics on Neonatal Abstinence Syndrome, accessed June 8, 2017, https://www.ncsacw.samhsa.gov/resources/opioid-use-disorders-and-me dication-assisted-treatment/neonatal-abstinence-syndrome.aspx. Jean Y. Ko. et al., “CDC Grand Rounds: Public Health Strategies to Prevent Neonatal Abstinence Syndrome,” Morbidity and Mortality Weekly Report (Centers for Disease Control and Prevention, March 10, 2017), accessed June 8, 2017, https://www.cdc.gov/mmwr/volumes/66/wr/mm6609a2.htm. National Institute on Drug Abuse, Principles of Substance Abuse Prevention for Early Childhood: A Research Based Guide (last updated March 2016), accessed June 8, 2017, https://www.drugabuse.gov/publications/principles-substance-abuse-prev ention-early-childhood/principles-substance-abuse-prevention-early-child hood. National Institute on Drug Abuse, Substance Use in Women (last updated September 2016), accessed June 8, 2017, https://www.drugabuse.gov/publications/research-reports/substance-use-i n-women/summary. Reddy, Uma M. J. M. Davis, Z. Ren, and M. F. Greene, “Opioid Use in Pregnancy, Neonatal Abstinence Syndrome, and Childhood Outcomes: Executive Summary of a Joint Workshop.” Obstetrics and Gynecology, vol. 130, issue 1 (July 2017). Substance Abuse and Mental Health Services Administration, A Collaborative Approach to the Treatment of Pregnant Women with Opioid Use Disorders. HHS Publications No. (SMA) 16-4978. Rockville, MD: Substance Abuse and Mental Health Services Administration, 2016. Substance Abuse and Mental Health Services Administration, “Advancing the Care of Pregnant and Parenting Women With Opioid Use Disorder and Their Infants: A Foundation for Clinical Guidance,” Rockville, MD: Substance Abuse and Mental Health Services Administration, 2016. Substance Abuse and Mental Health Services Administration, “Methadone Treatment for Pregnant Women.” HHS Publication No. (SMA) 14-4124 (Rockville, MD: Substance Abuse and Mental Health Services Administration, revised 2014). Appendix III: Comments from the Department of Health and Human Services Appendix IV: GAO Contact and Staff Acknowledgments GAO Contacts Acknowledgments In addition to the contact named above, Rashmi Agarwal, Assistant Director; Amy Leone, Analyst-in-Charge; Melissa Duong; Krister Friday; Jacquelyn Hamilton; Giao N. Nguyen; and Laurie Pachter made key contributions to this report. Related GAO Products Drug Control Policy: Information on Status of Federal Efforts and Key Issues for Preventing Illicit Drug Use. GAO-17-766T. Washington, D.C.: July 26, 2017. Medicaid Expansion: Behavioral Health Treatment Use in Selected States in 2014. GAO-17-529. Washington, D.C.: July 21, 2017. VA Health Care: Actions Needed to Ensure Medical Facility Controlled Substance Inspection Programs Meet Agency Requirements. GAO-17-242. Washington, D.C.: February 15, 2017. Highlights of a Forum: Preventing Illicit Drug Use. GAO-17-146SP. Washington, D.C.: November 14, 2016. Opioid Addiction: Laws, Regulations, and Other Factors Can Affect Medication-Assisted Treatment Access. GAO-16-833. Washington, D.C.: September 27, 2016. Drug Enforcement Administration: Additional Actions Needed to Address Prior GAO Recommendations. GAO-16-737T. Washington, D.C.: June 22, 2016. Office of National Drug Control Policy: Progress toward Some National Drug Control Strategy Goals, but None Have Been Fully Achieved. GAO-16-660T. Washington, D.C.: May 17, 2016. Veterans Justice Outreach Program: VA Could Improve Management by Establishing Performance Measures and Fully Assessing Risks. GAO-16-393. Washington, D.C.: April 28, 2016. State Marijuana Legalization: DOJ Should Document Its Approach to Monitoring the Effects of Legalization. GAO-16-1. Washington, D.C.: December 30, 2015. Prescription Drugs: More DEA Information about Registrants’ Controlled Substances Roles Could Improve Their Understanding and Help Ensure Access. GAO-15-471. Washington, D.C.: June 25, 2015. Mental Health: Better Documentation Needed to Oversee Substance Abuse and Mental Health Services Administration Grantees. GAO-15-405. Washington, D.C.: May 12, 2015. Prenatal Drug Use and Newborn Health: Federal Efforts Need Better Planning and Coordination. GAO-15-203. Washington, D.C.: February 10, 2015. Medicare Program Integrity: CMS Pursues Many Practices to Address Prescription Drug Fraud, Waste, and Abuse. GAO-15-66. Washington, D.C.: October 24, 2014. Office of National Drug Control Policy: Office Could Better Identify Opportunities to Increase Program Coordination. GAO-13-333. Washington D.C.: March 26, 2013. Child Welfare: States Use Flexible Federal Funds, But Struggle to Meet Service Needs. GAO-13-170. Washington, D.C.: January 30, 2013.

Why GAO Did This Study As the opioid crisis has increased in recent years, so has the number of pregnant women who use opioids, which can result in NAS. A recent peer-reviewed study found that cases of NAS have grown nearly five-fold between 2000 and 2012 and that most infants with NAS are covered under Medicaid. The Comprehensive Addiction and Recovery Act of 2016 includes a provision for GAO to examine NAS in the United States and related treatment services covered under Medicaid. This report 1) describes the hospital and non-hospital settings for treating infants with NAS and how Medicaid pays for services, 2) describes recommended practices and challenges for addressing NAS, and 3) examines HHS's strategy for addressing NAS. GAO reviewed HHS documentation and interviewed HHS officials. GAO also conducted site visits to four states—Kentucky, Vermont, West Virginia, and Wisconsin—selected based on several factors, including incidence rates of NAS and geographic variation. GAO interviewed stakeholders from 32 organizations, including health care providers and state officials in the selected states. What GAO Found The prenatal use of opioids or other drugs can produce a withdrawal condition in newborns known as neonatal abstinence syndrome (NAS). Health care providers, state officials, and other stakeholders told GAO that most infants with NAS are treated in the hospital—such as in a neonatal intensive care unit—though some may be referred to a non-hospital setting—such as a neonatal withdrawal center with nursery rooms—to complete their treatment. The table below provides more information on settings for treating infants with NAS and on how Medicaid pays for services in these settings. According to stakeholders GAO interviewed and literature reviewed, there are several recommended practices and challenges associated with addressing NAS. The most frequently recommended practices included prioritizing non-pharmacologic treatment to infants—treatment that does not involve medications—such as allowing the mother to reside with the infant during treatment; educating mothers and health care providers on treatment of NAS, among other things; and using a protocol in the hospital or non-hospital setting for screening and treating infants with NAS. The most frequently cited challenges included the maternal use of multiple drugs—or polysubstance use—as it can exacerbate NAS symptoms; stigma faced by pregnant women who use opioids; hospital staff burden and limited physical capacity to care for infants with NAS; limited coordination of care for mothers and infants with NAS; and gaps in research and data on NAS, such as research on the long-term effects of the condition. In May 2017, the Department of Health and Human Services (HHS) published a strategy document that makes key recommendations to address NAS. The Strategy recommends, for example, that health care providers receive continuing education on managing and treating infants with NAS and promote non-pharmacologic treatment. According to HHS officials, these recommendations will inform planning and policy across the department. However, HHS has yet to determine how and when the recommendations will be implemented, including establishing priorities; the roles and responsibilities of other federal, state, and public stakeholders; implementation timeframes; and methods for assessing progress. HHS officials told GAO that they expect to develop an implementation plan sometime in 2017 but had no timeline for doing so. Without a plan that clearly specifies how HHS will implement the Strategy and assess its progress, the department increases the risk that its recommendations for addressing NAS will not be implemented. What GAO Recommends HHS should expeditiously develop a plan for implementing the recommendations included in its strategy related to addressing NAS. HHS concurred that it should expeditiously address NAS, but noted implementation of the strategy is contingent on funding.

gao_GAO-18-12

gao_GAO-18-12_0

Background An estimated 481,000 workers were employed in the animal slaughtering and processing industry in 2016, according to the Current Population Survey, which is jointly sponsored by DOL’s Bureau of Labor Statistics and the U.S. Census Bureau. There were 5,282 meat and poultry plants in the United States, of which 4,133 conducted processing only, 14 conducted slaughter only, and 1,135 conducted both slaughter and processing, as of February 2017, according to FSIS (see fig. 1). More than 30 million beef cattle, 117 million hogs, 243 million turkeys, and 8 billion chickens were slaughtered in the United States in 2016, according to USDA’s National Agricultural Statistics Service data. As of June 2017, almost 7,500 FSIS inspectors worked in meat and poultry plants, according to FSIS. These inspectors are generally exposed to the same types of hazards as plant employees. Meat and poultry plants are generally designed for an orderly flow from point of entry of the living animal to the finished food product (see fig. 2). Typically, the animal is brought to the meat or poultry plant and taken to the kill floor area, where the animal is rendered unconscious and slaughter occurs. Workers and machines behead and eviscerate the animal, among other things, after which it is chilled for several hours. FSIS inspectors ensure that the carcass meets federal food safety standards. Workers and machines then process the carcass and may break it into small portions that can be transported directly to supermarkets. Slaughter and processing of meat and poultry require workers to perform a high number of repetitive motions. Although plants have increased automation, much of the work is still done by hand using saws, knives, and other tools. Federal Roles Related to Meat and Poultry Worker Safety and Health OSHA helps ensure safe and healthful working conditions for workers in the meat and poultry industry and other industries, in part by setting and enforcing workplace safety and health standards. To carry out its responsibilities under the Occupational Safety and Health Act of 1970, as amended (OSH Act), OSHA establishes workplace safety and health standards; conducts inspections; investigates complaints from workers and reports of fatalities and severe injuries at worksites; and offers cooperative programs, training, and outreach, among other efforts. OSHA is responsible for enforcing private sector employers’ compliance with these standards in about half the states, while the remaining states have assumed that responsibility under a state plan approved by OSHA. These “state-plan states” adopt and enforce their own standards (which must be “at least as effective” in providing safe and healthful employment conditions as the federal standards). With respect to federal employers, federal agencies are generally required to establish and maintain a comprehensive and effective occupational safety and health program that is consistent with OSHA’s standards. OSHA is generally responsible for inspecting federal employers in all states, including state-plan states. As part of its enforcement, OSHA conducts on-site inspections of federal and non-federal employers, collecting evidence through methods such as observation, document review, and interviews. Steps in an inspection may include an opening conference, a walkaround by inspectors, worker interviews, and a closing conference. Based on evidence developed during the inspection, OSHA evaluates whether the employer has violated any safety or health standards. The inspection may result in issuance of a citation if appropriate, and possible appeals by the employer (see fig. 3). Although OSHA does not fine federal agencies, it does monitor these agencies and conducts federal workplace inspections in response to workers’ reports of hazards. Since workers at meat and poultry plants include both plant employees and FSIS employees, OSHA officials may inspect FSIS, the plant employer, or both when it receives a complaint or referral about hazards at the plant. OSHA conducts both programmed and unprogrammed inspections. Programmed inspections are planned based on injury incidence rates, previous citation history, or random selection. For example, OSHA’s emphasis programs focus inspections on a particular safety or health hazard or a specific industry. Unprogrammed inspections are conducted in response to imminent danger, fatalities, worker complaints, referrals, and catastrophic events (such as hospitalizations). FSIS inspects each meat and poultry carcass at the majority of meat and poultry plants throughout the United States. The Federal Meat Inspection Act and the Poultry Products Inspection Act give FSIS responsibility for ensuring the safety and wholesomeness of meat and poultry that enter interstate commerce. As a federal employer, FSIS is also required to establish and maintain a comprehensive and effective occupational safety and health program for its employees that is consistent with OSHA’s standards. However, OSHA (or a state agency in a state-plan state) is responsible for overseeing the safety and health of non-federal plant workers. FSIS’s Environmental, Safety, and Health Group administers FSIS’s occupational safety and health program and investigates safety concerns of FSIS inspectors. Within the Department of Health and Human Services, NIOSH conducts occupational safety and health research and workplace evaluations, and makes recommendations to prevent worker injuries and illnesses. In 2016, we reviewed NIOSH evaluations on hazards in the meat and poultry industry such as those associated with musculoskeletal disorders (MSD), chemical exposure, and pathogens and animals, and recommended in our report that NIOSH conduct a study of the injuries and illnesses experienced by meat and poultry sanitation workers. In 1994, after a workplace fire in 1991 that killed 25 poultry workers in North Carolina, OSHA and FSIS signed an MOU on how the two agencies could work together on worker safety and health at meat and poultry plants. The MOU outlines the policies and procedures the agencies agreed to use, including a process for FSIS to refer serious hazards to OSHA, plans for OSHA and FSIS to develop and implement training on hazard recognition for FSIS staff, an agreement to coordinate on the development of standards and share information on common concerns, and plans for evaluating implementation of the MOU. Federal Role in Chemical Safety at Meat and Poultry Plants Meat and poultry plants use chemicals such as antimicrobials to reduce potential contamination on food and machinery during processing. Antimicrobials may be sprayed directly on meat or poultry, or may be used to clean machinery. FSIS officials and worker advocates have raised worker safety concerns about peracetic acid, an antimicrobial chemical that is being used by the meat and poultry industry for both of these purposes. Peracetic acid has recently become the antimicrobial of choice, according to an FSIS official and a representative from an advocacy group. An FSIS official told us that this was because it is cheap and effective at reducing potential contamination on food. In addition, it is safe to use on food because it generally degrades before consumption, according to FDA officials. FDA, FSIS, EPA, and OSHA all play a role in regulating the use of chemicals at meat and poultry plants. Under the Federal Food, Drug, and Cosmetic Act (FFDCA), as amended, FDA approves food additives, which include antimicrobial food additives, to ensure the food to which they are applied is safe for human consumption. Antimicrobial food additives such as peracetic acid are applied to meat or poultry to reduce the incidence of human illness from food-borne pathogens, such as Salmonella and Campylobacter. FSIS reviews new ingredients and new technology, including new substances or new applications of substances, to determine whether they are safe and suitable for use in meat and poultry products. FSIS’s review includes an assessment of whether the substance could affect food safety, FSIS regulations, inspection procedures, or the safety or health of FSIS inspection personnel. FSIS coordinates its reviews of new ingredients with FDA’s reviews, in accordance with an MOU between FSIS and FDA, most recently amended in January 2000. EPA is responsible for regulating chemicals that meet the definition of a pesticide under the FFDCA and the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA), as amended. Peracetic acid meets the definition of an “antimicrobial pesticide” regulated by EPA when it is used to disinfect, sanitize, or inhibit the growth of microorganisms on surfaces and machinery used in meat and poultry plants. OSHA may regulate the use of chemicals as part of its responsibility for overseeing workplace safety and health. For example, the hazard communication standard requires chemical manufacturers and importers to develop Safety Data Sheets that describe the chemicals’ hazards and include information on safe handling, among other things. The standard also requires employers to ensure their employees have access to these sheets and to provide training on handling these chemicals appropriately. Each agency has a different review or oversight responsibility. The same chemical may undergo different types of review, depending on its intended use in meat or poultry plants. For example, as part of EPA’s pesticide registration process, EPA conducts risk assessments to estimate the nature and probability of harmful effects on the environment and human health, which may include people who may be exposed to the pesticides through their work. FDA’s review of antimicrobial food additives, as mentioned above, focuses on safety to consumers, and does not include a worker safety component. OSHA Increased Meat and Poultry Industry Inspections Since 2005, but Faces Challenges Identifying Worker Concerns OSHA Increased Its Annual Inspections Since 2005 and Provided Compliance Assistance Inspections OSHA’s inspections of the meat and poultry slaughter and processing industry increased from 177 in 2005 to 244 in 2016, due primarily to an increase in poultry inspections (see fig. 4). Officials explained that the increase in meat inspections from 2009 to 2010 and poultry inspections from 2008 to 2012 were associated with increases in complaints during those time periods. OSHA officials said that all inspections decreased in 2013 partly due to the federal government shutdown that year. They added that poultry inspections increased from 2013 to 2016, which officials attributed to the introduction of new severe injury reporting requirements, as well as several emphasis programs focusing inspections on the poultry industry. OSHA consistently conducted more meat than poultry inspections, due to the greater number of meat plants than poultry, according to OSHA officials. In states with state OSH plans, inspection numbers increased from 183 inspections in 2005 to 212 in 2016, due primarily to an increase in meat inspections (see fig. 5). State-plan states conducted almost three times as many total meat inspections as poultry from 2005 through 2016. The number of state poultry inspections has remained steady over the time period, in contrast with the increase seen in federal OSHA poultry inspections. OSHA officials said they did not believe there is an overarching explanation for the trend in state-plan state inspections, because each state plan is independently run, and added that publicly available data show the increase in meat inspections from 2008 to 2010 could have been driven by a large increase in programmed inspections conducted by state-plan states during that time, along with increases in several types of unprogrammed inspections in 2010. in the meat and poultry industry. These programs instruct inspectors to investigate potential hazards such as chemical exposure, noise, and ergonomic hazards. Two of the four programs also focus on issues such as bathroom access, temperature of the plant, and employer recording of injuries and illnesses, to check for recordkeeping violations. OSHA inspections incorporated three of these emphasis programs in 2016; the remaining program did not have inspections that year. OSHA’s most frequently used national and regional emphasis programs in meat and poultry plants in 2016 are shown below (see fig. 6). failing to inspect and test equipment failing to investigate an ammonia leak, and to investigate ammonia vapor releases from pipes and valves within 48 hours without input from at least one employee having experience and knowledge specific to the process being evaluated, and failing to ensure the findings of the process hazard analysis were resolved in a timely manner and actions documented failing to provide workers with an appropriate respirator for chlorine and sodium hydroxide releases failing to provide workers adequate training and information regarding the hazards of ammonia lack of control procedures to protect workers from electrical hazards and being struck or caught by machine parts exposing employees to trip hazards and potential electrical hazards by obstructing passageways failing to conduct fire, chemical release, and extreme weather evacuation drills. OSHA initiates unprogrammed inspections in response to required employer reporting of fatalities or severe injuries and complaints or referrals from sources such as employees, union representatives, media reports, or others. Unprogrammed inspections may include issues covered by a relevant emphasis program as well. OSHA’s unprogrammed inspections of the meat and poultry industry recently increased sharply— from 95 in 2014 to 210 in 2015—mainly due to the new severe injury reporting requirements (see fig. 7). According to agency officials, OSHA decreased the number of planned programmed inspections in order to reallocate resources as the number of unprogrammed inspections increased. If OSHA determines that a meat or poultry plant has violated a workplace safety or health standard, it may cite the plant, specifying which standard or standards were violated (see fig. 8). The most frequently cited standard for employers in the meat and poultry industry, the control of hazardous energy (lockout/tagout), relates to safely shutting down a machine, and ensuring it remains shut off, while the machine is being serviced. OSHA inspects safety controls related to this standard as part of its emphasis program for amputations. In cases where an applicable standard does not exist, OSHA may use the general duty clause of the OSH Act to cite a plant for exposing its employees to a hazard. For example, OSHA does not have a specific standard related to ergonomic hazards, which may cause MSDs. Workers we interviewed in all five states said they frequently experience pain related to postures or movements, and medical experts we interviewed said that meat and poultry workers experience high rates of MSDs. Citing the general duty clause can be challenging and resource intensive due to the high burden of proof necessary to establish each element of the violation, such as the difficulty in showing that work hazards caused an injury, according to OSHA officials. In 2016, OSHA proposed initial meat and poultry plant penalties with a median of $7,000 and assessed final penalties with a median of $4,900 for inspections where violations were found (see fig. 9). Proposed penalties may be reduced after employers contest them before an administrative law judge, or as a result of negotiating penalty amounts with OSHA through an informal settlement process. A representative of one worker advocacy group we interviewed said lowering penalties weakens OSHA’s deterrence capabilities. OSHA officials and one worker advocate said that allowing companies to negotiate lower penalties can benefit workers because it may result in companies agreeing to create safety programs or finding other solutions that improve worker safety. One OSHA official said that citations may affect a company’s workers’ compensation insurance rate, magnifying the financial impact of the violation. According to OSHA officials, initial and final penalties increased in 2010 due to administrative adjustments that had the effect of raising penalties on average. We previously reported that, according to an OSHA official, OSHA increased penalties in 2010 after it determined that penalties were too low to deter employer violations. In addition, officials said that a few large penalties raised average penalties in 2010-2013. OSHA compliance assistance efforts during the years 2005-2016 included worker outreach through local foreign consulates, support for training meat and poultry workers, administering employer recognition programs, and supporting state consultation programs that provide technical assistance to small and medium-sized businesses. OSHA has not comprehensively tracked its compliance assistance activities in the past, but officials told us the agency launched a database module that started tracking these activities in fiscal year 2017. Recent examples of OSHA’s compliance assistance efforts related to meat and poultry plants include the following: OSHA officials stated that in fiscal years 2011-2015, states provided 558 on-site consultation visits, largely funded by OSHA, to small and medium-sized meat and poultry plants. The visits provide confidential safety and occupational health advice to small and medium-sized businesses in all states across the country, according to OSHA, with priority given to high-hazard worksites. These on-site consultation programs, at no cost to employers, work with employers to develop or maintain injury and illness prevention programs, which included assisting employers on identifying potential hazards to prevent injuries, according to OSHA officials. OSHA officials stated that, as of July 2016, six meat and poultry plants were participating in the Safety and Health Achievement Recognition Program, which recognizes small and medium-sized businesses that have used OSHA's on-site consultation program services and operate an exemplary injury and illness prevention program, according to OSHA officials. OSHA officials stated that, as of July 2016, eleven meat and poultry plants participated in OSHA’s Voluntary Protection Programs (VPP), which aim to recognize employers that implement effective safety and health management systems and maintain worker injury and illness rates below average for their industry. OSHA has published guidance and other resources related to safety and health in the meat and poultry industry, such as a 2013 publication on preventing musculoskeletal injuries in poultry processing, and a poultry processing safety and health topics web page. OSHA provided grants for worker safety and health education to nonprofit organizations through the Susan Harwood Training Grant Program. These grants target underserved or low-literacy workers and workers in high-hazard industries. For example, in fiscal year 2016, OSHA awarded a grant to the Western North Carolina Workers Center to train poultry workers on topics including personal protective equipment, hazard mapping, ergonomics, and sanitation worker safety. OSHA officials stated that the agency conducted outreach with poultry industry representatives to discuss common hazards, such as MSDs and infectious pathogens, among others. For example, according to OSHA officials, in support of a Regional Emphasis Program on poultry processing, OSHA’s Dallas Regional Office conducted workshops in the southwestern United States in December 2015 to share safety and health information with employers in the poultry processing industry. OSHA worked with groups such as unions, trade or professional organizations, and educational institutions through its Alliance program to develop compliance assistance tools and share information with employers and workers to help prevent injuries, illnesses, and fatalities in the workplace. For example, OSHA officials said that the agency has formed alliances with foreign consulates to reach workers with limited English proficiency. Also, OSHA’s Omaha Area Office has an alliance with local organizations to help protect workers in the meat packing industry. OSHA Faces Challenges Identifying Worker Concerns, Responding to Medical Mismanagement, and Gaining Entry to Some Plants Identifying Worker Concerns OSHA faces challenges identifying and addressing meat and poultry worker safety and health concerns because workers may be reluctant to speak with inspectors, according to workers we interviewed in four states, as well as worker advocates. Workers we interviewed in four states said they fear dismissal or other punishment if they complain to OSHA or their state OSH agency about their workplace concerns, such as sustaining injuries or being discouraged from using the bathroom. We reported in 2016 that meat and poultry worker vulnerability may hinder reporting of work-related illnesses and injuries, according to federal officials and worker advocacy groups. In particular, these officials and advocates said that some meat and poultry workers may be less likely to report or seek treatment for injuries and illnesses because of their vulnerable status as undocumented or foreign-born workers and because of their economic vulnerability. Meat and poultry workers may also be reluctant to share information with OSHA at their workplace, as on-site interviews often do not allow workers to remain anonymous, even when conducted in private, according to workers in one state, as well as worker advocates we interviewed. According to OSHA officials, OSHA generally conducts worker interviews on-site during inspections. Officials added that, when OSHA conducts on- site interviews, inspectors tell plant supervisors which workers they wish to speak with, so the supervisors can find replacements for these workers on the production line. Therefore, the supervisor knows the identities of interviewed workers, even if the interview itself is private. Officials added that if workers cannot be pulled from the line, they are sometimes interviewed in front of other workers as they continue working. The OSH Act prohibits employers from retaliating against employees for filing complaints with OSHA. However, OSHA officials, workers from two states, and worker advocates we spoke with noted that workers may feel more comfortable sharing concerns about hazards if they are interviewed off-site. The OSHA Field Operations Manual, which sets forth OSHA’s enforcement policies and procedures, allows inspectors to interview workers in locations other than the workplace, and states that, “a free and open exchange of information between OSHA inspectors and employees is essential to effective inspections.” OSHA’s performance goals in DOL’s Strategic Plan include preventing discrimination against workers who report hazards. According to OSHA officials, they will try to schedule an interview off-site if an employee expresses discomfort or if a union arranges it. Officials stated that they do not automatically offer off- site interviews to each employee; rather, inspectors should consult with their Area Directors before offering to conduct an interview off-site. However, OSHA officials told us that inspectors interview meat and poultry workers off-site infrequently, since off-site interviews can be challenging and take additional time, as workers may be difficult to contact or may have ceased working with the company. OSHA also may be challenged to find an acceptable venue when the employee is available. They added that conducting interviews off-site is more feasible in cases when unions or worker advocacy groups have facilitated these meetings, and that interviewing workers on the production line may be advantageous in some cases, as it allows workers to clarify some uncertainties by showing the inspector how their work is done. According to federal internal control standards, agencies should use quality information to achieve their objectives. Although OSHA officials stated that OSHA has taken steps to enable the collection of quality information from workers, such as conducting a representative number of interviews and refraining from reporting information from specific interviews to employers, officials acknowledged that some workers may feel more comfortable sharing concerns about hazards if they are interviewed off- site. Taking additional steps to encourage workers to disclose sensitive concerns, such as by considering off-site interviews or exploring other options to obtain the information anonymously, would help OSHA learn details about hazards, injuries, and illnesses during an inspection and provide additional information to help improve the agency’s efforts to identify or address conditions that endanger worker safety and health. In particular, OSHA may not be aware of the scope of bathroom access issues, which meat and poultry workers we interviewed in all five states said was a concern, because the agency’s reliance on interviewing workers on-site may cause it to miss concerns of workers who are afraid to speak up. In addition, OSHA inspectors do not always ask specifically about bathroom access, and workers who experience bathroom access problems may not volunteer this information. OSHA’s sanitation standard provides that “toilet facilities, in toilet rooms separate for each sex, shall be provided in all places of employment,” based on the number of employees of each sex. According to OSHA guidance, this standard requires employers to make toilet facilities available so that employees can use them when they need to, and may not impose unreasonable restrictions on employee use of the facilities. OSHA guidance also states that denial or delay of bathroom access can result in various serious health effects, such as urinary tract infections, constipation, abdominal pain, and hemorrhoids. Meat and poultry workers may be denied timely bathroom breaks because they work in an assembly line environment, which generally requires workers to be replaced if they leave their station. Workers we interviewed in all five states said their requests to use the bathroom are often delayed or denied, and workers in two states said they fear punishment if they ask to use the bathroom too frequently or complain about lack of bathroom access to their supervisors or to OSHA. Worker advocates we spoke with reported hearing similar concerns on a frequent basis and four worker advocacy groups in different regions of the country reported concerns related to the timeliness of workers’ access to bathrooms based on non-generalizable interviews of poultry workers. For example, workers we interviewed in three states said they had suffered negative health effects, such as kidney problems, from delayed or denied bathroom breaks. One worker said she refrained from eating or drinking until she had completed her shift, to avoid needing a bathroom break. Also, workers we interviewed in all five states said that long lines at the bathroom further limited bathroom access. Meat and poultry industry representatives we interviewed said that bathroom access is not a problem because companies provide bathroom access when needed. They said companies take different approaches to ensuring bathroom access, such as having a supervisor fill in for a worker who leaves the line, establishing scheduled breaks, or allowing workers to leave the line as needed, even without a replacement. However, according to worker advocates, supervisors may vary in implementing plant policy and may feel pressure to fulfill production quotas. One industry representative told us they believe some supervisors in meat and poultry plants deny bathroom access in order to maximize production output. OSHA officials said they did not believe lack of bathroom access was a widespread problem in the meat and poultry industry. However, OSHA officials said they have not compared bathroom access practices in the meat and poultry industry with other industries involving moving production lines because they vary by establishment even within a single industry. OSHA issued a citation in March 2016 to a meat plant related to bathroom access, although that citation is currently being contested by the employer, and is pending as of September 15, 2017, according to officials. From 2005 through 2016, OSHA issued three additional citations to meat and poultry plants related to bathroom access; however, these citations were withdrawn after the employers reached formal or informal settlements with OSHA. OSHA guidance for inspecting poultry plants allows inspectors to ask specifically about bathroom access when there are complaints about it or prior problems, or in the context of specific regional emphasis programs, such as the poultry emphasis program in the southeast United States. In addition, OSHA’s poultry processing regional emphasis programs in regions IV and VI require the inspector to assess the adequacy of toilet and sanitary facilities, and of worker access to them. If there are no prior complaints or relevant emphasis programs, OSHA officials said inspectors ask workers about any other concerns, but do not always specifically ask about bathroom access. Officials said that requiring inspectors to investigate bathroom access would divert inspectors’ limited resources from higher-priority hazards and could result in companies’ claiming that the line of questioning is unsubstantiated. OSHA requires inspectors at poultry plants to consistently investigate other specific hazards, such as ergonomics hazards. According to OSHA officials, the agency selected these hazards based on prior inspection and illness and injury data showing the hazards to be widespread in the industry. Officials contrasted these with the small number of citations issued related to bathroom access. However, given that workers whom we asked about bathroom access during off-site interviews in all five states said that bathroom access is a problem, and worker advocates we interviewed stated it was as well, it is possible that OSHA is missing instances of this hazard, resulting in incomplete data to guide its inspections. According to federal internal control standards, managers should use quality information to achieve the agency’s objectives. While officials stated they believe that inspectors’ open-ended questions will prompt workers to share any concerns they have with bathroom access, workers may not volunteer this information unless specifically asked. For example, workers may not be aware that they have the right to access bathrooms and so may not realize that such information would be of interest to OSHA, according to one worker advocate we interviewed. Gathering additional information about whether meat and poultry workers experience delayed or denied access to bathrooms could help OSHA determine the extent of the problem and be better positioned to protect worker safety and health. Addressing Medical Mismanagement at Plant Health Units OSHA officials told us that addressing medical mismanagement at meat and poultry plants is challenging because of the complex issues involved and OSHA’s limited oversight of plants’ health unit staff. Specifically, they said that ensuring proper certification for medical providers is the responsibility of state authorities. In hazard alert letters to four meat and poultry plants, OSHA noted its concern that plant health unit staff were or may have been inappropriately supervised and working beyond the scope of their medical license. Officials said OSHA contacted state authorities who oversee health unit staff in one state about licensing concerns, and they planned to contact additional states. OSHA officials we interviewed expressed concern about meat and poultry workers’ access to plant first aid stations or health units and the quality of medical treatment workers receive. OSHA issued one general duty clause citation and four hazard alert letters to five meat and poultry plants in 2015-2016 related to medical mismanagement issues, which describe OSHA’s findings or concerns about inappropriate medical treatment, lack of worker access to health care, underqualified practitioners, and challenges to reporting (see sidebar). In the citation, OSHA found that the plant delayed care for injured workers, stating these actions could result in risk of further injury or exacerbated pain, among other conditions. In a 2015 hazard alert letter to a poultry plant, OSHA noted that it appeared the plant used its first aid station to prevent injuries from appearing on the plant’s OSHA log, such as by failing to refer workers to a physician for evaluation or treatment when appropriate. In addition, OSHA noted that a number of workers were fired after suffering MSDs, sometimes on the same day of the MSD occurrence, and further noted workers’ fears of being fired for visiting the first aid station. OSHA recommended voluntary improvements to the plant’s medical management practices. In a 2014 hazard alert letter to another poultry plant, OSHA identified practices that it determined were contrary to good medical practice for managing work-related MSDs, including prolonged treatment by nursing station staff without referral to a physician. The letter included one example in which a worker made over 90 visits to the nursing station before referral to a physician. Meat and poultry workers we interviewed in all five states reported problems with on-site medical care; for example, workers said their supervisor or plant nurse may not take appropriate steps when a worker is injured or ill, such as not referring the worker to a doctor or failing to move the worker to a different work station on the line. Worker advocates we spoke with reported hearing similar concerns. One worker we interviewed said that she experienced severe pain in her wrist and visited the on-site medical unit over the course of 3 months before they referred her to a doctor affiliated with the plant, during which time she continued to work. When the doctor did not find any problems on her X-ray, she went to a doctor unaffiliated with the plant, who found a bone fracture. Meat and poultry workers in three states also said that fear of being reprimanded or losing their jobs sometimes compels them to refrain from accessing care at a plant health unit, or from complaining about inadequate medical care. Workers in one state said they are penalized every time they visit their plant health unit. Amputation Leads to OSHA Detection of Medical Mismanagement and Other Hazards Following the amputation of a worker’s finger at a poultry plant in 2016, the Occupational Safety and Health Administration (OSHA) conducted inspections and cited the plant for violations related to: deficiencies with the procedures meant to stressors as they performed tasks requiring repetitive, forceful motion for extended periods of time, often in awkward positions failure to comply with generally accepted good engineering practices with respect to exhaust systems, ammonia sensors, and alarms, exposing workers to the hazards posed by a potential ammonia release failure to provide free personal protective equipment to workers failure to repair or replace damaged electrical equipment, exposing workers to the risk of electrical shock. According to one industry representative, plants do not have enough guidance on how to ensure their health units are properly staffed and operated. OSHA issued guidance in 1999 about occupational health professionals’ qualifications and scope of practice, as well as a 2006 best practices guide on the fundamentals of a workplace first aid program. However, OSHA officials told us these guidance documents do not address many of the medical management issues they are currently observing in plant health units, which include lack of supervision of medical personnel, personnel working outside their scope of practice, out- of-date health unit protocols, inappropriate response to injuries and illness, lack of quality assurance, poor worker access to health units, and inadequate recordkeeping. OSHA officials told us that the agency has recently begun updating its guidance related to health units to help clarify employers’ responsibilities with regard to the personnel in these units and the services they provide. However, these updates have not been completed, according to officials. Federal internal control standards call for agencies to externally communicate the information needed to achieve their objectives. By updating and issuing its guidance, OSHA could help plant health units be better positioned to provide appropriate care to injured and ill workers. OSHA also issued hazard alert letters recommending the plant take steps to address the following hazards: medical management practices that prevent appropriate standards of care, increase the likelihood of workers developing serious musculoskeletal disorders, restrict referrals to physicians, and discourage employees from reporting symptoms and injuries. In some cases, plant management may deny entry to OSHA inspectors attempting to conduct an inspection, and resolving these issues can create delays. Although the OSH Act authorizes OSHA inspectors to enter plants “without delay” at reasonable times to conduct inspections, employers have the right to refuse entry, in which case OSHA may seek an inspection warrant. If the employer denies entry after OSHA obtains a warrant, OSHA determines its response on a case-by-case basis. Denials of OSHA inspector entry to meat and poultry plants increased in 2016. All 15 denials in that year occurred in region IV, which includes the Southeast United States. In contrast, from 2005-2015, there were 16 denials of entry in the meat and poultry industry. The denials in 2016 took place in Georgia (6), Alabama (5), Florida (2) and Mississippi (2). According to OSHA officials, the agency experienced denials in all 15 cases when it inspected a plant in response to a complaint or referral and moved to expand the inspection to incorporate its regional emphasis program for the poultry industry. OSHA has not been able to expand its inspections in any of these cases, according to OSHA officials. These denials of entry have the potential to limit OSHA’s understanding of worker safety and health in plants during the days or months prior to gaining entry, and addressing denials is resource-intensive, according to OSHA officials. There is currently ongoing litigation in a case in which OSHA was inspecting a plant after an employee was burned by an electrical fire. OSHA attempted to expand the inspection under a relevant emphasis program, and the plant contested the expansion in court. OSHA officials said that the case is pending as of September 15, 2017, and they will consider the outcome of the case when determining their response to any similar denials of entry in the future. OSHA and FSIS Have Not Fully Implemented MOU That Outlines Collaboration on Worker Safety and Health OSHA and FSIS’s main vehicle for collaboration on worker safety and health is their 1994 MOU, but efforts to implement and evaluate this agreement have been limited. The MOU outlines the policies and procedures the agencies agreed to use, including (1) a process for FSIS to refer serious hazards facing plant workers or FSIS inspectors to OSHA, (2) plans for OSHA and FSIS to develop and implement training for FSIS staff in hazard recognition, and (3) an agreement to coordinate standards development and exchange information on matters of common concern. In 2005, we found that agency efforts to implement this MOU had lapsed, and we recommended that OSHA and FSIS revisit and update certain aspects of their MOU, as discussed below. OSHA and FSIS have taken some steps to implement the policies and procedures outlined in the MOU. However, we found issues with the MOU’s implementation in these three areas, hampering achievement of the MOU’s goals. Further, OSHA and FSIS have not evaluated the implementation of the MOU. Making Referrals to OSHA The 1994 MOU calls for FSIS inspectors—who may observe hazards to both plant workers and inspectors—to refer serious workplace hazards to OSHA, via FSIS headquarters. Serious hazards are defined in the MOU as those for which there is a substantial probability that death or serious physical harm could result. The two agencies have established a process for these referrals, but according to FSIS officials its inspectors are reluctant to make them, as discussed below. Until 2014, FSIS inspectors were to refer these hazards to OSHA by sending a referral to OSHA via FSIS headquarters, but, according to FSIS officials, inspectors rarely made referrals under the former system. In 2014, FSIS established a procedure for its inspectors to notify OSHA directly of serious workplace hazards that may affect both FSIS inspectors and plant workers and issued a notice that provides instructions for inspectors to use OSHA’s nationwide public toll-free number to report such hazards. Because the agencies are not able to track all of these referrals, as callers can remain anonymous, it is not possible to assess the extent to which FSIS inspectors are making them, according to OSHA officials. OSHA data show that since 2012, OSHA had received 14 complaints and 2 referrals about FSIS, of which 5 phone calls were from FSIS current or former employees, but these data are likely incomplete. According to FSIS officials, its inspectors may be reluctant to make these referrals because they fear it could trigger an OSHA inspection of FSIS. As a federal employer, FSIS is responsible for ensuring its own employees are protected from plant hazards, and is subject to OSHA inspection and notification of safety and health hazards faced by its employees. OSHA data show that from 2005 to 2016, OSHA inspected FSIS in meat and poultry plants 105 times, of which 14 were in response to complaints and referrals. FSIS occupational safety and health specialists said these inspections can be a drain on their resources because they are time-consuming and there are very few FSIS safety and health specialists to respond to them. FSIS employs three occupational safety and health specialists, along with one team lead, resulting in a ratio of one specialist for every 3,100 employees, according to FSIS officials. FSIS inspectors we contacted did not report communicating with OSHA, and stated that they share any worker safety concerns they might have with their management or with plant contacts. FSIS occupational safety and health specialists told us that FSIS requests technical assistance from OSHA to address hazards that may affect both plant workers and FSIS inspectors. However, they said that OSHA may inspect FSIS instead of providing assistance, even though OSHA has other ways of offering technical assistance to federal agencies. For example, FSIS occupational safety and health specialists told us that when they reached out to OSHA for assistance with hazards posed by peracetic acid, OSHA instead inspected FSIS for what FSIS considered to be unrelated hazards. According to OSHA officials, their enforcement team is obligated to respond to complaints and referrals, including calls to their toll-free number, and may inspect FSIS if there is a report of a hazard at a meat or poultry plant to which FSIS employees are exposed. OSHA officials noted that FSIS employees should not be reluctant to report hazards because OSHA inspections can protect FSIS workers. OSHA officials told us that they are able to provide assistance with hazards if FSIS contacts OSHA’s area offices, but if in the process, OSHA learns about a potential hazard that is FSIS’s responsibility, then OSHA may investigate the agency. FSIS officials told us they did not believe this process would ensure a quick enough response by OSHA to provide FSIS assistance with urgent hazards that could be harming FSIS inspectors and plant workers. Training FSIS Staff OSHA and FSIS agreed under the 1994 MOU to provide training to FSIS inspectors so that they could recognize serious workplace hazards faced by plant workers and FSIS inspectors. OSHA officials told us they developed such training for FSIS in the mid to late 1990s, but according to FSIS officials the course was too excessive and burdensome for FSIS inspectors, whose main responsibility is food safety. In 2005, we recommended that OSHA and FSIS revisit and update their MOU to ensure that FSIS inspectors receive training in recognizing and referring workplace hazards to OSHA. The two agencies did not update their MOU, but FSIS officials told us the agency strengthened its training of FSIS inspectors and OSHA officials told us that the agency planned to cooperate with FSIS to encourage revitalization of FSIS’s inspector training. According to FSIS officials, in 2013, FSIS began requiring its inspectors whose duties were not limited to being on the line to take AgLearn 8500, an FSIS course on identifying and reporting hazards that was reviewed by OSHA. This course—which is now available only on CD—is required for inspectors who do not work on the line and is optional for those who do. In 2014, OSHA provided three training sessions on identifying workplace hazards to FSIS managers, according to OSHA officials. However, FSIS was not able to provide information on whether or how the managers who received the training had shared what they learned with FSIS inspectors because it did not track this information. Coordinating Standards Development Line Speed Meat and poultry slaughter and processing generally occurs along a “disassembly line,” on which workers and machines produce various cuts of meat. These lines can include live hang in poultry plants, evisceration lines, and “cone” lines where deboning occurs. The Food Safety and Inspection Service (FSIS) sets maximum evisceration line speed in order to ensure its inspectors can effectively perform their inspection procedures. According to FSIS officials, FSIS does not regulate the speed of other lines, which may run slower than evisceration lines due to complex worker tasks. The Occupational Safety and Health Administration (OSHA)— which is responsible for overseeing worker safety and health—does not play a role in regulating line speed, according to FSIS and OSHA officials. GAO reported in 2016 on concerns that high line speeds may exacerbate existing hazards that can cause musculoskeletal disorders (MSD). OSHA and National Institute for Occupational Safety and Health (NIOSH) officials told us that line speed—in conjunction with forceful exertions, awkward postures, and other factors—affects the risk of MSDs. When plants increase line speed, they may address worker safety by increasing staffing or creating new lines. speed may affect worker safety (see sidebar). FSIS officials told us that OSHA provided comments after the proposed poultry modernization rule was published in the Federal Register. According to the fall 2016 unified regulatory agenda, FSIS is working on a proposed rule to amend the federal meat inspection regulations to establish a new inspection system for hog slaughter establishments. FSIS officials told us they consulted with OSHA officials about the possible worker safety implications of the proposed rule on hog slaughter prior to the rule being sent to OMB. However, they also stated that—contrary to the OSHA officials cited above—they believed the OMB review process was sufficient for addressing any worker safety implications in rules proposed by FSIS. Evaluating the Implementation of the MOU FSIS and OSHA agreed to jointly evaluate the effectiveness and impact of the actions taken under the MOU—in part by tracking the number of FSIS referrals to OSHA, inspections made in response to these referrals, and the number and types of hazards cited in these inspections—and to make adjustments to the MOU as appropriate. According to FSIS and OSHA officials we spoke with, this has not been done. Neither OSHA nor FSIS was able to tell us why these evaluations did not take place. For example, the MOU states that when training has been completed, OSHA and FSIS will analyze field-level evaluations to assess whether the training has raised FSIS inspectors’ awareness and reporting of serious workplace hazards. OSHA officials said they do not know if FSIS used the training materials they provided to FSIS to train FSIS field inspectors. FSIS officials said the training OSHA provided was too time-consuming, so they did not use it. Further, FSIS officials told us the agency does not formally survey staff who complete AgLearn 8500 because it is available only on CD. Our prior work has noted that developing mechanisms to monitor, evaluate, and report results can help enhance and sustain collaboration. Evaluating the implementation of the MOU and making any needed changes, including setting specific timeframes for periodic evaluations of actions taken under the MOU, would help ensure the goals of collaboration are fully met. Gaps in Federal Review, Safety Information, Measurement, and Research May Put Meat and Poultry Workers at Risk of Harm from Certain Chemicals Federal Reviews of Certain Chemicals Leave Gaps Related to Plant Worker Safety Federal reviews of certain antimicrobial chemicals before they are used in meat and poultry plants leave gaps with respect to worker safety and health. FSIS inspectors and workers in meat and poultry plants are exposed to antimicrobial chemicals every day, as they are commonly used during all work shifts, both on animal and bird carcasses and on work surfaces and machinery. In general, the potential for chemical exposure is greater for plant workers than for FSIS inspectors. According to officials we spoke with at various plants, plant workers handle these chemicals by receiving shipments, opening containers, and filling machines with the chemical, among other ways, while FSIS inspectors are generally not present at various times that workers are using the chemical, most notably, when the plant is being cleaned between shifts. Depending on a chemical’s intended use, it may or may not undergo a federal review of the risks it poses to worker safety and health before it is used in the plant. The regulation of chemicals used in meat and poultry plants is complex, as several federal agencies have their own specific areas of jurisdiction with regard to their oversight. OSHA does not conduct reviews of chemicals before they are used in the workplace, according to OSHA officials. OSHA officials stated that the agency is limited from taking such an approach, because doing so would overwhelm the agency’s resources. In addition, OSHA is charged with oversight of workers in multiple industries—not just the meat and poultry industry—which would make it difficult for them to utilize a review process that examines all chemicals before they are used in the workplace. Antimicrobial chemicals intended for use as sanitizers in plants to clean machines and surfaces are generally subject to EPA’s pesticide registration process, which considers user or worker safety (see fig. 10). This review does not generally include antimicrobial chemicals applied directly to meat and poultry in plants. When antimicrobials are proposed for use directly on meat or poultry to combat foodborne pathogens, FDA and FSIS both conduct reviews before they are used in the plant, but neither review specifically focuses on plant worker safety or health. FDA’s review of antimicrobial food additives is focused on ensuring they are safe for consumers to eat. FSIS’s review is focused on ensuring that the antimicrobials do not affect the safety of meat and poultry products or interfere with inspections and that they comply with other FSIS regulations; it also includes an assessment of any adverse effects on FSIS inspector safety and health as they perform their duties. Since these federal reviews do not generally take into consideration the occupational risk of chemicals to plant workers, who make up the majority of personnel in a plant, these chemicals could be used in plants directly on meat or poultry to combat foodborne pathogens without a federal assessment of their potential effects on plant worker safety and health or how these effects may be prevented or addressed. As a result, plant workers may be put at risk of chemical hazards. During its review of antimicrobial chemicals proposed for use directly on meat or poultry, FSIS receives information from chemical manufacturers that could be relevant to chemical safety for both FSIS plant-level officials and meat and poultry workers. FSIS occupational safety and health officials told us that the information they request goes beyond what is included in the chemicals’ Safety Data Sheets, and may include directions for use or safety information that is specific for dilution levels and conditions of use at plants. However, this information is not shared with OSHA, NIOSH, at the local level with FSIS in-plant inspectors, or with plant management because FSIS does not have a process for doing so. OSHA and FSIS occupational safety and health officials and an industry representative we interviewed told us that this information would be useful to them when it is available. OSHA officials told us that information on chemical hazards, employee exposure, and safety controls and practices would enable it to strengthen its response to protecting all workers from these chemical hazards and develop outreach and technical assistance for the meat and poultry industry. An FSIS safety and health official told us that this information would have been useful at one plant, because neither plant management nor FSIS inspectors at the plant had received information to adequately protect FSIS employees from the effects of peracetic acid, and there had been complaints from inspectors about the chemical. In addition, NIOSH conducts occupational safety and health research, among other things, and could benefit from such information. Federal internal control standards call for agencies to internally and externally communicate the necessary quality information to achieve the entity’s objectives. By FSIS establishing a process to regularly share the worker safety information it collects during reviews of new chemicals—internally with FSIS inspectors and externally with plant management, OSHA, and NIOSH—the federal government will be better positioned to use existing resources to support the safety and health of plant workers and FSIS inspectors. Gaps in Inspector Safety Information As discussed above, FSIS conducts reviews of new ingredients and technologies, including antimicrobial chemicals, proposed for use on meat and poultry products8. However, there may be information gaps in FSIS’s examination of the potential risks these new chemicals may pose to inspector safety and health. As part of this review, FSIS requests information from chemical manufacturers or plants describing how the new chemicals will not adversely affect the safety and health of FSIS inspectors. As FSIS’s Environmental, Safety, and Health Group reviews this information, other program areas within FSIS also review the submission to determine whether the chemical is otherwise safe and suitable under the conditions of its intended use—i.e., that it will not adversely affect product safety, violate FSIS regulations, or interfere with inspection procedures. If FSIS determines that the chemical will not have these effects, the agency will issue a letter of “no objection” for the use of the new chemical. It is unclear whether FSIS consistently reviews these chemicals to ensure they will not adversely affect inspector safety and health because the agency does not have a robust process for tracking and sharing information needed to make this determination among the various program areas within the agency participating in the review. Officials in FSIS’s Environmental, Safety, and Health Group told us that they often initially receive inadequate information to make this determination, despite new guidance developed in 2015 on the type of information that chemical manufacturers and plants may submit to enable FSIS to evaluate potential adverse effects to inspector safety. In cases where they do not receive sufficient information, the Environmental, Safety, and Health Group will ask the FSIS program area that is leading the review to request additional information from the manufacturer or plant. However, FSIS does not have a process that seamlessly tracks the worker safety information it receives as part of its review process, and FSIS occupational safety and health officials told us it is not clear whether submissions contain complete inspector safety information before a “no objection” letter is issued. In response, other FSIS officials told us that they would not approve a new chemical until they have adequate information that shows it will not adversely affect the safety and health of FSIS inspectors, among other things. Improving communication within FSIS about this review process is one goal of FSIS’s 2017 Annual Plan. To help implement this goal, FSIS formed a working group in April 2017 that is developing a draft directive to facilitate improved coordination among the program areas involved in the review process, including the Environmental, Safety, and Health Group. More specifically, the draft directive provides procedures and protocols and describes an electronic system for tracking information submitted. According to FSIS officials coordinating reviews, the electronic system will replace the current manual system and will be accessible to all program areas involved in the review process. Further, the draft of a “no objection” letter will be distributed to the program areas involved in the review to ensure that all remaining outstanding questions or issues related to the notification have been addressed prior to issuing the letter to the submitter. According to an FSIS official, the agency plans to finalize and issue the draft directive by the end of calendar year 2017 and anticipates converting to the electronic tracking system in fiscal year 2018. Gaps in Measurement OSHA and FSIS officials told us that they have faced challenges responding to complaints about air quality in meat and poultry plants, because it is hard to measure airborne peracetic acid. According to OSHA, FSIS, and NIOSH officials, there is no sufficiently reliable method to measure peracetic acid in plants, in part because peracetic acid is not stable and breaks down quickly. As a result, it is harder to assess the extent of worker exposure to this chemical and plan for an appropriate response. Some plants currently use monitors to sample for the components of commercial peracetic acid (acetic acid and hydrogen peroxide); however, the effects of peracetic acid exposure on workers can be different than those caused by either of these individual chemicals or by mixtures of peracetic acid with other chemicals. In 2013, OSHA’s Salt Lake Technical Center began working to develop a validated sampling and analytical method that would permit measurement of airborne peracetic acid with a high degree of confidence. Work on the method continues, according to OSHA officials. NIOSH has begun evaluating a range of commercially available peracetic acid monitors and is planning to evaluate an air sampling method for peracetic acid. The lack of a reliable way to measure peracetic acid could also affect any efforts by OSHA to develop a permissible exposure limit (PEL), a type of workplace safety and health standard that officials said would enable the agency to more easily cite employers for exposing their workers to peracetic acid hazards, compared to using the general duty clause. According to OSHA officials, the process for developing PELs is arduous, and peracetic acid is one of many chemicals without such a limit or with one that is outdated. In response to our 2012 report, which found OSHA’s standard-setting process to be challenging and lengthy, OSHA and NIOSH developed an MOU to support their research on developing potential standards. Gaps in Peracetic Acid Research In March 2017, NIOSH announced its intent to initiate a study of workplace uses of and occupational exposure to peracetic acid, but this study will not examine the safety and health hazards this chemical may pose if it is combined with other chemicals, as can happen in slaughter plants. The NIOSH study aims to develop an immediately dangerous to life or health (IDLH) value and an effective workplace measurement method, among other things. While the focus of this research is the characterization of workplace exposure to peracetic acid, the study is not intended to address the extent and consequences of mixing peracetic acid with other substances, which can occur in several ways in meat and poultry plants. As carcasses move from one stage of processing to another, peracetic acid can come into contact with other substances, such as when there are spills or in drainage systems. FSIS officials, a worker advocate, and plant workers we interviewed expressed concern that the mixing of chemicals can create new safety and health risks for workers. For example, an FSIS official said that an inspector at a poultry plant complained about effects from airborne chemicals that appeared to be related to the location of her work station, directly over a drain in which multiple substances were pooling. In 2011, 152 workers at an Arkansas poultry plant reported being hospitalized from effects of chlorine gas created after a supervisor added sodium hypochlorite (bleach) into a container holding a residual acidic antimicrobial solution, creating a chemical reaction. NIOSH officials told us they are aware that chemicals can be used in plants alongside peracetic acid and result in a mixed exposure, and that this may be a serious problem. Although the focus of the current peracetic acid study is primarily on the health effects of and exposures to peracetic acid alone, NIOSH officials said that NIOSH has the capability to assist in characterizing worker exposures of concern, and could consider such research in a follow-on study, depending on available resources. In addition, NIOSH officials told us that the agency will consider whether potential health hazards exist from other chemicals in the environment, particularly if they interfere with measuring peracetic acid exposures and assessing health effects in workers. Moreover, officials told us that their current study could provide the basis of follow-on research into other workplace chemical hazards, including mixtures. In 2004, NIOSH recognized mixed exposures as a priority area for the occupational safety and health research community and identified significant gaps and research needs. According to the report, workers from agriculture, construction, mining, and other industries are commonly exposed to combinations of chemical substances, biological or physical agents, and other stressors, and knowledge is limited regarding the potential health effects of mixed exposures. Identifying these effects can help characterize worker exposure and develop hazard controls that take into account the components of the mixtures. According to NIOSH officials, mixed exposures continue to be important to study because they may represent a health hazard to workers, and employers should prevent or control workplace exposures to such mixtures. By considering the addition to the agency’s research agenda of a proposal to examine peracetic acid’s use in combination with other chemicals, NIOSH will be better able to characterize worker exposure to such scenarios and develop controls to reduce this hazard for workers. Conclusions While OSHA’s enforcement efforts in the meat and poultry industry have increased since we reported in 2005, worker safety and health problems persist and improvements are needed in identifying worker concerns, strengthening federal collaboration, and protecting workers from certain chemicals. Workers we spoke with reported they are reluctant to report injuries, illnesses, and hazards because they fear losing their jobs. There is a mismatch between concerns we heard from workers and the problems reported by OSHA, particularly in the area of bathroom access. Taking additional steps to encourage workers to disclose sensitive concerns and gathering additional information to determine the scope of bathroom access issues could enable OSHA to better identify worker safety and health concerns. OSHA’s efforts to address medical mismanagement at plants—which has resulted in poor medical care for workers—could be improved by issuing updated guidance for employers on how to manage their health units. Collaboration between OSHA and FSIS is limited and has improved little since we recommended in 2005 that the two agencies strengthen their 1994 MOU on worker safety. Since FSIS is already present in many plants, the federal government is missing out on a cost-effective opportunity to further protect the safety and health of both plant workers and FSIS inspectors by leveraging resources in this fiscally constrained environment. Evaluating the implementation of the MOU and making any needed changes would help ensure the agencies improve their collaboration. With regard to chemicals, there are gaps in information sharing and research that have heightened the risk of chemical hazards for plant workers and FSIS inspectors. In particular, FSIS collects information on how to protect its inspectors from new chemicals, but it does not have a process to share this information with its own inspectors, plant management, OSHA, or NIOSH. By FSIS establishing a process to regularly share the worker safety information it collects during reviews of new chemicals, the federal government will be better positioned to use existing resources to support the safety and health of plant workers and FSIS inspectors. Finally, NIOSH’s plan to conduct a study on peracetic acid will likely yield useful information for meat and poultry worker safety, but it is not intended to address the potential consequences of mixing peracetic acid with other substances, which can occur in several ways in meat and poultry plants. By considering the addition to the agency’s research agenda of a proposal to examine peracetic acid’s use in combination with other chemicals in meat and poultry plants, NIOSH will be better able to characterize worker exposure to such scenarios and develop controls to reduce this hazard for workers. Recommendations for Executive Action We are making seven recommendations, including four to OSHA, two to FSIS, and one to NIOSH. Specifically: The Assistant Secretary of Labor for Occupational Safety and Health should take additional steps to encourage workers to disclose sensitive concerns during OSHA inspections of meat and poultry plants; for example, by considering additional off-site interviews or exploring other options to obtain information anonymously. (Recommendation 1) The Assistant Secretary of Labor for Occupational Safety and Health should gather more information, such as by asking workers during meat and poultry plant inspections, to determine the extent to which bathroom access is a problem and how to address any identified issues. (Recommendation 2) The Assistant Secretary of Labor for Occupational Safety and Health should update its guidance for employers on how to manage their health units to address the challenges of managing these units. (Recommendation 3) The Assistant Secretary of Labor for Occupational Safety and Health should work with FSIS to assess the implementation of the MOU and make any needed changes to ensure improved collaboration; and set specific timeframes for periodic evaluations of the MOU. (Recommendation 4) The FSIS Administrator should work with OSHA to assess the implementation of the MOU and make any needed changes to ensure improved collaboration; and set specific timeframes for periodic evaluations of the MOU. (Recommendation 5) The FSIS Administrator should develop a process to regularly share the worker safety information it collects during its review of new chemicals with FSIS inspectors, plant management, OSHA, and NIOSH. (Recommendation 6) The Director of NIOSH should consider including in the agency’s research agenda a proposal for examining the extent of peracetic acid’s use in combination with other chemicals in meat and poultry plants, and any safety and health hazards these combinations may pose to workers. (Recommendation 7) Agency Comments and Our Evaluation We provided a draft of this report to the U.S. Department of Labor (DOL), the U.S. Department of Agriculture (USDA), the U.S. Department of Health and Human Services (HHS), and the Environmental Protection Agency (EPA) for their review. DOL’s Occupational Safety and Health Administration (OSHA), USDA’s Food Safety and Inspection Service (FSIS), and HHS provided written comments that are reprinted in appendixes II, III, and IV, respectively. In an e-mail dated October 5, 2017, an EPA audit liaison indicated that EPA had no comments. OSHA did not state whether it concurred or not with the four recommendations made to it. USDA expressed concern with the draft report’s characterization of FSIS’s collaborative efforts and also described planned actions to address the two recommendations we made to it. HHS agreed with the one recommendation we made to it. DOL and HHS provided technical comments, which we incorporated as appropriate. With respect to our first recommendation that OSHA take additional steps to encourage workers to share information during meat and poultry inspections, OSHA stated that it fully supports the idea of continuous improvement of its processes that would expand its ability to identify and address hazards before an injury, illness, or fatality occurs. However, OSHA noted that it would be challenging to conduct offsite interviews in terms of witness cooperation, resources, and inspector safety. We continue to believe that OSHA should take steps to enhance reporting by meat and poultry workers. Our report describes meat and poultry workers’ reluctance to report injuries, illnesses, and hazards to OSHA because of their fear of employer retaliation. OSHA’s Field Operations Manual highlights the importance of a free and open exchange of information between OSHA inspectors and employees for conducting effective inspections. Conducting additional offsite interviews is one way to encourage employee reporting. However, there may be alternative additional steps OSHA could take to better position it to encourage workers to disclose sensitive concerns, consistent with our recommendation. With respect to our second recommendation that OSHA gather additional information to determine the extent to which bathroom access is a problem in meat and poultry plants, OSHA stated it could not commit to routinely asking about bathroom access at each meat and poultry inspection. OSHA stated that each inspection requires a flexible approach to address unique worksite hazards. It further stated that OSHA does not routinely ask questions about any potential hazards that go beyond the scope of a complaint inspection, unless those hazards are in plain sight. However, as noted in the report, OSHA does require inspectors at poultry plants to consistently investigate other specific hazards, such as ergonomics hazards. Our report highlights the challenges meat and poultry workers may face gaining timely access to bathrooms. However, workers might not volunteer access information to OSHA. Our work identified a mismatch between the concerns we heard from workers and the problems reported by OSHA. Better understanding the scope of bathroom access problems would better position OSHA to respond appropriately. Further, OSHA may choose to address this issue without routinely asking workers about bathroom access, such as by selectively querying workers based on criteria determined by the agency. With respect to our third recommendation to update its guidance for employers on management of plant health units, OSHA stated that it intends to revisit its guidance. With respect to our fourth and fifth recommendations for OSHA and FSIS to work together to assess the MOU’s implementation, make changes to improve collaboration, and set timeframes for periodic evaluations of the MOU, neither agency stated whether it agreed or not. OSHA stated that meat and poultry plants provide an opportunity for the two agencies to work collaboratively to identify employee hazards and promote safety and health, but OSHA did not comment specifically on the recommendation. FSIS stated that it already has directives in place to recognize and report hazards affecting FSIS employees, and acknowledged that the MOU was designed to additionally have FSIS employees report hazards affecting plant employees due to the regular presence of its inspectors in plants. FSIS noted that in collaborating with OSHA, FSIS will need to ensure its primary mission is not compromised by undertaking activities that take time and resources away from its food safety inspection responsibilities. We continue to believe that strengthening the MOU and developing a mechanism to regularly evaluate it would help ensure that the goals of the MOU are met, and that leveraging FSIS’s presence in plants provides the federal government with a cost-effective opportunity to protect worker safety and health. With respect to our sixth recommendation that FSIS regularly share the worker safety information it collects during its review of new chemicals with FSIS inspectors, plant managers, OSHA, and National Institute for Occupational Safety and Health (NIOSH), FSIS stated that the agency already has a process for sharing chemical safety information with its inspectors. However, FSIS has not provided us with evidence that it has shared the worker safety information it collects related to new chemicals, such as safety information that is specific for dilution levels and conditions of use at plants, as noted in the report. FSIS also stated that it would take certain steps to share information about approval of chemicals with other agencies such as OSHA and NIOSH, but the steps identified did not include sharing worker safety information. Incorporating worker safety information would further help enhance this information sharing. FSIS further stated that some of the information collected during its review of new chemicals may be proprietary. In addition, FSIS also expressed concern with how we characterized its collaboration with OSHA and NIOSH on worker safety. Specifically, in reference to the report’s discussion of the development of the poultry inspection modernization rule, FSIS stated that it consulted with and included OSHA and NIOSH during the appropriate step of the rulemaking process, and that the agency followed the Administrative Procedure Act in proposing the rule. We do not intend to suggest any deficiencies with FSIS’s rulemaking procedures. Rather, our report points out possible opportunities for earlier and enhanced collaboration with OSHA on standards development. FSIS also requested that GAO include information in the report about the directive FSIS issued to implement the annual attestation on work-related conditions required by the poultry modernization final rule, and that the agency is sharing the information it receives as part of this process with OSHA. We have incorporated this information into the report. In its written comments, HHS agreed with our seventh recommendation that it consider including in NIOSH’s research agenda a proposal for examining the extent of peracetic acid’s use in combination with other chemicals in meat and poultry plants, and any safety and health hazards these combinations may pose to workers. As agreed with your office, unless you publicly announce the comments of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies of this report to the appropriate congressional committees; the secretaries of Labor, Agriculture, and Health and Human Services; and the Administrator of EPA. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-7215 or brownbarnesc@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix V. Appendix I: Objectives, Scope, and Methodology This report (1) describes the efforts the Occupational Safety and Health Administration (OSHA) in the U.S. Department of Labor (DOL) has made to help ensure meat and poultry workers’ safety and health, and assesses what, if any, challenges OSHA faces in carrying out these efforts; (2) examines how OSHA and the U.S. Department of Agriculture’s (USDA) Food Safety and Inspection Service (FSIS) have collaborated to help ensure meat and poultry worker safety and health; and (3) assesses any factors that may affect OSHA and FSIS efforts to protect meat and poultry workers from chemical hazards. The estimated total employment for the animal slaughtering and processing industry in this report is an annual average calculated from household data collected by the Current Population Survey (CPS) in 2016. The CPS is a probability sample and estimates derived from its data have sampling errors associated with them. We followed the DOL Bureau of Labor Statistics (BLS) technical guidance for estimating the standard error of annual average totals from CPS data. We express our confidence in the precision of our estimate as a 95 percent confidence interval. This is the interval that would contain the actual population value for 95 percent of the CPS samples that the U.S. Census Bureau could have drawn. To describe injury and illness rates in the meat and poultry industry, we analyzed and reported survey data from the BLS Survey of Occupational Injuries and Illnesses (SOII) for calendar years 2004 through 2015 (the most recent year for which data were available). The SOII provides estimates of the number and frequency (incidence rates) of workplace injuries and illnesses by industry and also by detailed case circumstances, such as injury type and event, and worker characteristics for cases that result in days away from work, based on data from logs kept by employers (survey respondents)—private industry and state and local governments. Survey respondents provide counts for all recordable injuries and illnesses under OSHA recordkeeping regulations. To report SOII data from the meat and poultry industry (using North American Industry Classification System (NAICS) code 31161 for the animal slaughtering and processing industry), BLS provided estimates of each industry’s injury and illness incidence rates and their associated relative standard errors. All estimates produced from the analysis of the SOII data are subject to sampling errors. We express our confidence in the precision of the results as a 95 percent confidence interval. This is the interval that would contain the actual population value for 95 percent of the samples the respective agency could have drawn. For estimates derived from BLS’s SOII data, we used the agency-provided relative standard errors to estimate the associated confidence intervals. All estimates we report have the associated 95 percent confidence interval provided. To assess the reliability of BLS SOII data, we reviewed documents related to the data sources, such as BLS’s Handbook of Methods, and we interviewed agency officials knowledgeable about these data. We found that SOII data were sufficiently reliable for our purposes in generally reporting estimated incidence rates of injuries and illnesses in the meat and poultry industry. To address all three objectives, we reviewed relevant federal laws and regulations and interviewed officials from OSHA and FSIS. We also visited four states—Arkansas, Georgia, Minnesota, and Texas—selected based on factors such as high production of meat or poultry; regional emphasis programs focusing on meat or poultry; presence of an OSHA regional or area office; presence of industry and worker advocate contacts; and access to meat or poultry plants participating in the Voluntary Protection Program or the Safety and Health Achievement Recognition Program. We used USDA statistics on the numbers of cattle, hogs, chicken, and turkeys slaughtered annually in the United States as a proxy for meat and poultry production for each state. As appropriate for each site visit, we met with either local OSHA or state Occupational Safety and Health (OSH) agency officials, as well as FSIS officials (including inspectors, supervisors, and an occupational safety and health official), industry representatives, experts in issues related to worker safety, and representatives of worker advocacy groups; and we visited four meat and poultry plants. At each plant, we met with plant management, FSIS management and inspectors, and plant safety and health staff, as available. The information gathered from these interviews is not generalizable to all plants or meat or poultry workers. We also interviewed and reviewed information from additional stakeholders, including experts in issues related to worker safety, as well as representatives of worker advocacy groups. We identified and interviewed these stakeholders based on previous work and on referrals from other stakeholders. We also attended worker safety conferences hosted by the meat industry, the poultry industry, and worker advocates. We also conducted group and individual interviews with meat and poultry workers in six locations in five states: Arkansas, Delaware, Nebraska, North Carolina, and Virginia. We interviewed between six and approximately 30 workers per state, totaling approximately 72 workers across all 5 interviews. We selected sites based on a variety of factors, such as states with a relatively high level of meat or poultry slaughter, according to USDA data; type of plant (meat or poultry); and geographic diversity. We also considered resource availability and the ability of supporting organizations to coordinate worker interviews. We coordinated with worker advocacy groups or worker centers to identify meat and poultry workers who were available and willing to meet with us. Interviews were conducted in English or Spanish. The information gathered from these interviews is not generalizable to all meat or poultry workers. To describe the efforts OSHA has made to help ensure meat and poultry workers’ safety and health and assess any challenges, we reviewed relevant documentation, such as agency guidance and information about enforcement and compliance assistance activities. We interviewed officials from OSHA and FSIS, as well as representatives of the meat and poultry industry. We also analyzed enforcement data from calendar years 2005-2016 from two OSHA databases: the OSHA Information System and OSHA Legacy Data. We examined data starting in 2005 because our previous report on OSHA inspections in the meat and poultry industry examined inspections data through 2004. We analyzed enforcement data on federal and state inspections of meat and poultry plants, including data on the type of inspection, violations found, standards cited, penalties assessed, and whether inspectors were denied entry into the plant. To analyze the number of inspections and the results of OSHA inspections of meat and poultry plants, we analyzed inspections of plants with NAICS codes 311611, 311612, and 311613 for meat plants, and NAICS code 311615 for poultry plants. To assess the reliability of the data, we reviewed relevant agency documentation, conducted electronic data testing, and interviewed agency officials knowledgeable about these data. Based on these reviews, we determined that the data were sufficiently reliable for our purposes. To assess OSHA’s efforts, we compared information we learned to internal controls from Standards for Internal Control in the Federal Government that call for agencies to use quality information and to internally and externally communicate the necessary quality information to achieve the entity’s objectives. To examine how OSHA and FSIS have collaborated to help ensure meat and poultry worker safety and health, we reviewed relevant documentation, such as information about OSHA’s and FSIS’s collaborative activities, and we interviewed officials from OSHA and FSIS. To analyze information on OSHA inspections of FSIS in meat and poultry plants, we used the most recent data available for calendar years 2005- 2016 from the OSHA Information System and OSHA Legacy Data. We also requested FSIS confirm which establishments pertained to the meat and poultry industry. In assessing agency efforts, we reviewed the 1994 memorandum of understanding (MOU) agreed to by OSHA and FSIS, and prior GAO reports that highlight interagency collaboration. We also compared information we learned from officials to internal controls from Standards for Internal Control in the Federal Government that call for agencies to internally and externally communicate the necessary quality information to achieve the entity’s objectives. To assess any factors that may affect OSHA and FSIS efforts to protect meat and poultry workers from chemical hazards, we reviewed relevant documentation, such as Environmental Protection Agency (EPA), Food and Drug Administration (FDA), and FSIS processes for reviewing new workplace chemicals, including FSIS’s Compliance Guideline Procedures for New Technology Notifications and Protocols. We interviewed officials from OSHA, EPA, FDA, and FSIS to understand how these reviews are carried out and the extent to which agencies coordinate and share information. We also interviewed representatives of the meat and poultry industry. We compared information we learned from our review of documents and interviews with officials to internal controls from Standards for Internal Control in the Federal Government that call for agencies to internally and externally communicate the necessary quality information to achieve the entity’s objectives. To understand efforts underway to develop tools to measure the presence of chemicals used in plants, we reviewed scientific information on chemicals, such as peracetic acid, and interviewed officials from OSHA’s Salt Lake Technical Center regarding validated sampling and analytical methods. Focus shifted to peracetic acid during the course of our review because it was identified by FSIS officials and worker advocates as a chemical commonly used in plants for which OSHA had no permissible exposure limit, and FSIS officials told us there were complaints the new chemical was causing illnesses. We reviewed National Institute for Occupational Safety and Health (NIOSH) health hazard evaluations to understand the extent of concerns related to chemicals, including peracetic acid. To identify any gaps in peracetic acid research, we reviewed documents, including NIOSH’s 2017 Request for Information on peracetic acid, as well as NIOSH’s research agenda and goals for studying the mixture of chemicals, including its 2004 Mixed Exposures Research Agenda. We also interviewed officials from NIOSH’s Education and Information Division. We conducted this performance audit from May 2016 to November 2017 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Comments from the Department of Labor Appendix III: Comments from the Department of Agriculture Appendix IV: Comments from the Department of Health and Human Services Appendix V: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Blake Ainsworth, (Assistant Director), Mary Denigan-Macauley, Eve Weisberg (Analyst-in-Charge), Rosemary Torres Lerma, Monika Gomez, Linda Collins, Erik Kjeldgaard, Cathy Roark, Susan Aschoff, James Bennett, Almeta Spencer, Sarah Cornetto, Monica Savoy, and Hiwotte Amare made significant contributions to this report. Also contributing to this report were Ivelisse Aviles, Carl Barden, Tim Bober, Kevin Bray, Marcia Crosse, John Mingus, Steve Morris, Ardith Spence, and Mark Ward. Related GAO Products Workplace Safety and Health: Additional Data Needed to Address Continued Hazards in the Meat and Poultry Industry. GAO-16-337. Washington, D.C.: April 25, 2016. Regulatory Guidance Processes: Selected Departments Could Strengthen Internal Control and Dissemination Practices. GAO-15-368. Washington, D.C.: April 16, 2015. Chemical Assessments: Agencies Coordinate Activities, But Additional Action Could Enhance Efforts. GAO-14-763. Washington, D.C.: September 29, 2014. Food Safety: USDA Needs to Strengthen Its Approach to Protecting Human Health from Pathogens in Poultry Products. GAO-14-744. Washington, D.C.: September 30, 2014. Food Safety: More Disclosure and Data Needed to Clarify Impact of Changes to Poultry and Hog Inspections. GAO-13-775. Washington, D.C.: August 22, 2013. Workplace Safety and Health: OSHA Can Better Respond to State-Run Programs Facing Challenges. GAO-13-320. Washington, D.C.: April 16, 2013. Workplace Safety and Health: Further Steps by OSHA Would Enhance Monitoring of Enforcement and Effectiveness. GAO-13-61. Washington, D.C.: January 24, 2013. Managing for Results: Key Considerations for Implementing Interagency Collaborative Mechanisms. GAO-12-1022. Washington, D.C.: September 27, 2012. Workplace Safety and Health: Multiple Challenges Lengthen OSHA’s Standard Setting. GAO-12-330. Washington, D.C.: April 2, 2012. Workplace Safety and Health: Better OSHA Guidance Needed on Safety Incentive Programs. GAO-12-329. Washington, D.C.: April 9, 2012. Workplace Safety and Health: Enhancing OSHA’s Records Audit Process Could Improve the Accuracy of Worker Injury and Illness Data. GAO-10-10. Washington, D.C.: October 15, 2009. OSHA’s Voluntary Protection Programs: Improved Oversight and Controls Would Better Ensure Program Quality. GAO-09-395. Washington, D.C.: May 20, 2009. Whistleblower Protection Program: Better Data and Improved Oversight Would Help Ensure Program Quality and Consistency. GAO-09-106. Washington, D.C.: January 27, 2009. Workplace Safety and Health: OSHA Could Improve Federal Agencies’ Safety Programs with a More Strategic Approach to Its Oversight. GAO-06-379. Washington, D.C.: April 21, 2006. Results-Oriented Government: Practices That Can Help Enhance and Sustain Collaboration among Federal Agencies. GAO-06-15. Washington, D.C.: October 21, 2005. Workplace Safety and Health: Safety in the Meat and Poultry Industry, While Improving, Could Be Further Strengthened. GAO-05-96. Washington, D.C.: January 12, 2005. Workplace Safety and Health: OSHA’s Voluntary Compliance Strategies Show Promising Results, But Should Be Fully Evaluated before They Are Expanded, GAO-04-378. Washington, D.C.: March 19, 2004. Workplace Safety and Health: OSHA Can Strengthen Enforcement through Improved Program Management. GAO-03-45. Washington, D.C.: November 22, 2002. Workplace Safety and Health: OSHA Should Strengthen the Management of Its Consultation Program. GAO-02-60. Washington, D.C.: October 12, 2001. Food Safety: Weaknesses in Meat and Poultry Inspection Pilot Should Be Addressed Before Implementation. GAO-02-59. Washington, D.C.: December 17, 2001. Worker Protection: Better Coordination Can Improve Safety at Hazardous Material Facilities. GAO-01-62. Washington, D.C.: October 26, 2000. Community Development: Changes in Nebraska’s and Iowa’s Counties with Large Meatpacking Plant Workforces. GAO/RCED-98-62. Washington, D.C.: February 27, 1998.

Why GAO Did This Study Meat and poultry slaughter and processing is one of the most hazardous industries in the United States. GAO was asked to review federal efforts to help ensure meat and poultry worker safety and health. This report (1) describes the efforts OSHA has made to help ensure worker safety and assesses any challenges to these efforts, (2) examines how OSHA and FSIS have collaborated to ensure worker safety, and (3) assesses factors that may affect OSHA and FSIS efforts to protect workers from chemical hazards. GAO analyzed OSHA inspection data from 2005—when GAO last reported on this issue—through 2016. GAO also interviewed OSHA staff in headquarters and six field offices; officials at four other federal agencies; worker advocates; and industry representatives. GAO visited four plants and interviewed workers at six sites in five states selected based on factors such as meat or poultry production. What GAO Found The Department of Labor's Occupational Safety and Health Administration (OSHA) increased its annual inspections of the meat and poultry industry from 177 in 2005 to 244 in 2016. OSHA officials told GAO that this increase was related to several new enforcement programs focusing on the poultry industry, as well as new reporting requirements that prompt additional inspections. However, OSHA faces challenges identifying and addressing worker safety concerns because workers may be reluctant to contact OSHA for fear of employer retaliation, although employers are prohibited from doing so by federal law. If workers are afraid to share concerns, OSHA may not be able to identify or address conditions that endanger them. In particular, OSHA may not be aware of the scope of problems workers could face gaining timely access to bathrooms. When asked by GAO, workers in five selected states cited bathroom access as a concern and said they fear speaking up at work, where OSHA inspectors typically interview them. Taking additional steps to encourage workers to disclose sensitive concerns and gathering additional information to determine the scope of bathroom access issues could enable OSHA to better identify worker safety and health concerns. OSHA's and the Department of Agriculture's Food Safety and Inspection Service's (FSIS) main vehicle for collaboration on worker safety is their 1994 memorandum of understanding (MOU), but efforts to implement and evaluate the MOU have been limited. The MOU outlines plans for collaboration, such as referrals of plant hazards to OSHA by FSIS inspectors, training of FSIS staff, and information sharing. OSHA and FSIS have taken some steps to implement the policies and procedures outlined in the MOU. However, GAO found issues with the MOU's implementation in these three areas, hampering achievement of the MOU's goals. For example, according to FSIS officials, FSIS inspectors may be reluctant to make referrals to OSHA about hazards in plants because they fear it could trigger an OSHA inspection of FSIS. Further, the agencies have not evaluated the implementation of the MOU. Evaluating the implementation of the MOU and making any needed changes would help ensure the goals of the MOU are met and further protect the safety and health of both plant workers and FSIS inspectors. Gaps in federal efforts create challenges to protecting workers from certain chemical hazards. For example, depending on a chemical's intended use, it may not undergo a federal review of the risks it poses to worker safety and health before it is used in a plant. FSIS collects information on how to protect its inspectors from new chemicals, but it does not have a process to share this information with OSHA or plants, among others, so that plant workers can be similarly protected. By FSIS establishing a process to regularly share the worker safety information it collects, the federal government will be better positioned to use existing resources to support the safety and health of plant workers and FSIS inspectors. What GAO Recommends GAO is making seven recommendations, including that OSHA encourage workers to disclose sensitive concerns and gather bathroom access information; OSHA and FSIS strengthen their MOU; and FSIS share worker safety information. OSHA had concerns about two of these recommendations and did not address one. FSIS expressed concerns but described planned actions to address the recommendations. GAO believes the recommendations should be fully implemented.

gao_GAO-18-113

gao_GAO-18-113_0

Background Role of Fighter Pilots Fighter pilots staff both operational and non-operational positions, and fighter pilots alternate between these positions throughout their career. Operational positions include both flying (i.e., combat pilot or instructor pilot positions) and non-flying positions, such as a close air support duty officer in an Air Operations Center or an air controller in a ground infantry unit. In flying positions, fighter pilots operate aircraft that are critical to achieving and maintaining air dominance during combat operations and include Air Force, Navy, and Marine Corps fixed-wing fighter and attack aircraft with air-to-air, air-to-ground, and electronic warfare missions. These aircraft operate during the first days of a conflict to penetrate enemy air space, defeat air defenses, and achieve air dominance, allowing follow-on ground, air, and naval forces freedom to operate within the battle space. Once air dominance is established, fighter aircraft continue to strike ground targets for the remainder of the conflict. Some fighter aircraft are also essential to protecting the homeland by responding to potential airborne and ground-based threats. Fighter pilots are assigned a variety of tasks when they are in an operational squadron. As well as studying for flights, flying, and debriefing, fighter pilots must also perform other squadron duties, such as coordinating squadron travel to external training locations, scheduling daily flights, or overseeing squadron maintenance departments. In addition to these duties, fighter pilots are required to complete common military training (i.e., training that is required for all military personnel and is not linked to a particular occupation). In May 2017, we reported that common military training comprises more than half of mandatory training requirements in the military services (not including additional training that the military services may require for specific groups of servicemembers, such as fighter pilots). Non-operational positions are generally non-flying positions and include assignments to headquarters or combatant command positions. Certain non-operational positions can only be filled by qualified pilots. For example, certain positions require fighter pilots due to the need for specialized technical knowledge, such as writing operational manuals for fighter aircraft. Other non-operational positions are more general in nature and are divided among officer communities in a military service. For example, Navy officials told us that certain shore assignments—positions that do not involve deployment—can be staffed by officers from aviation, submarine, or surface warfare communities. Aircraft Operated by Fighter Pilots DOD’s current fighter aircraft fleet is comprised of both legacy and new aircraft (see fig. 1). The legacy aircraft include Air Force F-16, F-15, A-10, and F-22A and Navy and Marine Corps F/A-18A-D, EA-6B, and AV-8B. Most of these aircraft were purchased in the 1970s and 1980s and are more than 25 years old on average. DOD has been recapitalizing this aging legacy fleet by acquiring and fielding new aircraft, namely the Navy’s F/A-18E/F and EA-18G and the joint service F-35. The Departments of the Air Force and the Navy are operating many of their fixed-wing aircraft well beyond their original designed service lives, and some of these legacy aircraft are confronted with sustainment challenges that affect their availability. In 2017, senior Air Force and Navy officials testified before the House Armed Services Committee regarding, among other things, the maintenance and sustainment issues relating to aging aircraft that are affecting the readiness of their forces. Processes for Determining and Staffing Fighter Pilot Authorizations According to Air Force, Navy, and Marine Corps guidance, the military services are to determine personnel requirements for military units. Service officials told us that this process includes squadron requirements—that is, the number of operational positions in a fighter pilot squadron that a military service has determined should be staffed by a qualified fighter pilot. Squadron requirements are primarily based on the missions the squadrons are expected to fulfill, and the military services use a variety of inputs to determine fighter pilot squadron requirements. These inputs include the projected operations of fighter squadrons, analyses of the amount of workload in the squadrons, the number of aircraft assigned to the squadrons, and the planned ratio of fighter pilots to aircraft. The military services are to determine the required number of fighter pilots to staff squadrons and meet operational mission needs and to document these in squadron staffing documents. The military services also determine the rank that pilots should have when staffing specific positions in a squadron—for example, Marine Corps F/A-18 squadron staffing documents specify the rank that should be held by pilots leading specific departments such as those for safety, operations, and maintenance. According to service officials, the military services then staff squadron requirements to the extent possible based on the number of those requirements funded by Congress and the number of trained and qualified personnel available to be staffed to those positions (see fig. 2). We refer to these funded positions as authorizations. Military service workforce planning documents acknowledge that, after this process, a squadron’s staffing level may be lower than the established squadron requirements—a readiness risk that the military services manage by assigning a higher priority to the staffing of certain positions, such as those in deployed squadrons. The military services vary in how they define when gaps between authorizations and staffing levels become a shortage. Specifically, fighter pilot staffing levels of 85 to 99 percent of authorizations could be considered a shortage, depending on the military service. For example, Air Force officials told us that their established practice is that pilot communities with less than 100 percent of authorizations are considered to be insufficiently staffed. Navy officials told us that they have a shortage when they are unable to fully staff deploying squadrons. Marine Corps personnel documents reflect that Marine Corps communities with less than 85 percent of authorizations are considered “unhealthy.” The process of staffing fighter pilots is managed in the Air Force by the Air Force Personnel Center, in the Navy by the Navy Personnel Command, and in the Marine Corps by the Deputy Commandant for Manpower and Reserve Affairs. According to service guidance, the Secretaries are to review squadron requirements, and this review is required every 2 years for the Air Force and the Marine Corps and every 5 years for the Navy. Further, DOD guidance states that staffing requirements are driven by workload and shall be established at the minimum levels necessary to accomplish mission and performance objectives. The Military Services Had Fewer Fighter Pilots Than Authorizations Due to a Variety of Contributing Factors; Initiatives Are Underway to Increase Fighter Pilot Staffing Levels According to Air Force, Navy, and Marine Corps data, each military service had fewer fighter pilots than authorizations from fiscal years 2013 through 2017, and the Air Force and the Marine Corps project that these gaps will continue for several years. According to service officials, because of low numbers of fighter pilots, the military services are unable to staff all operational fighter pilot positions. According to the military services, deploying squadrons have been fully staffed with fighter pilots, due to staffing approaches that include extending deployments and augmenting deployed squadrons with fighter pilots from other squadrons. Service officials identified multiple factors that have led to low numbers of fighter pilots, including challenges in training and retaining fighter pilots. To increase fighter pilot numbers, the military services are taking a variety of actions. The Air Force, the Navy, and the Marine Corps Had Fewer Fighter Pilots Than Authorizations According to Air Force, Navy, and Marine Corps data, each military service had fewer fighter pilots than authorizations (i.e., funded positions) from fiscal years 2013 through 2017. Specifically, the Air Force and the Marine Corps had fewer fighter pilots than authorizations for most years from fiscal years 2006 through 2017. The magnitude of these gaps has grown since fiscal year 2006 and is projected to continue through at least fiscal year 2023. The Navy had fewer fighter pilots than authorizations in fiscal years 2013 through 2017. According to service officials, these gaps between fighter pilot numbers and authorizations have prevented the military services from fully staffing operational positions, including in non- deployed squadrons and training units. The Air Force Generally Had Fewer Fighter Pilots Than Authorizations Since Fiscal Year 2006, Including for Operational Positions Since Fiscal Year 2014 According to Air Force pilot staffing level and authorizations data for fiscal years 2006 through 2017, the Air Force had fewer fighter pilots than authorizations for 11 of 12 years from fiscal years 2006 through 2017. This gap grew from 192 fighter pilots (5 percent of authorizations) in fiscal year 2006 to 1,005 (27 percent) in fiscal year 2017. According to briefing documents prepared by the Air Force, this gap is concentrated among fighter pilots with fewer than 8 years of experience. The Air Force forecasts that the fighter pilot gap will persist over time, even as the Air Force takes steps to train more fighter pilots and improve retention. Figure 3 shows the Air Force fighter pilot staffing levels and authorizations for fiscal years 2006 through 2017. For information on trends for all Air Force fixed-wing aircraft pilot communities, see appendix II. According to Air Force data, the Air Force generally had sufficient fighter pilots to staff operational positions for fighter pilots for fiscal years 2006 through 2013. Air Force officials added that during that period, Air Force fighter pilot gaps were generally limited to non-operational positions, such as staff assignments at Air Force headquarters or combatant commands. However, our analysis found that the Air Force has been unable to fully staff operational positions since fiscal year 2014. The gap between staffing levels and operational positions increased from 39 fighter pilots (1 percent of authorizations) in fiscal year 2014 to 399 (13 percent) in fiscal year 2017. The Navy Had Fewer Fighter Pilots Than Authorizations for Operational Positions in Fiscal Years 2013 through 2017 According to Navy fighter pilot staffing levels and authorizations data for fiscal years 2013 through 2017, the Navy had fewer fighter pilots than authorizations for each of these fiscal years. Specifically, in fiscal year 2013 the Navy had a gap of 57 fighter pilots (12 percent) at the first tour milestone (i.e., a fighter pilot’s first operational tour at sea completed between 3 and 6 years of service), and this gap grew to 136 fighter pilots (26 percent) in fiscal year 2017 (see fig. 4). Navy officials told us that they believe current gaps in the fighter pilot community could increase through mid-2019. For information on trends for all Navy fixed-wing aircraft pilot communities, see appendix III. According to Marine Corps pilot staffing levels and authorizations data for fiscal years 2006 through 2017, the Marine Corps had fewer fighter pilots than authorizations during that time frame. This gap grew from 63 fighter pilots (6 percent of authorizations) in fiscal year 2006 to 262 (24 percent) in fiscal year 2017. Further, according to Marine Corps data, the gap is concentrated in the Marine Corps’ junior fighter pilot population (i.e., those fighter pilots below the rank of Officer-4—a major). The Marine Corps forecasts that the fighter pilot gap will decrease over time as the Marine Corps phases out legacy aircraft and takes steps to improve retention. Figure 5 shows the Marine Corps active component fighter pilot staffing levels and authorizations for fiscal years 2006 through 2017. For information on trends for all Marine Corps fixed-wing aircraft pilot communities, see appendix IV. In addition, Marine Corps data showed that the Marine Corps was unable to fully staff fighter pilot operational positions since fiscal year 2016. The gap between staffing levels and operational positions increased from 12 fighter pilots (1 percent of authorizations) to 57 (7 percent) in fiscal years 2016 through 2017. The Military Services Used Various Staffing Approaches to Mitigate the Impact of Low Numbers of Fighter Pilots on Deploying Squadrons Although all of the military services had fewer fighter pilots than authorizations in fiscal years 2013 through 2017, service officials stated that deploying squadrons have been fully staffed with fighter pilots. Service officials reported using various approaches to continue to fully staff deploying fighter squadrons, including (1) prioritizing staffing fighter pilots to flying positions that require fighter pilot-specific technical skills; (2) using senior pilots to staff junior positions; and (3) having pilots deploy for longer and more frequently than planned, including on deployments with other squadrons. For example, Navy officials told us that approaches such as extending fighter pilots’ deployments allowed them to reduce the fiscal year 2017 first tour fighter pilot gap from 136 pilots (26 percent) to 75 pilots (15 percent). However, squadron leaders and fighter pilots told us that these approaches are having a negative impact on the fighter pilot workforce. Specifically, squadron leaders and fighter pilots told us that the high pace of operations for senior fighter pilots has limited their availability to train junior pilots, which has constrained the military services’ ability to increase the number of pilots with specific qualifications. In addition, fighter pilots told us that increased frequency of individual deployments cause instability for their families and lead to career dissatisfaction. Additionally, as we have previously reported, a high tempo of operations has increased the challenge of aviation squadrons to rebuild readiness. For example, according to Air Force officials, high deployment rates for Air Force squadrons have resulted in less time for squadrons to complete their full training requirements because high deployment rates mean that there are fewer aircraft available for training at home stations. Service officials report that they can also mitigate low numbers of fighter pilots by leveraging surpluses in other pilot communities. For example, as outlined in Air Force documents supporting pilot retention bonuses, the Air Force has staffed mobility pilots (i.e., cargo transport and aerial refueling pilots) to instructor pilot positions for certain basic flying training that would otherwise be staffed by fighter pilots. The Navy can also staff certain Department Head positions designated for fighter pilots with non- pilot aviators from that community. According to military service data, fighter pilot communities generally have the largest gaps among all military fixed-wing pilot communities. For example, in fiscal year 2017 the Air Force had 73 percent of the fighter pilots it needed, while the bomber community, which had the second largest gap among Air Force fixed- wing pilot communities, had 85 percent of the pilots it needed. Service Officials Cited Multiple Factors That Have Contributed to Low Numbers of Fighter Pilots According to service officials, squadron leaders, and fighter pilots, multiple inter-related factors have reduced each military service’s number of fighter pilots. Factors cited include reductions to active duty end strength, aircraft readiness challenges, and declining retention. Reductions to Active Duty Military End Strength Reductions to active duty military end strength have contributed to reductions in fighter pilot staffing levels. Service officials told us that reductions to military service end strength targets as part of the 2008 drawdown of forces in Iraq and Afghanistan and funding reductions related to the Budget Control Act of 2011 led to reductions in the fighter pilot workforce. For example, the Air Force offered 54 fighter pilots early retirement incentives in fiscal years 2014 through 2015, while the Marine Corps offered 49 fighter pilots early retirement options between fiscal years 2013 through 2016. Further, as we have previously reported, the Air Force used fighter pilots to meet the initial demand for UAS operators. Air Force officials told us that they removed 206 of those pilots from the fighter pilot community in fiscal years 2011 through 2012. Reduced force structure has also decreased the opportunities for fighter pilots to gain experience in their aircraft. For example, the Air Force reported that the number of total Air Force fighter squadrons (including the reserve components) declined from 134 in fiscal year 1989 to 55 in fiscal year 2017 (a 59-percent decrease), and as such fewer squadrons are available to provide newly trained fighter pilots with flying experience. Aircraft Readiness Challenges Reduced aircraft availability has affected fighter pilots’ ability to meet flight hour targets. Service leaders told us that this has resulted in delays in training new pilots with the necessary qualifications to participate in certain missions. Specifically, according to November 2017 testimony, Air Force, Navy, and Marine Corps leaders reported that fighter pilots do not meet military service flight hour targets—in part due to reduced aircraft availability. For example, Navy and Marine Corps leaders testified that Navy and Marine Corps F/A-18 pilots average 13.5 and 12.7 flight hours per month, respectively, compared with goals of 20.1 and 15.7 hours per month. A senior Air Force leader testified before Congress that Air Force fighter pilots average about 16 flight hours per month. In June 2017 we reported on readiness challenges in Air Force and Marine Corps aviation squadrons. The military services have trained fewer fighter pilots than targeted over the last decade. In fiscal years 2007 through 2016, the Air Force trained 12 percent fewer new fighter pilots than the targeted amount, while the Navy and the Marine Corps each trained 8 percent fewer new fighter pilots than the targeted amount. Fighter pilots told us that the need to prioritize the staffing of experienced pilots to deploying squadrons has limited the number of experienced personnel available to train newer pilots at home stations. Recent Safety Concerns Regarding Onboard Systems in Naval Aircraft In April 2017, the Navy paused all basic flight training on the T-45 aircraft due to safety concerns regarding the oxygen supply and atmospheric pressurization systems aboard the training aircraft. The Navy and the Marine Corps share basic flight training resources, including training for fixed-wing aircraft pilots on T-45 aircraft. Navy and Marine Corps officials told us that if these basic training squadrons are unable to produce newly trained fighter pilots on schedule, this can lead to a decreased number of new fighter pilots in both military services. The F/A-18 has also been affected by problems with onboard oxygen supply systems leading to hypoxia, which can occur when aircrews receive insufficient or contaminated oxygen on board the aircraft. In August 2017 the Navy established a team to lead its effort to research and prevent these problems in fixed- wing aircraft. numbers of trained pilots. In addition, aircraft readiness challenges led the Navy to pause flight training on the T-45 training aircraft in April 2017 due to safety concerns regarding the oxygen supply systems aboard the training aircraft. Navy officials reported that gaps in first tour operational positions designated for all fixed-wing aircraft pilot communities could grow from 86 pilots in fiscal year 2017 to about 100 in fiscal year 2019. Fighter Pilot Retention Challenges Declining retention has also contributed to low fighter pilot numbers. Our analysis of Air Force and Navy bonus retention data shows that retention of experienced fighter pilots has declined in recent years. We found that the number of Air Force fighter pilots that have signed retention contracts decreased from 63 percent in fiscal year 2013 to 35 percent in fiscal year 2017 (see fig. 6). This decline has continued even as the Air Force increased its maximum aviation bonus contract from $125,000 in fiscal year 2012 to $225,000 beginning in fiscal year 2013, the highest amount offered by any of the military services. According to Navy retention data, the Navy pool of fighter pilots eligible for the Department Head milestone (i.e., a mid-career operational leadership tour for different aspects of squadron management for pilots with between about 11 and 13 years of service) has shrunk over time. Navy officials told us that, as a result, the percentage of fighter pilots selected for the Department Head milestone has increased. For example, the Department Head selection rate for Navy F/A-18 pilots increased from 49 percent in fiscal year 2012 to 100 percent in fiscal year 2017. Further, the Navy did not meet its goals for fighter pilots signing retention bonuses at the Department Head milestone in fiscal years 2013 through 2017. For example, the Navy fell short of its retention bonus target of 73 fighter pilots by 38 pilots (48 percent of the target) for fiscal year 2017. In comparison, the surveillance and transport pilot community met or exceeded its target of pilots who signed a bonus contract 2 out of 5 years during that same period, while the maritime patrol pilot community met or exceeded its target 4 out of 5 years. Figure 7 shows the Navy Department Head active component fixed-wing pilot retention bonus take rate for fiscal years 2013 through 2017. Squadron leaders and fighter pilots we met with attributed declining retention to the staffing approaches being used by the military services to mitigate fighter pilot gaps and fully staff deployed squadrons. For example, squadron leaders told us that assigning senior fighter pilots to junior positions hurts retention by reducing leadership opportunities believed to be necessary for promotion. Fighter pilots also told us that quality of life has decreased as a result of longer and more frequent deployments. Further, fighter pilots told us that understaffing fighter pilots in operational units has resulted in an increased workload per pilot and lower quality of service for non-deployed fighter pilots in those units. The Air Force Has Implemented Initiatives to Help Increase Fighter Pilot Numbers and the Navy and the Marine Corps Are Formulating Initiatives to Address Overall Retention Concerns The Air Force has developed and implemented initiatives to help increase fighter pilot numbers, and the Navy and the Marine Corps are developing initiatives to address overall retention concerns. The Air Force established a dedicated team to identify and develop initiatives specifically to address its reported fighter pilot shortage, and this effort has resulted in over 35 implemented initiatives. The Navy and the Marine Corps have not formulated initiatives specifically for fighter pilots, but have identified actions to address retention concerns. Navy and Marine Corps officials stated that, because the military services can still staff authorizations for deployed squadrons, they do not believe their staffing levels of fighter pilots have reached a critical shortage. However, Navy and Marine Corps personnel management officials we met with told us that they are closely monitoring trends in fighter pilot retention, and they believe that decreased retention in the near future may exacerbate fighter pilot gaps in their military services. The military services’ initiatives are summarized below. The Air Force established a dedicated effort to address fighter pilot workforce challenges, and many initiatives from this effort have been implemented. Specifically, in March 2016, the Chief of Staff of the Air Force directed the initiation of an effort to address the Air Force fighter pilot shortage. The Air Force created a Fighter Enterprise Tiger Team in March 2016, and began formulating initiatives to address the fighter pilot shortage that the Air Force identified. For example, as the result of one initiative, 126 contractors have been placed in fighter squadrons to assist with administrative tasks and reduce workload for fighter pilots, and additional contractor support is in the process of being added. Also, in the fall of 2016 the Air Force reinstated its award program to recognize fighter pilots for superior performance. According to a member of the Air Force’s Fighter Enterprise Tiger Team, the awards are non-monetary, but because they are merit- based they can help fighter pilots to be more competitive when being assessed for promotion. In February 2017, the Air Force effort was expanded from a focus on fighter pilots to include all rated personnel and renamed the Aircrew Crisis Task Force. The 37 initiatives implemented by the Air Force as of November 2017 as a result of the Fighter Enterprise Tiger Team and Aircrew Crisis Task Force efforts are presented in appendix V. The Navy is formulating a service-wide strategy—referred to as Sailor 2025—which includes over 40 initiatives to address retention issues throughout the Navy. We identified 10 initiatives from Sailor 2025 that may address some of the retention issues raised in our discussion groups with fighter pilots—such as dissatisfaction with the assignments and promotion processes. For example, Navy officials told us they are developing staffing software to manage assignments and make the process more transparent and flexible. The Navy is also testing a new performance evaluation system to more accurately evaluate sailor performance. In addition, the Navy has adjusted the existing aviation bonus program by increasing the maximum bonus amount for fighter pilots from $25,000 to $30,000 per year for fiscal year 2018. To increase the number of available fighter aircraft, the Navy has also established a Rhino Readiness Recovery team— referring to the Navy’s term for the F/A-18 E-F Super Hornet aircraft— to identify and address readiness challenges in that community. Navy officials told us they believe their approaches are sufficient to address any potential future Navy fighter pilot gaps. In November 2017, the Marine Corps reinstated the aviation bonus program last offered by the Marine Corps in fiscal year 2011. The Marine Corps is offering 2-year contracts totaling $40,000 to fighter pilots who have completed their service obligation—except for those fighter pilots assigned to the EA-6B aircraft. However, the Marine Corps is not in the process of developing any non-monetary initiatives to address pilot retention. Rather, the Marine Corps is addressing aircraft readiness challenges—an issue consistently raised by fighter pilots in our discussion groups—by establishing four lines of effort to increase the number of available fighter aircraft for fighter squadrons. Marine Corps officials told us that they have begun implementing multiple initiatives for those lines of effort. For example, one initiative is focused on improving availability of aircraft spare parts by increasing their funding and modernizing the spare parts supply chain. Another initiative is focused on growing the maintenance workforce and retaining experienced aircraft maintainers. Specifically, the Marine Corps is offering retention bonuses to experienced aircraft maintainers. The Military Services Have Not Reevaluated Fighter Pilot Squadron Requirements to Reflect Changing Conditions, Increased Workload, and Emerging UAS Requirements Fighter pilots and squadron leaders told us that the fighter pilot occupation has significantly changed in recent years, but the military services have not reevaluated fighter squadron requirements. Fighter pilots and squadron leaders from each of the military services consistently told us that the fighter pilot occupation has significantly changed in recent years due to changes in fighter aircraft tactics and technology, additional training requirements, and the removal of administrative support positions from squadrons. The fighter pilots added that these changes have led to an unsustainable increase in workload. As discussed earlier, squadron requirements—the number of fighter pilots needed to meet operational mission needs—are calculated by the military services using a variety of inputs, including workload. Once these requirements are funded by Congress, they are an “authorization.” Service guidance requires squadron requirements to be reevaluated on a 2-year schedule (5 years for the Navy) and to be updated as conditions change. For the Air Force, guidance defines staffing requirements as the staffing needed to accomplish a job, mission, or program, and notes that staffing should be sized to reflect the minimum essential level to accomplish the required workload. The Office of the Administrative Assistant to the Secretary of the Air Force (Resources), along with Major Command manpower staffs establish staffing standards and, at a minimum, by policy are to reevaluate these standards for applicability and updates every 2 years, or earlier if dictated by significant workload or mission changes. The Air Force Manpower Analysis Agency determines staffing resource requirements and provides staffing and management consultation services to Air Force functional communities for improved resource utilization and enhanced mission effectiveness and efficiency. The Air Force could not provide specifics on the most recent updates to squadron requirements, because such data were stored in a database that managed requirements on a position-by-position basis, rather than aggregated by squadron. Air Force pilots and squadron leaders consistently told us that changing conditions in fighter squadrons, such as a higher pace of changes to tactics and technology, increased training requirements, and reduced administrative support, have increased fighter pilot workload. However, Air Force officials told us that metrics that inform squadron requirements (i.e., crew ratios, the targeted ratio of pilots to aircraft) have not been increased because the Air Force is instead prioritizing the effort to recapitalize its fleet of fighter aircraft. Separately from reevaluating squadron requirements, Air Force officials told us that they have implemented changes to address workload concerns cited by fighter pilots—such as adding contractor staff in squadrons to provide administrative support, as part of initiatives to address fighter pilot shortages they have identified. According to Air Force officials, the Air Force is currently reassessing non-operational requirements for fighter pilots (i.e., non-flying positions at headquarters organizations). Air Force officials told us that this reassessment is focused on determining which non-operational requirements currently assigned to fighter pilots could be assigned to other types of officers or pilots, to reduce the overall number of fighter pilot requirements. For the Navy, squadron requirements are dependent on the current wartime requirements developed by the Navy, referred to as the Required Operational Capability and Projected Operational Environment of a particular squadron, aircraft configuration, specified operating profile, computed workload, and established doctrinal constraints. The June 2017 update to the relevant Navy guidance reduced the frequency of the reviews of these requirements from every 2 years to every 5 years, with updates as required by those officials responsible for specific units. Navy officials told us that reviews are to be completed every 2 years, but updates are only made to squadron staffing documents if specific events occur, such as additional aircraft being assigned to a squadron. Navy officials added that they believe squadron requirements are accurate and updated with sufficient frequency. However, Navy pilots and squadron leaders consistently told us that changing conditions in fighter squadrons, such as a higher pace of changes to tactics and technology, increased training requirements, and more frequent individual deployments, have increased fighter pilot workload. However, the Navy has not recently updated squadron requirements to reflect such changes. Specifically, Navy fighter pilot squadron requirements are outlined in 15 Navy squadron staffing documents, and as of November 2017, 9 out of 15 of those documents had not been updated within the last 2 years. For the Marine Corps, guidance states that reviews to optimize force structure will be conducted every 2 years, taking into consideration new and emerging requirements. Marine Corps pilots and squadron leaders consistently told us that changing conditions in fighter squadrons, such as a higher pace of changes to tactics and technology, reduced aircraft availability, and more frequent individual deployments, have increased fighter pilot workload. However, the Marine Corps has not updated squadron requirements to reflect such changes. Specifically, Marine Corps fighter pilot squadron requirements are outlined in four fighter squadron staffing documents, and, as of November 2017, none had been updated within the last 2 years. Marine Corps squadron leaders and fighter pilots told us that updates to squadron requirements are not being conducted for squadrons of legacy aircraft, but added that they believe updates are warranted due to the continued delays in fielding the F-35 and resulting extensions to the planned service of legacy platforms. Marine Corps officials told us that they have not updated squadron requirements because (1) Marine Corps fighter pilot authorizations and staffing levels are below squadron requirements, so any increase to squadron requirements would require a significant increase to fighter pilot authorizations, and (2) the Marine Corps has faced challenges obtaining technical assistance to conduct workload assessments in fighter squadrons. The Navy Manpower Analysis Center conducts workload reviews of squadrons, based on specific events such as changes to the amount of time needed for maintaining the specific type of aircraft. Navy and Marine Corps fighter pilots we met with told us that they have had difficulties maintaining fighter jets in their squadrons, and Navy and Marine Corps leaders have made similar statements in congressional testimony. Further, Marine Corps officials told us that they have had difficulty utilizing the Navy Manpower Analysis Center to update workload analyses for their fighter squadrons, as they believed the center prioritizes work for Navy organizations. Navy Manpower Analysis Center officials told us that there is no formal requirement for their organization to conduct analyses for the Marine Corps, but that they respond to such requests on an ad-hoc basis. They added that the Marine Corps Combat Development Command has formal responsibility for updating Marine Corps workload analyses. We have previously reported that the size and data collection capacity of the Navy Manpower Analysis Center has limited the Navy’s capacity to carry out periodic workload reassessments, which may be a contributing factor to chronic under-staffing of ship crews. The Military Services Have Not Incorporated Plans for Increased Reliance on UAS into Fighter Pilot Workforce Projections According to a DOD planning document, funding for UAS platforms was expected to grow by 17 percent in fiscal years 2014 through 2018. Moreover, in 2015, the Secretary of the Navy directed the establishment of a Deputy Assistant Secretary of the Navy for Unmanned Systems and announced that the F-35 will likely be the last cockpit-operated strike fighter aircraft the Department of the Navy will buy or fly. The Chief of Naval Operations announced in 2017 that the future of the Navy includes UAS systems as an integral part of the future fleet and must be purchased in large numbers to expand naval presence in key areas. For example, the Navy is developing a UAS platform—the MQ-25 Stingray— which is intended to replace that portion of the F/A-18 fighter aircraft’s mission set that involves re-fueling other F/A-18 aircraft. Also, in 2015 the Secretary of the Air Force stated that UAS pilots were flying on average about four times more hours than pilots in cockpit- operated aircraft. Further, a document outlining the Air Force’s vision for UAS for fiscal years 2009 through 2047 notes that systems will work in tandem with cockpit-operated aircraft, for example to attack air-defense systems, and that autonomous technologies will potentially allow one pilot to direct multiple aircraft, leading to personnel efficiencies. Although the impact of UAS on the fighter pilot workforce appears to be significant, the Air Force, the Navy, and the Marine Corps have not accounted for the planned increased use of UAS to complete missions similar to those carried out by fighter aircraft, and the potential impact of these changes on fighter pilot requirements. Specifically, Air Force and Navy officials told us that their military services have not conducted an assessment of the impact of future UAS operations on fighter pilot requirements. While the UAS platforms that are expected to overlap with fighter aircraft missions will not be fielded until the mid-2020s, the length of time required to develop an experienced fighter pilot compels the military services to begin incorporating these planned changes to fighter pilot requirements promptly. For example, Navy fighter pilots who are entering initial training in 2018 will not have completed their active duty service obligation (currently 8 years after Navy pilots complete flight training) when the MQ-25 Stingray system is expected to be fielded in 2026. A key tenet of human capital planning is determining existing and future skills and competencies, and associated workforce gaps. Without steps by the military services, to include reevaluating workload and taking into account the impact that the planned use of UAS will have on the fighter pilot workforce, the military services will not fully know the extent and nature of gaps between fighter pilot numbers and authorizations and how to best target actions to address these gaps. Conclusions Fighter pilots are critical to achieving and maintaining air dominance during combat operations. To achieve that mission, the military services must have appropriate numbers of qualified fighter pilots. Service officials report that no unit is deploying without 100 percent of its fighter pilots, and they believe that they will continue to be able to meet their operational missions. Nevertheless, the Air Force, the Navy, and the Marine Corps, are reporting fewer fighter pilots than authorizations, and they project that these gaps will continue through at least fiscal year 2023. Without re- evaluating fighter pilot requirements, it will be difficult for the military services to accurately determine the number of fighter pilots needed to complete missions and help ensure success in combat. Specifically, without updating squadron requirements to reflect the growing administrative burden on fighter pilots in non-deployed squadrons, the currently identified differences between fighter pilot numbers and authorizations may be understated. By contrast, without updating future fighter pilot requirements to take into account changing roles and missions—in particular the increasing role of UAS in combat operations— forecasted fighter pilot gaps may be overstated. In short, reevaluating fighter pilot requirements is a key first step to help the military services clearly determine the magnitude of the gaps and target strategies to meet their personnel needs. Recommendations for Executive Action We are making the following three recommendations: The Secretary of the Air Force should ensure that the Director of Operations and the Air Force Manpower Analysis Agency reevaluate fighter pilot squadron requirements, to include updating current assumptions of fighter pilot workload, and assessing the impact of future incorporation of UAS platforms into combat aviation. (Recommendation 1) The Secretary of the Navy should ensure that the Chief of Naval Operations reevaluate fighter pilot squadron requirements, to include updating current assumptions of fighter pilot workload, and assessing the impact of future incorporation of UAS platforms into combat aviation. (Recommendation 2) The Secretary of the Navy should ensure that the Commandant of the Marine Corps and the Deputy Commandant for Aviation reevaluate fighter pilot squadron requirements. (Recommendation 3) Agency Comments We provided a draft of this report to DOD for review and comment. We had initially recommended that the Marine Corps also assess the impact of UAS platforms on fighter pilot squadron requirements, but removed that portion of the third recommendation because Marine Corps officials told us that Marine Corps UAS squadrons will continue to be resourced with operators through the accession process and Marine Corps UAS operator requirements do not affect either pilot inventories or fighter pilot workload. In its written comments, reproduced in appendix VI, DOD concurred with our recommendations, citing its commitment to addressing manpower, personnel, and training challenges for the fighter pilot community and broader aviation and aviation support capabilities. DOD also provided technical comments, which we incorporated as appropriate. We are sending copies of this report to appropriate congressional committees; the Secretary of Defense; the Secretary of the Air Force; the Secretary of the Navy; and the Commandant of the Marine Corps. In addition, the report will be available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact Brenda S. Farrell at (202) 512-3604 or farrellb@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix VII. Appendix I: Scope and Methodology To assess the extent to which the military services had differences in the number of fighter pilots compared to authorizations, as well as contributing factors and service initiatives to address the differences, we obtained and analyzed data on authorizations designated for pilots and corresponding staffing levels of pilots for all fixed-wing, cockpit-operated (hereafter referred to as fixed-wing) aircraft communities in the Air Force, the Navy, and the Marine Corps. We did not include the Army in the scope of our review because the Army does not operate fighter aircraft. For the Air Force and the Marine Corps active component, we compared pilot staffing levels with authorizations for all fixed-wing aircraft communities for fiscal years 2006 through 2017. We obtained and analyzed projected authorizations and staffing levels for the same pilot communities for fiscal years 2018 through 2023. We further obtained and reviewed Marine Corps data on fighter pilot operational position staffing targets and staffing levels for fiscal year 2017 and similar Air Force data for fiscal year 2018. We also compared Air Force and Marine Corps reserve component fighter pilot staffing levels with authorizations for fiscal years 2006 through 2017. For the Navy, we obtained and analyzed data on authorizations designated for active and reserve component pilots for fiscal years 2006 through 2017, and staffing levels for all Navy fixed-wing aircraft communities in the active component for fiscal years 2011 through 2017. However, the Navy’s authorization data did not specify how many fighter pilots were assigned to non-flying assignments because the Navy does not fully assign non-flying authorizations to specific communities, unlike the Air Force and the Marine Corps. Therefore, we were unable to conduct an analysis comparing total Navy fighter pilot staffing levels with authorizations, as we did for the Air Force and the Marine Corps. We instead obtained and analyzed Navy data on differences between authorizations and pilot staffing levels for first operational tour, Department Head, and Command positions for all fixed-wing aircraft pilot communities in the active component. We also obtained and analyzed Navy data on differences between staffing targets and pilot staffing levels for the Navy Reserve for fiscal year 2017, the only year of data available. We also obtained and analyzed Navy retention data for pilots eligible for Department Head assignments, a mid-career milestone in fixed-wing communities for fiscal years 2011 through 2017. Retention data for the Department Head milestone are made available in annual aviation continuation pay reports to Congress. Fiscal year 2018 retention data will be available in fiscal year 2019. We further obtained and analyzed the number of fighter pilots the Air Force, the Navy, and the Marine Corps trained in fiscal years 2007 through 2016. To assess the reliability of the data we obtained, we reviewed corroborating documentation, analyzed the data for inconsistencies, and interviewed service officials about the reliability of the data. We determined that the data we used were sufficiently reliable to describe the trends in personnel staffing levels and authorizations for the time period included in our scope. We met with DOD and service officials to discuss the results of our analysis and factors that may have contributed to low numbers of fighter pilots. We also collected and reviewed service documentation regarding the factors they identified. We interviewed service officials and reviewed documentation to identify any initiatives taken or planned to increase fighter pilot numbers. In addition, we selected a non-generalizable sample of locations where fighter pilots are stationed (see table 1). We selected these locations based on geographic diversity (one location for each military service in both the eastern and western portions of the contiguous United States), a diversity of types of fighter aircraft, and a mix of squadron types at the locations (i.e., operational squadrons, training squadrons, and reserve component squadrons). In selecting locations we also considered the availability of pilots due to conflicts with deployment or training events. At each location, we moderated one to two discussion groups with fighter pilots for a total of 13 discussion groups ranging from between 3 and 20 pilots per group. We held separate sessions with junior and senior pilots at all locations, except for at Naval Air Station Oceana, Marine Corps Air Station Cherry Point, and Marine Corps Air Station Miramar, due to pilot availability. We also interviewed unit leadership at these locations (i.e., wing and squadron commanders and executive officers) to obtain their perspective on the status of the fighter pilot workforce. While these discussion groups and interviews allowed us to learn about many important aspects of the fighter pilot workforce from the perspective of fighter pilots and squadron leaders, they were designed to provide anecdotal information and not results that would be representative of all the department’s more than 5,000 fighter pilots as of fiscal year 2017. To assess the extent to which the military services have reevaluated squadron requirements for the number of fighter pilots needed, including the consideration of UAS pilot requirements, we reviewed service guidance to determine the frequency with which elements of fighter pilot squadron requirements are to be reevaluated, reviewed service documentation, and interviewed service officials to determine the extent to which these elements had been reevaluated on schedule, reviewed service documentation regarding the planned mix of cockpit-operated and remotely-operated aviation platforms for future operations, and discussed with service officials the extent to which these plans are incorporated into forecasts of fighter pilot squadron requirements. We conducted this performance audit from November 2016 to April 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Comparison of Air Force Pilot Staffing Levels and Authorizations for Fixed- Wing, Cockpit-Operated Aircraft The Air Force uses pilots from both the active and reserve components to staff fixed-wing, cockpit-operated (hereafter referred to as fixed-wing) aircraft pilot positions that Congress authorizes and funds through appropriations. These Air Force pilots staff a mix of operational and non- operational positions. Operational positions include both flying (e.g., combat pilot or instructor pilot positions) and non-flying positions (e.g., close air support duty officer positions in an Air Operations Center) that directly support combat operations. Non-operational positions are generally non-flying, and include assignments to headquarters or combatant command positions, some of which can be staffed by other types of Air Force officers. This appendix compares Air Force pilot staffing levels with authorizations for operational and non-operational positions for all fixed-wing aircraft communities for fiscal years 2006 through 2017. Air Force fixed-wing community pilots operate fighter, bomber, mobility, surveillance, and special operations aircraft. Fighter pilots operate tactical aircraft that engage in air-to-air and air-to-surface attacks and include the A-10, F-15, F-16, F-22A, and F-35 aircraft. Bomber pilots operate aircraft to deliver munitions and include the B-1, B-2, and B-52 aircraft. Mobility pilots operate aircraft used for aerial refueling and troop and cargo transport and include the C-17 and KC-135 aircraft. Surveillance pilots operate aircraft used for surveillance and reconnaissance to support ground troops and include the E-8 and U-2 aircraft. Special operations pilots operate aircraft that provide close-air support for ground troops, including the AC-130. Air Force officials report that they can staff certain operational positions designated for fighter pilots with pilots from other pilot communities that have surpluses. For example, Air Force officials told us they can staff mobility or surveillance pilots (communities which both had a surplus of pilots in fiscal year 2017) to certain basic flying training instructor pilot positions that would otherwise be staffed by fighter pilots. Figure 8 shows the Air Force active component fixed-wing aircraft community pilot staffing levels and authorizations for fiscal year 2017. According to Air Force data for the active component, the Air Force had fewer fighter pilots than authorizations in 11 of 12 years from fiscal year 2006 through fiscal year 2017. This gap grew from 192 fighter pilots (5 percent of authorizations) in fiscal year 2006 to 1,005 (27 percent of authorizations) in fiscal year 2017. Figure 9 shows the comparison of the Air Force’s active component fighter pilot staffing levels with authorizations for fiscal years 2006 through 2017. According to Air Force data for the reserve components the Air National Guard and the Air Force Reserve, the Air Force had fewer fighter pilots than authorizations every fiscal year from fiscal year 2006 through 2017. For example, the Air Force reported that the reserve components had a gap of 271 fighter pilots (17 percent of authorizations) in fiscal year 2017. Figure 10 illustrates the gap between staffing levels and authorizations for Air Force fighter pilots in the reserve components for fiscal years 2006 through 2017. According to Air Force data for the active component, the Air Force had fewer bomber pilots than authorizations in fiscal years 2014 through 2017. This gap grew from 11 bomber pilots (1 percent of authorizations) in fiscal year 2014 to 135 (15 percent of authorizations) in fiscal year 2017. Figure 11 shows the comparison of the Air Force’s active component bomber pilot staffing levels with authorizations for fiscal years 2006 through 2017. According to Air Force data for the active component, the Air Force had more mobility pilots than authorizations from fiscal year 2006 through fiscal year 2017. This surplus peaked at 1,637 mobility pilots (132 percent of authorizations) in fiscal year 2011, and declined to 264 (105 percent of authorizations) in fiscal year 2017. Figure 12 shows the comparison of the Air Force’s active component mobility pilot staffing levels with authorizations for fiscal years 2006 through 2017. According to Air Force data for the active component for fiscal years 2006 through 2017, the Air Force had fewer surveillance pilots than authorizations in fiscal years 2012 and 2013. In fiscal year 2017, the surplus was 220 surveillance pilots (128 percent of authorizations). Figure 13 shows the comparison of the Air Force’s active component surveillance pilot staffing levels with authorizations for fiscal years 2006 through 2017. According to Air Force data for the active component, the Air Force had fewer special operations pilots than authorizations from fiscal year 2006 through fiscal year 2017. Special operations pilot staffing levels and authorizations have increased substantially from fiscal year 2009 through fiscal year 2017. Further, the gap between the staffing levels and authorizations decreased from 342 special operations pilots (29 percent of authorizations) in fiscal year 2009 to 227 (14 percent of authorizations) in fiscal year 2017. Figure 14 shows the comparison of the Air Force’s active component special operations pilot staffing levels with authorizations for fiscal years 2006 through 2017. Appendix III: Comparison of Navy Pilot Staffing Levels and Authorizations for Fixed- Wing, Cockpit-Operated Aircraft The Navy uses pilots from both the active and reserve component to staff fixed-wing, cockpit operated (hereafter referred to as fixed-wing) aircraft pilot positions that Congress authorizes and funds through appropriations. These Navy pilots staff a mix of operational and non-operational positions. Operational positions include both flying (i.e., combat pilot or instructor pilot) and non-flying positions that directly support combat operations. Non-operational positions are generally non-flying positions, and include assignments to positions at headquarters or in the combatant commands that can be staffed by other types of Navy officers. The Navy does not separate non-operational fighter pilot authorizations from authorizations for other pilots. As a result, this appendix only presents Navy pilot staffing levels for those communities for fiscal years 2011 through 2017, and compares Navy pilot staffing levels to specific operational positions. Specifically, we compared authorizations and pilot staffing levels for Navy first operational tour, Department Head, and Command positions for all fixed-wing, cockpit-operated aircraft communities in the active component for fiscal years 2013 through 2017. Navy fixed-wing community pilots operate fighter, surveillance and transport, and maritime patrol aircraft. Navy fighter pilots operate tactical aircraft for air defense and support. Fighter aircraft include both legacy and Super Hornet variants of the F/A-18, as well as newer tactical aircraft such as the EA-18G developed for electronic warfare and the F- 35. Surveillance and transport pilots operate turboprop aircraft, including the E-2D for surveillance and airborne early warning missions and the C- 2A for troop and cargo transport between aircraft carriers and shore bases. Maritime patrol pilots operate jet aircraft for missions including anti-submarine warfare and anti-surface warfare, and include aircraft such as the P-8A. According to Navy personnel data, Navy fighter pilot staffing levels decreased from 1,707 fighter pilots in fiscal year 2015 to 1,548 (a 9- percent decrease) in fiscal year 2017. Figure 15 compares changes in the Navy fighter, surveillance and transport, and maritime patrol pilot staffing levels for fiscal years 2011 through 2017. According to Navy fixed-wing pilot staffing levels and authorizations data for the first tour milestone (i.e., pilots’ first operational tours at sea for pilots generally with between 3 and 6 years of service), the Navy was unable to fully staff fighter pilot, surveillance and transport, and maritime patrol operational positions from fiscal years 2013 through 2017. The fighter pilot gap grew from 57 fighter pilots (12 percent of authorizations) in fiscal year 2013 to 136 fighter pilots (26 percent) in fiscal year 2017. The surveillance and transport pilot gap varied from 29 pilots (20 percent of authorizations) in fiscal year 2013 to 30 pilots (23 percent) in fiscal year 2017, while the maritime patrol community pilot gap decreased from 112 pilots (23 percent of authorizations) in fiscal year 2013 to 4 pilots (1 percent) in fiscal year 2017. Figure 16 compares the Navy active component fighter pilot, surveillance and transport, and maritime patrol communities’ first tour staffing levels and authorizations for operational positions for fiscal years 2013 through 2017. According to Navy fixed-wing aviator staffing levels and authorizations data for the Department Head milestone (i.e., a mid-career operational leadership tour for different aspects of squadron management for pilots with between about 11 and 13 years of service), the military service had more fighter, surveillance and transport, and maritime patrol aviators than authorizations for fiscal years 2013 through 2017. However, the surplus of fighter aviators compared with authorizations decreased from 68 aviators (133 percent of authorizations) in fiscal year 2013 to 28 aviators (114 percent) in fiscal year 2017. Figure 17 compares the Navy active component fighter, surveillance and transport, and maritime patrol aviator communities’ Department Head staffing levels and authorizations for operational positions for fiscal years 2013 through 2017. According to Navy fixed-wing aviator staffing levels and authorizations data for the Command milestone (i.e., a leadership tour for Commanders, including squadron commander, for aviators with between about 17 and 19 years of service) the number of fighter, surveillance and transport, and maritime patrol aviators compared with authorizations increased from fiscal years 2013 through 2017. For example, while the fighter pilot community had fewer aviators than authorizations in fiscal year 2013 (a gap totaling 2 percent of authorizations), it had a surplus of aviators in fiscal year 2017 (2 percent above authorizations). Figure 18 compares the Navy active component fighter, surveillance and transport, and maritime patrol aviator communities’ Command milestone staffing levels and authorizations for fiscal years 2013 through 2017. Appendix IV: Comparison of Marine Corps Pilot Staffing Levels and Authorizations for Fixed-Wing, Cockpit Operated Aircraft The Marine Corps uses pilots from both the active and reserve components to staff fixed-wing, cockpit-operated (hereafter referred to as fixed-wing) aircraft pilot positions that Congress authorizes and funds through appropriations. These Marine Corps pilots staff a mix of operational and non-operational positions. Operational positions include both flying (e.g., combat pilot or instructor pilot positions) and non-flying positions (e.g., air controller in a ground infantry unit) that generally support combat operations. Non-operational positions are generally non- flying and include assignments to headquarters or combatant command positions, some of which can be staffed by other types of Marine Corps officers. This appendix compares Marine Corps pilot staffing levels with authorizations for operational and non-operational positions for all fixed- wing aircraft communities for fiscal years 2006 through 2017. Marine Corps fixed-wing community pilots operate fighter, tiltrotor, and tanker aircraft. Fighter pilots operate tactical aircraft for air defense and close air support and attack missions, and include the EA-6B, AV-8B, F/A-18, and F-35. Tiltrotor pilots operate the MV-22—an aircraft that operates as a helicopter for takeoffs and landings and, once airborne, converts to a turboprop aircraft—and is used to transport combat troops and equipment. Tanker pilots operate the KC-130, an aircraft used for in- flight refueling and transport of troops and equipment. According to Marine Corps data for the active component, the Marine Corps had fewer pilots than authorizations in all of its fixed-wing communities in fiscal year 2017. The Marine Corps forecasts this gap will persist through at least fiscal year 2023. Figure 19 shows the Marine Corps active component fixed-wing pilot staffing levels and authorizations for fiscal year 2017. According to Marine Corps data for the active component, the Marine Corps had fewer fighter pilots than authorizations from fiscal year 2006 through fiscal year 2017. This gap grew from 63 fighter pilots (6 percent of authorizations) in fiscal year 2006 to 322 fighter pilots (24 percent) in fiscal year 2017. Figure 20 shows the comparison of the Marine Corps’ active component fighter pilot staffing levels and authorizations for fiscal years 2006 through 2017. According to Marine Corps data for the active component, the Marine Corps had fewer fighter pilots than operational positions in fiscal years 2016 and 2017. Figure 21 shows the comparison of the Marine Corps’ active component fighter pilot staffing levels with operational positions for fiscal years 2006 through 2017. According to Marine Corps data for the reserve component, the Marine Corps Reserve, a community that the Marine Corps uses to augment its available staffing levels of active duty pilots, had more fighter pilots than authorizations for 8 of 12 years in fiscal years 2006 through 2017. The Marine Corps had a gap of seven reserve component fighter pilots (6 percent of authorizations) for fiscal year 2017. Figure 22 shows the Marine Corps Reserve fighter pilot staffing levels and authorizations for fiscal years 2006 through 2017. According to Marine Corps data for the active component, the Marine Corps had fewer junior fighter pilots—those pilots between grades Officer-1 and Officer-3—than authorizations from fiscal year 2006 through fiscal year 2017. Marine Corps officials told us that, as a result, the Marine Corps assigns pilots at the Officer-4 grade to staff positions designated for junior pilots. For example, in fiscal year 2017 the Marine Corps needed an additional 309 junior fighter pilots (48 percent of authorizations) to fill all authorizations. Figure 23 shows the comparison of the Marine Corps active component junior and senior fighter pilot staffing levels and authorizations for fiscal year 2017. According to Marine Corps data for the active component, the Marine Corps had fewer tiltrotor pilots than authorizations from fiscal year 2006 through fiscal year 2017. Tiltrotor pilot staffing levels and authorizations have increased substantially from fiscal year 2006 through fiscal year 2017. Further, the gap between the staffing levels and authorizations has decreased from 137 tiltrotor pilots (70 percent of authorizations) in fiscal year 2006 to 322 tiltrotor pilots (34 percent of authorizations) in fiscal year 2017. Figure 24 shows the comparison of the Marine Corps active component tiltrotor pilot staffing levels and authorizations in fiscal years 2006 through 2017. According to Marine Corps active component data, the Marine Corps had fewer tanker pilots than authorizations from fiscal year 2006 through fiscal year 2017. This gap decreased from 86 tanker pilots in fiscal year 2006 (22 percent of authorizations) to 18 tanker pilots (5 percent of authorizations) in fiscal year 2017. Figure 25 shows the comparison of the Marine Corps active component tanker pilot staffing levels and authorizations for fiscal years 2006 through 2017. Appendix V: Initiatives Implemented by the Air Force to Address Reported Fighter Pilot Shortages In March 2016, the Chief of Staff of the Air Force created the Fighter Enterprise Tiger Team to address the fighter pilot shortage that the Air Force identified. In February 2017, the Air Force effort was expanded to include all rated personnel and renamed the Aircrew Crisis Task Force. According to Air Force officials, the task force has focused on the following areas to improve fighter pilot retention: work/life balance, quality of service, and monetary compensation. In August 2017 the Aircrew Crisis Task Force held a Dedicated Aircrew Retention Team Summit that included organizing discussion groups with pilots to obtain information on retention challenges. According to Air Force officials, in September 2017, the task force presented 25 of the 44 recommendations developed at the summit to the Chief of Staff of the Air Force and as of November 2017 the Chief of Staff had decided to implement 2 of them immediately and conduct additional analysis on the other 23. The recommendations being analyzed include reducing the length and number of deployments for fighter pilots and converting some non-flying fighter pilot positions to UAS pilot positions. As of November 2017, these efforts had resulted in 37 implemented initiatives (see table 2). Appendix VI: Comments from the Department of Defense Appendix VII: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, key contributors to this report were Lori Atkinson, Assistant Director; Vincent Buquicchio, Timothy Carr, Mae Jones, Foster Kerrison, Amie Lesser, Michael Silver, and Nell Williams. Related GAO Products Military Personnel: Actions Needed to Better Position the Navy and the Marine Corps to Support Expanding Unmanned Systems Operations. GAO-18-162. Washington, D.C.: February 6, 2018. Department of Defense: Actions Needed to Address Five Key Mission Challenges. GAO-17-369. Washington, D.C.: June 13, 2017. DOD Training: DOD Has Taken Steps to Assess Common Military Training. GAO-17-468. Washington, D.C.: May 23, 2017. Navy Force Structure: Actions Needed to Ensure Proper Size and Composition of Ship Crews. GAO-17-413. Washington, D.C.: May 18, 2017. High Risk Series: Progress on Many High-Risk Areas, While Substantial Efforts Needed on Others. GAO-17-317. Washington, D.C.: February 15, 2017. Military Compensation: Additional Actions Are Needed to Better Manage Special and Incentive Pay Programs. GAO-17-39. Washington, D.C.: February 3, 2017. Unmanned Aerial Systems: Air Force and Army Should Improve Human Capital Planning for Pilot Workforces. GAO-17-53. Washington, D.C.: January 31, 2017. Unmanned Aerial Systems: Further Actions Needed to Fully Address Air Force and Army Pilot Workforce Challenges. GAO-16-527T. Washington, D.C.: March 16, 2016. Unmanned Aerial Systems: Actions Needed to Improve DOD Pilot Training. GAO-15-461. Washington, D.C.: May 14, 2015. Air Force: Actions Needed to Strengthen Management of Unmanned Aerial System Pilots. GAO-14-316. Washington, D.C.: April 10, 2014. Tactical Aircraft: DOD’s Ability to Meet Future Requirements Is Uncertain, with Key Analyses Needed to Inform Upcoming Investment Decisions. GAO-10-789. Washington, D.C.: July 29, 2010. Unmanned Aircraft Systems: Comprehensive Planning and a Results- Oriented Training Strategy Are Needed to Support Growing Inventories. GAO-10-331. Washington, D.C.: March 26, 2010. Human Capital: Key Principles for Effective Strategic Workforce Planning. GAO-04-39. Washington, D.C.: December 11, 2003. Military Personnel: Actions Needed to Better Define Pilot Requirements and Promote Retention. GAO/NSIAD-99-211. Washington, D.C.: August 20, 1999.

Why GAO Did This Study Fighter pilots operate aircraft that are critical to achieving and maintaining air dominance during combat operations. The military services invest significant time and funding to train, compensate, and retain fighter pilots. According to Air Force officials, it costs between $3-$11 million and takes approximately 5 years to develop an individual fighter pilot to lead combat missions. Senate Report 114-255 included a provision for GAO to review the Department of Defense's (DOD) management of the fighter pilot workforce. GAO's report (1) assesses the extent to which the military services had differences in the number of fighter pilots compared to authorizations, and describes any contributing factors as well as initiatives to address the differences, and (2) assesses the extent to which the military services had reevaluated squadron requirements for the number of fighter pilots needed, including consideration of UAS pilot requirements. GAO analyzed military service personnel data, documentation on service initiatives to address factors contributing to fighter pilot shortages, and service documentation of requirements; met with a non-generalizable sample of fighter pilots at seven locations; and interviewed DOD and service officials. What GAO Found The Air Force, the Navy, and the Marine Corps had gaps between the actual numbers of fighter pilots and authorizations (i.e. funded positions) in fiscal years (FY) 2013 through 2017. In FY 2017 the Air Force's gap was the widest at 27 percent of authorizations (see fig. below) and is projected to continue through FY 2023. The Marine Corps' gap grew from 6 percent in FY 2006 to 24 percent in FY 2017; it is concentrated in fighter pilots below the rank of major. While the Navy did not have comparable data, it had a gap at fighter pilots' first operational tours that grew from 12 percent in FY 2013 to 26 percent in FY 2017, and Navy officials stated it could increase through mid-2019. Service officials attributed these gaps to aircraft readiness challenges, reduced training opportunities, and increased attrition of fighter pilots due to career dissatisfaction. To help increase fighter pilot numbers, the military services are taking actions, including increasing the amounts of financial incentives to retain pilots. The military services have not recently reevaluated squadron requirements to reflect increased fighter pilot workload and the emergence of unmanned aerial systems (UAS). According to service guidance, squadron requirements are to be reviewed on a 2-year schedule and to be updated as conditions change (in June 2017 the Navy revised its guidance to extend its schedule from 2 years to 5 years). However, service officials acknowledged that they have not updated all squadron requirements within the last 2 years. These officials stated that the requirements have not been reevaluated because existing conditions do not warrant the change. However, fighter pilots and squadron leaders interviewed at locations GAO visited consistently stated that the typical workload has significantly increased in recent years due to, among other things, changes in fighter aircraft tactics and technology and reductions to administrative support in squadrons. Further, the military services have not assessed the effect of increased reliance on UAS on fighter pilot requirements. The Air Force's vision for UAS notes that systems will work in tandem with cockpit-operated aircraft and that autonomous technologies will potentially lead to personnel efficiencies. Without re-evaluating squadron requirements to reflect current and emerging conditions, the nature of the gap may be inaccurate and thus make it difficult for the military services to target strategies to meet their personnel needs. What GAO Recommends GAO recommends that the Air Force, the Navy, and the Marine Corps reevaluate fighter pilot squadron requirements. DOD concurred with the recommendations.

gao_GAO-18-522

gao_GAO-18-522_0

Background SEC Examinations of FINRA The Office of Compliance Inspections and Examinations (OCIE) administers SEC’s nationwide examination and inspection program for registered SROs, broker-dealers, transfer agents, clearing agencies, investment companies, and investment advisers. OCIE conducts examinations and inspections to improve compliance, prevent fraud, monitor risk, and inform policy. Individual groups in OCIE have oversight responsibility for the various registered entities. The FINRA and Securities Industry Oversight (FSIO) program within OCIE performs examinations of FINRA and the Municipal Securities Rulemaking Board, an SRO that regulates the municipal bond market. As part of its FINRA oversight activities, FSIO conducts four types of reviews that may involve Section 964 areas. Program inspections are reviews of FINRA operations and program areas (for example, FINRA’s review of applications by broker-dealers seeking to become members). Oversight examinations are single, stand-alone examinations of specific examinations that FINRA conducts of its member firms. FSIO initiates an oversight examination when its examinations of a broker- dealer find deficiencies FSIO believes should have been identified by FINRA in its own examination of the broker-dealer. Thematic oversight examinations are a series of oversight examinations that evaluate FINRA’s review of a particular regulatory area across a number of its member firms. Tips, complaints, and referrals are allegations or statements of concern about possible violations of securities laws or risky conduct received by SEC. FSIO reviews FINRA-related tips, complaints, and referrals by evaluating facts and circumstances and conducting background research. The reviews may result in FINRA-related inspections or examinations or may be used for inspection planning purposes. To help identify the FINRA programs and topics that it will review, FSIO uses a risk-based approach that includes an annual assessment of high- risk areas and consideration of the areas specified in Section 964. According to SEC staff, FSIO also conducts ongoing monitoring of FINRA’s activities through reviews of information provided by FINRA and meetings with FINRA officials. Standards Useful for Assessing Examination Policies and Procedures Generally accepted government auditing standards define performance audits as those that provide findings or conclusions based on an evaluation of sufficient, appropriate evidence against criteria. Performance audit objectives can include assessments of program effectiveness, economy, and efficiency; internal control; compliance; and prospective analyses. SEC’s examinations of SROs share many of the attributes of performance audits, including their objectives. For example, examinations (including inspections) of FINRA enable FSIO staff to evaluate compliance with applicable laws and regulations; FINRA rules, regulations, or by-laws; or both. Although SEC is not required to follow the auditing standards when examining SROs, these standards and guidance provide a framework for conducting high-quality reviews that can serve as useful criteria in evaluating a regulatory agency’s examination or inspection programs. Areas of generally accepted government auditing standards relevant to SRO examinations include independence, competence, quality control and assurance, planning, supervision, evidence, documentation, and reporting: Independence refers to the audit organization and individual auditor’s need to be independent and include documentation proving independence. Competence refers to the extent to which audit staff collectively should possess adequate professional competence and technical knowledge, skills, and expertise. Quality control and assurance refers to a system of quality control that an organization should establish that is designed to provide the organization with reasonable assurance that its personnel comply with professional standards and legal requirements. Planning includes creating a written audit plan for each audit. Supervision requirements include sufficient guidance and direction to the staff assigned to the audit to address the audit objectives and follow applicable requirements, while staying informed about significant problems encountered, reviewing the work performed, and providing effective on-the-job training. Evidence refers to sufficient, appropriate evidence to provide a reasonable basis for the auditor’s findings and conclusions. Audit documentation requirements state that auditors must prepare documentation related to planning, conducting, and reporting for each audit. Finally, communication of the results entails auditors issuing audit reports. SEC’s FINRA Oversight Activities Covered Key Areas Identified in the Dodd- Frank Act Since fiscal year 2015, SEC examinations related to FINRA included reviews of all areas identified in Section 964. We determined that FSIO completed at least one examination covering each of the Section 964 areas since fiscal year 2015 (see table 1). In total, FSIO began or completed 61 examinations (program inspections, oversight examinations, and thematic oversight examinations) related to FINRA programs and operations in that period. Some examinations evaluated other aspects of FINRA’s programs and operations (those not specifically identified in Section 964), such as market surveillance and restitution for harmed investors. FSIO examinations either focused on a single Section 964 area or considered multiple areas. Some examinations focused specifically on a single Section 964 area. For example, in 2017 FSIO reviewed FINRA’s arbitration program, which provides retail investors a venue for resolving disputes with their brokers. Other examinations considered one or more of the areas as part of a broader scope. For instance, a program inspection completed in 2016 touched on FINRA’s arbitration services, cooperation with state securities regulators, transparency, and other topics. Another program inspection involved governance, policies on former employees, and other topics. FSIO examinations most frequently covered FINRA examinations (41 of 61). Nearly all of the oversight examinations reviewed at least some aspect of FINRA examinations. In two cases, the oversight examinations also covered another area—review of advertising by FINRA members. Selected Guidance Used for FINRA Governance Inspections Generally Was Consistent with Relevant Auditing Standards We found that OCIE policies and procedures used for examining FINRA since fiscal year 2015 generally were consistent with the requirements of generally accepted government auditing standards. SEC uses an examination manual to conduct its SRO examinations. We previously found that OCIE policies and procedures (including the prior version of the manual) generally were consistent with the requirements of the auditing standards that we determined were most relevant to assessing examination policies and procedures: independence, competence, quality control and assurance, planning, supervision, evidence, documentation, and reporting. We compared the current and prior versions of the examination manual. More specifically, we selected requirements for planning, prefieldwork scoping, and communicating findings from the current manual and compared those with similar sections in the prior version of the manual. We found that the new version includes the same material as the prior version while also incorporating additional guidance in certain areas. The planning section of the current version includes two additional requirements on the inclusion of non-National Examination Program staff. The communicating findings section of the current version included two additional requirements related to extensions of time to respond to disposition letters. Two of the four additional requirements were generally consistent with government auditing standards, and the remaining two additional requirements were minor adjustments that did not materially change the requirements. Therefore, we deemed the selected sections of the current version of the manual to also be consistent with the auditing standards. Inspections of FINRA Governance Were Consistent with SEC Internal Guidance OCIE (and from 2016, FSIO) program inspections of FINRA governance in fiscal years 2015–2017 were consistent with internal examination guidance. OCIE identified five inspections in that period that related to FINRA governance. Each of the inspections focused on one of the following areas: (1) code of conduct, (2) executive and employee compensation practices, (3) investment portfolio, (4) compliance resource provider program, and (5) the funding mechanism for its regulatory services agreement. FINRA’s code of conduct imposes restrictions on employees’ investments and requires financial disclosures that are uniquely related to its role as a securities regulator. The code also outlines FINRA’s ethical commitments and expectations and provides guidance on what employees must do to meet them. FINRA’s executive and employee compensation practices consist of salary and incentive compensation determined by FINRA’s Management Compensation Committee using operational, strategic, and financial factors, in addition to individual performance. FINRA’s investment portfolio is governed by a policy based on the degree of risk appropriate for FINRA assets, as applied by its board to its investment objectives. In the compliance resource provider program, FINRA worked with organizations to offer firms compliance-related products and services at a discounted price or with additional features. According to FINRA staff, this program was discontinued in May 2017 and replaced with FINRA’s Compliance Vendor Directory. FINRA’s regulatory service agreements are designed to provide market surveillance, financial surveillance, examinations, investigations, and disciplinary services to other entities, including the New York Stock Exchange LLC and the Chicago Board Options Exchange. For our review, we judgmentally selected the most relevant requirements from the section of the examination manual related to planning inspections and the most relevant requirements from the section of the manual related to communicating inspection findings to determine if OCIE conducted the inspections in accordance with its guidance. The planning section covers planning examinations and prefieldwork scoping and requires the examination team to discuss the results of background research and determine an appropriate scope for the examination as early as possible. The communicating findings section requires entities to be provided with timely and concise communications on the results. It also discusses how examination staff should take further actions for those findings that could involve notifications to other regulators. We reviewed relevant inspection-related documentation (including scope memorandums, disposition letters, emails, and information extracted from an examination database) and compared them against the selected requirements to determine if the guidance was followed. We tallied our results with a scorecard methodology (see fig. 1). We found that all five inspections we reviewed met all requirements applicable to that particular inspection. For example, across all the inspections, OCIE examiners held prefieldwork meetings and documented and received approval for the scope of the examinations. Additionally, all five inspections met the required 180-day completion deadline and closed the inspection with a disposition letter. In cases in which requirements were not applicable, the reasons generally were that a triggering event had not occurred and no further action was needed. For instance, the scope was not modified in any of the inspections, so the requirement for approval of such modifications did not apply. Furthermore, none of the inspections included non-National Examination Program staff (such as personnel from SEC’s Enforcement Division), and so requirements surrounding participation by those groups did not apply. Agency Comments We provided a draft of this report to SEC for their review and comment. In its comment letter, which is reprinted in appendix II, SEC concurred with our findings and appreciated our attention to the issues discussed in the report. SEC also provided technical comments on the draft report, which we incorporated as appropriate. We are sending copies of this report to the appropriate congressional committees, the Securities and Exchange Commission, the Financial Industry Regulatory Authority, Inc., and other interested parties. In addition, this report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-8678 or clementsm@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix III. Appendix I: Objectives, Scope, and Methodology This report (1) determines if the Securities and Exchange Commission’s (SEC) oversight of the Financial Industry Regulatory Authority, Inc.’s (FINRA) operations and programs since fiscal year 2015 included the 10 areas specified in Section 964 of the Dodd–Frank Wall Street Reform and Consumer Protection Act (Dodd-Frank Act), (2) evaluates the extent to which selected SEC internal guidance on conducting examinations of FINRA follows generally accepted government auditing standards, and (3) evaluates the extent to which examinations of FINRA’s governance practices followed SEC’s internal guidance. To assess whether oversight of FINRA by SEC’s Office of Compliance Inspections and Examinations (OCIE) included the Section 964 areas, we requested and reviewed documentation for all examinations since fiscal year 2015 (from October 2014 through April 2018) that OCIE staff identified as relating to Section 964 areas. We use the term “examination” to include program inspections, two types of oversight examinations, and oversight activities stemming from tips and referrals. The documentation included scope memorandums, deficiency letters, and closing letters to the file for OCIE examinations. We evaluated whether the documentation indicated that an examination’s scope and findings covered one or more Section 964 areas or included other areas related to FINRA oversight that were not specified in Section 964. To determine the extent to which OCIE’s internal guidance on conducting examinations of FINRA followed generally accepted government auditing standards, we compared SEC’s examination manual against generally accepted auditing standards. We reviewed selected sections of the current version of the manual and the earlier version. We judgmentally selected the two sections that most directly related to our focus on self- regulatory organization (SRO) inspections, which focused on preparing for examinations and communicating examination findings. Other areas of the examination manual that were not relevant focused on administration and organizational issues. We relied on our work that found that the earlier version of the manual followed the auditing standards and also interviewed pertinent staff within OCIE to discuss the guidance and why it did or did not include certain elements. We analyzed any differences between the versions to determine whether changes or additions in the current version of the manual also followed auditing standards. In addition, we interviewed FINRA staff to gain a general understanding of how OCIE staff work with them to conduct examinations. To determine the extent to which OCIE’s program inspections of FINRA’s governance in fiscal years 2015–2017 followed OCIE’s internal guidance, we used a scorecard methodology to compare inspections of FINRA’s governance with the examination manual and draft updates. We only reviewed the extent to which examinations followed specified guidelines and did not evaluate the analysis, findings, or disposition of the examinations. We created a checklist of relevant elements from the examination manual by judgmentally selecting 6 requirements from the planning inspections section of the manual and 11 requirements from the communicating findings section of the manual that were most applicable to our focus on the actual SRO inspection process. Other requirements that we deemed less relevant include examinations of exempt reporting advisers and the process for approving examination documents. The planning section of the manual covers planning examinations and prefieldwork scoping and requires the examination team to discuss the results of background research and determine an appropriate scope for the examination as early as possible. The communication of examination findings section requires entities to be provided with timely and concise communications on the results. It also discusses how examination staff should take further actions for those findings that could involve notifications to other regulators. We then reviewed different types of inspection-related documentation to determine whether the guidance was followed. For instance, we assessed certain inspection requirements, such as compliance with changing the scope of the inspection, based on formal written documentation such as scope memorandums and disposition letters. We assessed other requirements (such as whether prefieldwork team meetings were held) based on informal documentation, such as email appointments. We also relied on other internal documentation, which included the examination tracking database, which is used to certify compliance with a requirement to complete an inspection within 180 days from the completion of audit work. Two analysts then independently compared the elements against documentation for the five OCIE inspections to determine the extent to which the inspections documented the requirements outlined in the examination manual. Analysts assigned a rating of “yes” if the element was found in the inspection materials we reviewed, “no” if there was no mention of the element in the inspection materials we reviewed, “partially” if the element was not fully addressed in the inspection materials we reviewed, and “n/a” if the element was not applicable to the inspection. We also interviewed pertinent staff within OCIE to discuss the guidance and why it did or did not include certain elements. We conducted this performance audit from November 2017 to July 2018 in accordance with generally accepted government auditing standards. These standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Comments from the Securities and Exchange Commission Appendix III: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Karen Tremba (Assistant Director), Jon D. Menaster (Analyst in Charge), Kevin Averyt, Farrah Graham, Marc Molino, Akiko Ohnuma, Barbara Roesmann, and Jessica Sandler made key contributions to this report.

Why GAO Did This Study The securities industry is generally regulated by a combination of federal and industry oversight. FINRA, a self-regulatory organization, is responsible for regulating securities firms doing business with the public in the United States. SEC oversees FINRA's operations and programs. Section 964 of the Dodd-Frank Act includes a provision for GAO, following an initial report, to triennially review and report on aspects of SEC's oversight of FINRA. GAO issued its first report in May 2012 ( GAO-12-625 ) and its second report in April 2015 ( GAO-15-376 ). This report (1) determines if SEC's oversight of FINRA included the 10 areas specified in Section 964 of the Dodd-Frank Act and (2) evaluates the extent to which selected SEC internal guidance for examinations of FINRA follows generally accepted government auditing standards and the extent to which SEC's examinations of FINRA's governance practices followed SEC internal guidance. GAO reviewed all SEC examinations relating to a Section 964 area completed since fiscal year 2015 (including five that were governance-related), reviewed certain SEC procedures used to examine self-regulatory organizations against Government Auditing Standards , and compared completed inspections against SEC guidance. GAO also interviewed SEC and FINRA staff. What GAO Found Since fiscal year 2015, Securities and Exchange Commission (SEC) examinations of the Financial Industry Regulatory Authority, Inc. (FINRA) covered each of the 10 areas specified in Section 964 of the Dodd-Frank Wall Street Reform and Consumer Protection Act (Dodd-Frank Act), such as governance, funding, and transparency. The most commonly covered area was FINRA examinations of its members. Selected SEC guidance used to examine FINRA, including requirements for planning, prefieldwork scoping, and communicating findings, was consistent with generally accepted government auditing standards, and SEC inspections of FINRA were consistent with SEC's guidance. The five governance-related inspections of FINRA that GAO reviewed were consistent with SEC guidance for planning examinations and communicating findings (see fig.). Not all the requirements were applicable (because in certain instances completion of one requirement eliminated the need to satisfy others). What GAO Recommends GAO is not making any recommendations. SEC agreed with GAO's findings.

gao_GAO-18-408T

gao_GAO-18-408T_0

Background With a staff of over 47,000 members, the Coast Guard operates a multimission fleet of 201 fixed and rotary-wing aircraft and over 1,400 boats and ships. Operational control of surface and air assets is divided into two geographic Areas (Pacific and Atlantic), within which are nine Districts consisting of 37 sectors and the stations within them. The Coast Guard’s program oversight, policy development, and personnel administration are carried out at the Coast Guard’s headquarters. As shown in table 1, the Coast Guard is responsible for 11 statutory missions identified in the Homeland Security Act of 2002, as amended. The Coast Guard manages these missions through six mission programs, also listed in table 1. As part of its marine safety mission, for example, the Coast Guard conducts, among other activities, safety inspections and vessel accident investigations, including those involving commercial fishing vessels, which are part of an industry with one of the highest death rates of any industry in the United States. For each of its 11 missions, the Coast Guard has developed goals and targets to assess and communicate agency performance. The Coast Guard’s performance assessment process also includes identifying performance gaps and implementing corrective actions to address unmet performance goals. As part of its process, the Coast Guard is to establish targets for the current and subsequent 2 fiscal years, according to Coast Guard officials. Each target is set by the Coast Guard, but according to the Coast Guard’s Annual Performance Report (APR), some are derived from external factors. For example, DHS requires that Coast Guard set a 100 percent target for the percent of people in imminent danger saved in the maritime environment. Further, several of the Coast Guard’s assets used to conduct these missions are approaching the end of their intended service lives. As part of its efforts to modernize its assets used to carry out various missions, the Coast Guard has begun acquiring new vessels, such as the National Security Cutter, the Fast Response Cutter as well as other assets. However, concerns surrounding the affordability of this effort remain as the Coast Guard continues to pursue multiple new acquisitions without long-term planning to guide the affordability of its acquisition portfolio. Figure 1 shows the Coast Guard’s Fast Response Cutter and National Security Cutter. Coast Guard Actions that Could Improve Data Quality We previously reported on actions the Coast Guard could take to ensure that, among other things, it addresses limitations posed by incomplete data, the use of unrealistic asset performance data, and limitations with its performance goal data, for more effective program management. Examples of data limitations that we have recommended that the Coast Guard take action on are below. Improve completeness of mission data. In December 2017, we found that several different federal agencies play a role in overseeing and promoting commercial fishing vessel safety, including the Coast Guard. As part of its marine safety activities, the Coast Guard conducts, among other activities, safety inspections and vessel accident investigations. Commercial fishing has one of the highest death rates of any industry in the United States and vessel disasters are the leading cause of fatalities among fishers, according to the National Institute for Occupational Safety and Health. However, our December 2017 review found that more information is needed to calculate vessel safety statistics that could enhance the Coast Guard’s knowledge about accident, injury, and fatality trends involving commercial fishing vessels. The Coast Guard collects some data on commercial fishing vessels that operate in federal waters—including a vessel’s length and construction date—but data on the population of the active U.S. commercial fishing vessel fleet are not complete. Between 2006 and 2015, the Coast Guard investigated 2,101 commercial fishing vessel accidents that were identified as occurring in federal waters. While the number of accidents generally increased over this time period, the number of injuries and fatalities declined over the same 10-year period. However, we could not assess the number of accidents, injuries, and fatalities by fishery— meaning the area in which a certain type of fish (e.g., shrimp, salmon, crab) is caught—because the Coast Guard’s data is not complete. Further, we were unable to calculate the rates of commercial fishing vessel accidents, injuries, and fatalities, because reliable data on certain information needed to do so—including the total number of vessels that are actively fishing and the fishery or region in which the vessel operates—are either not maintained or are not collected by the Coast Guard or other federal agencies. Having this information could be useful to carrying out the Coast Guard’s marine safety mission, which includes enforcing laws to prevent death, injury and property loss in the marine environment. We recommended in our December 2017 report that the Coast Guard ensure that data it collects during commercial fishing vessel incident investigations is accurately captured. We also recommended that the Coast Guard work with stakeholders to form a working group to determine an efficient means to establish a reliable estimate of the population of active commercial fishing vessels. The Coast Guard agreed with both recommendations, and in February 2018 informed us that it is in the process of developing additional data fields to capture more information, such as the fishery in which the commercial fishing is involved, and is engaging stakeholders to establish an appropriate working group. We will continue to monitor these actions. Use more realistic asset performance data. In our May 2016 report on Coast Guard strategic planning, we found that the Coast Guard did not provide field units with realistic goals for allocating assets, by mission. We reported that the Coast Guard’s strategic allocations of assets were based on unrealistic assumptions about the performance capacity of its assets and did not reflect asset condition and unscheduled maintenance. This was due, in part, to the Coast Guard not including information from its field units on the actual performance of its assets. For example, agency officials noted that one of its classes of cutters was 50 years old and these cutters were hampered by mechanical failures requiring emergency dry dock repairs, which resulted in reduced availability to carry out their missions during the year. In another example, a field unit stated that based on historical use, it planned for 575 hours per vessel for one type of cutter instead of the 825 hours performance capacity. Because actual asset use has consistently fallen below asset performance capacities, there is not a direct alignment between the Coast Guard’s strategic operational goals and its prospects for achieving those goals. As a result, the headquarters’ strategic intent, which is based on asset capacity rather than actual performance, did not provide the field with strategic, realistic goals for allocating assets by mission. Agencies should use quality information that is appropriate, current, complete, accurate, accessible, and timely to achieve objectives and address related risks. We recommended that the Coast Guard incorporate field unit input, such as information on assets’ actual performance, to inform more realistic asset allocation decisions. The Coast Guard concurred with this recommendation, and in February 2018 informed us that it plans to address this recommendation through changes to two process documents that are under revision, with an expected completion date in March 2018. Improve performance goal data. In our October 2017 review of Coast Guard performance goals, we reported that the Coast Guard and DHS identified limitations with two of the seven selected performance goals we reviewed, including the five year average number of recreational boating deaths and injuries. In particular, officials believe that many recreational boating injuries that do not require hospitalization are not reported to the Coast Guard. These officials believe that the amount of underreporting may vary over time due to changes in industry trends, making it difficult to accurately determine actual injury rates and program performance. We determined that the data for this performance goal was not sufficiently reliable for the purposes of our reporting objectives due to these likely limitations. We found that the Coast Guard did not report the possible extent of these limitations with this performance goal in its fiscal year 2016 APR. For the other performance goal, the Coast Guard and DHS identified limitations with the number of detected incursions of foreign fishing vessels violating U.S. waters, which is publicly reported in DHS’s APR. DHS’s review of this performance goal, reported in August 2015, raised questions about the validity of this goal—that is, whether it provides a useful measure of the Coast Guard’s performance. Specifically, the review noted that this performance goal is intended to measure a deterrence effect, but doing so is inherently difficult and may lead to contradictory interpretations of performance. In October 2017, we found that the data for this performance goal was sufficiently reliable for our reporting objective purposes, but questions remain about its validity. Reliable data is not a useful indication of performance unless it is also a valid representation of the goal being addressed. DHS officials reported that they did not include a discussion of the limitations for this performance goal in DHS’s fiscal year 2015 APR because the performance goal met the minimum threshold for data reliability despite the goal’s limitations. Coast Guard officials reported they were aware of these limitations and were working with DHS and OMB to improve the performance goal and implement corrective actions within 1 to 2 years. We recommended that the Coast Guard assess the extent to which documentation on performance data reliability contains appropriate information on known data reliability limitations and update these documents, as needed, based on the results of the assessment. The Coast Guard concurred and in February 2018, informed us that it had taken initial actions to address our recommendation. However, our preliminary review of these actions indicates that further action will be needed to fully address our recommendation, such as documenting and reporting the limitations of performance data. Coast Guard Actions that Could Improve Transparency of Data for Reporting on Its Mission Performance and Capital Planning Our previous reports have identified areas in which the Coast Guard could improve the transparency of its data used for reporting on its mission performance and planning. Improve transparency of data on mission performance. In our October 2017 report on performance goals, we found that the Coast Guard’s APR has not been released publicly since 2011 due to a previous DHS leadership decision. Consequently, there has not been full visibility over performance across all of the Coast Guard’s missions. For example, one of the Coast Guard’s missions—defense readiness—has no goals that are publicly reported or shared with Congress, even though measures related to defense readiness are included in the Coast Guard’s APR. Coast Guard officials stated that they could see the benefit of publicly releasing their APR; however, DHS’s decision to limit the number of performance goals shared publicly has so far deterred the Coast Guard from pursuing the public release of its APR. DHS officials told us that the department is concerned about conflicting information that a component’s APR might present because it is vetted and produced separately from the DHS APR. However, the lack of transparency regarding performance data shared publicly and with Congress can result in an incomplete picture of mission performance and can limit effective oversight of Coast Guard operations. As a result, the public and Congress may be unable to determine the extent to which the Coast Guard is meeting its missions. We recommended that future Coast Guard APRs be available on the Coast Guard’s public website. The Coast Guard concurred with this recommendation and in February 2018, the Coast Guard informed us that it had completed its 2017 APR and are determining an appropriate approach for making it publicly available. Improve capital planning transparency. In our previously issued work on the Coast Guard’s annual 5-year capital investment plan (CIP), we found that the CIP does not consistently reflect current total cost estimates or the effects of tradeoffs that are made as part of the annual budget cycle. We made several recommendations in recent years intended to help the Coast Guard plan for future acquisitions and the difficult tradeoff decisions it will likely face. The Coast Guard generally concurred with these recommendations and is in various stages of implementation. For example, in 2017 we reported that we have made recommendations that DHS and the Coast Guard take several actions to gain an understanding of what the Coast Guard needs to meet its mission within its likely acquisition funding levels. These recommended actions included the Coast Guard: (1) conducting a comprehensive portfolio review across all its acquisitions to develop revised baselines that meet mission needs and reflect realistic funding scenarios and (2) developing a 20-year plan that identifies all necessary recapitalization efforts and any fiscal resources likely necessary to complete these efforts. For example, in 2014 we recommended the Coast Guard develop a 20-year fleet modernization plan that identifies all acquisitions needed to maintain the current level of service and the fiscal resources needed to acquire them. Without these efforts, the Coast Guard will continue, as it has in recent years, to plan its future acquisitions through the annual budgeting process, an approach that has led to delayed and reduced capabilities. In 2016, the Coast Guard revised its 2005 Mission Needs Statement, which provides a basic foundation for long-term investment planning that is to serve as the basis for evaluating the effectiveness of various fleet mixes, and inform the Coast Guard’s CIP. However, the 2016 Mission Needs Statement did not identify specific assets the Coast Guard needs to achieve its missions, nor did it update the annual hours it needs from each asset class to satisfactorily complete its missions. The Coast Guard has stated it is developing a 20-year Long-term Major Acquisition Plan, but it has not stated when the plan will be completed or what will be included in this plan, such as potential trade-offs that could be made across the Coast Guard’s portfolio of acquisitions to better meet mission needs within realistic funding levels. A long-term plan with a tradeoff analysis would facilitate a full understanding of the affordability challenges facing the Coast Guard while it builds the Offshore Patrol Cutter. DHS concurred with our 2014 recommendation, but it is unclear when the Coast Guard plans to complete the 20-year plan. Chairman Hunter, Ranking Member Garamendi, and members of the sub- committee, this completes my prepared statement. I would be happy to respond to any questions you may have at this time. GAO Contacts and Staff Acknowledgments If you or your staff members have any questions about this testimony, please contact Nathan Anderson at (202) 512-3841 or andersonn@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this statement. Other individuals making key contributions to this work, and the underlying reports on which it is based, include Dawn Hoff (Assistant Director); Andrew Curry (Analyst-in-Charge); Chuck Bausell; David Bieler; Richard Cederholm; John Crawford; Timothy J. DiNapoli; Michele Fejfar; Laurier R. Fish; Peter Haderlein; Eric Hauswirth; Laura Jezewski; Tracey King; Benjamin Licht; Marie A. Mak; Gary Malavenda; Diana Moldafsky; Heidi Nielson; Meg Ullengren; and Kayli Westling. This is a work of the U.S. government and is not subject to copyright protection in the United States. The published product may be reproduced and distributed in its entirety without further permission from GAO. However, because this work may contain copyrighted images or other material, permission from the copyright holder may be necessary if you wish to reproduce this material separately.

Why GAO Did This Study The Coast Guard, a component of DHS, serves as the principal federal agency responsible for maritime safety, security, and environmental stewardship in U.S. ports and waterways. To ensure that the Coast Guard is effectively fulfilling its missions, agency managers must have accurate information and base decisions on sound analyses for effective program management. This statement discusses Coast Guard actions needed to (1) improve the quality of data used for program management and (2) improve the transparency of its data for reporting on mission performance and planning. This statement is based on relevant products GAO issued from June 2014 through December 2017 on Coast Guard strategic planning and management issues, as well as related recommendation follow-up conducted through February 2018. GAO reviewed applicable laws, regulations, policies and guidance. GAO also interviewed Coast Guard officials responsible for administering these programs and obtained information on how they used data to inform decisionmaking. GAO interviewed a range of stakeholders, including federal and industry officials. What GAO Found GAO's prior work recommended multiple actions to improve the Coast Guard's program management by improving the quality of data it uses to manage and report on its mission performance. Specifically, GAO recommended actions such as collecting more complete data and clarifying the data limitations to facilitate more effective program management. For example, in December 2017, GAO found that more information is needed to calculate vessel safety statistics that could enhance the Coast Guard's knowledge about accident, injury, and fatality trends involving commercial fishing vessels. Having more complete information could be useful to carrying out its marine safety mission, and GAO recommended, among other things, that the Coast Guard ensure that data collected during commercial fishing vessel incident investigations is accurately captured. In 2018, the Coast Guard reported taking initial steps to capture more accurate data. GAO's prior work also identified areas where the Coast Guard could improve the transparency of the data it uses for reporting on its mission performance as well its capital planning purposes. For example, in an October 2017 report on performance goals, GAO found the Coast Guard's Annual Performance Report (APR) has not been released publicly since 2011. Consequently, there has not been full visibility over performance across all of the Coast Guard's missions. Coast Guard officials stated that a decision by Department of Homeland Security (DHS) leadership to limit the number of performance goals shared publicly had deterred the Coast Guard from public release of its APR. GAO recommended that APRs be available on the Coast Guard's website; the Coast Guard plans to publicly release future APRs. In addition, previous GAO reports found that the Coast Guard's annual 5-year capital investment plan, which projects acquisition funding needs for the upcoming 5 years, did not consistently reflect current total cost estimates or the effects of tradeoffs made as part of the annual budget cycle. GAO made recommendations to help the Coast Guard plan for future acquisitions and the difficult trade off decisions it will face given funding constraints. The Coast Guard agreed, but it is unclear when it will complete the 20-year plan. What GAO Recommends GAO is not making new recommendations in this statement but has made them to the Coast Guard and DHS in the past on improving its program management through, among other things, better quality and more transparent data. DHS and the Coast Guard agreed with these recommendations and reported actions or plans to address them.

gao_GAO-19-175

gao_GAO-19-175_0

Background Missions, Roles, and Responsibilities of Federal Agencies in Our Review Federal agencies carry out a variety of missions, including protecting and defending government buildings, public lands, and natural resources, as well as federal employees, elected officials, and visitors to federal sites. Agencies with FLEOs are also charged with investigating civil and criminal violations of federal laws. Inspectors General, which may also have FLEOs, are independent and objective units within agencies that are charged with combatting waste, fraud, and abuse within the programs and operations of their agencies. Table 2 lists the agencies within our review (20 agencies included in our review of spending data and 5 agency components included in our review of inventory controls) and describes their law enforcement missions. For more information about the data each agency provided, see appendix II. Types of Firearms, Ammunition, and Selected Tactical Equipment For the purposes of our review, a firearm is any weapon that is designed to expel a projectile by the action of an explosive or that may be readily converted to do so. Some firearms are single-shot, while others may be semi-automatic (requires a separate pull of the trigger to fire each cartridge) or fully automatic (can shoot automatically more than one shot, without manual reloading, by a single function of the trigger). Ammunition includes its component parts, such as cartridge cases, primers, bullets, or propellant powder designed to be used in a firearm. Ammunition can be used in multiple types of firearms, based on the size. For example, 9mm caliber ammunition used in pistols can also be used in certain types of fully automatic firearms. See figure 1 below for more information about the types of firearms FLEOs may use. In addition to firearms and ammunition, federal agencies may also have a variety of tactical equipment available to their officers to support their law enforcement roles. For example, officers engaged in counterdrug activities may use armored vehicles for drug raids in rural areas or night- vision equipment to maintain surveillance of drug activities. Officers that work in counterterrorism and border security may use helicopters or other aircraft, as well as armored or tactical vehicles, to patrol or surveil locations. See figure 2 for examples of selected tactical equipment in our review. Purchase Data Reported in FPDS-NG and USASpending.gov Federal Procurement Data System-Next Generation (FPDS-NG) is a comprehensive web-based tool for agencies to publicly report contract transactions, including firearms, ammunition, and tactical equipment purchases. The public can download FPDS-NG data on contract actions from the USASpending.gov website, and this data set enables users to examine spending in multiple categories across government agencies. The contracting officer, who awards a contract or order against an existing contract, has responsibility for accurately recording the individual contract action information in FPDS-NG. Agencies are responsible for developing a process for recording contract actions and monitoring results to ensure their timely and accurate reporting in FPDS-NG, and must submit certifications about the accuracy of contract reporting to the General Services Administration. The Federal Acquisition Regulation (FAR) and the FPDS-NG Government User’s Manual require that each transaction record include the name of the funding agency—the agency that provided the obligated funds for the transaction. The FPDS-NG Government User’s Manual also requires a product or service code (PSC) that reflects the product or service procured. If more than one PSC applies, the PSC that represents the predominance of the dollars obligated should be selected. Generally, the FAR requires that agencies report contract actions with a total estimated value greater than $3,500 to FPDS-NG. Generally, contract actions that do not meet the $3,500 threshold may also be reported, but the FAR does not require agencies to do so. Available Data Show that Selected Agencies Reported Spending at Least $1.5 Billion on Firearms, Ammunition, and Tactical Equipment from Fiscal Years 2010 through 2017 The 20 federal agencies in our review reported data from their internal record-keeping systems on the amount they spent on firearms, ammunition, and selected tactical equipment. These agencies reported spending at least $38.8 million on firearms, $325.9 million on ammunition, and $1.14 billion on tactical equipment—at least $1.5 billion in total—from fiscal years 2010 through 2017. For detailed information about the data each agency provided, see appendix II. Firearms Spending The 20 agencies in our review reported spending a total of at least $38.8 million on firearms for their FLEOs from fiscal years 2010 through 2017, based on available spending data they provided from their internal record- keeping systems. The amount each agency reported spending on firearms over the 8-year period ranged from $106,000, in the case of the Social Security Administration’s Office of the Inspector General (SSA OIG) to $4 million at U.S. Customs and Border Protection (CBP). Of the 20 agencies in our review, 18 agencies also reported the number of firearms they bought. These agencies reported buying a total of at least 44,551 firearms during this time. The quantity of firearms each of these 18 agencies reported buying over the 8-year period ranged from at least 311 at SSA OIG to at least 8,500 at the Bureau of Alcohol, Tobacco, Firearms, and Explosives (ATF). Agencies buy a variety of firearms in support of their law enforcement missions. From fiscal years 2010 through 2017, agencies reported buying pistols, rifles and shotguns, and three of the agencies—NPPD, ICE, and U.S. Secret Service—also reported buying revolvers. Seventeen agencies reported buying semi-automatic firearms, while eight agencies reported buying fully automatic firearms and ten agencies reported buying single-shot firearms. See figure 3 for more information about the types of firearms that agencies reported buying over the 8-year period. Agency officials told us there were several reasons why they buy firearms, such as to update their entire firearms inventory, to replace malfunctioning firearms, or to test new models of firearms. Agencies typically do not update their firearms inventory often because firearms can last many years when properly serviced and maintained. This is reflected in agencies’ spending data, which generally show periodic larger orders of firearms and more frequent smaller orders. For example, ATF reported buying several thousand pistols in both fiscal years 2012 and 2017, and fewer than 1,000 pistols and rifles in total in the intervening years. Similarly, BIA reported buying several hundred firearms in 2010, 2012, and 2014, and fewer than 200 in total in the remaining 5 years of our review. When firearms near the end of their useful life, agencies can choose to retire or replace them. Additionally, agencies frequently reported buying three or fewer firearms at a time, and officials from one agency we spoke with said that they may buy a single firearm at a time in order to test out new models for future consideration. Ammunition Spending The 20 agencies in our review reported spending a total of at least $325.9 million on ammunition for their FLEOs from fiscal years 2010 through 2017, based on available spending data from agencies’ internal record- keeping systems. The amount each agency reported spending on ammunition over the 8-year period ranged from $309,000 (SSA OIG) to $128 million (CBP). Of the 20 agencies in our review, 16 agencies also reported the number of rounds of ammunition they bought. The 16 agencies reported buying a total of at least 767 million rounds of ammunition during this time. The number of rounds of ammunition each of these agencies reported buying over the 8-year period ranged from at least 846,000 rounds (SSA OIG) to at least 429 million rounds (CBP). These agencies most frequently reported buying handgun ammunition, and .40 caliber was the most frequently reported caliber. See figure 4 for more information about the types of ammunition that agencies reported buying over the 8-year period. Agency officials we spoke with said the quantity of ammunition they buy annually varies within each agency based on factors such as ammunition usage in previous years, the number of officers qualifying to use a firearm each year, the skill level of officers, the type of training conducted, and their budget each fiscal year. Specifically, agencies require officers to pass certain firearms qualifications standards in order to maintain their proficiency—typically quarterly for pistols, and biannually for rifles and shotguns. Officers must qualify on each firearm they are authorized to carry, and some agencies may have additional training requirements throughout the year. For example, HHS Office of Inspector General (OIG) officials said that, in addition to quarterly qualifications, officers also complete eight additional training modules each year that cover topics that include responding to multiple assailants, use of cover, and reactive shooting techniques. HHS OIG officials noted that they may add additional training if needed, and officials take this into account when determining the type and amount of ammunition they order each year. Tactical Equipment Spending Of the 20 agencies in our review, 17 provided spending data for their tactical equipment. The 17 agencies reported spending a total of at least $1.14 billion on tactical equipment for their FLEOs from fiscal years 2010 through 2017, based on available spending data from agencies’ internal record-keeping systems. The amount each agency reported spending on tactical equipment over the 8-year period ranged from $10,000 (SSA OIG) to $609 million (CBP). We cannot report the total quantities of tactical equipment agencies bought because agencies reported quantity data using different units of measurement. For example, when we requested data on the number of camouflage uniforms agencies bought, agencies used “1” to refer to a variety of clothing, such as a single pair of pants or a full set of uniforms. A few agencies accounted for a significant portion of the total reported spending on tactical equipment. Specifically, four agencies—CBP, U.S. Marshals Service, Federal Bureau of Investigation (FBI), and Drug Enforcement Administration—reported spending at least $755 million in the manned aircraft category, or 66 percent of all reported tactical equipment spending for all agencies. See figure 5 for the types of tactical equipment agencies reported in spending data. The 17 agencies that reported buying tactical equipment most frequently reported buying aiming devices, such as sights and scopes, and specialized image enhancement devices, such as thermal cameras or night-vision goggles. Few agencies reported buying tactical and weaponized vehicles, aircrafts, and vessels. See figure 6 for more information about the types of equipment reported in each agency’s spending data. BIA and Forest Service Publicly- Available Data Differed Somewhat from their Internal Data, While ICE’s Data Differed Substantially For the three agencies we reviewed—BIA, Forest Service, and ICE— publicly-available purchase data from USASpending.gov on firearms and ammunition did not consistently match the internal agency data we reviewed. Table 3 shows the total dollar value of the firearms and ammunition obligations that each agency reported to us, alongside the dollar value of the obligations in the publicly-available data. Differences between the agency-reported values and the values shown in the publicly-available data ranged from less than 1 percent to approximately 700 percent of the values reported by the agencies. Of the three agencies that we reviewed, ICE had the largest discrepancies between the agency-reported and publicly-available values. ICE reported to us $2,539,585 in firearms obligations and $47,965,399 in ammunition obligations for fiscal years 2010 through 2017; however, the publicly- available data for ICE for the same time period shows $19,728,786 in firearms obligations—about eight times greater than what ICE reported to us—and $146,198,549 in ammunition obligations—about three times the amount that ICE reported to us. According to our analysis, some of the difference between the ICE- provided and publicly-available obligations in USASpending.gov results from other DHS agencies using ICE contracts to make firearms and ammunition purchases, and ICE not properly identifying the funding agency for those purchases in the Federal Procurement Data System- Next Generation (FPDS-NG), the database from which USASpending.gov draws contracting data. In these cases, agency officials told us that under a process known as “strategic sourcing,” ICE performs the procurement functions and is reimbursed by the purchasing agency. The Federal Acquisition Regulation (FAR) requires that the agency that “provided the predominant amount of funding for the contract action” be recorded in FPDS-NG. The FPDS-NG Government User’s Manual also specifies that users record the agency that “provided the obligated funds”—that is, the agency that purchased the item or service. However, when ICE records the transaction data in FPDS-NG, ICE lists itself as the funding agency for firearms and ammunition transactions. For example, in the publicly-available records, a fiscal year 2013 purchase of pistols from the manufacturer totaling $847,960 in obligations shows ICE as the funding agency, but the transaction description states: “to purchase pistols for FPS .” As a result, ICE appears to be the funding agency for more firearm and ammunition transactions in the publicly-available data than in the data ICE reported to us. ICE officials explained that their contracting officers manually enter ICE as the funding agency in FPDS-NG. They interpret the FPDS-NG Government User’s Manual guidance to allow designation of ICE as the funding agency, since payment for the purchase is made from an ICE account even when those funds are reimbursed by the agency that actually receives the purchased product. However, the FPDS-NG Government User’s Manual specifically requires the identification of the funding agency and distinguishes between the agency that makes the payment and the agency that ultimately provides the funds for the purchase. FPDS-NG guidance also clarifies that when one agency buys on behalf of another, the agency that is requiring the purchase should be recorded as the funding agency, not the payment office. Because ICE recorded other agencies’ purchases as its own in the publicly-available data, it significantly inflated the apparent dollar value of its firearms and ammunition purchases. As a result of ICE not accurately recording the correct funding agency information in FPDS-NG, the public does not have accurate information on the value of firearms and ammunition purchases made by ICE, and the agencies that make purchases using ICE contracting services. As we have previously reported, data need to be presented in a way that meets the needs of the end users—both policy makers and the public—if USASpending.gov is to fulfill its purpose of increasing accountability and transparency in federal spending. Improving the accuracy of the reported funding agency can better help the public understand and use federal data, and increase accountability and transparency of these sensitive purchases. Data on obligations for the publicly-available purchase records had smaller discrepancies when compared with purchases of firearms and ammunition reported to us by the BIA and Forest Service. The BIA’s ammunition obligations in the agency- and publicly-reported data sets closely matched, with a discrepancy of less than 1 percent overall from fiscal years 2010 through 2017. The Forest Service’s firearms obligation amounts in the agency- and publicly-reported data sets also closely matched, with a discrepancy of less than 1 percent overall from fiscal years 2010 through 2017. However, the dollar value of the Forest Service’s ammunition obligations recorded in the publicly-available data was approximately 25 percent greater than the value of the ammunition purchases in the data that they provided to us. Forest Service officials explained that the publicly-reported data includes ammunition purchased for non-law enforcement purposes—such as protecting Forest Service employees from wildlife attacks, controlling invasive species, or euthanizing injured animals—and those purchases were excluded from the data that they provided to us because the scope of this review focuses on purchases for FLEOs. The type and amount of information recorded in the publicly-available data also contribute to discrepancies between agency-provided and publicly-available purchase records. As a result, the publicly-available data may comply with the data reporting requirements enumerated in the FAR and in the FPDS-NG Government User’s Manual and still differ from the agency-provided data. Differences between publicly-available data and agency purchase records include: The product or service code (PSC) selected in the publicly-available data may not reflect all of the items in a purchase. According to the FPDS-NG Government User’s Manual, the PSC selected for a purchase should reflect the items that constitute “the predominance of the dollars obligated,” and only one PSC may be associated with a purchase. Therefore, when a purchase is assigned a firearms or ammunition PSC, but the purchase includes non-firearms or non- ammunition items as well, the total obligated amount of the purchase will be associated only with the selected PSC in the publicly-available data. This may result in over- or under-reporting the value of the obligations for firearms or ammunition purchases. For example, in fiscal year 2014, BIA purchased 103 shotguns, 220 tactical lights, and other equipment in a single transaction. The total value of the obligation was $145,970, of which the shotguns constituted 50.4 percent of the purchase ($73,549). Consistent with FPDS-NG guidance, the entire purchase was categorized as “Guns, through 30mm” in the publicly-available data, even though almost half of the purchase was for non-firearms items, thereby over-reporting the obligated value of the firearms purchased—in this case, effectively doubling the apparent obligated value of the firearms purchased while omitting the obligated value of the non-firearms items that were part of the purchase. Conversely, a purchase categorized as “Optical Sighting and Ranging Equipment” obligated for $2,971 included a line item for $500 of ammunition. BIA included the $500 ammunition purchase in the data provided to us, but that amount was included under the equipment PSC in the publicly-available data in keeping with FPDS-NG guidance. This excluded the purchase from the publicly-available data that we reviewed and under-reported the obligated value of ammunition purchases by BIA. The available PSCs in the publicly-available data do not distinguish between firearm parts and fully functional firearms. Several purchases associated with firearms PSCs included descriptive information indicating that the purchase was for firearms parts. ICE officials also confirmed that an order described as firearm “parts” could include fully functional firearms. The officials noted that whether a particular purchase included fully functional firearms, firearms parts, or both cannot be determined without the statement of work, and the statement of work is not part of the publicly-available data. By including purchases of both fully functional firearms and firearm parts in the same category, publicly-available data may inflate the obligated value of functional firearms purchases. Firearms and ammunition purchases may not be assigned a related PSC in the publicly-available data. In cases where agencies assigned a non-firearms or ammunition PSC to a firearms or ammunition purchase, those purchases were excluded from the publicly-reported data that we analyzed. For example, a Forest Service purchase of rifles and sights which obligated $50,799 was assigned a PSC for “Assemblies Interchangeable Between Weapons In Two or More Classes” in the public data, rather than a firearms- or ammunition- specific PSC. Another Forest Service purchase for rifles which obligated $23,457 was assigned a PSC for “R&D-Defense System: Weapons (Basic Research).” Agencies are not required to report purchases of $3,500 or less to FPDS-NG. Because purchases of $3,500 or less generally are not required to be reported to FPDS-NG, these purchases may be reported inconsistently or not at all in the publicly-available data. Ammunition is often purchased by the selected agencies in small quantities and may cost $3,500 or less. For example, Forest Service officials noted that such small ammunition purchases may be made using a purchase card, and their internal data included at least 130 such purchases. In addition, the publicly-available data do not include a field for agencies to report quantity information associated with purchases. Therefore, the number of firearms or rounds of ammunition that an agency purchased are not available in the publicly-available purchase data. HHS, EPA, and IRS Reported Varying Levels of Inventories of Firearms, Corresponding Ammunition and Limited Tactical Equipment HHS, EPA, and IRS law enforcement components, our case studies, in total reported inventories of five types of firearms—all with corresponding types of ammunition—and nine types of tactical equipment. According to officials in all components, their inventories of these items can be attributed to a variety of factors, including the missions and responsibilities of their FLEOs, the number of FLEOs, and the office’s schedule for disposing of and acquiring inventory. Table 4 summarizes the types and quantities of firearms, ammunition, and tactical equipment reported at case study components as of November 2017. For additional information on case study components, see appendix II. Firearms. The numbers and types of firearms the components in our review reported having in their inventories varied. As of November 2017, all components reported inventories of pistols and shotguns, five components reported rifles, and three reported fully automatic firearms. Officials noted that they make decisions about what to have in their inventories based on factors such as their number of FLEOs and mission needs. All components issued pistols to FLEOs to carry, in accordance with their statutory authority. Officials stated that these firearms are to be carried on duty so FLEOs are prepared for potentially dangerous circumstances, such as serving warrants on armed individuals. Similarly, qualified FLEOs in all components can temporarily carry rifles or shotguns in anticipation of, or in response to, high-risk situations, such as active shooter threats or arrests of suspects who are believed to be dangerous. We found that components had more pistols per FLEO than shotguns or rifles per FLEO, which reflects components’ preferences to issue pistols to officers as their duty weapons. For example, six components reported having roughly a 2 to1 or 3 to 1 pistol-to-officer ratio, while EPA OIG agents had a 5 to1 ratio. Case study components generally reported having more pistols than FLEOs because every FLEO is assigned at least one pistol. On the other hand, six components reported having about 1 shotgun or fewer per every two FLEOs. However, EPA Office of Enforcement and Compliance Assurance (OECA) had a 1.4 to 1 shotgun- to-FLEO ratio. According to EPA OECA officials, their shotgun-to-agent ratio is higher than other agencies because of two factors: 1) EPA OECA historically had more agents, which made their shotgun-to-agent ratios lower than when they acquired the shotguns in use today, and 2) EPA OECA sends additional unassigned shotguns to natural disaster response locations to pre-position them for EPA OECA agent use. Components in the Food and Drug Administration (FDA) and IRS reported keeping a relatively smaller number of shotguns, which they said they only deploy for high-risk investigations. Among the five components with rifles, rifle-to- FLEO ratios ranged widely—from less than 1 to 10 to 6 to 10 —due to differences in the number of FLEOs and mission needs. For example, officials with IRS Police, which had four rifles for nine officers, stated that they only use the rifles for continuity of operations drills and annual qualifications. Ammunition. As of November 2017, case study components reported inventories of ammunition ranging from 14,706, in the case of the IRS Police, to approximately 5 million rounds held by IRS CI. (See table 4 for all components’ inventories of ammunition.) Each law enforcement agency independently decides how much ammunition to allocate to its firearm-carrying personnel for training and qualification. Component officials noted that ammunition inventories constantly fluctuate throughout the year, based on factors such as the amount used for qualification and training purposes, and the timing of ammunition shipments. Officials from all components stated that their ammunition inventories can quickly change by thousands of rounds depending on training and qualification timing. For example, according to NIH officials, between November 2017 and February 2018, officers used 5,110 rounds of rifle ammunition during training, and subsequently NIH acquired 14,400 rounds of rifle ammunition. To help ensure they have sufficient ammunition on hand to support the training and operational needs of their FLEOs, components may maintain inventories of ammunition to last for several months. The length of time between ordering and receiving ammunition orders can be lengthy, sometimes up to 1 year, according to officials. Therefore, components order ammunition in large quantities to ensure there is enough available for training and qualification purposes. Tactical Equipment. Six case study components reported inventories of 9 of the 18 types of tactical equipment reviewed. Breaching equipment and aiming devices were the most frequently reported kinds of tactical equipment at these six components: four components reported breaching equipment and four reported aiming devices in inventories. Three components reported other equipment. EPA OIG reported inventories of silencers and tactical lighting; however, in February 2018 EPA OIG officials told us they began to transfer their silencers to another federal agency because officials decided that they were no longer necessary to meet their mission. Among the agencies in our review, NIH reported having pyrotechnics and large-caliber launchers. NIH officials said these items were necessary for assisting other law enforcement agencies in the event that extreme circumstances, such as a terrorist attack or riot, occurred in the area. Case Study Components’ Inventory Controls Vary and Selected Offices Generally Followed Their Own Procedures HHS, IRS, and EPA case study components have inventory controls in place for firearms, ammunition, and tactical equipment. Components generally followed their procedures at selected locations to track and secure FLEOs’ firearms. In addition, all 12 case study components’ offices that we visited were in compliance with their ammunition and tactical equipment inventory controls. All components had some controls governing ammunition and equipment, though the specific controls varied by component. Case Study Components Have Controls to Track and Secure Firearms HHS, IRS, and EPA case study components have controls in place for tracking, verifying, and securing FLEOs’ firearms. Through our observations, we found these components are generally following their inventory and security procedures at selected locations. According to Standards for Internal Control in the Federal Government, agencies should design control activities to respond to risks related to vulnerable assets. To ensure these controls operate effectively, management can take steps such as periodically counting and comparing such assets to control records, and establishing physical control to secure and safeguard vulnerable assets. Other examples of these controls include security for and limited access to assets, such as equipment that might be vulnerable to risk of loss or unauthorized use. In addition, we have identified areas that have been consistently recognized as important for effective inventory management that align with these controls. These areas include recording and tracking firearms inventory data and maintaining, controlling, and accounting for firearms inventories, among other things. Firearms Tracking. At each of these components, firearms are considered sensitive items and are tracked in electronic data systems or using paper records. At HHS OIG, there is a separate firearms tracking system, and other components track firearms in their overall property management system. A barcode or serial number is used to track the firearm through the life-cycle of the weapon—which includes initial assignment, changes in assignment (to a different agent or to storage), and weapon disposal. According to each component’s policy, every firearm has a bar code or serial number associated with it, and each firearm is assigned to an agent or placed in storage. When an agency receives a firearm, the serial number or bar code is entered into the agency’s firearms inventory system and upon assignment to an agent updated with the agent’s name. Typically, firearms in storage are assigned to the primary firearms instructor or the firearms coordinator in the inventory system to ensure accountability. At all case study components the primary firearms instructors and firearms coordinators are the persons responsible for managing the firearms inventory of an office and ensuring firearms are properly tracked—these are considered ancillary duties for these individuals, in addition to their regular responsibilities as FLEOs. We observed demonstrations or documentation of these tracking records at each office we visited and found them to be generally in accordance with office policies. Firearms Verification. Each component has a process whereby at least once a year officials conduct a firearms inventory to ensure that issued and stored firearms match with records in the office’s inventory tracking system. Figure 7 describes a general process that all components we reviewed follow to verify their firearms inventory. Five components we examined require this annual firearms inventory process to be conducted in person and entail the visual inspection of the serial number and condition of the firearm. However, FDA and EPA OECA FLEOs are permitted to send an email containing their firearm serial number, a photo of the firearm, or both to the official conducting the inventory to virtually verify they are in possession of their firearm. Officials stated that this saved them the expense of FLEOs in remote locations traveling for annual inventory checks. In addition to regularly verifying inventories of their firearms, five case study components conduct periodic checks to verify the accuracy of firearm inventory data. During these checks, headquarters or other inspection officials review data recorded about inventories, storage controls, and proper maintenance of weapons. For example, IRS CI and HHS OIG conduct regular internal reviews of firearms inventories and records that they use to identify data errors and make recommendations to improve data quality. We found that recent checks at the case study components have rarely identified issues related to recorded firearms data, and the majority of the identified issues were related to minor data errors, such as incorrectly recorded assignments of guns to FLEOs, locations of guns, and serial numbers. For example, in May 2017 EPA OECA’s check of one area office found two firearms listed as being in storage were actually in service. All components with these administrative errors corrected them as they became aware of them, according to the audit reports and agency officials. However, during our review of EPA OIG inventory control practices for firearms, we found that EPA OIG headquarters did not have a management review process in place for firearms inventory, which contributed to examples of inaccurate firearms inventory data. Specifically, we found 6 out of the 12 EPA OIG offices reported inventories with at least one firearm that did not match the location and individual to which it was assigned. For example, we found that the Special Agent in Charge at one EPA OIG’s field office verified that 10 firearms were physically in the office when actually they had been shipped to headquarters 11 months prior to the inventory date. According to EPA OIG officials, the errors we found were largely due to EPA OIG headquarters not updating records when agents transferred firearms from one office to another. As a result of our review, headquarters is implementing practices to improve future data quality. Specifically, EPA OIG officials said they reconciled their inventory data and had field offices with inaccurate inventories prepare memos to reflect their actual inventories. EPA OIG also updated firearms procedures to include headquarters increased oversight of firearms audits and inspections. In practice, EPA OIG officials stated that this will entail an annual reconciling of inventory memos sent by field offices with inventory data records maintained at EPA OIG headquarters. Thus, EPA OIG can more reliably track the location and agents responsible for firearms, ensuring proper weapons control procedures and accountability. Firearms Storage. All case study components had policies in place that required secure storage of firearms. All seven components’ policies specify that this must include a locked container, such as a file cabinet or desk safe for issued firearms or a firearm vault or safe for unissued firearms. In all 12 of the components’ offices we visited, we observed unissued firearms stored in a designated room that was kept locked with limited access. Specifically, we observed firearms storage rooms secured by keycard access, alarms, cameras, and other security devices. At all of the components’ offices we visited, only FLEOs with a need to enter the rooms had authorization to do so. For example, only FLEOs in management roles and firearms instructors had access to the firearms storage room at the IRS CI and HHS OIG offices we visited, according to officials at each. In the secure firearms storage rooms at all of the locations we visited, we observed firearms that were further protected by cabinets, cages, safes, vaults, or combinations of these devices. We also observed some instances of firearms being stored securely in safes outside of the designated firearms storage room. For example, two EPA OIG and OECA offices we visited had spare shotguns and related equipment in biometric safes in offices to be used in response to active shooter threats. Qualified FLEOs in these two offices can use their fingerprints to open the safes and respond to such threats. Based on our observations, the offices we visited for the six components were in compliance with their firearms storage policies. However, we found the IRS Police office in Martinsburg to be out of compliance with policy that limits access to stored firearms to the armorer and chief of security and its policy that restricts storage in locked cabinets to five firearms. Based on our finding, IRS Police officials said they will update their policy to reflect its current practice of allowing all IRS Police officers access to stored firearms to respond to active shooter threats and acquired an additional safe to store firearms in June 2018. For each of the components we examined in our review, there were no instances of firearms being lost or stolen from any office, according to component data and officials. Case Study Components Vary in Their Controls Over Ammunition and Tactical Equipment All 12 offices for case study components that we visited were in compliance with their ammunition and tactical equipment inventory controls and procedures. All components had policies to treat ammunition and equipment as items with some level of control, though the specific controls varied by component. Ammunition. At the components’ offices we visited, we observed various methods for tracking ammunition inventories, such as electronic logs, physical logs, or visual inspection. In general, case study components’ policies require ammunition tracking through an ammunition log updated as ammunition is used, a regular inventory, or both. Components track ammunition to ensure they have enough on hand for training and qualifications, according to component officials. Some components, such as EPA OIG, maintain tight controls over their ammunition use by tracking ammunition use by lot number and the number of bullets used by each gun in physical logs. EPA OIG officials stated that they used this level of precision for three reasons: to track how many bullets are fired through each firearm for maintenance purposes, to quickly identify ammunition in inventory that may be affected by recalls, and to have a high degree of accountability. HHS OIG tracks ammunition using an electronic log, which they said they update whenever ammunition is removed for training and qualifications. IRS CI currently limits access to ammunition and entrusts use of force coordinators with independently managing their ammunition inventory to meet the training needs of their area of responsibility. One IRS CI use of force coordinator said that she relies on her experience and judgment to keep track of ammunition use at her field office without formally recording it. IRS CI is the only component in our review that does not have a documented policy to track ammunition or conduct a regular inventory; however, IRS CI officials indicated they are in the process of establishing a nationwide policy to track ammunition and conduct a regular ammunition inventory. Officials at two case study component offices said they provide their FLEOs extra boxes of ammunition to use for practice at a firing range outside of work hours. In general, four of the seven case study components did not track the real- time amount of ammunition they had on hand; rather, component officials said they had a general sense of the amount of ammunition in order to know if they had enough for training and qualifications, and when they would need to reorder ammunition. All case study components’ policies require secure storage of ammunition. In all 12 of the components’ offices we visited, we observed that ammunition was stored in their firearms storage rooms, and all offices took the additional step of securing the ammunition further in locked safes. For example, IRS CI requires ammunition storage in a security cabinet or a security room, preferably separately from firearms. Both of the IRS CI offices that we observed stored ammunition separately from firearms and with limited access. In the offices for two components, NIH and HHS OIG, we observed ammunition secured in a separate room from the firearms. Tactical equipment. Case study components do not control the tactical equipment in their inventory in the same ways that they control firearms and ammunition. Case study components that possessed aiming, breaching, and tactical lighting equipment did not have policies guiding their storage because they do not generally consider them to be as dangerous as a firearm or valuable enough to be tracked. However, components that possessed silencers, large-caliber launchers, pyrotechnics, or other items did have policies to control these items. Specifically, IRS CI policy requires the electronic tracking and inventory of night-vision equipment, optical equipment, and vehicles. NIH policy directs that large-caliber gas launchers be stored in the armory and pyrotechnics be stored in their bunker with ammunition. IRS CI, NIH, and EPA OIG all tracked this equipment during their annual inventory verification. Case study components did not report any instances of loss or theft of the tactical equipment in their inventories. Storage of tactical equipment varied and corresponded with case study components’ use of items. For example, we observed silencers and aiming devices stored on or near firearms at EPA OIG and NIH because they are accessories that attach to firearms. We observed breaching equipment to be stored either in the firearms storage room, vehicles, or elsewhere in the offices we visited. For example, at EPA OIG’s headquarters office, breaching equipment was stored in the firearms room because officials stated that they were only likely to use breaching equipment as part of a planned operation. However, officials at the FDA and NIH offices we visited said that their breaching equipment was stored in vehicles so it could be more readily accessible if they needed to use it. Conclusions Accurate reporting of firearms and ammunition is critical for accountability and transparency of these sensitive purchases. While reporting such purchases with precision is difficult, the execution of this responsibility impacts the public’s access to information about which agencies purchase what types of firearms and ammunition, and the amount that they spent on those purchases. However, ICE did not properly identify the funding agency in FPDS-NG for purchases where other DHS agencies used ICE contracts to procure firearms and ammunition. This inflated its publicly-available data to show a significantly higher obligated dollar value of purchases than it actually purchased. Because ICE does not accurately report the agency that funded the purchase to FPDS-NG, the public does not have accurate information on how much ICE and the agencies that make purchases using ICE procurement services have obligated for firearms and ammunition. Data need to be presented in a way that meets the needs of the end users—both policymakers and the public—if USASpending.gov is to fulfill its purpose of increasing accountability and transparency in federal spending. Improving the accuracy of the reported funding agency can better help the public understand and use federal purchase data, and increase accountability and transparency for these sensitive purchases. Recommendation for Executive Action To improve the accuracy of publicly-available purchase information, the Director of ICE should update ICE’s contracting process to include the name of the appropriate funding agency in data entered into FPDS-NG for firearms and ammunition purchases. (Recommendation 1) Agency Comments We provided a draft of the sensitive product to DHS, DOI, DOJ, EPA, HHS, SSA OIG, Treasury, USDA, and VA for review and comment. Agencies provided technical comments, which we incorporated as appropriate. DHS also provided written comments on the sensitive report, which are reproduced in full in appendix III. In its written comments, DHS concurred with our recommendation and described the actions ICE plans to take in response. We are sending copies of this report to interested congressional committees, the Secretaries of the Department of the Interior, the Department of Homeland Security, the Department of Agriculture, the Department of Justice, the U.S. Department of Veterans Affairs, the Social Security Administration, the Department of the Treasury, the Environmental Protection Agency, and the U.S. Department of Health and Human Services. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-8777 or goodwing@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix IV. Appendix I: Objectives, Scope, and Methodology This report provides additional information on our objectives, scope, and methodology. Specifically, our objectives were to examine the following questions: 1. What do available data show about spending on firearms, ammunition, and selected tactical equipment made by federal agencies with 250 or more federal law enforcement officers from fiscal years 2010 through 2017? 2. To what extent have selected agencies accurately reported purchases of firearms and ammunition in publicly-available data on USASpending.gov? 3. What types and quantities of firearms, ammunition, and selected tactical equipment do the Department of Health and Human Services (HHS), Environmental Protection Agency (EPA), and Internal Revenue Service (IRS) have documented to be in their inventory systems, as of November 2017? 4. What inventory controls and procedures are in place at HHS, EPA, and IRS, and to what extent do these agencies follow these procedures at selected locations? This report is the public version of a sensitive report that we issued in October 2018. HHS, IRS and the Transportation Security Administration (TSA) deemed some of the information in our October report to be sensitive, which must be protected from public disclosure. Therefore, this report omits sensitive information about the number of FLEOs at the TSA, an illustration of how HHS’s National Institutes of Health Police secures its firearms, and the number and types of some firearms, ammunition, and tactical equipment in NIH’s and IRS’s inventory. Although the information provided in this report is more limited, the report addresses the same objectives as the sensitive report and uses the same methodology. To address our first question, we obtained available spending data on firearms, ammunition, and certain tactical equipment from 20 agencies from the departments named in the Chief Financial Officers Act that employed 250 or more federal law enforcement officers (FLEOs) at any point from fiscal years 2010 through 2017. Specifically, we identified applicable agencies by reviewing Office of Personnel Management employment data and contacting agency officials to verify the employment of 250 or more FLEOs during the timeframe we reviewed. We excluded military branches from our review, such as all Department of Defense (DOD) branches and the U.S. Coast Guard, which is part of the Department of Homeland Security (DHS). Table 5 shows the 20 agencies, within eight departments, included in our scope for this question. Because there was no definitive list of what is considered tactical equipment, we developed a list of equipment to include in our review. To do so, we reviewed the National Firearms Act List, the Law Enforcement Equipment Working Group Recommendations, the DOD’s list of Controlled Property, and the Special Weapons and Tactics Gear used by the New York, Los Angeles and Houston police departments. We then selected and categorized the tactical equipment that appeared in two or more of these lists to include in our review. The 18 categories of tactical equipment we created were: (1) silencers, (2) explosive devices, (3) large-caliber weapons (>.50 caliber, excluding shotguns), (4) armored vehicles, (5) weaponized aircraft, vessels, or vehicles, (6) camouflage uniforms, (7) manned aircraft, (8) unmanned aerial vehicles, (9) tactical vehicles, (10) command and control vehicles, (11) pyrotechnics and specialized munitions, (12) breaching apparatus, (13) riot batons, (14) riot helmets, (15) riot shields, (16) tactical lighting (excludes basic flashlights), (17) specialized image enhancement devices (such as thermal imaging devices and night vision gear), and (18) aiming devices (such as scopes and tripods). To more closely describe the types of equipment frequently reported in the large-caliber weapons and breaching apparatus categories, we refer to them in the report as large-caliber launchers and breaching equipment, respectively. To collect data from the 20 agencies within the scope of our review for this objective, we developed a data collection instrument that requested spending data of firearms, ammunition, and selected tactical equipment from agencies’ internal record-keeping systems from fiscal years 2010 through 2017. For this objective, we requested the types of information that agencies reported using the data collection instrument: the date of each purchase; descriptive information on what was bought, including the caliber or gauge of firearms and ammunition; the quantity of items bought; the amount spent and whether those amounts were estimates; the type of record-keeping system used by the agency and any limitations associated with it or challenges compiling the information we requested; descriptions of changes or updates to the system that may have affected the data; and the contracting office(s) that were responsible for buying these items for the agency. We asked agencies not to include data on any items they received without spending funds, so agencies may have received more firearms, ammunition, or equipment in their inventories than what they reported in their spending data. For example, agencies may have received these items through interagency transfers, which may have no cost to receiving agencies. For the first objective, we analyzed the data agencies reported in the “amount spent” column, and we use the terms “spending” and “spent” to refer to these data. In cases where agencies only reported the amount they obligated for a purchase on the data collection instrument, we confirmed with agencies that those amounts reflected the amount they spent on the purchase and that we could use those amounts in our analysis. We pre-tested the instrument with Veterans Health Administration and U.S. Immigration and Customs Enforcement (ICE), whose officials provided feedback on the feasibility of providing the requested data. Based on the feedback we received from the pre-test, we revised and finalized the instrument and requested that the 20 agencies provide spending data on firearms, ammunition, and tactical equipment from fiscal years 2010 through 2017 from their internal record-keeping systems. To assess the reliability of the spending data, we conducted tests for missing data and obvious errors, reviewed relevant documentation, interviewed agency officials about their spending records and data reporting practices, and followed up with agency officials as needed. We re-categorized agency data that appeared to be miscoded. For example, we received data on specialized munitions and large-caliber launchers that agencies categorized as ammunition and firearms, respectively. Based on the descriptive data that agencies had provided, we re- categorized those items. We also adjusted the data to ensure consistency in their format, such as consistent entry of dates and use of categories of tactical equipment. Agencies in our review submitted a range of detail about their firearms when reporting their data, and we could not determine to what extent firearms were fully automatic. Some firearms have selector switches that allow the user to switch between semiautomatic and fully-automatic capabilities, while others are limited to shooting three-round bursts with each pull of the trigger. As such, we reported all firearms that are capable of firing multiple rounds with the single pull of the trigger as fully-automatic firearms. We also combined ammunition intended for use in machine guns with rifle ammunition because machine guns shoot rifle-caliber ammunition, and we did not have confidence that every agency accurately distinguished which rifle- caliber ammunition was intended for use in machine guns and what was reserved for rifles. We found the data sufficiently reliable for the purpose of reporting the minimum thresholds of total amounts agencies spent and the numbers of firearms and rounds of ammunition they purchased during fiscal years 2010 through 2017. However, we found the data were not reliable for reporting the number of tactical equipment items purchased or reporting further comparative analysis. Agency officials reported various challenges in compiling the data we requested and we identified some data limitations, as described in table 6. To address our second question, we selected three agencies—the Bureau of Indian Affairs (BIA), the U.S. Forest Service, and the U.S. Immigration and Customs Enforcement (ICE)—to assess the extent to which their purchases of firearms and ammunition were accurately reflected in publicly-available data. We requested that agencies report internal purchase data for firearms and ammunition that included, among other things, the amount obligated for these items, a unique transaction identifier (called the Procurement Instrument Identifier), and the product or service code for firearms and ammunition purchases from fiscal years 2010 through 2017. We compared the amounts these three agencies obligated for firearms and ammunition purchases, as reported to us in their data collection instruments, with the obligation data that are publicly- available data on USASpending.gov for those three agencies. We obtained publicly-available data from USASpending.gov, which includes purchase data from the Federal Procurement Data System-Next Generation (FPDS-NG), using product or service codes (PSC) that identify contracts for firearms or ammunition purchases. We also reviewed our related work and Inspector General reports on the quality of USASpending.gov data. We obtained records for purchases made in fiscal years 2010 through 2017. We did not include equipment purchases because the publicly-available data lack product or service codes that would allow us to reliably identify those records. To select the three agencies for inclusion in this analysis, we started with the 20 agencies with at least 250 FLEOs in our scope. From those, we selected agencies that provided us with records of their firearms and ammunition purchases and that USASpending.gov listed as the funding agency for one or more firearms or ammunition purchase (12 agencies). From those, we selected three agencies based on the total dollar value of purchases reported in USASpending: one small (BIA), one medium (Forest Service), and one large (ICE), based on natural breaks in dollar values and not selecting multiple agencies from the same department. We then compared obligations data provided to us by each of the three agencies against obligations in the publicly-available purchase records using the Procurement Instrument Identifier to match records across agency-provided and publicly-available purchase data. We additionally corroborated these obligations by comparing fields related to the date of purchase, purchase value, and vendor. We reviewed a portion of purchase records to compare and interviewed agency officials about differences in the publicly-available and agency-provided data. For firearms, we included the following PSCs: 1005 - Guns, through 30mm 1010 - Guns, over 30mm up to 75mm 1015 - Guns, 75mm through 125mm 1020 - Guns, over 125mm through 150mm 1025 - Guns, over 150mm through 200mm 1030 - Guns, over 200mm through 300mm 1035 - Guns, over 300mm We excluded several weapons PSCs from our analysis that described weapons other than firearms, such as 1040 – Chemical Weapons and Equipment. Any excluded PSC, particularly 1095 – Miscellaneous Weapons, may have been used as the PSC to categorize a purchase that included firearms and those purchases would be excluded from the publicly-available records that we examined. Similarly, for ammunition, we included records associated with the following PSCs: 1305 – Ammunition, through 30mm 1310 – Ammunition, over 30mm up to 75mm 1315 – Ammunition, 75mm through 125mm 1320 – Ammunition, over 125mm Like firearms, we excluded records that were associated with non- firearms ammunition, such as 1336 – Guided Missile Warheads and Explosive Components. Any excluded PSC, particularly 1395 - Miscellaneous Ammunition, may have been used as the PSC to categorize a purchase that included firearms ammunition and those purchases would be excluded from the publicly-available records that we examined. To address our third and fourth questions, we reviewed inventory information and controls for case study federal law enforcement components within U.S. Department of Health and Human Services (HHS), Environmental Protection Agency (EPA), and Internal Revenue Service (IRS). These components are as follows: EPA Office of Inspector General (OIG), Office of Enforcement and Compliance Assurance (OECA); Food and Drug Administration (FDA), National Institutes of Health (NIH), HHS OIG, and IRS. Within IRS, two offices employ FLEOs: Criminal Investigation (CI) and Police Officer Section (Police). Accordingly, we can draw conclusions only about these components. We obtained and analyzed inventory data and other available documentation provided by these components regarding their current inventory as of November 2017 of firearms, ammunition, and tactical equipment. This included firearms in storage as well as those in FLEO possession. To assess the reliability of the inventory data, we reviewed components’ documentation related to data management, especially policies to ensure that items are properly entered and removed from the system. In addition, we reviewed components’ recent purchases to verify that purchases were inventoried and reviewed their purchase data for any limitations that may affect their quality. We reviewed components’ acquisitions through DOD’s excess property program, known as the Law Enforcement Support Office (LESO) or 1033 program, to ensure these items were inventoried. We did not conduct physical inventories during site visits, and therefore did not actually count components’ physical inventories. To examine these components’ firearms, ammunition, and tactical equipment inventory controls, we reviewed their policies describing storage protocols and inventory control procedures, and we interviewed components’ officials to better understand these policies and procedures in practice. We also compared these policies with applicable Standards for Internal Control in the Federal Government and key areas that we have identified as important for effective inventory management. We reviewed internal and external inspection and Inspector General reports related to the controls over firearms, ammunition, and tactical equipment at these components to identify any reported deficiencies and actions taken or planned to address those deficiencies. We also conducted site visits to components’ offices selected based on a variety of factors, including the number of agencies with component offices in each city we visited, data discrepancies at field offices, and reports of loss or theft at these offices. The IRS Police facility we visited in Martinsburg, West Virginia is the only location for this component. In all, we visited 12 offices. During site visits, we observed officials demonstrating inventory inspection, inventory data entry, and access and other security controls. In addition, we interviewed officials responsible for maintaining and inventorying firearms, ammunition, and certain tactical equipment. The observations and information we obtained from the offices visited cannot be generalized to other locations for these components, but provide insights about the components’ controls for firearms, ammunition, and tactical equipment. We conducted this performance audit from June 2017 to October 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. We subsequently worked with the agencies addressed in this report from October 2018 to December 2018 to prepare this public version of the original sensitive report for release. This public version was also prepared in accordance with these standards. Appendix II: Spending on Firearms, Ammunition, and Tactical Equipment by Agency This appendix summarizes agency-provided spending data for the 20 agencies with 250 or more federal law enforcement officers and the 5 additional agency components that were included in our review. This appendix also identifies the challenges that agency officials identified in collecting the requested data and the limitations we identified while analyzing their data. To compile this information, we asked the agencies to provide data on their spending on firearms, ammunition, and selected tactical equipment from fiscal years 2010 through 2017. We reviewed the spending data provided by agencies to assess their accuracy and completeness, and followed up with agency officials as needed. Some agencies reported limited access to spending records because of their storage format or physical location, and we asked agencies to report the data that were accessible to them within the timeframe of this review. We found the data sufficiently reliable for the purpose of reporting the minimum thresholds of total amounts agencies spent and the numbers of firearms and rounds of ammunition they bought during the 8-year period in our scope. However, we found that data were not reliable for reporting the number of tactical equipment items purchased or for comparing the data across and within agencies. For more information on the methodology we used to collect these data and the challenges agencies faced compiling these data, see appendix I. Officials also provided information about the missions of their agencies or components, and the roles and responsibilities of their law enforcement officers. We included inventory data as of November 2017 for the agency components that were included in our review of inventory controls. U.S. Forest Service Protects the public, employees, natural resources, and property under the jurisdiction of the Forest Service by enforcing the applicable laws and regulations that affect the National Forest System. Officers are responsible for conducting enforcement and investigations of criminal and civil offenses that affect the management of the National Forest System. Officers engage in public safety patrol operations, investigations of significant criminal offenses, community policing programs, natural disaster response, and law enforcement services at large group events, among other things. Agency stores purchase records for six years, which limited access to older data. Ammunition purchases and record-keeping are mostly decentralized among field offices, and officials could not provide the level of detail requested for all purchases. Upgrades to the record-keeping system may have compromised data from previous systems. Firearms spending data Minimum values by year Dollars (in thousands) No known limitations. Ammunition spending data Minimum values by year Dollars (in thousands) Tactical equipment spending data U.S. Department of Health and Human Services Food and Drug Administration (FDA) FDA’s Office of Criminal Investigations protects public health and furthers the FDA’s mission by investigating suspected criminal violations of the Federal Food, Drug, and Cosmetic Act and other related laws. No known challenges. Officers conduct investigations related to criminal violations of the Food Drug and Cosmetic Act, which include conducting search and seizure warrants, transporting prisoners following arrest, conducting undercover operations, and other hazardous duties as necessary. Firearms spending data Values by year Dollars (in thousands) Ammunition spending data Minimum values by year Dollars (in thousands) Tactical equipment spending data Inventory as of November 2017 U.S. Department of Health and Human Services National Institutes of Health (NIH) Protects our country’s scientific research and the NIH research community, ensures that the mission of NIH is not impeded by personal attacks, loss of assets, criminal activity or acts of terrorism. Officers with the Division of Police are responsible for protecting property, employees and visitors; screening visitors entering NIH facilities; monitoring onsite equipment, cameras and alarms; operating the visitor badging system on and off campus; safeguarding selected buildings; patrolling areas of the NIH; providing traffic enforcement; conducting intelligence gathering and reporting; providing dignitary protection; and preparing warrants and arresting suspects. No known challenges. Firearms spending data Values by year Dollars (in thousands) Ammunition spending data Minimum values by year Dollars (in thousands) Tactical equipment spending data Type by year No known limitations. Inventory as of November 2017 NIH determined inventory information to be sensitive. U.S. Department of Health and Human Services (HHS) Office of Inspector General (OIG) No known challenges. Special agents conduct criminal investigations related to fraud, waste, and abuse within HHS’ hundreds of programs, which can involve surveillance, undercover operations, search warrants, and arrest warrants. U.S. Customs and Border Protection (CBP) Safeguards America’s borders, protects the public from dangerous people and materials, and enables legitimate trade and travel. Officers are responsible for preventing terrorists and weapons from entering the country, enforcing laws at ports of entry, and preventing the illegal trafficking of people, narcotics and contraband. Record-keeping system was implemented in 2016. Prior records required manual review. Records of purchases made at local field offices may be unavailable. U.S. Immigration and Customs Enforcement (ICE) Enforces federal laws governing border control, customs, trade and immigration to promote homeland security and public safety. responsible for conducting investigations to protect critical infrastructure industries that are vulnerable to sabotage, attack or exploitation. Some purchases are decentralized among field offices. Hard copy records required manual review and were difficult to access. Older records stored at field offices may not have been available. Firearms spending data Minimum values by year No known limitations. Ammunition spending data Minimum values by year Dollars (in millions) Tactical equipment spending data U.S. Department of Homeland Security National Protection and Programs Directorate (NPPD) NPPD’s Federal Protective Service prevents, protects, responds to and recovers from terrorism, criminal acts, and other hazards threatening the U.S. Government’s workforce, critical infrastructure, services, and the people who receive these services. No known challenges. Officers with the Federal Protective Service are responsible for enforcing all federal laws and regulations on and off federal property; investigating, mitigating, and defeating threats to federal facilities and the people who work within or visit those facilities; and providing integrated security, law enforcement, and protective intelligence capabilities to ensure the Federal Government functions securely. Firearms spending data Values by year Dollars (in thousands) Ammunition spending data Values by year Dollars (in thousands) Tactical equipment spending data No known limitations. U.S. Department of Homeland Security Transportation Security Administration (TSA) TSA’s Federal Air Marshal Service (FAMS) detects, deters and defeats criminal and terrorist activities that target our nation’s transportation systems. TSA’s Office of Inspection (OOI) ensures the integrity, efficiency, and effectiveness of TSA’s workforce, operations, and programs through objective audits, covert testing, inspections, and criminal investigations. No known challenges. Officers with FAMS are deployed on U.S. aircraft worldwide to protect airline passengers and crew against the risk of criminal and terrorist violence, and perform investigative work to proactively fight terrorism. Officers with OOI investigate allegations of misconduct by TSA employees and contractors, conduct inspections of TSA operations, and evaluate effectiveness of security systems through covert testing and audits. Firearms spending data Values by year No known limitations. Ammunition spending data Dollars (in millions) Tactical equipment spending data U.S. Department of Homeland Security U.S. Secret Service (USSS) Ensures the safety and security of the President, the Vice President, their families, the White House, the Vice President’s Residence, national and visiting world leaders, former U.S. Presidents and events of national significance; and protects the integrity of our currency and investigates crimes against our national financial system committed by criminals around the world and in cyberspace. Officers are responsible for executing security operations that prevent, deter, and mitigate identified threats and vulnerabilities; and conducting investigations to identify, locate and apprehend individuals and criminal organizations targeting the nation’s critical financial infrastructure and payment systems. Record-keeping system changed in 2015 and older records may be incomplete. Officials were unable to provide records of equipment purchases because purchasing is decentralized among field offices and records required manual review; they provided current inventory records of the equipment instead. Firearms spending data Minimum values by year Dollars (in thousands) Officials reported that some cost data were estimated based on available information The percentages do not total 100 because of rounding. Ammunition spending data Minimum values by year Dollars (in millions) Tactical equipment spending data U.S. Department of the Interior Bureau of Indian Affairs (BIA) BIA’s Office of Justice Services upholds tribal sovereignty and customs while supporting tribal justice systems, and corroboratively ensures the safety of Indian communities by protecting life and property, enforcing laws, and maintaining justice and order. Officers are responsible for patrolling designated Indian reservations, providing local law enforcement, responding to calls for emergency response, investigating crimes, transporting prisoners to and from tribal court appearances, gathering and analyzing criminal intelligence, and collaborating with state and federal task forces. The current financial management system was implemented in 2013, and records from the previous system may be incomplete. Firearms spending data Minimum values by year Dollars (in thousands) No known limitations. Ammunition spending data Dollars (in thousands) Because of incomplete data from the agency, we determined the data are not reliable for reporting in this category. Tactical equipment spending data U.S. Department of the Interior U.S. Fish and Wildlife Service (FWS) Firearms spending data Minimum values by year Dollars (in thousands) Ammunition spending data Dollars (in thousands) Because of incomplete data from the agency, we determined the data are not reliable for reporting in this category. Tactical equipment spending data U.S. Department of the Interior National Park Service (NPS) events within the NPS; and providing protection for dignitaries and visiting foreign heads of state. NPS’ Visitor and Resource Protection Directorate—under which fall the Law Enforcement, Security, and Emergency Services and U.S. Park Police—protects the safety and health of NPS visitors, partners, and staff, as well as our natural and cultural resources. Financial management system implemented in 2013, and older records did not retain the cost data fields. Some records kept as paper copies and were decentralized among field offices. Firearms spending data Minimum values by year Dollars (in thousands) Ammunition spending data Dollars (in thousands) Tactical equipment spending data U.S. Department of Justice Bureau of Alcohol, Tobacco, Firearms and Explosives (ATF) Mission Protects the public from crimes involving firearms, explosives, arson, and the diversion of alcohol and tobacco products; regulates lawful commerce in firearms and explosives; and provides worldwide support to law enforcement, public safety, and industry partners. Officers are responsible for reducing violent crime by targeting firearms traffickers, violent criminal organizations, armed violent offenders, and career criminals; investigating and arresting individuals and organizations that illegally supply firearms to prohibited individuals; and deterring the diversion of firearms from lawful commerce into the illegal market with enforcement strategies and technology. Retention policy is 6 fiscal years, so older records may have been unavailable. There may be errors associated with manual data entry. Firearms spending data Minimum values by year Dollars (in thousands) firearms in 2017. At the time of our review, the funds were obligated but not yet spent. Ammunition spending data Dollars (in thousands) No known limitations. Tactical equipment spending data No known limitations. U.S. Department of Justice Federal Bureau of Investigation (FBI) Firearms spending data Minimum values by year No quantity or type data reported. Ammunition spending data Minimum values by year Firearm of intended use for all years No quantity or firearm of intended use data reported. Tactical equipment spending data U.S. Department of Justice Federal Bureau of Prisons (BOP) Protects society by confining offenders in prisons and community-based facilities and provides work and other self-improvement opportunities to assist offenders in becoming law-abiding citizens. Officers are responsible for ensuring the security of federal prisons, providing inmates with needed programs and services, and modeling mainstream values. Officers help protect public safety and provide security and safety to the staff and inmates in prison facilities. All non- custody staff are trained to assume the duties of Correctional Officers. Officials reported limited data—most records prior to 2016 were stored in file cabinets at prison facilities or in warehouses and would have required significant time and resources to review. Purchasing is decentralized across prison facilities, which limited the availability of data, and inconsistent data entry procedures among officials may limit the reliability of data. Drug Enforcement Administration (DEA) illicit drug trafficking; coordination with law enforcement officials on drug enforcement efforts and to reduce availability of illicit drugs. Firearms purchases were embedded in contracts and required manual review. Firearms spending data Values by year Dollars (in thousands) Ammunition spending data Dollars (in millions) Tactical equipment spending data U.S. Department of Justice U.S. Marshals Service (USMS) Officers provide for the security of federal courts and execute and enforce federal court orders, apprehend fugitives and non-compliant sex offenders. Officers also transport federal prisoners from arrest to incarceration, manage and disposes of assets subject to forfeiture, and provide protection for government witnesses and their families. The record-keeping system changed in 2012, and there may be gaps in the data from 2013 and 2014 while the agency transitioned to the new system. Firearms spending data Minimum values by year Dollars (in thousands) Ammunition spending data Firearm of intended use for all years No quantity data provided. Tactical equipment spending data U.S. Department of the Treasury The Internal Revenue Service (IRS) IRS’ Criminal Investigation serves the American public by investigating potential criminal violations of the Internal Revenue Code and related financial crimes in compliance with the law. Officers enforce tax laws and support tax administration to ensure compliance with the law and combat fraud. Investigations focus on tax fraud, abusive tax schemes, identity theft, public corruption, virtual currency, cyber-crimes, and narcotics-related financial crimes. Some requested data were not retained in the system. Agency could not provide equipment purchase data. Ammunition, equipment, and purchases $3,500 or less are decentralized among field offices. The Internal Revenue Service (IRS) Police IRS’ Police Force at the Enterprise Computing Center in Martinsburg, West Virginia provides protection for the people, property and processes of this location, which houses 10 of IRS’ 19 critical tax processing functions. No known challenges. Officers patrol the facility and have authority to serve warrants and make arrests. Firearms spending data No reported firearms purchases between fiscal years 2010 through 2017. Dollars (in thousands) No known limitations. Tactical equipment spending data Inventory as of November 2017 U.S. Department of the Treasury The Treasury Inspector General for Tax Administration (TIGTA) Provides independent oversight of Internal Revenue Service (IRS) activities and addresses threats arising from lapses in IRS employee integrity, violence directed against the IRS, and external attempts to corruptly interfere with federal tax administration. Officers are responsible for conducting investigations that protect the integrity of the IRS; detecting and preventing fraud and other misconduct within IRS programs; investigating allegations of criminal violations and administrative misconduct by IRS employees; and protecting IRS against external attempts to corrupt or threaten its employees. After fiscal year 2016, record-keeping system no longer tracked purchases under $3,000. Firearms spending data Minimum values by year Dollars (in thousands) No known limitations. Ammunition spending data Dollars (in thousands) Tactical equipment spending data No known limitations. U.S. Department of the Treasury The U.S. Mint No known challenges. Officers with the Mint Police protect life and property, prevent, detect, and investigate criminal acts, collect and preserve evidence, make arrests, and enforce federal and local laws. Office of Enforcement and Compliance Assurance (OECA) EPA’s criminal enforcement program focuses on criminal conduct that threatens people’s health and the environment; enforces the nations’ laws by investigating cases, collecting evidence, conducting forensic analyses; and provides legal guidance to assist with prosecutions. No known challenges. Agents enforce the nation’s laws by investigating cases, collecting evidence, conducting forensic analyses and providing legal guidance to assist with prosecutions. Firearms spending data No reported firearms purchases from fiscal years 2010 through 2017. Dollars (in thousands) Tactical equipment spending data Inventory as of November 2017 U.S. Environmental Protection Agency (EPA) Office of Inspector General (OIG) Helps the agency protect the environment in a more efficient and cost effective manner by performing audits, evaluations, and investigations of EPA and its contractors; promoting economy and efficiency; and preventing and detecting fraud, waste, and abuse. Law enforcement agents conduct criminal investigations of financial fraud involving EPA programs or funds; employee misconduct; intrusion into EPA computers; threats against EPA employees, contractors, facilities and assets; assaults on EPA employees or contractors and other acts of violence in EPA facilities; impersonating EPA officials; counterfeiting or misuse of insignia, logos or credentials; and theft of property or funds within EPA facilities. No known challenges. Office of the Inspector General (OIG) OIG’s Office of Investigations conducts and coordinates investigative activity related to fraud, waste, abuse, and mismanagement in SSA programs and operations. No known challenges. Officers with the Office of Investigations investigate wrongdoing by applicants, beneficiaries, contractors and third parties, and employees; conduct joint investigations with other law enforcement agencies; share responsibility for investigating threats or violence against SSA employees and facilities; and assist in the investigation of terrorism cases and other cases involving national security. Firearms spending data Values by year Dollars (in thousands) No known limitations. Ammunition spending data Dollars (in thousands) No known limitations. Tactical equipment spending data No known limitations. Department of Veterans Affairs (VA) Veterans Health Administration (VHA) Protects veterans by enforcing federal law at VA medical facilities (and some National Cemetery and Benefits locations) and by serving as initial response forces to active threat incidents. Officers protect veterans, visitors, and staff on department facilities and grounds; investigate serious incidents on VA controlled property; and provide personal protection to the Secretary and Deputy Secretary of the VA. Purchasing and record-keeping are decentralized among facilities, and there is no agency-wide system for recording purchases. Purchase card records were not easily accessible or identifiable. Firearms spending data Minimum values by year Dollars (in thousands) We identified missing data for some cost and date fields Officials reported some cost data were estimated based on available information The percentages do not total 100 because of rounding. Ammunition spending data Minimum values by year Dollars (in thousands) Tactical equipment spending data Type by year Appendix III: Comments from the Department of Homeland Security Appendix IV: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the above contact, Adam Hoffman (Assistant Director) and Michelle Serfass (Analyst-in-Charge) managed this assignment. Christoph Hoashi-Erhardt, Allison Gunn, and Kelsey Burdick made significant contributions to this report. David Alexander, David Blanding Jr., Willie Commons III, Eric D. Hauswirth, Julia Kennon, Susan Hsu, Diana Maurer, Wayne McElrath, and Kevin Reeves also contributed.

Why GAO Did This Study Federal law enforcement agencies purchase firearms, ammunition, and tactical equipment, such as riot shields, to support their missions. GAO was asked to review these purchases for federal law enforcement agencies, and inventory controls at HHS, EPA, and IRS specifically. This report examines, among other objectives (1) firearms, ammunition, and selected tactical equipment spending by federal agencies with 250 or more FLEOs from fiscal years 2010 through 2017; (2) the extent to which select agencies accurately reported purchases of firearms and ammunition in publicly-available data; and (3) inventory controls in place at HHS, EPA, and IRS. GAO obtained available data on purchases from 20 agencies and from USASpending.gov, and reviewed inventory information and controls within HHS, EPA, and IRS. GAO also conducted site visits to HHS, EPA, and IRS offices to observe inventory controls, selected based on data discrepancies or reports of loss or theft, among other factors. This is a public version of a sensitive report that GAO issued in October 2018. Information that HHS, IRS, and the Transportation Security Administration deemed sensitive has been omitted. What GAO Found The 20 federal law enforcement agencies in GAO's review reported spending at least $38.8 million on firearms, $325.9 million on ammunition, and $1.14 billion on tactical equipment—at least $1.5 billion in total—from fiscal years 2010 through 2017, based on data agencies provided to GAO. The internal agency data on firearms and ammunition purchases for the Bureau of Indian Affairs, U.S. Forest Service, and U.S. Immigration and Customs Enforcement (ICE) did not always match data that were publicly available on USASpending.gov—a government source for federal contract data. In particular, the dollar value of firearms purchases by ICE in USASpending.gov was approximately 8 times greater than the value of the purchases reported by ICE to GAO. Some differences result from other agencies using ICE contracts to make firearms and ammunition purchases, and ICE not properly identifying the funding agency for those purchases in the system that supplies data to USASpending.gov. Because ICE does not accurately report the agency that funded these purchases, the public does not have accurate information on how much ICE—and the agencies that make purchases using ICE contracts—have spent on firearms and ammunition. This decreases accountability and transparency of federal purchases, which is in conflict with the intended purpose of this system. Department of Health and Human Services (HHS), the Environmental Protection Agency (EPA), and the Internal Revenue Service (IRS) have inventory controls for tracking, verifying, and securing federal law enforcement officers' (FLEOs) firearms. GAO observed these agencies' law enforcement components and found them to be generally following their inventory and security policies at selected locations. In instances where agencies were not in compliance with their policies, the agencies made corrections during the course of GAO's review. Each component has a process whereby at least once yearly officials review the firearms inventory to ensure that firearms match with records in the office's inventory system. The figure below illustrates a general process that all components GAO reviewed follow to verify their firearms inventory. Ammunition and tactical equipment inventory controls varied because agencies generally did not consider these items to be as sensitive as firearms. Examples of these controls include security for, and limited access to, equipment that might be vulnerable to risk of loss or unauthorized use, such as silencers or pyrotechnics. What GAO Recommends GAO recommends that the Director of ICE update ICE's contracting process to provide the name of the agency funding the purchase of firearms and ammunition to improve the accuracy of publicly available data. ICE concurred with the recommendation.

gao_GAO-18-297

gao_GAO-18-297_0

Background Home Equity Building A homeowner can build home equity immediately by making a down payment on their home, assuming the down payment is not financed separately as a loan. Throughout the life of a mortgage, homeowners can continue to build equity (1) by making regular mortgage payments to reduce the principal amount outstanding, (2) by making additional payments to further reduce the principal amount outstanding, and (3) through appreciation in their home’s value. Additionally, the components of a mortgage (discussed below) may affect the pace of home equity building. Throughout this report, for the purposes of illustrating home equity building, we assumed that a home’s value remained unchanged from the time of the loan origination. However, home values are highly contingent on market conditions and other factors that are beyond a homeowner’s control. For example, although homes can appreciate in value, homes also can depreciate in value, which can have a negative effect on homeowners’ equity. Homeowners also could lose money on their home if they sold it shortly after purchasing because principal reduction in the initial years of a mortgage is relatively small and the benefit of any home value appreciation would be limited. Additionally, selling a home incurs transaction costs, such as realtor commissions. To avoid losing money on a home sale, homeowners would need to sell their home at an amount higher than their purchase price plus transactions costs. For example, if a homeowner buys a home for $250,000 (all fees included) and plans to sell it 3 years later, assuming transaction costs of 10 percent (or $25,000), the homeowner would have to sell the home for at least $275,000 to break even, meaning an annual appreciation in home value of more than 3 percent. If the home’s value did not appreciate at that rate, or depreciated, the homeowner would lose money on the sale. Mortgage Loan Components The majority of American families achieve homeownership by taking out a loan—a mortgage—to cover at least some of the purchase price. The primary components of a mortgage loan are the following: Term (duration). The most common term is 30 years. According to the Urban Institute, the 30-year fixed-rate mortgage represented approximately 90 percent of the fixed-rate purchase mortgages (that is, not for refinancing an existing mortgage) originated every month from January 2010 through July 2017, and 15-year fixed-rate purchase mortgages represented about 6 percent. Down payment. Most mortgage lenders require borrowers to make a down payment (of 3 percent or more of the purchase price, depending on the mortgage) that is applied to the purchase price of the home. A down payment also helps a borrower build home equity, assuming the down payment is not financed as a separate loan. Interest rate. Lenders charge borrowers a percentage of the mortgage amount, in exchange for providing funds to buy a home. An interest rate can be fixed or adjustable for the life of the mortgage (adjustable-rate mortgage or ARM). Because a fixed-rate mortgage’s interest rate does not change regardless of prevailing rates, a borrower’s payments for principal and interest remain the same for the life of the mortgage. In contrast, an adjustable-rate mortgage’s interest rate, for which the initial interest is generally lower than for a fixed-rate mortgage, will adjust at agreed-upon intervals. As a result, adjustable-rate mortgage payments can increase or decrease depending on the changes in interest rates and terms of the loan. Payment frequency and amount. Payments are generally made on a monthly basis. Fixed- and adjustable-rate mortgages generally have fully amortizing payment schedules—that is, the regularly scheduled payments will fully pay down the principal and interest over the life of the mortgage, with the amounts allocated to reducing principal and interest changing over time (see fig. 1). The Federal Role in Mortgage Markets The U.S. markets for single-family housing finance include a primary market, in which lenders make (originate) or refinance mortgage loans, and a secondary market, in which mortgage loans are purchased from lenders and packaged into securities—known as mortgage-backed securities—that are sold to investors. The federal government participates in the primary and secondary mortgage markets. In the primary market, federal agencies provide homeownership assistance programs and products intended for increasing access to and affordability of homeownership. Relevant federal agencies and a government-sponsored enterprise that provide homeownership assistance and their primary housing-related policy goals include the following: Department of Housing and Urban Development provides housing assistance to low-and moderate-income families and promotes urban development. Federal Housing Administration (FHA) seeks to broaden homeownership, strengthen the mortgage marketplace, and increase access to credit by providing mortgage insurance. Public and Indian Housing helps ensure safe, decent, and affordable housing through programs such as housing choice vouchers. Community Planning and Development seeks to develop viable communities and provide decent housing and a suitable living environment through block grant assistance. Department of Veterans Affairs assists service members, veterans, and eligible surviving spouses of veterans to become homeowners through guaranteeing and issuing (in limited circumstances) mortgages for home purchases. Rural Housing Service (RHS), which is an agency within USDA, insures and guarantees housing loans for home purchases, repair, and rental housing development. Federal Home Loan Banks help provide liquidity to each bank’s member financial institutions to support housing finance and community investment. FHLBank members include commercial banks, thrifts, and credit unions. FHLBanks provide 10 percent of their earnings for affordable housing programs, including grants for affordable housing for households with incomes at or below 80 percent of the area median. Federal homeownership assistance programs can be categorized in terms of the products or services they offer or the mechanisms they use. The categories include mortgage guarantees and insurance, down- payment assistance, vouchers, and direct loans (discussed in more detail later in this report). In addition to these categories of homeownership assistance, tax expenditures, such as exclusions, exemptions, deductions (including the mortgage interest deduction), credits, deferrals, and preferential rates, can promote homeownership. For example, homeowners can take advantage of tax deductions (by choosing to itemize deductions on their tax returns) to help lower their taxable income. Taxpayers who itemize deductions may deduct qualified interest they pay on their mortgage. Taxable income may be reduced by the amount of interest paid on first and second mortgages of up to $750,000 for homes purchased generally after December 15, 2017. Additionally, taxpayers generally may deduct up to $10,000 for state and local taxes, including property taxes paid by homeowners on their homes. Participation in the secondary mortgage market occurs through the following entities: Fannie Mae and Freddie Mac. Fannie Mae and Freddie Mac are government-sponsored enterprises (enterprises)—congressionally chartered, for-profit, shareholder-owned companies. They are the two largest participants operating in the secondary mortgage market. Generally, Fannie Mae and Freddie Mac purchase mortgage loans that meet certain criteria for size, features, and underwriting standards—known as conforming loans—from lenders. In purchasing loans, the enterprises provide market liquidity, so lenders can provide more loans to borrowers. Ginnie Mae. Ginnie Mae is a wholly-owned government corporation. Ginnie Mae guarantees the timely payment of principal and interest on mortgage-backed securities supported by pools of loans backed by government-insured mortgages, including mortgages insured by FHA, VA, and USDA. Mortgage-Related Regulations In a process called underwriting, mortgage lenders evaluate the creditworthiness of potential borrowers in making mortgage loans, among other things. Amid concerns that risky mortgage products and poor underwriting standards contributed to the recent housing crisis, Congress included mortgage reform provisions in the Dodd-Frank Wall Street Reform and Consumer Protection Act (Dodd-Frank Act). The Dodd-Frank Act generally requires lenders to determine consumers’ ability to repay home mortgage loans before extending credit and provides a presumption of compliance with the ability-to-repay requirement for qualified mortgages. The ability-to-repay regulations set forth lenders’ responsibilities to determine a borrower’s ability to repay a residential mortgage loan, and special payment calculation rules apply for loans with balloon payments, interest only payments, or negative amortization. The regulations require lenders to make a reasonable and good faith determination of a consumer’s reasonable ability to repay a loan. The regulations establish a safe harbor and a presumption of compliance with the ability-to-repay rule for certain qualified mortgage loans (QM). The rule generally prohibits loans with negative amortization, interest-only payments, or balloon payments from being qualified mortgages, and limits the points and fees a lender may charge borrowers on a qualified loan. The regulations establish general underwriting criteria for qualified mortgages. For example, under QM requirements borrowers generally cannot exceed a maximum monthly debt-to-income ratio of 43 percent, unless the loan is eligible for sale to an enterprise. If a mortgage loan meets the requirements of a QM loan, it is eligible for the safe harbor and the lender is deemed to have complied with the ability-to-pay requirement unless the loan is a higher priced mortgage loan. A higher priced mortgage loan that otherwise meets the definition of a QM is presumed to have complied with the ability-to-pay requirements, but the presumption can be rebutted if the consumer proves that the lender did not make a good faith and reasonable determination of the consumer’s ability to repay. Additionally, federal mortgage insurance is included in the determination of whether an FHA-insured loan is a higher priced mortgage loan. Federal Homeownership Assistance Programs Can Have Equity- Building Effects, but Are Not Specifically Designed to Accelerate Equity Building Existing federal homeownership assistance programs use features and mechanisms that can have equity-building effects, but the programs are not specifically designed to accelerate equity building. The programs can assist homeowners to build equity over time by providing access to homeownership, but the programs do not have an explicit focus on accelerating the ongoing pace of paying down the loan principal faster than a 30-year fixed-rate mortgage. Rather, the overall focus of the programs is on providing affordable access to homeownership, according to officials of relevant agencies and entities and based on their mission goals. For example, the goal of FHA’s mortgage insurance program is to facilitate access to affordable mortgages for home buyers who might not be well-served by the private market. FHA implements this goal by providing insurance to lenders to facilitate access to mortgage financing for lower-income home buyers. See table 1 for examples of federal homeownership assistance programs, by major program types and potential for affecting equity building, either at a point in time or throughout the life of a mortgage. Federal Mortgage Insurance and Loan Guarantees Increase Market Liquidity Federal mortgage insurance and guarantee programs increase market liquidity, which ultimately expands access to homeownership. The federal government commits to pay part or all of a loan’s outstanding principal and interest loss to a lender or other mortgage holder if the borrower defaults. Because they obtain insurance or a guarantee against the possibility of loss from borrower default, lenders are more willing to provide loans to borrowers who might not otherwise be served by the private market, allowing more homeowners—particularly lower-income borrowers—an opportunity to build home equity. FHA offers mortgage insurance and RHS and VA provide loan guarantees. For example, FHA will insure loans with a down payment as low as 3.5 percent from most borrowers, and conventional mortgages will allow down payments as low as 3 percent. FHA-insured loans also have more lenient credit requirements that particularly benefit minority households and first-time home buyers who might otherwise find it difficult or more expensive to take out a mortgage. Among federal mortgage insurance programs, FHA has the highest volume of mortgages insured. Federal Down-Payment Assistance Programs Can Have Equity-Building Effects Federal and federally mandated programs that provide funding for grants and loans for down-payment assistance can have equity-building effects. Although accelerated equity building is not the policy goal of these programs, down-payment assistance can lower the barrier to homeownership for some lower-income home buyers so that the equity-building effects of homeownership can accrue. Examples of programs include the following: HUD’s HOME Investment Partnership Program is a block grant program that provides funding to states and localities to be used exclusively for affordable housing activities to benefit low-income households. Funds can be used for down-payment assistance for eligible low-income home buyers. According to HUD data, more than 75 percent of low-income home buyers who have received assistance from the HOME program have used HOME funds for purchasing a home (which includes down-payment assistance) since the program’s inception in 1992, directly contributing to homeowner equity building. HUD’s Community Development Block Grant (CDBG) program also provides funding to eligible states and localities for community and economic development efforts, including housing assistance. Eligible uses of home-buyer assistance include grants for down payments and closing costs. In fiscal year 2016, CDBG funds provided direct housing assistance for down payment and closing costs to 2,483 households. FHLBanks contribute funding to the Affordable Housing Program (AHP), which can provide grants for down-payment assistance through either the AHP competitive or set-aside program. Member financial institutions of the FHLBanks can apply for the set-aside funds and then distribute the funds as grants to eligible households. Set-aside grants may be no greater than $15,000 per household, and at least one-third of the FHLBanks’ annual set-aside allocation must be used for eligible first-time home buyers. According to FHFA, the FHLBanks funded about $77 million for down-payment or closing-cost assistance in 2016 (almost 90 percent of total set-aside program funding). The down-payment assistance grants have an immediate equity-building effect. RHS and HUD administer self-help grant programs that provide opportunities for very-low and low-income home buyers to purchase subsidized homes: Program participants help construct homes in exchange for subsidies, including down-payment assistance. RHS officials told us that the home buyer’s labor serves as a down payment for the home, providing the home buyer with equity at the time of purchase. RHS’s program also includes a subsidized interest rate determined by the home buyer’s income, as well as a 33-year mortgage duration that can be extended up to 38 years, to reduce the monthly mortgage payment and make the loan as affordable as possible. HUD officials raised concerns about the extent to which down-payment assistance promotes home equity building. For example, some mortgages with down-payment assistance can be associated with higher delinquency rates. Specifically, HUD officials pointed to data indicating that FHA has experienced higher loan delinquency rates for loans with down-payment assistance. As with any homeownership-assistance programs or mortgages, the potential for home equity building requires a homeowner to sustain and pay down the mortgage. In addition to down-payment assistance, HOME, CDBG, and AHP funds can be used for buying down the mortgage interest rate. Interest-rate buy-downs have accelerated equity-building effects throughout the life of the mortgage because a higher proportion of monthly mortgage payments are applied to the mortgage principal. However, agency and enterprise officials and housing experts with whom we spoke said the down payment is the biggest barrier to homeownership, and in the current environment of low interest rates, buy-downs of interest rates are not common. In addition to federal programs, some state housing finance agencies also provide down-payment assistance grants and loans that have accelerated equity-building effects. For example, the Minnesota Housing Finance Agency provides a monthly payment loan (in addition to the mortgage) of up to $12,000 to be used for down payments or closing costs. The monthly payment loan has an interest rate equal to the rate on the borrower’s first mortgage, and the loan can be paid back over a 10-year period. According to Minnesota Housing Finance Agency officials, by making payments directly on the monthly payment loan, the borrower is effectively accelerating equity building on that part of the home purchase because of the shorter term compared to a 30-year mortgage. Federal Voucher Program Can Facilitate Equity Building HUD’s Housing Choice Voucher Program provides assistance in helping a homeowner pay for monthly mortgage and other homeownership expenses, which facilitate homeownership and equity building. Vouchers are administered locally by public housing agencies, but not all public housing agencies participate in the program. A home buyer would have to apply for a housing choice voucher with a participating public housing agency to use the funding for a mortgage instead of rent. First- time homeowners who meet income limits and receive homeownership counseling can qualify for the program. The payment assistance generally continues as long as the family resides in the home, and the maximum term for the assistance is 15 years if the home purchase is financed with a mortgage longer than 20 years. According to HUD, about 11,000 homeowners were receiving assistance from the Homeownership Voucher Program as of September 2017, about 0.5 percent of all vouchers. Federal Direct Loans Can Provide Access to Homeownership RHS and VA both offer direct loans for home purchases to eligible borrowers who may otherwise be unable to obtain financing in the private marketplace, providing access to homeownership and equity building. RHS offers direct loans to borrowers in rural areas with incomes of generally not more than 80 percent of the area median income. Loan funds can be used to build, repair, renovate, or relocate a home, or to purchase and prepare sites, including providing water and sewage facilities. RHS provided 7,089 direct loans for single-family homes in fiscal year 2016. VA provides direct home loans to eligible Native American veterans to finance the purchase, construction, or improvement of homes on federal trust land, or to refinance a prior direct loan to reduce the interest rate. According to VA, 13 direct loans were provided to Native Americans in fiscal year 2016. Options and Mechanisms That Accelerate Equity Building Present Trade-offs for Homeowners and Lenders Borrowers have options to accelerate equity building that include obtaining shorter-term mortgages, making more frequent or additional payments, or choosing a mortgage product available in the private mortgage market designed to accelerate equity building. These options accelerate equity building by affecting the key components of a mortgage—term (duration), down payment, interest rate, or payment frequency or amount. The advantages of building equity faster can include using home equity as a financial cushion in emergencies, like unexpected medical expenses. However, there are trade-offs to these options, such as higher monthly payments for shorter-term mortgages. Additionally, stakeholders identified key trade-offs and considerations in introducing new products and mechanisms for accelerating home equity building that could affect the success of the products or mechanisms. Home buyers and Homeowners May Take Actions on Their Own to Accelerate Equity Building Home buyers and homeowners may take actions on their own to accelerate home equity building. For example, home buyers can choose a 15- or 20-year mortgage rather than a 30-year mortgage. The shorter- term product will increase the relative pace of equity building. In July 2017, almost 6 percent of all new purchase mortgage originations were for 15-year fixed-rate mortgages, according to the Urban Institute. However, shorter-term loans may present trade-offs for borrowers, which we discuss later in the report. Homeowners also can make extra mortgage payments to further reduce the principal balance, which can accelerate equity building and shorten the mortgage term. For example, according to our analysis, a homeowner making an extra monthly payment of $100 on a 30-year fixed-rate mortgage for $225,000 would accelerate equity building and reduce the mortgage duration by more than 4 years (see fig. 2). Homeowners generally have the flexibility to make extra payments at their discretion and could discontinue the extra payments at any time if they need the funding for other priorities. Homeowners also can refinance their mortgage to take advantage of lower interest rates, shorter mortgage terms or both. Lower-interest and shorter-term loans can help build equity faster. About 27 percent of mortgage refinances were for 15-year fixed-rate mortgages in October 2017, according to enterprise data reported by FHFA. However, refinancing (similar to purchase loans) incurs transaction costs (see table 2). A lender may offer low- or no-cost refinancing, but likely would charge a higher interest rate in exchange for lowering or eliminating fees. Additionally, other payments might be required at closing (which would be out-of-pocket expenses unless they were financed), including upcoming mortgage insurance and property taxes. Also, homeowners who refinance to take advantage of lower interest rates could extend their mortgage term or choose to cash out some of the existing home equity, thereby eliminating the potential for accelerated equity-building effects in refinancing. See figure 3 for a comparison of how different refinancing options can affect home equity building. A Recently Introduced Mortgage Product Accelerates Equity Building through Shorter Terms and Lower Interest Rates The Wealth Building Home Loan (WBHL) is a relatively new private- sector mortgage product that incorporates a number of features specifically designed to accelerate equity building (see fig. 4). The WBHL, which has been offered commercially on a limited basis for about 3 years, has shorter mortgage terms (15 or 20 years), can have a fixed or adjustable rate, and allows the interest rate to be bought down. A lower interest rate would allocate a greater portion of each monthly payment to reduce mortgage principal and also reduce the amount of the monthly payments. Moreover, the WBHL allows for no down payment (including allowing the financing of closing costs). The no down-payment feature is designed to facilitate access to homeownership. According to lenders we spoke with who offer WBHLs, allowing for no down payment is the key feature that distinguishes the WBHLs from standard 15- or 20-year mortgage loans available in the private-sector mortgage marketplace. Consistent with what we heard from lenders, officials from Fannie Mae and Freddie Mac told us loans that do not require a down payment generally are not available in the private-sector mortgage marketplace. Additionally, because of the low or no down-payment features, lenders we spoke with who offer WBHLs typically require private mortgage insurance, which is provided by a major mortgage insurer. As shown in figure 4, the monthly mortgage payments of a WBHL can increase substantially, compared with the payments of a 30-year fixed- rate mortgage. Some lenders we interviewed offer WBHLs with the option to buy down the interest rate, and some require a minimum buy-down. One lender requires borrowers to pay 2 points (or 2 percent of the mortgage loan amount), which buys down one-half of a percentage point of the interest rate. Another lender offers a 15-year loan with an option to pay 3 points to buy down the interest rate to 1.75 percent for the first 7 years. Rates increase to 5 percent for the remaining 8 years. The lender also offers a 20-year loan with the option to pay 2 points to buy down the interest rate to 2.99 percent for the first 7 years. Rates increase to 5.25 percent for the remaining 13 years. Although the option to buy down the interest rate has been advanced as a feature that accelerates equity building, some lenders we interviewed said that borrowers tend to pay the minimum required points only, because borrowers generally prefer to pay as little cash as possible at loan origination. Additionally, some lenders and other stakeholders have said that, in a low interest-rate environment, the incentive for borrowers to buy down the mortgage interest rate is greatly reduced. A Proposed Mortgage Product May Accelerate Equity Building through an Equity Savings Account Another mortgage product that we identified during our review—the Fixed-Payment Cost-of-Funds Index (Fixed-COFI) Mortgage—has been proposed by two economists, but has not yet been offered by private- sector lenders. This type of mortgage is intended to provide another option for consumers that encourages equity building and limits exposure for borrowers and lenders to interest rate fluctuations. The Fixed-COFI would allow borrowers with little or no money down to obtain an adjustable-rate mortgage that features a fixed monthly mortgage payment and an equity savings account. Funds in the equity savings account could be used to pay down the mortgage principal, thereby accelerating home equity building. According to the economists of this proposed product, the low to no down-payment feature may help individuals with little to no savings access homeownership, particularly those who live in high-cost areas where the rent payment is comparable to a mortgage. In addition to the borrower’s fixed monthly payments, the Fixed-COFI mortgage also would determine how the borrower’s fixed payments would be allocated, including to the equity savings account. The borrower’s fixed monthly payment would be fully amortizing and be calculated based on prevailing rates for a 30-year fixed-rate mortgage at the time of loan origination. But the interest portion of the payment due to the lender would be separately calculated each month, based on a rate derived from COFI plus a gross margin to account for lenders’ costs and insurance risk premiums. Each month, the difference between the borrower’s fixed payment and interest due the lender based on the COFI rate plus a gross margin would determine if any funds from the borrower’s payment would be added to the equity savings account. The funds allocated to the equity savings account are designed to be used to pay down the principal. However, the ways in which the home equity funds could be used to pay down mortgage principal depend on the terms of each loan. If the home equity account were depleted, lenders might cover any payment shortfalls and seek insurance reimbursements. In addition, the accelerated equity-building effect of the Fixed-COFI mortgage product would rely on the historical difference between the COFI rate and 30-year fixed rate (see fig. 5). If the difference between the rates narrowed, the savings allocated to the equity savings account would lessen, and equity-building effects would be reduced. That is, in months in which the COFI rate plus the gross margin was lower than the 30-year fixed rate used to calculate the monthly payments, the difference between the COFI-based and fixed amounts would be deposited into a home equity savings account. In months in which the fixed payment would not cover the interest payment (because the COFI rate plus the gross margin is higher than the 30-year fixed rate used to calculate the fixed monthly payment), funds could be withdrawn from the equity savings account to cover any shortfall. If the equity savings account had a zero balance, the lender could seek an insurance payout. According to the economists, some details of the Fixed-COFI contract can be modified for different rules concerning refinancing and savings. For example, a borrower and a lender can agree to how and when funds in the home equity savings account could be applied to pay down the mortgage principal. However, the Fixed-COFI mortgage contract would place limits on a borrower’s options to refinance—for instance, only in the case of the loss of a job—because it is designed to protect borrowers and lenders from fluctuations in interest rates. If interest rates drop significantly, benefits from the rate decrease for a borrower with a Fixed-COFI mortgage would be limited as compared with the benefits of a borrower with a 30-year fixed-rate mortgage who refinances. For example, the additional savings from lower interest rates for the borrower with a Fixed-COFI mortgage could only be used to pay down the mortgage principal. In contrast, although refinancing has costs, borrowers with a traditional 30-year fixed-rate mortgage would be able to refinance to take advantage of the lower rate and reduce their monthly payment. They could use the resulting difference in monthly payments from the new, refinanced loan to pay down mortgage principal, build up savings, or for any other purposes. Advantages of Accelerated Home Equity Building Include a Financial Safeguard For some homeowners, building home equity faster can provide financial benefits. Home equity can serve as a financial asset to fund retirement, education expenses, or absorb financial emergencies like the loss of a job. All else being equal, having more home equity also can help sustain homeownership through a downturn in the housing market. For example, default rates are generally higher for loans with higher loan-to-value (LTV) ratios. Although some accelerated equity-building options are designed to be originated with high LTV ratios (in some cases exceeding 100 percent), the accelerated equity-building effect can lower the LTV ratio at a faster pace than for a 30-year fixed-rate mortgage. As shown in figure 6, according to our analysis, LTV ratios can converge after about 5 years for a 15-year fixed-rate mortgage with a high LTV and a 30-year fixed-rate mortgage with a higher down payment. More specifically, in about 5 years a 15-year fixed-rate loan with an LTV ratio of 103 percent at origination will reach the same LTV ratio as a 30-year fixed-rate loan with an LTV ratio of 80 percent at origination. Borrowers under both mortgage scenarios would have accrued close to 30 percent equity in about 5 years, assuming no change in the home’s value. Lenders and proponents of accelerated equity building with whom we spoke said that having substantial equity in a home provides more options for remediation in the event the homeowner encounters difficulties making mortgage payments. For instance, a lender with whom we spoke said that having more equity in a home provides a borrower with a better opportunity to refinance to get a better interest rate and also extend their loan term, both of which would lower their monthly payment. Two lenders with whom we spoke also said that accelerated equity-building options can provide financial discipline and serve as a forced savings mechanism by, for example, paying additional principal on the mortgage. In addition, proponents of accelerated equity building have suggested that homeowners with more equity at stake may have more incentive to stay in their home because they have more invested in the home. In addition to building equity, borrowers with shorter-term mortgages or those opting to make extra payments on 30-year mortgages would reduce overall loan expenditures—relative to the interest they would pay on a 30- year loan (see fig. 7). However, the overall higher mortgage payments can make these options less affordable for lower-income borrowers or limit financial flexibility, as discussed below. Trade-offs for Home Buyers and Homeowners Include Limited Access to Shorter-Term Loans and Reduced Affordability Limited Access Accelerated equity-building products, such as a 15-year fixed-rate mortgage or a WBHL, may not be accessible for all borrowers, partly due to tighter credit requirements. Officials from a state housing finance agency told us that minimum credit score requirements for some WBHLs limit access for borrowers with lower credit scores, which includes many lower-income borrowers. For example, a private mortgage insurer for WBHLs requires a minimum credit score of 680, compared with the minimum for the state housing finance agency of 640 for 30-year fixed- rate mortgages. The average score for WBHLs insured by the private mortgage insurer is 749. Moreover, requirements for a minimum debt-to-income ratio may also limit lower-income borrowers’ ability to access WBHLs or 15-year fixed- rate loans. According to a private mortgage insurer, the average income of borrowers for WBHLs it insures is 177 percent of county median income. As mentioned previously, the QM rule generally requires home buyers to have a debt-to-income ratio of 43 percent or less. As we previously reported, although QM regulations are not expected to significantly affect the overall mortgage market, some researchers have estimated that QM regulations could adversely affect certain lower- income home buyers, particularly those living in high-cost areas. The higher monthly payments of shorter-term loans can result in debt-to- income ratios significantly above the 43 percent limit, as illustrated in table 3. In areas where housing costs are high, research suggests that lower- income home buyers are more likely to have high debt-to-income ratios. Higher monthly payments for accelerated equity-building mortgages could make some of these borrowers ineligible for those types of loans, or essentially limit those borrowers to significantly smaller loans, as discussed in the following section. Reduced Affordability and Financial Flexibility The biggest barrier to homeownership is affordability, which includes having enough savings for a down payment as well as sufficient monthly income to sustain a mortgage, according to agency officials and stakeholders with whom we spoke. For example, 53 percent of adults were unable to save any money in 2016 and 13 percent of adults had difficulty paying their bills at least once in 2016 because of income volatility, according to the Federal Reserve. For the same loan amount, the monthly payments of a 15-year mortgage can be more than 40 percent greater than the monthly payments of a 30-year mortgage, depending upon the current market interest rates. The higher monthly payments may make shorter-term loans unaffordable for many low- income home buyers or leave borrowers with less discretionary income to cover other obligations, including paying off higher-interest debt, putting some of them at greater risk of defaulting on monthly mortgage payments. The higher monthly payments of shorter-term loans thus reduce homeowners’ financial flexibility. In contrast, experts and stakeholders highlighted the greater flexibility a 30-year mortgage affords homeowners, including for situations where individuals may experience instability or fluctuations in their income. For example, though some home buyers may have adequate income over the course of a year to afford monthly mortgage payments, fluctuations in monthly income can affect a homeowner’s ability to sustain a higher monthly mortgage payment. However, a 30-year fixed-rate mortgage may enable a homeowner to make additional payments to build equity faster and still maintain a lower monthly payment than a 15-year mortgage. As shown in the scenario in figure 7 above, a homeowner could pay off a 30-year mortgage in 15 years by making additional monthly payments. Reduced Home Purchasing Power The higher monthly payment required of a shorter-term mortgage can reduce a home buyer’s purchasing power. As seen in table 4, a longer- term mortgage allows for a substantially higher home purchase price for the same monthly payment for principal and interest. Borrowers are likely to qualify for smaller loan amounts for shorter-term mortgages because of the effect of the higher monthly payments (of shorter-term mortgages) on their debt-to-income ratio. Some proponents of accelerated equity-building loans advertise that the monthly payment amounts of shorter-term and 30-year fixed-rate mortgages are comparable, with minimal loss in purchasing power. This might be the case if the loan amount for the shorter-term mortgage were less than the loan for the 30-year fixed-rate mortgage, as illustrated in table 4. However, determining loss of purchasing power based on the monthly payments of two mortgages with different loan amounts may not provide an equivalent comparison. Potential Lower Lifetime Wealth Shorter-term mortgages can reduce lifetime wealth. This is because the difference between the higher monthly payments and the monthly payments of a 30-year mortgage could have been invested elsewhere to produce a higher return—assuming an individual has the financial knowledge and discipline to invest the funds. The higher required monthly payments of a 15-year mortgage can ensure a larger investment in home equity. However, some research suggests that, depending on market conditions and the risk appetite of a homeowner, purchasing a house with a 30-year fixed-rate mortgage can provide a higher lifetime return on investment compared to a 15-year fixed-rate mortgage because the difference between the monthly payments can be invested at a rate of return that likely would be higher than the difference in mortgage interest rates between 30- and 15-year mortgages. In addition, homeownership may not always be the most effective means of building household wealth. For example, in some circumstances individuals may achieve greater household wealth through renting rather than buying a home. Individuals for whom rental payments would be less than mortgage payments for a comparable home can invest the difference and build greater wealth—if the return on their investment exceeded the return associated with the appreciation of the value of a home. However, factors such as an individual’s financial literacy and risk tolerance, and overall market conditions can affect the success of any investment strategy, including investing in a home or in any alternatives. Trade-offs for Lenders Include Market Uncertainty For lenders, shorter-term mortgages generally reduce credit risk—the likelihood of loss with default—compared with longer-term loans. In addition, lenders with whom we spoke said that borrowers choosing shorter-term loans (such as WBHLs) generally have good credit and high incomes, further reducing credit and default risk. However, market uncertainties related to the lack of a secondary market and performance data could limit lenders’ willingness to offer accelerated equity-building products. Lack of a Secondary Market Products like WBHLs are not currently eligible for purchase by Fannie Mae and Freddie Mac. According to Fannie Mae and Freddie Mac, WBHLs are not currently traded in the secondary mortgage market because of factors such as the low volume of transactions and the high LTV ratio. Lenders with whom we spoke who offer WBHLs generally have been holding the loans in their own portfolio, which can expose them to credit risk and interest-rate risk. Some of the lenders told us they only offer adjustable-rate WBHLs, to reduce interest-rate risk. But homeowners could experience a rate shock when the interest rate adjusts. For example, according to our analysis, if a WBHL for $250,000 adjusted the interest rate after 7 years, the monthly payment could increase by more than $200 (13 percent). The rate adjustment also might increase credit risk for lenders, because some borrowers then might be less able to sustain the monthly payments. Some lenders may be unwilling to take on these risks, which could limit the availability of accelerated equity-building mortgages in the market. However, lenders with whom we spoke have been exploring options to sell loans that have “seasoned”—for example, after the LTV ratio of the loan reached 97 percent—on the secondary market. Risk Premium Mortgages with LTV ratios of 96.5 percent or more (those that have 3.5 percent or less in down payment) also would be ineligible for some federal guarantee programs. Generally, high-LTV loans have a greater risk of default, and lenders with whom we spoke who offer WBHLs all require private mortgage insurance for those loans. Lenders and private mortgage insurers may price WBHLs at a premium—for example, through higher fees, interest rates, or insurance premiums—to account for the risk, which may add to the costs of monthly payments and make these mortgages less affordable for some borrowers. Lack of Performance Data Because WBHLs are new (introduced in 2014) to the marketplace, there are not enough data on loan performance to adequately assess payment delinquency and default risk. The number of lenders currently offering WBHLs is limited. According to the American Enterprise Institute, about $100 million of WBHLs have been originated since 2014. Mortgage insurers with whom we spoke provided a similar estimate. Lenders told us that the performance of their WBHLs is strong but may not offer a meaningful indicator of future performance if the loans were to become more widely available (because WBHLs currently tend to attract less-risky borrowers). According to lenders and housing experts with whom we spoke, performance data on similar loans, such as fixed-rate 15-year mortgages, cannot be readily used to project performance for WBHLs because WBHLS are not strictly comparable (they have higher LTV ratios). Other Trade-offs Include Limited Promotion and Adoption Due to Product Complexity Stakeholders, including agency officials, also identified key trade-offs and considerations in introducing new products and mechanisms to accelerate equity building, such as how product complexity and reduced market liquidity could affect the success and the costs to borrowers of the products or mechanisms. These trade-offs and considerations apply to proposed products such as the Fixed-COFI as well as to actions or mechanisms for accelerating equity building, such as making mortgage payments on a biweekly basis and paying off a percentage of the loan principal in a shorter term (such as financing 20 percent of the principal in 5 years). Some stakeholders said that new products that have unfamiliar or complex features, such as the Fixed-COFI mortgage’s underlying adjustable rate and equity savings account, could be difficult for lenders, borrowers, and investors to understand, which could limit the promotion and adoption of such products. In addition, administering new products or mechanisms to accelerate equity building could have additional complications, such as how to schedule and credit biweekly payments. For example, lenders or servicers may not have a structure in place to properly credit additional payments on a biweekly basis and may hold the extra payment until the end of the month, negating the accelerated equity- building effect of the extra payment. Moreover, a few stakeholders said that lenders or servicers may charge additional fees for processing biweekly mortgage payments. Stakeholders and agency officials also noted that any new mortgage product would not (at least initially) be eligible for securitizing and trading in the secondary market. As a result, a new product would not be as liquid as current products securitized and sold in the secondary market by Fannie Mae or Freddie Mac, such as 30-year fixed-rate mortgages. Because of the lack of market liquidity for new products, lenders may charge a premium, making the products less affordable for lower-income borrowers. Agency Comments We provided a draft of this report to HUD, FHFA—and FHFA also provided copies to Fannie Mae and Freddie Mac, FHLBanks, Agriculture, and VA for their review and comment. HUD, FHFA, FHLBanks, and Agriculture provided technical comments on the report draft, which we incorporated where appropriate. We are sending copies of this report to appropriate congressional committees, the Secretary of HUD, the Director of FHFA—who provided copies to the President and Chief Executive Officer of Fannie Mae and the Chief Executive Officer of Freddie Mac, the President of the FHLBank of Des Moines (coordinating for the FHLBanks), the Secretary of Agriculture, the Secretary of Veterans Affairs, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-8678 or GarciaDiazD@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix III. Appendix I: Objectives, Scope, and Methodology This report describes (1) how federal homeownership assistance programs affect home equity building, and (2) options, including private- sector mortgage products, through which borrowers can accelerate home equity building and the trade-offs of these options for both borrowers and lenders. We define accelerated equity building as any mortgage product or feature that accelerates the pace of principal reduction on a mortgage debt, relative to a 30-year fixed-rate mortgage. We used the 30-year fixed-rate mortgage as our point of comparison because it is the most common type of mortgage product and represents the market standard. To describe how federal homeownership assistance programs affect home equity building, we reviewed relevant federal statutes, regulations, and agency program policies and guides and other resources to identify relevant homeownership assistance programs from the Departments of Housing and Urban Development (HUD), Veterans Affairs (VA), and Agriculture (USDA); and Fannie Mae, Freddie Mac, and the Federal Home Loan Banks (collectively, the enterprises). We reviewed prior GAO reports on federal homeownership assistance programs and the U.S. housing finance system. We also reviewed relevant academic papers and literature discussing homeownership and equity building. We interviewed agency and enterprise officials to discuss the relevant homeownership assistance programs and policy goals, including the extent to which products or mechanisms used in the programs affect or accelerate home equity building, and the role of the secondary mortgage market in providing market liquidity for new mortgage products. In addition to federal agencies and the enterprises, we interviewed officials from two state housing finance agencies. Some stakeholders we interviewed recommended the two state housing agencies because the agencies likely placed a greater focus on accelerating home equity building. To describe the options (or mortgage products) borrowers have to accelerate home equity building, including any trade-offs, we used databases such as ProQuest and searched for and reviewed papers and literature published from 2007 to 2017 by individuals who discussed options to accelerate home equity building. We also attended two housing conferences and met with housing experts and stakeholders from academia, housing advocacy organizations, and industry, including mortgage lenders and insurers, selected because they made proposals to increase homeownership or build home equity faster, wrote on homeownership issues, were recommended by government officials, or were involved in providing mortgage products designed to accelerate equity building. From interviews with industry stakeholders and housing conferences we attended, we identified two products: (1) the Wealth Building Home Loan (WBHL), which has been introduced in the marketplace, and (2) the Fixed-Payment Cost-of-Funds Index (COFI) Mortgage, which has been proposed but is not currently offered by any lenders. We reviewed and analyzed relevant academic papers and literature on the advantages and trade-offs of options to accelerate equity building. We also conducted interviews with academics, experts, industry stakeholders (including mortgage lenders and insurers), and organizations to discuss advantages and trade-offs of accelerated equity-building products, and the role of the secondary market in providing market liquidity for new mortgage products. We selected academics, experts, and industry stakeholders and organizations who proposed accelerated equity-building mortgage products, had written on homeownership and wealth building issues, or whom officials of federal agencies and the enterprises or our other interviewees recommended. We also attended housing conferences, which provided additional suggestions for publications to review and academics and stakeholders to interview. Furthermore, to illustrate methods to accelerate home equity building and compare the effects of different mortgage products on home equity building, we developed hypothetical mortgage scenarios. For the mortgage scenarios, we used Excel’s payment function to calculate the amortization schedule of the mortgages in our hypothetical scenarios. The payment function is a standard formula that calculates monthly payment schedules based on inputting interest rate, number of payment periods over the life of a mortgage, and the present value of the mortgage. The scenarios we developed were only several possible scenarios out of the many that we could have chosen. We identified specific mortgage features in papers and literature by individuals who proposed mortgage products designed to accelerate home equity building. For illustration purposes, we used an average of the monthly interest rates published in Freddie Mac’s Primary Mortgage Market Survey for September and October 2017, as well as current market rates advertised by private mortgage lenders, to inform our selection of interest rates for our scenarios. We conducted this performance audit from January 2017 to March 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: CoreLogic Home Equity Data, by State This appendix provides regional data on equity building that we obtained from CoreLogic. CoreLogic is a publicly traded company that provides data, analytics, technology, and services related to the mortgage industry, among other things. The data in figure 8 show the percentage of homeowners in each state who have 20 percent equity or less in their homes. The level of home equity can be affected by a number of factors, including the age of the loan, the amount of principal paid down, and home market values. We did not assess the reliability of CoreLogic’s data. Appendix III: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Andrew Pauline (Assistant Director), Kun-Fang Lee (Analyst in Charge), Steve Brown, Raheem Hanifa, Jeff Harner, Jill Lacey, Barbara Roesmann, Jessica Sandler, MaryLynn Sergent, Jena Sinkfield, Anne Stevens, and Jim Vitarello made key contributions to this report.

Why GAO Did This Study The federal government has a number of programs to help increase access to affordable homeownership for first-time buyers and lower-income households, including programs that provide guarantees for certain types of mortgages and funding that can be used for down-payment assistance. Generally, homeowners can build home equity by making payments on a mortgage to reduce the outstanding principal (assuming home value does not depreciate). Recently, there has been interest in mortgage products that accelerate home equity building. GAO was asked to explore options for building equity through homeownership. This report discusses (1) how federal homeownership assistance programs affect home equity building; and (2) options, including private-sector mortgage products, through which borrowers can accelerate home equity building and the trade-offs of these options for both borrowers and lenders. GAO analyzed relevant laws and program guidance of federal homeownership assistance programs. GAO attended housing conferences and interviewed relevant federal and state agency officials, academics, and industry stakeholders, including mortgage insurers and lenders, to identify existing and proposed accelerated equity-building products and mechanisms and to better understand the benefits and trade-offs of accelerated equity building. GAO also developed examples of mortgage scenarios to illustrate the trade-offs of accelerated equity building. Federal agencies provided technical comments, which were incorporated where appropriate. What GAO Found Federal homeownership assistance programs generally are not designed to accelerate equity building (home equity is the difference between the value of a home and the amount owed on a mortgage). For example, programs that offer grants for down-payment assistance can provide a one-time boost to home equity. However, these programs are not specifically designed to accelerate equity building—that is, increasing the pace of paying off principal more quickly than would be the case with a 30-year fixed-rate mortgage. Instead, the focus of federal programs is on providing affordable access to homeownership, including through grants, loans, and mortgage insurance or guarantees. For instance, federal mortgage insurance programs help provide market liquidity by protecting lenders from losses, in turn increasing access to credit and homeownership, and ultimately, the opportunity for equity building for home buyers. Borrowers have options to accelerate equity building that include obtaining shorter-term mortgages, making more frequent or additional payments, or choosing a mortgage product designed to accelerate equity building. For example, a mortgage product introduced by private lenders in 2014—the Wealth Building Home Loan (WBHL)—has features designed to accelerate equity building, including shorter terms (15 or 20 years) and the option to buy down the interest rate. The product also allows for no down payment. However, these products have trade-offs, including the following: Shorter-term loans build home equity (in terms of principal reduction) at a faster rate, but require higher monthly payments (see fig.). Payments for a 15-year fixed-rate mortgage can be more than 40 percent higher than for a 30-year fixed-rate mortgage. Higher payments may make mortgages less affordable or limit access for lower-income borrowers. For example, higher payments may result in a higher debt-to-income ratio for some home buyers, which may prevent them from qualifying for a mortgage unless they buy a less expensive home. In contrast, all else equal, loans with a shorter term generally have reduced credit risk—the likelihood of a home buyer defaulting on a mortgage—for lenders. Note: Monthly mortgage payments do not include property tax or any type of insurance. Interest rates used are generally consistent with market rates in September and October 2017.

gao_GAO-18-250

gao_GAO-18-250_0

Background This section provides information on (1) BLM headquarters, state, and field offices; (2) the lifecycle of oil and gas wells; (3) BLM’s bonding regulations; and (4) BLM’s 2012 well review and 2013 bond adequacy review policies. BLM Headquarters, State, and Field Offices BLM is responsible for issuing leases for private entities to develop oil and gas resources on and under roughly 700-million acres of (1) BLM land, (2) other federal agencies’ land, and (3) private land where the federal government owns the mineral rights. According to BLM, approximately 32-million acres were leased for oil and gas operations at the end of fiscal year 2015. BLM also oversees oil and gas operations on 56-million acres of Indian lands. BLM administers its programs through its headquarters office in Washington, D.C.; 12 state offices; 38 district offices; and 127 field offices. Of these, 10 state offices and 33 field offices manage oil and gas programs, and these are located primarily in the Mountain West, the center of much of BLM’s oil and gas development. BLM headquarters develops guidance and regulations for the agency, and the state, district, and field offices manage and implement the agency’s programs. Because BLM has few acres of land in the eastern half of the United States, the Eastern States State Office, in Washington, D.C., is responsible for managing land in 31 states, and the remaining state offices generally conform to the boundaries of one or more states. Figure 1 shows the boundaries of the 12 BLM state offices. Lifecycle of Oil and Gas Wells Once operators obtain federal oil and gas leases and drill wells, those wells can be actively producing, inactive, or reclaimed. An orphaned well is a well that BLM determined has no responsible or liable party and for which there is insufficient bond coverage for reclamation. This situation may occur, for example, when an operator has declared bankruptcy. Shut-in and temporarily abandoned wells are examples of types of inactive wells that can become orphaned. Shut-in wells are physically and mechanically capable of producing oil or gas in paying quantities or capable of service use. For example, an operator may put a well in shut- in status if it has not been connected to a sales line or the line is too far away and it is not economical to connect to at this time. Temporarily abandoned wells are another type of inactive well that is not physically or mechanically capable of producing oil or gas in paying quantities but that may have value for a future use. Figure 2 depicts the lifecycle of oil and gas wells overseen by BLM. BLM’s Bonding Regulations The Mineral Leasing Act of 1920, as amended, requires that federal regulations ensure that an adequate bond is established before operators begin preparing land for drilling to ensure complete and timely reclamation of the land. Accordingly, BLM regulations require operators to submit a bond to ensure compliance with all of the terms and conditions of the lease, including, but not limited to paying royalties, plugging wells, and reclaiming disturbed land. BLM regulations generally require operators to have one of the following types of bond coverage: individual lease bonds, which cover all of an operator’s wells under one lease, and the minimum amount is set at $10,000; statewide bonds, which cover all of an operator’s leases in one state, and the minimum amount is set at $25,000; or nationwide bonds, which cover all of an operator’s leases in the United States, and the minimum amount is set at $150,000. BLM can accept two types of bonds: surety bonds and personal bonds. A surety bond is a third-party guarantee that an operator purchases from a private insurance company approved by the Department of the Treasury. The operator is required to pay a premium to the surety company to maintain the bond. These premiums can vary depending on various factors, including the amount of the bond and the assets and financial resources of the operator. If operators fail to reclaim the land they disturb, the surety company can either pay BLM the amount of the bond to help offset reclamation costs, or in some circumstances, BLM may allow the surety company to perform the required reclamation. A personal bond must be accompanied by one of the following financial instruments: certificates of deposit issued by a financial institution whose deposits are federally insured, granting the Secretary of the Interior authority to redeem it in case of default in the performance of the terms and conditions of the lease; cashier’s checks; negotiable Treasury securities, including U.S. Treasury notes or bonds, with conveyance to the Secretary of the Interior to sell the security in case of default in the performance of the lease’s terms and conditions; or irrevocable letters of credit that are issued for a specific term by a financial institution whose deposits are federally insured and meet certain conditions. If operators fail to reclaim the land they disturb, BLM will redeem the certificate of deposit, cash the check, sell the security, or make a demand on the letter of credit to pay the reclamation costs. BLM’s 2012 Well Review and 2013 Bond Adequacy Review Policies In response to our previous recommendations that BLM develop a comprehensive strategy to improve monitoring agency performance in conducting well reviews and bond adequacy reviews, BLM issued a 2012 well review policy and a 2013 bond adequacy review policy. These policies contain directives for conducting reviews when wells and bonds meet certain criteria. The well review policy directs: that field office officials evaluate every shut-in well at least once every 5 years; that field office officials review all wells that have been inactive for 25 years or longer and that have no anticipated beneficial use by March 29, 2013; that if field office officials determine that there are wells that are not capable of producing oil or gas in paying quantities or have no beneficial use, officials are to send the operator a written order directing the operator to demonstrate that these wells are capable of producing oil or gas in paying quantities or have a future beneficial use, or the operator is to submit plans to reclaim the wells; that each state office submit to BLM headquarters a consolidated annual report recording well reviews; and that the annual report identify the leases that were reviewed and the wells that were reviewed on each lease, and describe what follow-up action the field office official conducting the review performed. The bond adequacy review policy directs: that field offices perform bond adequacy reviews on all bonds at least once every 5 years or whenever a bond review is warranted; that field offices verify and tie all federal wells to their appropriate bond number and enter bond information and bond adequacy review data into AFMSS; that field offices perform adequacy reviews on all bonds using specific instructions and a worksheet that assigns points for three risk factors: (1) status of wells covered by the bond (share of inactive wells, deep wells, and wells with marginal production); (2) operator-specific compliance history; and (3) reclamation stewardship diligence; that if the field office official performing the review determines that the bond amount is insufficient, the official is to take the necessary steps to determine the appropriate bond amount and increase the bond; that if the bond being reviewed is a statewide or nationwide bond, field offices are to review the wells within their field office jurisdiction; and that each BLM state office with an oil and gas program submit a semi- annual bond adequacy review report to BLM headquarters. BLM’s Actual Reclamation Costs and Potential Oil and Gas Well Liabilities Have Likely Increased, but the Agency Does Not Systematically Track These BLM’s actual costs incurred to reclaim orphaned wells and potential liabilities have likely increased for fiscal years 2010 through 2017 based on our analysis of available information. Precisely how the agency’s actual reclamation costs and potential liabilities have changed is unclear because BLM does not systematically track them at an agency-wide level. BLM headquarters officials we interviewed told us that they did not have any information on actual costs incurred to reclaim orphaned wells and stated that BLM’s data systems were not designed to track incurred reclamation costs. In addition, AFMSS provides a snapshot of orphaned wells as identified at the time that the data are queried and does not provide data for prior time periods. Because BLM headquarters does not record actual reclamation costs incurred at an agency-wide level, we requested documentation for the reclamation costs incurred by 13 selected BLM field offices for fiscal years 2010 through July 2017. This documentation identified about $2.1 million in reclamation costs incurred over this period, or an average of about $267,600 per year by these 13 field offices. We estimate that total actual reclamation costs for all field offices are likely to be higher than this amount as other field offices may have also reclaimed orphaned wells during this period. In January 2010, we found that, for all field offices across the agency, BLM spent about $3.8 million from fiscal years 1988 through 2009, or an average of about $171,500 per year. Comparing the average costs incurred by the 13 selected field offices to the data we previously reported demonstrates that actual total reclamation costs incurred have likely increased since 2010. In addition to actual costs increasing, potential liabilities are also likely to have increased, though BLM does not systematically track information on potential liabilities that might result from an increase in the number of orphaned wells. Potential liabilities include costs that the agency may incur to reclaim wells that operators fail to reclaim. We believe these costs have also increased because the number of known orphaned wells on federal and Indian lands managed by BLM has increased. We identified changes in the number of known orphaned wells since we last reported on this matter in January 2010. In January 2010, we found that BLM had identified and was managing 144 orphaned wells. Over half of those 144 wells (75) were still identified in AFMSS as orphaned as of July 2017, and the total number of identified orphaned wells on federal lands had increased from 144 to 219. Also, BLM officials from the 13 selected field offices identified about $46.2 million in estimated potential reclamation costs associated with orphaned wells and inactive wells that officials deemed to be at risk of becoming orphaned. Also concerning potential liabilities, our analysis of AFMSS data and OGOR production data through September 2016 found that BLM managed about 15,600 inactive wells, of which over 1,000 were inactive for 25 years or more. In contrast, a document provided to us by BLM headquarters indicates 325 wells had been inactive for 25 years or more as of around 2017. This document summarizes data from AFMSS queries conducted by BLM field and state offices at various times from 2013 through 2014 and queries conducted at various times from 2016 through 2017. BLM officials told us that this difference could be because AFMSS reports sometimes return conflicting data since the reports draw from current and historical statuses of wells from both AFMSS and OGOR. We combined AFMSS and OGOR data to identify the number of inactive wells because although BLM records the total number of wells on federal lands over time—a rough indicator of how potential reclamation costs may change—the agency does not systematically record more specific types of wells that may be at higher risk of becoming orphaned, such as inactive wells or wells that have been inactive for 25 years or more. Moreover, we identified inconsistencies between the data and the document provided to us by BLM headquarters summarizing the data. For example, BLM’s summary document did not include one state office, even though the data include that state office as having two wells that were inactive for 25 years or more in 2014. BLM’s summary document states that there had been a reduction in the number of wells that were inactive for 25 years or more between the times of the two data queries. However, because BLM does not systematically track the number of inactive wells, in particular those wells that are at high risk of becoming orphaned, the agency does not know how its potential liabilities may be changing. These liabilities include wells inactive for 25 years or more. Although we were unable to determine the full extent of the increase in BLM’s potential liabilities because BLM does not have the data needed for such an analysis, other factors also suggest such an increase. For example, there has been an increase in oil and gas development on federal lands, and therefore, there is the potential for an increase in the total number of wells on federal lands at risk of becoming orphaned and needing to be reclaimed in the future. BLM’s portfolio of oil and gas wells on federal lands has changed over the years, based on overall trends in the oil and gas industry. According to AFMSS data provided by BLM, the total number of wells on federal lands that are capable of production increased along with rising oil and gas prices, from about 89,600 wells in fiscal year 2010 to peaking to about 94,800 wells in fiscal year 2014. As oil and gas prices declined starting in 2014, the total number of wells capable of production also declined to about 94,100 wells in fiscal year 2016. In addition, declining oil and gas prices (by nearly half from 2010 through 2017) have placed financial stress on oil and gas operators, thereby increasing bankruptcies and the risk of wells becoming orphaned. For example, coalbed methane—natural gas extracted from coal beds—was economical to produce when natural gas prices were higher and thousands of coalbed methane wells were drilled on federal lands. However, coalbed methane production has declined because the spread of shale gas production has driven down natural gas prices. Officials we interviewed in one BLM field office told us that the drop in natural gas prices contributed to an increasing number of bankruptcies for operators of coalbed methane wells. Our analysis of AFMSS data suggests that there were thousands of inactive coalbed methane wells as of October 2017. To the extent that market conditions remain unfavorable for coalbed methane production, BLM’s potential future reclamation costs may increase if any operators of these wells go bankrupt or are otherwise unwilling or unable to pay the full costs of reclamation, leaving these wells orphaned. According to federal internal control standards, management should use quality information, which should be complete, to achieve the entity’s objectives. However, BLM does not systematically or comprehensively track the agency’s actual costs incurred to reclaim orphaned wells and the information necessary to determine potential liabilities, including indicators of potential future reclamation costs, such as the number of inactive wells, orphaned wells, and estimates of reclamation costs for orphaned wells. BLM headquarters officials said that they sometimes check AFMSS to see how many orphaned wells there are, but without doing so systematically and recording the results of these checks, it is not possible to determine how the agency has been making progress in managing the number of orphaned wells. EPAct 2005 requires that the costs of reclaiming orphaned wells be recovered from persons or entities providing a bond or other financial assurance. Without systematically and comprehensively tracking actual reclamation costs incurred and the information necessary to determine potential liabilities including the numbers of orphaned wells and inactive wells over time, BLM cannot ensure that it has sufficient bond coverage or other financial assurances to minimize the need for taxpayers to pay for the costs of reclaiming orphaned wells. The Extent to which BLM Implemented Its Well Review Policy and Bond Adequacy Review Policy Directives Is Unclear The extent to which BLM has implemented its well review policy and bond adequacy review policy is unclear. Specifically, we were unable to fully assess the extent to which BLM’s field and state offices have implemented directives included in these policies because of inconsistent well review information, inaccurate well and bond data in AFMSS, and inadequate monitoring of well and bond policies’ implementation. Inconsistent well review information. We were unable to fully assess the extent to which BLM implemented some directives in the well review policy because the well review information reported by field offices differed across the agency. For example, officials we interviewed at the 13 selected BLM field offices had different understandings of what specific actions constitute a well review, and therefore differed in their understanding of which wells were to be included in the annual reports for documenting well reviews. Specifically, officials from 11 out of 13 selected field offices told us that a well review consisted of actions—such as reviewing a well’s status, conducting a physical inspection, and providing additional notices or letters to the well operator when a well is inactive. Officials in 2 other field offices told us that while they conduct similar actions, they consider the sole action of correcting data on a well’s status to constitute a well review. For example, a BLM official told us that one reported well review was conducted on a well that had been reclaimed in 1986 but that was not noted in AFMSS. The official told us that following this well review, they corrected the well status in AFMSS and noted that this well should not have been on the list of wells to review. While correcting well data helps improve the accuracy of AFMSS, when some offices count such corrections as well reviews and others do not, this variance results in inconsistent information in BLM’s annual well review reports. Such inconsistencies in what counts as a well review may be the result of a lack of clarity in BLM’s well review policy that does not specify what constitutes a well review. Unlike the bond adequacy review policy, which provides instructions to field offices on how to conduct a bond adequacy review and directs field offices to use a specific worksheet to calculate bond adequacy, the well review policy does not contain specific instructions on what actions field offices are to take to conduct a well review, such as how to count reviews or report them. A January 2018 report by the Department of the Interior’s Office of Inspector General (OIG) similarly found that BLM’s well review policy does not specifically outline how to conduct and document reviews of shut-in wells (shut-in wells, as noted earlier, are inactive wells that are physically and mechanically capable of producing oil or gas in paying quantities or capable of service use). Under federal standards for internal control, management should design control activities to achieve objectives and respond to risks; such activities include appropriate documentation of internal control in management directives, administrative policies, or operating manuals. Without developing and communicating specific instructions outlining what actions constitute a well review for annual- reporting purposes, BLM cannot have reasonable assurance that its field offices are conducting and reporting on well reviews in a consistent manner. Inaccurate well and bond data in AFMSS. Our ability to assess the extent to which BLM implemented its well review and bond adequacy review policies was impeded by inaccuracies in certain AFMSS data. BLM officials told us that some of the data in AFMSS on wells and bonds were not reliable. For example, BLM officials told us that there may be discrepancies between the bonds listed in AFMSS and the bonds listed in the Bond and Surety System, which is BLM’s official database for all oil and gas bonds. Officials told us that bonds may be missing from AFMSS because BLM field offices are responsible for manually entering the bond number from the Bond and Surety System into AFMSS. In addition, AFMSS data we reviewed contained other inaccuracies. Specifically, the data we reviewed contained future dates for when wells were completed, or capable of production, when some wells last changed statuses, and when some well reviews were reportedly conducted. BLM officials told us that AFMSS allows users to enter future dates, which can result in inaccurate data. Having inaccurate dates for wells’ statuses and wells’ reviews is problematic because it means it is not possible to assess whether reviews are being conducted as directed by BLM policy. For example, BLM’s well review policy directs field offices to review each shut-in well every 5 years. BLM’s performance against this directive cannot be assessed without reliable information on when wells become shut-in and when well reviews are conducted. In written responses to our request for information, BLM officials stated that AFMSS has some edit checks, but the accuracy of the data entered into AFMSS is dependent on field office officials responsible for data entry. BLM officials stated that AFMSS has some electronic safeguards, such as certain number fields only accepting numbers. In addition, AFMSS has dropdown menus and checkboxes to narrow the parameters of certain data being entered. However, there are no edit checks to prevent field offices from inputting future status dates. In addition, BLM’s data administration and management handbook establishes that data stewards are to, among other things, establish target quality levels, data quality plans (including audits and other quality assurance steps), and certify the quality of the data. BLM officials stated that they have national level AFMSS data stewards and information-technology data stewards. However, BLM officials stated that the agency has not defined AFMSS target quality levels and did not provide any data quality plans. Officials stated that BLM headquarters conducts annual data reviews and will periodically review sample well files to detect data inconsistencies and errors. In addition, BLM officials stated that field offices are responsible for certifying the accuracy of the data they enter into AFMSS, and BLM headquarters is responsible for providing oversight. However, BLM headquarters officials did not provide documentation of any data certifications or data reviews, raising concerns over the extent of this oversight. Under federal standards for internal control, management should design control activities, including control activities used in information processing, to achieve objectives and respond to risks. Examples of such control activities include: conducting edit checks of data entered, accounting for transactions in numerical sequences, and comparing file totals with control accounts. Without taking steps to improve AFMSS data quality, such as by conducting more edit checks and having data stewards certify the quality of the data, BLM cannot have reasonable assurance that management has the accurate information it needs to track whether field offices are conducting well and bond adequacy reviews as intended. In its January 2018 report, the OIG found similar issues related to the accuracy of AFMSS data. Specifically, the OIG found that AFMSS data were unreliable due to inaccurate well status information. The OIG also found that BLM officials update AFMSS manually during a well review or as needed, as opposed to automating the data, meaning that information about the status of individual wells in AFMSS and data used for BLM’s annual well report are not timely. The OIG recommended that BLM develop and implement a quality control process to identify inaccurate or incomplete data in AFMSS. BLM concurred with this recommendation. Inadequate monitoring of well and bond policies’ implementation. BLM headquarters has taken some actions to monitor the implementation of its well and bond adequacy review policies across the agency, but its efforts have been limited, and the agency cannot ensure that its policy directives have been fully implemented. For example, BLM headquarters officials told us that headquarters relies on national well review and bond adequacy review reports to monitor the extent to which field offices are conducting well and bond adequacy reviews. These well and bond adequacy review reports provide some information on how BLM field offices conducted their reviews during a given year, but the reports as previously mentioned above have data limitations and do not consistently record a field office’s progress in meeting the policies overall. For example, annual well review reports list the wells field offices reviewed in a given year, but do not compare this statistic to a list of the wells that each field office should have reviewed. Similarly, field offices’ bond adequacy review reports list the bonds that the field offices reviewed in a given year. However, the reports do not compare the bonds reviewed to a list of bonds each field office should have reviewed. In addition, our analysis of 58 selected bonds reported as reviewed across the 13 selected field offices found that 4 bonds—about 7 percent—were not reviewed, even though field offices had reported that they had conducted the reviews. The bond adequacy review policy directs field offices to review all bonds once every 5 years or whenever a bond review is warranted. Therefore, the bond adequacy review reports on their own provide insufficient information for BLM headquarters to monitor progress about whether field offices are fully implementing the directive. We also identified discrepancies between the annual well review and semi-annual bond adequacy review reports that state offices submitted to BLM headquarters and the information in headquarters’ national summary, which consolidates the state office information. These discrepancies limit the usefulness of the national summary for monitoring the extent to which field offices are conducting well and bond adequacy reviews as directed by the policies. For example, 3 out of 10 state offices reported a different number of bond adequacy reviews completed in their fiscal year 2016 state reports than what was reported in BLM’s fiscal year 2016 national report. Similarly, 6 out of 9 state offices reported a different number of completed well reviews in their fiscal year 2016 state report than what was reported in BLM’s fiscal year 2016 national report. Similarly, the OIG’s January 2018 report found that BLM can only report its progress in reviewing wells that have been inactive for 25 years or more by using field office spreadsheets, coupled with AFMSS data. The report stated that using spreadsheets and AFMSS data have made it difficult, however, for BLM to demonstrate proper oversight. BLM’s headquarters officials had to ask state office officials how many wells had been reviewed and then had to summarize those results in a spreadsheet. The OIG recommended that BLM monitor and track reviews of shut-in wells in a management system. BLM concurred and stated that AFMSS and an update to AFMSS that is under development were the appropriate databases for monitoring and tracking well reviews. Overall, we found that BLM’s current approach to monitoring the agency’s progress in implementing its well and bond adequacy review policies has been limited. We reviewed leading practices for monitoring the implementation of agency policies. These practices call for, among other things: (1) periodically collecting and analyzing data on performance indicators, (2) establishing procedures for ensuring the quality of data on performance indicators, (3) documenting that monitoring plans were executed, and (4) considering performance information in making management decisions. Without taking actions to strengthen its approach to monitoring, such as collecting and analyzing data on performance indicators and ensuring the quality of those data, BLM’s ability to assess the extent to which field offices are reviewing all inactive wells and determining the adequacy of all bonds is limited. Agency Officials and Stakeholders Identified Several Challenges BLM Faces in Managing Its Potential Oil and Gas Well Liabilities According to BLM officials and stakeholders we interviewed, BLM faces several challenges in managing its potential liabilities. In particular, BLM officials and stakeholders told us that one challenge in managing BLM’s potential liabilities was identifying and managing shut-in wells and preventing them from becoming orphaned. Another challenge identified was limited resources and competing priorities in reclaiming orphaned wells. Other challenges to managing BLM’s potential liabilities include difficulties in reviewing nationwide bonds, minimum bond amounts, and operators’ unresponsiveness. BLM Faces Challenges Identifying and Managing Shut-in Wells and Preventing Them from Becoming Orphaned BLM officials from 6 of the 20 BLM offices—including headquarters and selected state and field offices—and 2 of the 10 stakeholders told us that one of the challenges that BLM faces in managing its potential liabilities is identifying and managing shut-in wells. As previously mentioned, shut-in wells are inactive wells that are physically and mechanically capable of producing oil or gas in paying quantities or capable of service use. Since shut-in wells may become orphaned and therefore involve BLM resources to reclaim, identifying and managing them is a way for BLM to manage its potential liabilities. BLM’s 2012 well review policy directs field offices to review all shut-in wells on federal and Indian lands every 5 years and to ensure that shut-in wells no longer capable of production are reclaimed. However, operators are generally not required to notify BLM when they place a well in shut-in status. As a result, officials noted that it is difficult for field offices to identify all shut-in wells in order to review them. Officials from one field office told us that identifying when a well becomes shut-in is challenging unless inspectors are able to physically find the well. Even when wells have been identified to BLM as shut-in, some BLM officials at selected field offices said that they have few policy tools to manage shut-in wells. In reviewing the well review policy, we found that it contains certain directives for wells that are temporarily abandoned, including that an operator is to conduct well integrity testing prior to placing a well in temporarily abandoned status and a 30-day limit for how long operators can place wells in temporarily abandoned status without receiving BLM approval. However, the policy contains no similar directives related to testing or limited time frames for placing wells in shut- in status. As a result, BLM may be unable to identify and reduce its inventory of shut-in wells, including wells that have been in shut-in status for an extended period of time. In its January 2018 report, the OIG similarly found that the well review policy does not provide field offices the leverage to make an operator conduct integrity testing since the policy does not have instructions on the method, frequency, and way to proceed with a notice or order. Without having these test results available to them, the report found that BLM staff cannot be certain that an inactive well is environmentally sound and capable of production. The report recommended that BLM develop and implement guidance or update the well review policy to require integrity testing on inactive wells at specific periods. Strengthening the identification and management of shut-in wells could be particularly helpful in managing BLM’s potential liabilities because such wells have represented a large portion of orphaned wells. According to our analysis of AFMSS data, 138 of the 242 orphaned wells BLM manages were in shut-in status prior to becoming orphaned. Moreover, one of these wells had been in shut-in status since 1926. BLM’s Colorado and New Mexico state offices have taken steps to address the challenges associated with shut-in wells becoming orphaned. For example, in September 2016, BLM’s New Mexico state office issued a policy that directed operators to obtain BLM’s approval in order to place a well in shut-in status for more than 90 days and directed the operator to conduct periodic testing to verify that wells that have been inactive for more than 12 consecutive months remain capable of production. Under federal standards for internal control, management should design control activities—such as by clearly documenting internal control in management directives, administrative policies, or operating manuals—to achieve objectives and respond to risks. Without providing greater specificity in current policy or new supplemental guidance to all BLM field offices on how to identify and manage shut-in wells, the agency is at an increased risk of having unidentified shut-in wells, and wells that remain in shut-in status for extended periods of time, leading to increased potential liabilities if such wells become orphaned. BLM Faces Challenges Related to Limited Resources and Competing Priorities BLM officials and stakeholders told us that one of the challenges BLM faces in managing its potential liabilities is limited resources, including staff and funding, and competing priorities. Specifically, officials from 14 of the 20 BLM offices and 3 of the 10 stakeholders told us that BLM field offices have limited staff and therefore prioritize other work, such as processing drilling permits, over conducting well and bond adequacy reviews, which are used to manage potential liabilities. BLM prioritizes processing drilling permits over well and bond adequacy reviews in part because the agency is required by statute to process drilling permits within 30 days of receiving a complete application. BLM headquarters officials told us that processing permits is the agency’s highest priority activity and that they ask field offices for monthly progress reports with projected goals for processing permits within the next 90 days, and compare the offices’ accomplishments to agency targets. BLM headquarters officials told us that prioritizing processing permits increases the workload at the national, state-office, and field-office levels. Officials from one BLM state office told us that other challenges to managing its potential liabilities are staffing limitations and the time it takes to conduct bond adequacy reviews. These state office officials told us that bond reviews can take a long time to complete because some bonds are associated with several hundred wells. Similarly, officials from one field office stated that conducting bond adequacy reviews was time consuming and that they had only one staff member dedicated to conducting the reviews. In 2011, we found that a lack of resources and higher agency priorities were the primary reasons for why many BLM field office officials we interviewed had not conducted well and bond adequacy reviews or did not know the number of reviews they had conducted. In addition, officials from 6 of the 20 BLM offices and 1 stakeholder told us that another challenge BLM faces in managing its potential liabilities is prioritizing funding to reclaim orphaned wells. For example, an official from one state office told us that securing funding to reclaim orphaned wells is a challenge because BLM does not set aside funding to pay for reclamation costs. BLM officials in one field office told us that they had not received funding from BLM headquarters specifically for reclamation in over 10 years, despite managing a growing number of orphaned wells. An official from this field office told us that without dedicated funds from BLM headquarters for this purpose, the field office was unable to reclaim the orphaned wells. In addition, officials from another field office told us that time frames for competing and awarding contracts to perform reclamation work do not coincide with securing funding from BLM headquarters, and that funding has to be obligated by the end of the fiscal year. These officials explained that in one instance, by the time they obtained funding for well reclamation, it was too late to issue a contract for the work. EPAct 2005 requires the establishment of a program to reclaim orphaned, abandoned, or idled oil and gas wells on federal lands. As part of this program, BLM conducts well reviews and bond adequacy reviews. As discussed above, about half of the orphaned wells BLM identified in 2009 were not reclaimed and remained orphaned in 2017, and BLM officials cited funding as the issue. The Project Management Institute, Inc. has established a standard on program management. Under the standard, program resource management planning ensures that all required resources are made available for managers to enable the delivery of benefits for a program. Resource management planning involves identifying existing resources and the need for additional resources. The program manager analyzes the availability of each resource, in terms of both capacity and capability, and determines how these resources will be allocated to avoid over-commitment or inadequate support. Such planning, through a resource management plan, forecasts the expected resources across a program to allow the program manager to identify potential resource shortfalls or conflicts over the use of scarce or constrained resources. The plan is also to describe guidelines for making program resource prioritization decisions and resolving resource conflicts. Based on our discussions with BLM headquarters and field office officials, BLM does not have a resource management plan. For example, when we discussed resources for reclaiming orphaned wells with BLM headquarters officials, they told us that some BLM offices obtain funding from state funds established for reclaiming orphaned wells, but not all offices have been able to access such funds. If unable to secure funding from the states, offices may request funding from BLM headquarters for reclamation, and as mentioned previously, occasionally try to use unexpended funds left at the end of a fiscal year. In its comments on the draft report, Interior noted that BLM engages in annual work planning processes designed to facilitate agency resource allocation decisions. However, BLM overall does not have information on the federal resources needed to reclaim known orphaned wells. Without developing a resource management plan addressing resources needed for conducting well and bond adequacy reviews and reclaiming orphaned wells, BLM cannot have reasonable assurance that it is achieving the program’s objectives. Agency Officials and Stakeholders Identified Several Additional Challenges BLM Faces in Managing Its Potential Liabilities Agency officials and stakeholders cited additional challenges including BLM’s ability to review nationwide bonds, minimum bond amounts, and operator unresponsiveness. Reviewing nationwide bonds. Officials from 10 of the 20 BLM offices told us that they encountered challenges reviewing nationwide bonds because of a lack of coordination between BLM offices. The purpose section of the bond adequacy review policy states that field offices are to review bonds to determine whether the bond amount appropriately reflects the level of potential risk posed by the operator. However, the bond adequacy review policy also states in a directive that if the bond being reviewed is a nationwide or statewide bond, field offices are only to review the wells within their field office. Officials from one field office told us that without insights into an operator’s activities in the jurisdictions of other field offices, bond adequacy reviews do not cover when an operator has been cited with an Incident of Noncompliance or the number of inactive wells the operator may have in other jurisdictions. These field office officials said that it is important to communicate and coordinate with other field offices when there is a need to require an operator to secure a larger bond. For example, to require a well operator to increase the amount of its bond, BLM must show that the operator meets the point system’s threshold in the bond adequacy review’s calculation worksheet. Officials in one state office told us that under a nationwide or statewide bond, an operator might not reach the agency’s threshold for requiring a bond increase based on an operator’s activities in the jurisdiction of one field office but may meet the threshold if BLM’s bond adequacy review assessed all of the operator’s operations within a state or across the nation. Under federal standards for internal control, management should design control activities to achieve objectives and respond to risks, such as by clearly documenting internal controls, and having the documentation appear in management directives, administrative policies, or operating manuals. While BLM has documented its policy, the purpose of the policy to ensure that the bond amount appropriately reflects the level of potential risk posed by the operator conflicts with a directive of the policy that offices are only to review wells within their own jurisdiction. Officials told us that BLM is currently revising the bond adequacy review policy. As the agency revises its bond adequacy review policy, BLM has the opportunity to ensure that bond adequacy reviews reflect the overall risk presented by operators. By having the policy ensure that the reviews of nationwide and statewide bonds account for overall operator risk, BLM can have better assurance that it will reduce the likelihood of using taxpayer funds to pay to reclaim orphaned wells. Minimum bond amounts. Officials from 9 of the 20 BLM offices and 1 stakeholder told us that BLM faces challenges related to federal minimum bond amounts that in their opinion are too low. For example, officials from one BLM state office expressed concerns about operators with multiple wells covered by the minimum bond amounts, which the officials believed to be inadequate to cover total potential reclamation costs. Minimum bond amounts were set in the 1950s and 1960s and have not been updated to keep up with inflation. Specifically, the $10,000 minimum for individual bonds was established in 1960, and the bond minimums for statewide bonds ($25,000) and nationwide bonds ($150,000) were established in 1951. If adjusted to 2016 dollars, these amounts would be $63,613 for an individual bond, $189,825 for a statewide bond, and $1,138,952 for a nationwide bond. According to BLM headquarters officials, the agency does not require that operators provide full liability bonds. These officials told us that they believed that most operators would not be able to remain in business if bond amounts were based on estimated total reclamation costs. Operators’ unresponsiveness. Officials from 8 of the 20 BLM offices and 2 stakeholders told us that BLM faces challenges dealing with unresponsive operators when requiring operators to increase bond amounts or issuing Incidents of Noncompliance. For example, officials from one BLM state office told us that operators do not always respond to letters informing them of a requirement to secure an increase in their bond. Officials from another BLM state office told us that the agency can place operators on a noncompliance list prohibiting them from holding leases or conducting operations on federal lands. However, these officials also said that they have seen operators ask relatives to obtain leases in order to circumvent such prohibitions. Officials from one field office told us of one particular instance in which BLM had spent over 7 years attempting to enforce the requirements for reclamation activities. BLM had issued an Incident of Noncompliance, but the operator did not respond and instead reorganized as a separate corporate entity. Subsequently, the operator went bankrupt, requiring BLM to restart the communications process from the beginning with the newly formed entity. BLM officials told us that the agency has very little leverage when companies change their name or reorganize in an attempt to evade performing required reclamation activities. BLM headquarters officials told us that working with operators was a delicate balance, especially when oil and gas prices are down, and BLM field offices would benefit from conducting periodic operator outreach to have an open dialogue with the operators. Conclusions BLM is responsible for overseeing oil and gas development on federal lands and for balancing the sometimes competing priorities of encouraging oil and gas development, while ensuring that when wells run dry, operators return well sites to their original natural conditions. Federal laws, regulations, and BLM’s own policies call for the agency to take various actions to manage its potential oil and gas well liabilities and reclaim orphaned wells. However, BLM does not systematically or comprehensively track how much the agency has spent to reclaim orphaned wells or information, such as the number of orphaned wells and inactive wells over time, necessary to determine the agency’s potential liabilities. Without systematically or comprehensively tracking information on BLM’s well reclamation costs and indicators of potential future costs, its ability to monitor its progress and plan for its potential liabilities associated with orphaned wells is limited. In addition, implementation of BLM’s well and bond adequacy review policies by the field offices is hampered by officials having different understandings of what constitutes a well review. This variance is because BLM’s well review policy does not outline specific instructions on what actions field offices should take when conducting a well review. This situation results in inconsistent ways of conducting well reviews and annually reporting on them. Without developing and communicating specific instructions outlining what actions constitute a well review for annual-reporting purposes, BLM cannot have reasonable assurance that its field offices are conducting and reporting on well reviews in a consistent manner. Further, inaccuracies in certain AFMSS data, such as the dates that wells last changed statuses, raise questions about the quality of data BLM headquarters uses to determine the extent to which its offices are implementing the well review and bond adequacy review policies. BLM has not taken steps to improve AFMSS’ data quality such as through the use of additional edit checks to prevent field offices from inputting erroneous data or having data stewards certify the quality of the data. Without taking such steps, BLM cannot have reasonable assurance that management has accurate information it needs to track whether field offices are conducting well and bond adequacy reviews as intended. In addition, BLM’s approach to monitoring the implementation of its well and bond adequacy review policies is limited because the reports the agency uses to monitor implementation provide insufficient and at times conflicting information. Without taking actions to strengthen its approach to monitoring, such as collecting and analyzing data on performance indicators and ensuring the quality of those data, BLM’s ability to assess the extent to which field offices are reviewing all inactive wells and determining the adequacy of all bonds will continue to be limited. BLM officials and stakeholders identified several challenges that BLM faces in managing its potential oil and gas well liabilities, including identifying and managing certain inactive wells—specifically wells that are in shut-in status and that have the potential to become orphaned. This problem is because operators are generally not required to notify BLM when they place a well in shut-in status. Without providing greater specificity in current policy or supplemental guidance to all field offices, the federal government may face increased potential liabilities if shut-in wells become orphaned. In addition, BLM faces challenges related to limited resources and competing priorities, such as not setting aside funding to pay for reclaiming orphaned wells. Without developing a resource management plan addressing resources needed for conducting well and bond adequacy reviews and reclaiming orphaned wells, BLM cannot have reasonable assurance that it is achieving the program’s objectives. BLM also faces challenges related to conducting nationwide and statewide bond adequacy reviews because the bond adequacy review policy overall contains conflicting information on how field offices are to review bonds’ adequacy. BLM is currently revising the bond adequacy review policy and has an opportunity to ensure that the reviews of nationwide and statewide bonds reflect operators’ overall risks. Recommendations for Executive Action We are making the following seven recommendations to BLM: The Director of BLM should systematically and comprehensively track the actual costs BLM incurs when reclaiming orphaned wells and the information, including the number of orphaned wells and inactive wells over time, necessary to determine the agency’s potential liabilities. (Recommendation 1) The Director of BLM should develop and communicate specific instructions on what actions constitute a well review for annual-reporting purposes. (Recommendation 2) The Director of BLM should take steps to improve AFMSS data quality, for example, by conducting more edit checks and by having data stewards certify the quality of the data. (Recommendation 3) The Director of BLM should strengthen its approach to monitoring field offices’ implementation of the well review and bond adequacy review policies, such as by collecting and analyzing data on performance indicators and ensuring the quality of those data. (Recommendation 4) The Director of BLM should provide greater specificity in current policy or supplemental guidance to all BLM field offices on how to identify and manage all shut-in wells. (Recommendation 5) The Director of BLM should develop a resource management plan addressing resources needed for conducting well and bond adequacy reviews and reclaiming orphaned wells. (Recommendation 6) The Director of BLM should, in revising the bond adequacy review policy, ensure that the reviews of nationwide and statewide bonds reflect the overall risk presented by operators. (Recommendation 7) Agency Comments and Our Evaluation We provided a draft of this report to the Department of the Interior for review and comment. In its comments, reproduced in appendix II, Interior generally concurred with our recommendations. Interior stated that, following GAO’s 2011 report on potential oil and gas well liabilities, BLM implemented comprehensive policies to better manage and minimize the risks of idle and orphaned wells on federal and Indian lands. Interior agreed that there are areas where BLM can improve the accuracy of its data and further reduce the risks associated with idle and orphaned wells. Interior indicated that it will update and improve its existing policies and guidance consistent with the findings and recommendations in our report. In response to our sixth recommendation—that BLM develop a resource management plan addressing resources needed for conducting well and bond adequacy reviews and reclaiming orphaned wells—Interior stated that BLM conducts annual work planning processes which facilitate decisions regarding the allocation of agency resources and requested additional information clarifying how our recommendation fits into or differs from these. We expanded our description of resource management planning and added language regarding BLM’s annual work planning processes to the report. However, we were not able to review the scope or adequacy of BLM’s annual work planning processes as they relate to resource planning for well and bond reviews and reclaiming orphaned wells for this report. As agreed with your offices, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies to the appropriate congressional committees, the Secretary of the Interior, and other interested parties. In addition, the report will be available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-3841 or ruscof@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix III. Appendix I: Objectives, Scope, and Methodology This report examines (1) how BLM’s actual costs incurred to reclaim orphaned wells and potential oil and gas well liabilities have changed, if at all, for fiscal years 2010 through 2017; (2) the extent to which BLM has implemented its 2012 well review and 2013 bond adequacy review policies; and (3) BLM officials’ and stakeholders’ views on what challenges, if any, BLM faces in managing its potential liabilities. To examine how BLM’s actual reclamation costs incurred and potential oil and gas well liabilities have changed, we analyzed data in BLM’s Automated Fluid Minerals Support System (AFMSS) on oil and gas wells on federal and Indian lands, including inactive wells—which represent potential liabilities. We reviewed documentation provided by BLM and compared BLM’s policies and procedures on recording information on actual costs incurred to reclaim orphaned wells and potential liabilities against the information and communication standard outlined in Standards for Internal Control in the Federal Government. We selected and interviewed officials from 13 BLM field offices because, according to fiscal year 2016 data from the Department of the Interior’s Office of Natural Resources Revenue (ONRR) Oil and Gas Operations Report (OGOR) data system we analyzed, these offices were responsible for about 80 percent of all oil and gas wells managed by BLM. In addition, we interviewed officials from the 6 BLM state offices associated with the 13 selected field offices (see table 1). Findings from selected offices cannot be generalized to those we did not include in our review. However, because AFMSS does not contain information on actual costs incurred to reclaim orphaned wells, we obtained documentation of the actual reclamation costs that 13 selected BLM field offices incurred for fiscal years 2010 through July 2017. To analyze these costs, we reviewed purchase orders, invoices, and other documentation for actual reclamation work performed. We also obtained documentation, including spreadsheets with estimated potential reclamation costs that these 13 selected field offices faced as of July 2017. To assess the reasonableness of estimated reclamation costs, we reviewed estimates provided by officials from the selected field offices and compared those to historical actual costs that we previously reported in January 2010. We determined the overall estimated reclamation costs were sufficiently reasonable for providing a sense of the general magnitude of potential costs, though we did not assess the underlying inputs or assumptions used. The information we received is not generalizable to reclamation costs for other BLM offices that we did not review. We also analyzed AFMSS data on the number of wells capable of production on federal lands from fiscal years 2010 to 2016. The AFMSS database provides a snapshot of the time that the data are queried, and so does not include historical data over time. As such, to examine the number of inactive wells on federal and Indian lands and how long these have been inactive, we combined AFMSS data with data from the OGOR data system through September 2016. The Department of the Interior requires monthly OGORs from operators, which document and record the volume of oil and gas produced from wells on federal and Indian lands. From AFMSS, we identified the appropriate population of wells by selecting wells only located on federal and Indian lands, and excluded wells that were on state or private lands. Because we did not find data in AFMSS on how long a well had been in its last recorded status to be reliable, we analyzed production records from the OGOR data system. We also excluded data on wells that were in statuses in which there was no associated potential liability, such as wells pending an application for permit to drill. For each reporting date through September 2016, we aggregated data from multiple well completions to the 10-digit unique well identifier level. We then matched the unique well identifiers in AFMSS to those listed in the OGOR data system to enumerate inactive wells by duration of inactivity. For each reporting date, we designated wells with at least one completion showing non-zero production volumes or in drilling or monitoring status in the OGOR data system as active. We also designated a well as active at a certain date if AFMSS data indicated any of its completions were completed on that date. Otherwise we deemed wells where all completions had zero production reporting on a date as inactive for the corresponding period. In some cases, (i) no OGOR records existed with non-zero production volumes or drilling or monitoring well status and (ii) no AFMSS well completion date was provided, and so we calculated inactivity by using the earliest record date for that well in the OGOR data set. We discussed our methodology for calculating the number of wells with BLM officials. We compared the number of inactive wells from our analysis to those reported in BLM national and state reports to identify data inconsistencies. In addition, we analyzed AFMSS reports, as of July 2017, to analyze data on the number of orphaned wells. To assess the reliability of OGOR and AFMSS data, we reviewed agency documents, met with relevant agency officials, and performed electronic testing by verifying, for example, missing or out-of-range data values. We found the data for the number of inactive wells and how long they have been inactive as well as the data for the number of wells BLM has identified as orphaned to be sufficiently reliable for our purposes. To determine the extent to which BLM has implemented its 2012 and 2013 policies for conducting well reviews and bond adequacy reviews, we reviewed applicable laws and analyzed the well review and bond adequacy review policies. We reviewed information contained in BLM’s well review and bond adequacy review reports for fiscal year 2016 as well as data generated through AFMSS on bonds and wells as of October 2017. We were unable to fully assess BLM’s performance against the directives in the agency’s 2012 well review and 2013 bond adequacy review policies due to limited agency data and documentation as discussed in the report. Specifically, we identified data accuracy and consistency concerns with some of the data elements in the agency’s well review and bond adequacy review reports as well as some AFMSS data on wells and bonds, which we discuss in this report. We performed electronic testing by verifying out-of-range values, such as dates of well reviews conducted that were listed as being in the future. We also interviewed officials from BLM headquarters, the 13 selected field offices, and the 6 associated BLM state offices, to obtain information on the extent to which the selected offices implemented the 2012 and 2013 policy directives. We compared BLM’s procedures detailing how field offices are to count or report a well review as well as procedures for maintaining data quality against the control activities standard outlined in Standards for Internal Control in the Federal Government. We also compared BLM’s procedures for monitoring implementation of policy directives against leading practices for monitoring agency policies. We also reviewed documentation for a random, non-generalizable sample of 62 well reviews and 58 bond adequacy reviews, as reported by the 13 selected BLM field offices, for a total of 120 reviews. A GAO statistician selected a random sample of five well reviews for unique well numbers and five bond reviews of unique bond numbers that the 13 selected field offices had reviewed from the fiscal year 2016 well report and bond adequacy report. Due to variations in field offices’ reporting, some well and bond reviews from prior fiscal years were also included in the random selection. The Farmington field office also did not conduct any bond adequacy reviews in fiscal year 2016, and so we included bond reviews that the field office conducted in fiscal year 2015 in the random selection. In addition, the Pinedale and Rawlins field offices had not conducted any bond adequacy reviews in fiscal year 2016. As a result, we randomly selected additional reviews from fiscal year 2015 for those field offices. The Pinedale, Rawlins, and Colorado River Valley field offices conducted less than 5 bond reviews in each office in that fiscal year, so we selected and reviewed documentation in support of only those reviews they had conducted. We assessed the documentation to determine whether or not field offices conducted reviews and complied with selected directives of the well review and bond adequacy review policies. Information from our documentation reviews is not generalizable to all BLM field offices but provides illustrative examples of the information contained in BLM well and bond adequacy reviews. To examine BLM officials’ and stakeholders’ views on what challenges, if any, BLM faces in managing its potential oil and gas well liabilities, we conducted semi-structured interviews with officials from BLM headquarters, the 13 selected BLM field offices, and the 6 BLM state offices associated with these 13 field offices. In addition, we interviewed or obtained written responses from a standard set of questions from 8 representatives of stakeholder organizations. These representatives were knowledgeable about BLM’s oil and gas well management, and included academic, environmental, industry, and state organizations (see table 2). In addition, we spoke with knowledgeable officials from the Department of the Interior’s Office of Natural Resources Revenue (ONRR) and the Department of the Interior’s Office of Indian Energy and Economic Development, Division of Energy and Mineral Development. To identify knowledgeable stakeholders, we conducted a literature search, reviewed previous GAO reports, and obtained recommendations from BLM officials and stakeholders using a snowball technique in which an initial group of BLM officials and stakeholders we interviewed identified additional contacts to interview. From this list, we selected stakeholders who could provide a range of viewpoints. We generally asked the same questions during each interview but also discussed individual stakeholders’ perspectives, as appropriate. In our interviews, we asked officials and stakeholders what challenges, if any, BLM offices face in managing their potential oil and gas well liability. We also asked what challenges, if any, BLM offices face in conducting well reviews and bond adequacy reviews. To identify the challenges identified most often in the interviews, two analysts developed categories of challenges identified by BLM offices and stakeholders, and each analyst independently determined whether each BLM office and stakeholder had identified challenges that fit into these categories. The two analysts discussed and resolved any differences in their coding. The views of the BLM officials, stakeholders, and other agency personnel we interviewed are not generalizable to BLM officials, similar stakeholders, and other agency personnel who we did not interview. Lastly, we compared how BLM identified and managed certain inactive wells, as well as how it managed nationwide and statewide bonds, against the control activities standard outlined in Standards for Internal Control in the Federal Government and BLM’s resource management practices against certain requirements in the Energy Policy Act of 2005 (EPAct 2005) and leading practices by the Project Management Institute in The Standard for Program Management. We conducted this performance audit from November 2016 to May 2018, in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Comments from the Department of the Interior Appendix III: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Quindi Franco (Assistant Director), Marie Bancroft (Analyst-in-Charge), Richard Burkard, John Delicath, Cindy Gilbert, Shylene Mata, Celia Mendive, Dan Royer, Barbara Timmerman, Carolyn Voltz, Jack Wang, and Jina Yu made key contributions to this report.

Why GAO Did This Study In fiscal year 2016, private entities operated about 94,000 oil and gas wells on federal lands overseen by BLM. Once wells cease production, they can become inactive and potentially orphaned if an operator does not perform required reclamation and if an operator's bond is insufficient to cover the expenses. BLM considers oil and gas wells on federal and Indian lands and the associated leased lands as potential liabilities for the federal government because BLM may have to cover the costs of reclaiming well sites. To better manage its potential liabilities, BLM issued well and bond adequacy review policies in 2012 and 2013, respectively. GAO was asked to review how BLM manages its potential oil and gas well liabilities. This report examines, among other things: (1) how BLM's actual costs and potential oil and gas well liabilities have changed for fiscal years 2010 through 2017 and (2) the extent to which BLM has implemented its well and bond review policies. GAO analyzed BLM's policies and data and interviewed BLM officials and representatives from stakeholder organizations. What GAO Found GAO's analysis indicates that the Bureau of Land Management's (BLM) actual costs incurred and potential liabilities for reclaiming oil and gas wells have likely increased for fiscal years 2010 through 2017. However, the full extent of the increase is not known because BLM does not systematically track needed data. Based on GAO's analysis of data obtained from 13 of BLM's 33 field offices that manage oil and gas programs, the average annual reclamation cost was $267,600, an increase compared to the $171,500 annual average across all BLM offices that GAO reported in 2010. Similarly, GAO's analysis of BLM data found that the number of known orphaned wells, those that generally have no responsible or liable parties, for all field offices has increased from 144 in 2010 to 219 as of 2017. However, BLM's database that contains information on oil and gas wells on federal and Indian lands does not collect information on costs incurred or on potential liabilities that might result from an increase in the number of orphaned wells. Under federal internal control standards, management should use quality information to achieve the entity's objectives. Without systematically tracking such information, BLM does not have assurance that it has sufficient bonds or financial assurances to cover the costs of reclaiming orphaned wells. GAO was unable to fully assess the extent to which BLM field and state offices have implemented the agency's policies on reviewing wells and bond adequacy in part because of deficiencies in BLM's monitoring approach. For example, reports BLM headquarters used to monitor field offices' implementation of the policies have limitations. GAO identified discrepancies between the well and bond adequacy review reports that BLM state offices submitted to headquarters and the national summary consolidating states' information. Out of 10 state offices, 3 reported a different number of reviews completed in fiscal year 2016 than what BLM reported in its fiscal year 2016 national summary. Leading practices for monitoring the implementation of agency policies call for taking steps such as collecting and analyzing data on performance indicators. Without strengthening BLM's approach to monitoring, its ability to assess field offices' reviews of all inactive wells and determine the adequacy of all bonds is limited. What GAO Recommends GAO is making seven recommendations, including that BLM systematically track the agency's actual reclamation costs and potential liabilities and strengthen its approach to monitoring field offices' implementation of the well review and bond adequacy review policies. BLM agreed with GAO's recommendations.

gao_GAO-19-229

gao_GAO-19-229_0

Background The Navy currently has 51 attack submarines—comprising 33 Los Angeles class, 3 Seawolf class, and 15 Virginia class submarines (see fig. 1). Attack submarines are homeported at bases in the United States: in New London, Connecticut; Pearl Harbor, Hawaii; Norfolk, Virginia; San Diego, California; and Bangor, Washington; 4 are homeported overseas, in the U.S. territory of Guam. Submarine Safety Controls and Culture On April 10, 1963, the USS Thresher (SSN 593) sank during deep submergence tests off the coast of New England. One hundred and twelve officers and enlisted sailors and 17 civilians perished in the tragedy. The accident investigation concluded that the Navy did not have adequate procedures in place to prevent and respond to a catastrophic flooding incident. time, and that maintenance delays reduce the amount of time during which ships and submarines are available for training and operations. Submarine fleet and squadron officials emphasized the strict safety culture that permeates the submarine community. This emphasis on meeting safety certification criteria means that the Navy operates a supply-based submarine force that does not compromise on adherence to training and maintenance standards to meet combatant commander demands, according to these officials (see sidebar). Officials added that the Navy will delay deployment dates if necessary to ensure that these standards are met. As a result, deployed readiness is high and attack submarines are in excellent materiel condition as compared with the rest of the Navy fleet. The loss of the USS Thresher and its crew spurred the Navy to establish stringent safety requirements for submarines to prevent another loss at sea. Following the accident, the Navy established submarine safety certification criteria to provide maximum reasonable assurance that critical systems would protect the crew from flooding and allow the submarine to conduct an emergency surfacing should flooding occur. This program, known as SUBSAFE, is still in use today, to ensure that these critical systems receive a high quality of work and that all work is properly documented. According to the Navy, the SUBSAFE certification status of a submarine is fundamental to its mission capability, as it provides a thorough and systematic approach to quality, and to a culture that permeates the entire submarine community. According to Navy officials, since the SUBSAFE program was established in 1963, no SUBSAFE-certified submarine has ever been lost. The Navy has been unable to begin or complete the vast majority of its attack submarine maintenance periods on time resulting in significant maintenance delays and operating and support cost expenditures. Our analysis of Navy maintenance data shows that between fiscal year 2008 and the end of fiscal year 2018, attack submarines will have incurred 10,363 days of idle time and maintenance delays as a result of delays in getting into and out of the shipyards. Our analysis found that the primary driver affecting attack submarines are delays in completing depot maintenance. For example, of the 10,363 total days of lost time since fiscal year 2008, 8,472 (82 percent) were due to depot maintenance delays. As we previously reported, completing ship and submarine maintenance on time is essential to Navy readiness, as maintenance periods lasting longer than planned could reduce the number of days during which ships and crews are available for training or operations. Attack submarines also face delays in beginning maintenance when the public shipyards have no available capacity, in some cases forcing submarines to idle pierside because they are no longer certified to conduct normal operations. According to Navy officials, the SUBSAFE program—its program to ensure and certify submarine safety—requires submarines to adhere to strict maintenance schedules and pass materiel condition assessments before they are allowed to submerge. Attack submarines that go too long without receiving required maintenance are at risk of having their materiel certification expire. Should this certification expire, these submarines are restricted to sitting idle, pierside, while they wait until a shipyard has the capacity to begin their maintenance period (see fig. 2). We found that since fiscal year 2008, 14 attack submarines have spent a combined 61 months (1,891 days) idling while waiting to enter shipyards for maintenance. Idle time incurred while waiting to begin a maintenance period is often coupled with maintenance delays while at the shipyards, thus compounding total delays. We also found that the Navy incurs significant costs in operating and supporting submarines that are experiencing maintenance delays and idle time. We analyzed the operating and support costs the Navy incurs on average to estimate the costs of crewing, maintaining, and supporting attack submarines that are delayed in getting into and out of the shipyards. Using historical daily cost data the Navy adjusted for inflation, we estimated that since fiscal year 2008 the Navy has spent more than $1.5 billion in fiscal year 2018 constant dollars on attack submarines sitting idle while waiting to enter the shipyards, and on those delayed in completing their maintenance at the shipyards (see table 1). While the Navy would incur these costs regardless of whether the submarine was delayed, idled, or deployed, our estimate of $1.5 billion represents costs incurred from fiscal year 2008 through fiscal year 2018 for attack submarines without receiving any operational capability in return. While acknowledging the magnitude of these costs, Navy officials stated that there may be some benefits that could be realized from these operating and support costs since crews on idle attack submarines can conduct some limited training. Operating and support costs include payment of crew salaries, purchasing of spare parts, and conducting of maintenance, among other things, but they do not represent the full operational impact incurred by the Navy from the idle time and maintenance delays. For example, attack submarine depot-level maintenance requires the use of a drydock, and officials from the three public shipyards we visited told us that their drydock capacity was limited. A delayed attack submarine maintenance period can restrict the use of a drydock for much longer than originally anticipated, thereby preventing the shipyard from using that drydock to maintain other vessels, including other types of ships, or to conduct necessary repairs on the facilities. Navy Has Begun to Address Some Challenges Associated with Attack Submarine Maintenance Delays but Has Not Effectively Allocated Maintenance Periods to Limit Idle Time Navy Is Addressing Some Challenges at the Public Shipyards The Navy has started to address workforce shortages and facilities needs at the public shipyards. These efforts to address the Navy’s maintenance challenges are important steps, but they will require several years of sustained management attention to reach fruition. As we reported in September 2017, maintenance on ships and submarines may be delayed for numerous reasons, including workforce gaps and inexperience, the poor condition of facilities and equipment, parts shortages, changes in planned maintenance work, and weather. According to Navy officials, all of these issues continue to affect the Navy’s ability to complete attack submarine maintenance on time. According to officials, the Navy has begun to address some of these challenges. For example: The public shipyards have been hiring to address workforce shortages. The number of civilian full-time employees at the shipyards increased from 25,087 in 2007 to 34,160 in 2017, with a goal to reach 36,100 by 2020. Navy officials cautioned that this newly hired workforce is largely inexperienced and will require time to attain full proficiency. The Navy has released a plan to guide public shipyard capital investments. In September 2017 we reported that the Navy projected an inability to support 50 planned submarine maintenance periods over the ensuing 23 years, due to capacity and capability shortfalls at the public shipyards. We recommended that the Navy develop a comprehensive plan for shipyard capital investment. In February 2018 the Navy published its shipyard optimization plan, outlining an estimated $21 billion investment needed to address shipyard facility and equipment needs over 20 years to meet the operational needs of the current Navy fleet, but not the larger fleet size planned for the future. Navy Has Not Effectively Allocated Maintenance Periods among the Public and Private Shipyards to Limit Attack Submarine Idle Time While the public shipyards have operated above capacity for the past several years, attack submarine maintenance delays are getting longer and idle time is increasing. The Navy expects the maintenance backlogs at the public shipyards to continue. We estimate that, as a result of these backlogs, the Navy will incur approximately $266 million in operating and support costs in fiscal year 2018 constant dollars for idle submarines from fiscal year 2018 through fiscal year 2023, as well as additional depot maintenance delays. The Navy may have options to mitigate idle time and maintenance delays. For example, officials at the private shipyards—General Dynamics Electric Boat and Huntington Ingalls Industries-Newport News Shipbuilding—told us that they will have available capacity for repair work for at least the next 5 years. Although the Navy has shifted about 8 million man-hours in attack submarine maintenance to private shipyards over the past 5 years, it has done so sporadically, having decided to do so in some cases only after experiencing lengthy periods of idle time. According to private shipyard officials, the sporadic shifts in workload have resulted in repair workload gaps that have disrupted private shipyard workforce, performance, and capital investment—creating costs that are ultimately borne in part by the Navy. We believe that the Navy has not fully mitigated this challenge because it has not completed a comprehensive business case analysis to inform maintenance workload allocation across public and private shipyards, and to proactively minimize attack submarine idle time and maintenance delays. Such an analysis would help the Navy better assess private shipyard capacity to perform attack submarine maintenance and would help it incorporate a complete accounting of all costs, benefits, and risks, including: the large operating and support costs of having attack submarines sitting idle; the qualitative benefits associated with providing additional availability to the combatant commanders; and the potential for additional work at private shipyards to reduce schedule risk to submarine construction programs by allowing the yards to build and maintain a stable shipyard workforce. The April 2011 DOD Product Support Business Case Analysis Guidebook provides standards for DOD’s process for conducting analyses of costs, benefits, and risks. It states that data sources used to conduct a business case analysis should be comprehensive and should include both quantitative and qualitative values. It notes that benefits, such as the availability of a weapon system, may be qualitative in nature, and that DOD should evaluate all possible support options, to include government- and contractor-provided maintenance. Navy leadership has acknowledged that they need to be more proactive in leveraging private shipyard repair capacity, but officials cautioned that maintenance could cost more at a private shipyard than at a public shipyard. However, without a complete accounting of all costs, benefits, and risks, the Navy will remain unable to determine whether the cost of performing a maintenance period at a private shipyard would outweigh the mission benefits of having reduced idle time, additional operational availability, and the potential for reduced risk to submarine construction programs. Conclusions The nation’s investment in attack submarines provides the United States an asymmetric advantage to gather intelligence undetected, attack enemy targets, and insert special forces, among other capabilities. However, the Navy’s attack submarine fleet has suffered from persistent and costly maintenance delays. Although the Navy has several activities underway to reduce maintenance delays for the attack submarine fleet, it has not yet taken additional steps to maximize attack submarine readiness that fully address challenges such as the allocation of maintenance periods between public and private shipyards. Without addressing this challenge, the Navy will not achieve the full benefit of the nation’s investment in its attack submarines, and it risks continued expenditure of operating and support funding to crew, maintain, and support attack submarines that provide no operational capability because they are delayed in getting into and out of maintenance. Recommendation for Executive Action The Secretary of the Navy should ensure that the Chief of Naval Operations conducts a business case analysis to inform maintenance workload allocation across public and private shipyards; this analysis should include an assessment of private shipyard capacity to perform attack submarine maintenance, and should incorporate a complete accounting of both (a) the costs and risks associated with attack submarines sitting idle, and (b) the qualitative benefits associated with having the potential to both mitigate risk in new submarine construction and provide additional availability to the combatant commanders. Agency Comments We provided a draft of the classified version of the report to DOD for review and comment. That draft contained the same recommendation as this unclassified version as well as three additional recommendations DOD deemed sensitive. In written comments provided by DOD (reprinted in appendix II), DOD concurred with our recommendation stating that it has taken the first steps to take a more holistic view of submarine maintenance requirements and impacts across both the public and private shipyards. The Navy also provided technical comments, which we incorporated as appropriate. We are sending copies of this report to congressional committees; the Secretary of Defense; the Secretary of the Navy, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-3489 or pendletonj@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix IV. Appendix I: Scope and Methodology To assess the extent to which the Navy has experienced maintenance delays in its attack submarine fleet, we analyzed attack submarine maintenance delay and idle time data from Naval Sea Systems Command, and we reviewed prior GAO work on shipyard maintenance delays. The Navy determines depot maintenance delays by counting each day in which a submarine maintenance period extends beyond the planned completion date. Two Navy offices within Naval Sea Systems Command—that is, the Logistics, Maintenance, and Industrial Operations office and Program Executive Office Submarines—track days incurred from depot-level maintenance delays and idle time. To determine the total number of days of maintenance delays for each fiscal year within our scope, we subtracted the planned completion date from the actual completion date to produce the number of days of maintenance delays for each maintenance period for each submarine. We added together the days of maintenance delays across all attack submarines for each fiscal year, and then added the fiscal year totals to produce the overall total. Although the data included some maintenance periods that began before fiscal year 2008, we counted days of maintenance delays only from periods that were incurred in fiscal years 2008 through 2018. We also tracked the total number of days that the Navy completed maintenance periods ahead of schedule—that is, 153—but we noted these separately instead of subtracting them from the total number of days of maintenance delays. To estimate costs associated with these delays, we analyzed annual data from fiscal years 2011 through 2017 (the most current data available at the time of our review) from the Navy’s Visibility and Management of Operating and Support Costs system. We also reviewed prior work on determining the operating and support costs of Navy ships. The Navy calculates total operating and support expenditures for each attack submarine on an annual basis, as well as the yearly average expenditure for each attack submarine class, including Los Angeles class, Seawolf class, and Virginia class blocks one and two. For each class, we converted the Navy’s annual class averages into daily average costs by adding the annual class averages together for each year that data were available, fiscal years 2011 through 2017, then dividing that number by the total number of days. We then multiplied the daily class average by the total number of days of maintenance delays and idle time incurred by submarines within that class, according to our calculations outlined above, between fiscal year 2008 and fiscal year 2018, and we added these totals together to produce the total estimated operating and support cost for days of maintenance delays and idle time incurred during this period. The data did not include annual class average costs for fiscal years 2008, 2009, 2010, or 2018. However, the annual class averages for fiscal years 2011 through 2017 did not show significant variation, so we applied these averages to 2008, 2009, 2010, and 2018. To assess the extent to which the Navy has addressed any challenges and developed mitigation plans for any maintenance delays, we reviewed the Navy’s plans to address attack submarine maintenance delays and interviewed Navy headquarters, fleet, and squadron officials, attack submarine crews, and public and private shipyard officials to understand any plans to address attack submarine maintenance delays and idle time. We analyzed data on factors contributing to attack submarine maintenance delays, such as cannibalization rates. We visited three of the four public shipyards, including Pearl Harbor Naval Shipyard and Intermediate Maintenance Facility, Portsmouth Naval Shipyard, and Norfolk Naval Shipyard, to observe operations, training, and the condition of the facilities and equipment, and to interview officials about challenges affecting operational efficiency and performance. We also met with Navy maintainers at Naval Station Norfolk and Naval Submarine Base New London, and with the crew of the submarine tenders USS Frank Cable (AS-40) and USS Emory S. Land (AS-39) in Guam. We toured the two private shipyards that conduct attack submarine repair work—General Dynamics Electric Boat and Huntington Ingalls Industries-Newport News Shipbuilding—and interviewed executives at both locations. We also toured attack submarines and met with crew leadership, selected according to which submarines and crews were available for tours at each of the sites we visited. We visited the USS Boise (SSN 764) at Naval Station Norfolk and four attack submarines in depot-level maintenance: the USS Albany (SSN 753), the USS Jefferson City (SSN 759), the USS New Mexico (SSN 779), and the USS Springfield (SSN 761). We met with the crews of two attack submarines assigned to the operating forces at the time of our visit, the USS Missouri (SSN 780) and the USS North Dakota (SSN 784). We evaluated the Navy’s plans to address any challenges against criteria in federal standards for internal control, which state that agencies should evaluate performance in achieving key objectives and addressing risks; the Department of Defense’s business case analysis guidebook, which provides standards for the process used to conduct analyses of costs, benefits, and risks; the Project Management Book of Knowledge, which provides best practices for project management; and the Secretary of the Navy’s December 2017 Strategic Readiness Review, which calls for the early identification of systemic risks before problems occur. To assess the reliability of the data sources for conducting analyses to address all of the objectives in this report, we reviewed systems documentation and interviewed officials to understand system operating procedures, organizational roles and responsibilities, and error-checking mechanisms. We selected the time frames for each of the data series above after assessing their availability and reliability, to maximize the amount of data available for us to make meaningful comparisons. We assessed the reliability of each of the data sources. The Navy provided information based on our questions regarding data reliability, including information on an overview of the data, data-collection processes and procedures, data quality controls, and overall perceptions of data quality. The Navy provided documentation of how the systems are structured and what written procedures are in place to help ensure that the appropriate information is collected and properly categorized. Additionally, we interviewed Navy officials to obtain further clarification on data reliability, discuss how the data were collected and reported, and explain how we planned to use the data. We also conducted our own error checks to look for inaccurate or questionable data, and we discussed with officials any data irregularities we found. We conducted these assessments on the following data for attack submarines: Navy deployed and surge-ready submarines from fiscal years 2011 through 2018; maintenance timeliness from fiscal years 2000 through 2018; idle time from fiscal years 2008 through 2018; operating and support costs from fiscal years 2011 through 2017; and cannibalization rates from 2012 through 2017. Some of these data were used in prior reports, and their reliability had previously been assessed. After further assessing any data that we had not recently used, we determined that they were sufficiently reliable for the purposes of summarizing attack submarine readiness trends and related information. We interviewed officials, and where appropriate obtained documentation, at the following locations: Office of the Chief of Naval Operations Undersea Warfare Division (N97) Warfare Integration Division (N83) U.S. Fleet Forces Command Commander, Submarine Force, U.S. Atlantic Fleet Commander, Submarine Squadron 4 Commander, Regional Support Group Groton Commander, Submarine Force, U.S. Pacific Fleet Commander, Submarine Squadron 1 Commander, Submarine Squadron 7 Commander, Submarine Squadron 15 Naval Sea Systems Command (NAVSEA) Logistics, Maintenance, and Industrial Operations (NAVSEA 04) Program Executive Office, Submarines Attack Submarine Program Office (PMS 392) Submarine Maintenance Engineering, Planning, and Procurement (SUBMEPP) Supervisor of Shipbuilding, Conversion, and Repair (SUPSHIP) Newport News, Virginia Navy Education and Training Command Submarine Learning Facility Norfolk Navy Board of Inspection and Survey Norfolk Naval Shipyard, Norfolk, Virginia Pearl Harbor Naval Shipyard and Intermediate Maintenance Facility, Pearl Harbor, Hawaii Portsmouth Naval Shipyard, Kittery, Maine Newport News Shipbuilding, Virginia, operated by Huntington Ingalls Industries Electric Boat, Groton, Connecticut, operated by General Dynamics The performance audit upon which this report is based was conducted from August 2017 to October 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. We worked with DOD to prepare this unclassified version of the report for public release. This public version was also prepared in accordance with these standards. Appendix II: Comments from the Department of Defense Appendix III: Related GAO Products Report numbers with a C or RC suffix are Classified. Classified reports are available to personnel with the proper clearances and need to know, upon request. Columbia Class Submarine: Immature Technologies Present Risks to Achieving Cost Schedule and Performance Goals. GAO-18-158. Washington, D.C.: Dec. 21, 2017. Navy Readiness: Actions Needed to Address Persistent Maintenance, Training, and Other Challenges Affecting the Fleet. GAO-17-809T. Washington, D.C.: Sept. 19, 2017. Naval Shipyards: Actions Needed to Improve Poor Conditions that Affect Operations. GAO-17-548. Washington, D.C.: Sept. 12, 2017. Navy Readiness: Actions Needed to Address Persistent Maintenance, Training, and Other Challenges Facing the Fleet. GAO-17-798T. Washington, D.C.: Sept. 7, 2017. Military Readiness: DOD’s Readiness Rebuilding Efforts May Be at Risk without a Comprehensive Plan. GAO-16-841. Washington, D.C.: Sept. 7, 2016. Navy and Marine Corps: Services Face Challenges to Rebuilding Readiness. GAO-16-481RC. Washington, D.C.: May 25, 2016. (SECRET//NOFORN) Military Readiness: Progress and Challenges in Implementing the Navy’s Optimized Fleet Response Plan. GAO-16-466R. Washington, D.C.: May 2, 2016. Navy Force Structure: Sustainable Plan and Comprehensive Assessment Needed to Mitigate Long-Term Risks to Ships Assigned to Overseas Homeports. GAO-15-329. Washington, D.C.: May 29, 2015. Appendix IV: GAO Contact and Staff Acknowledgments Acknowledgments In addition to the contact named above, Suzanne Wren, Assistant Director; Chris Watson, Analyst in Charge; Herb Bowsher; Chris Cronin; Ally Gonzalez; Cynthia Grant; Carol Petersen; Amber Sinclair; and Cheryl Weissman made key contributions to this report.

Why GAO Did This Study According to the Navy, its 51 attack submarines provide the United States an asymmetric advantage to gather intelligence undetected, attack enemy targets, and insert special forces, among others. These capabilities make attack submarines some of the most–requested assets by the global combatant commanders. GAO was asked to review the readiness of the Navy's attack submarine force. This report discusses the extent to which the Navy (1) has experienced maintenance delays in its attack submarine fleet and costs associated with any delays; and (2) has addressed any challenges and developed mitigation plans for any maintenance delays. GAO analyzed readiness information from fiscal years 2008-2018, operating and support costs, maintenance performance, and other data; visited attack submarines and squadrons; and interviewed public and private shipyard and fleet officials. This is a public version of a classified report issued in October 2018. Information the Department of Defense deemed classified or sensitive, such as attack submarine force structure requirements and detailed data on attack submarine maintenance delays, has been omitted. What GAO Found The Navy has been unable to begin or complete the vast majority of its attack submarine maintenance periods on time resulting in significant maintenance delays and operating and support cost expenditures. GAO's analysis of Navy maintenance data shows that between fiscal year 2008 and 2018, attack submarines have incurred 10,363 days of idle time and maintenance delays as a result of delays in getting into and out of the shipyards. For example, the Navy originally scheduled the USS Boise to enter a shipyard for an extended maintenance period in 2013 but, due to heavy shipyard workload, the Navy delayed the start of the maintenance period. In June 2016, the USS Boise could no longer conduct normal operations and the boat has remained idle, pierside for over two years since then waiting to enter a shipyard (see figure). GAO estimated that since fiscal year 2008 the Navy has spent more than $1.5 billion in fiscal year 2018 constant dollars to support attack submarines that provide no operational capability—those sitting idle while waiting to enter the shipyards, and those delayed in completing their maintenance at the shipyards. The Navy has started to address challenges related to workforce shortages and facilities needs at the public shipyards. However, it has not effectively allocated maintenance periods among public shipyards and private shipyards that may also be available to help minimize attack submarine idle time. GAO's analysis found that while the public shipyards have operated above capacity for the past several years, attack submarine maintenance delays are getting longer and idle time is increasing. The Navy may have options to mitigate this idle time and maintenance delays by leveraging private shipyard capacity for repair work. But the Navy has not completed a comprehensive business case analysis as recommended by Department of Defense guidelines to inform maintenance workload allocation across public and private shipyards. Navy leadership has acknowledged that they need to be more proactive in leveraging potential private shipyard repair capacity. Without addressing this challenge, the Navy risks continued expenditure of operating and support funding to crew, maintain, and support attack submarines that provide no operational capability because they are delayed in getting into and out of maintenance. What GAO Recommends GAO recommends that the Navy conduct a business case analysis to inform maintenance workload allocation across public and private shipyards. The Department of Defense concurred with GAO's recommendation.

gao_GAO-18-380

gao_GAO-18-380_0

Background To help ensure that veterans are prescribed and use opioid pain medications in a safe and effective manner, VHA launched its OSI nationally in 2013. VHA established nine OSI goals in a December 2014 memorandum. For each of the goals, VHA directed its VISNs and their associated VHA medical facilities to take specified actions to meet the goals. (See table 1.) VHA has also developed opioid risk mitigation strategies for its providers to follow when prescribing opioid pain medications to veterans. These key strategies include one whose increased use is an explicit goal for the OSI and two that are requirements in VHA policy. 1. Annual urine drug screening for veterans on long-term opioid therapy. Providers should generally ensure that a urine drug screening has been conducted for veterans who are on long-term opioid therapy at least once in the 365 days prior to initiating or renewing an opioid prescription. Urine drug screening allows providers to monitor the types of drugs that are in a veteran’s system, including controlled and illicit substances. Increasing the use of urine drug screening is OSI goal two. 2. Annual prescription drug monitoring program (PDMP) query. PDMPs are state-run electronic databases used to track the prescribing and dispensing of prescriptions for controlled substances, identify suspected misuse or diversion (i.e., channeling drugs into illegal use), and identify trends in drug utilization. In 2016, VHA began requiring in policy that providers query state PDMPs at least once annually when prescribing opioids to determine whether their patients have received prescriptions for opioid medications or other controlled substances from non-VA providers. 3. Informed consent for long-term opioid therapy. In 2014, VHA began requiring in policy that providers educate their patients on the risks associated with the use of prescription opioids and to obtain veterans’ formal acknowledgment of these risks in writing prior to initiating long- term opioid therapy. Clinical Practice Guidelines for the Treatment of Chronic Pain In 2010, in coordination with DOD, VA developed clinical practice guidelines for its providers to use when prescribing opioids for chronic pain. These guidelines were updated in 2017. While clinical practice guidelines contain evidence-based recommendations, they are not required to be followed in all clinical situations; therefore, variations in practice may occur based on individual patient needs subject to the discretion of the provider. The 2017 VA/DOD clinical practice guidelines related to opioid therapy for chronic pain generally complement VHA’s OSI goals. For example, the guidelines recommend a conservative use of opioids for chronic pain and emphasize strategies to mitigate the risk of using opioids. The evidence- based clinical practice guideline recommendations include the following: 1. Use of non-pharmacological treatments. The guidelines advise not initiating opioid therapy for chronic pain. They also recommend alternatives to opioid therapy, such as non-opioid medications and non-pharmacological treatments. Non-pharmacological treatments for chronic pain include, for example, cognitive behavioral therapy and yoga. 2. Naloxone prescribing. Naloxone is a highly effective, potentially life- saving intervention for reversing opioid overdoses, and it can be prescribed to veterans as a preventive measure. Veterans who are prescribed naloxone can use it when experiencing an overdose or a family member can administer it on their behalf. According to the clinical practice guidelines, naloxone should be offered as an antidote to all patients at risk for an opioid overdose, including those who are in the process of tapering from opioids. The guidelines describe several significant risk factors which can indicate the prescribing of naloxone, including the duration and dose of opioids, current or history of depression or substance use disorder, and suicidality. 3. Appropriate follow-up visits with a provider. According to the guidelines, follow-up pain management visits should be scheduled at least every 1-4 weeks after any change in medication regimen and at least once every 1-3 months for the duration of the therapy to help ensure that the treatment plan is optimized. Under Its OSI, VHA Tracks Opioid Prescribing Patterns, Identifies Prescribing Outliers, and Educates Prescribers VHA officials told us that the main focus of the OSI is changing the prescribing patterns of providers to better align with evidence-based practices. We found that this has been carried out by VHA through three key efforts: tracking opioid prescribing rates and other trends, identifying irregular prescribing patterns, and educating providers on best practices through academic detailing. Tracking Opioid Prescribing Patterns. Under the OSI, VHA uses quarterly data derived from VHA’s electronic medical record system to monitor prescription opioid use among veterans, the related prescribing patterns of VHA providers, and the rates of urine drug screening for veterans receiving long-term opioid therapy. Specifically, VHA tracks the following four clinical indicators, known as the OSI metrics, for each of its medical facilities: 1. the percentage of patients dispensed an opioid, 2. the percentage of patients dispensed an opioid and a benzodiazepine, 3. the percentage of patients on long-term opioid therapy who received a urine drug screen within the previous year of having their prescription filled, and 4. the percentage of patients dispensed greater than or equal to 100 morphine milligram equivalents per day. Our analysis of quarterly OSI metric data shows that since the beginning of the OSI in the fourth quarter of fiscal year 2013 to the first quarter of fiscal year 2018 (the most recent data available at the time of our review), the percentage of veterans dispensed an opioid has decreased by 7 percentage points, or roughly 267,000 veterans, while the rate of urine drug screening for veterans on long-term opioid therapy increased significantly—by over 47 percentage points. The increase in the percentage of patients receiving a urine drug screening was driven more by a reduction in the total number of patients on long-term opioid therapy (about 197,000 veterans) than an increase in the number of patients receiving the screening (about 27,000 veterans). (See table 2.) Identifying Irregular Prescribing Patterns. As part of its monitoring of the OSI metrics, VHA Central Office has periodically identified VHA medical facilities and VHA providers who deviate from average prescribing rates across VHA. For example: Facilities. In 2014, VHA Central Office identified 39 of 140 medical facilities across 12 VISNs with relatively higher rates of opioid dispensing as outliers based on the OSI metric percentage of veterans dispensed an opioid. VHA Central Office notified the VISNs of these facilities and required each facility to submit a corrective action plan to VHA Central Office outlining the actions they would take to reduce opioid prescribing. Based on our analysis of VHA documents, we found that the identified facilities in the five VISNs selected for our review submitted information in response to VHA Central Office’s request. Providers. In February 2017, VHA Central Office identified 320 outliers out of 8,351 providers at 94 VHA medical facilities based on the relatively high proportion of their patients who were prescribed opioids. VHA Central Office directed the VHA medical facilities associated with these outlier providers to review their prescribing rates in the context of their clinical practice, and to report back with any feedback given or actions taken. According to officials from the five facilities in our review, outlier providers tended to be surgeons, pain management specialists, or physical rehabilitation providers who might be expected to prescribe opioids at a higher-than-average rate due to the nature of their specialty and the types of patients they treat. According to a VHA Central Office progress report, the facilities provided feedback and follow-up actions for 319 out of 320 outlier providers. In May 2017, VHA identified a second round of 303 outliers out of 8,505 providers; 187 of these providers were previously identified as outliers in February 2017. According to one VHA Central Office official, as of September 2017, VHA was reviewing these outlier data and will evaluate whether VHA facilities will be asked to conduct further reviews of these prescribers. Educating Providers through Academic Detailing. To help change the prescribing patterns of providers, VHA has also implemented a system- wide academic detailing program to educate providers and improve the delivery of evidence-based health care at facilities. In 2015, VHA required each VISN to establish such a program to improve performance on all OSI metrics. According to VHA Central Office officials, academic detailers are responsible for reviewing facility-level data on the prescribing patterns of providers and identifying potential areas of improvement. Detailers can educate providers with higher-than-average prescribing rates—such as those outliers identified in February and May 2017 by VHA Central Office—to help ensure providers are delivering safe and effective care for pain. According to VHA, as of January 2018, academic detailers have conducted over 20,000 opioid-related visits to VHA providers. According to VHA officials, their data also show that academic detailing results in greater patient safety for veterans taking opioids. For example, compared with those who did not receive academic detailing visits, providers who did receive such visits experienced (1) greater reductions in the proportion of their patients on high-dose opioids, (2) reductions in their patients’ average morphine milligram equivalent daily dosage, and (3) increases in their naloxone prescribing rates. VHA Has Made Progress on OSI Goals, but Performance Measurement Limitations Exist for Some Goals and Certain CARA Provisions Have Not Been Fully Implemented VHA Has Made Progress on Many 2014 OSI Goals, but for Some Goals, Performance Measurement Limitations Prevent an Accurate Determination of their Completion Based on our analysis of VHA information, we found evidence suggesting that the agency has accomplished six of the nine 2014 OSI goals. For example, the agency has seen increases in the use of urine drug screening for veterans on long-term opioid therapy, and it has developed provider tools to identify veterans at a higher risk for adverse events while using opioids. For several goals, although VHA did not implement the actions required in all those instances, the agency provided us with information or data demonstrating that the goals had effectively been met. However, for three OSI goals, it is unclear if the goal has been fully met because VHA lacks documentation showing that it has implemented the required action under the goal or the required action is still in progress. (See table 3.) (See appendix I for a more detailed description of VHA’s known efforts and data related to each goal). When asked about the lack of documentation for two of its OSI goals (goals four and seven), VHA officials told us that relevant documentation could not be produced. This lack of documentation is inconsistent with federal internal control standards. Specifically, according to federal internal control standards, management should evaluate and document the results of monitoring. By not documenting the actions it is taking under each of its OSI goals, VHA lacks assurance that these actions have been implemented by the VISNs or VHA medical facilities. As a result, VHA does not know whether it has fully met these OSI goals. Moreover, for the OSI goal related to establishing safe and effective VISN tapering programs for veterans using opioids and benzodiazepines (goal four), VHA officials told us that they addressed this goal by issuing national tapering guidance, including a provider reference guide in 2014, an opioid taper decision tool in 2016, and the VA/DOD clinical practice guidelines in 2017. However, these actions do not appear to be sufficient for meeting the goal as it is currently written, because issuing national guidance alone does not ensure that safe and effective tapering programs are established. Furthermore, VHA did not specify how safety and effectiveness within a tapering program would be measured, nor did the agency specify a deadline for the required action as described in December 2014. According to federal internal control standards relating to the establishment and review of performance measures and indicators, government agencies should use appropriate information to adequately assess performance, including establishing milestones or numerical targets, as appropriate. Without clearly defined and measurable outcomes, VHA cannot fully assess its progress towards meeting this OSI goal. VHA Has Not Implemented Certain CARA Requirements Related to Monitoring Veterans Prescribed Opioids We also found that VHA has not implemented two CARA requirements intended to improve opioid safety for veterans. First, CARA requires that VHA’s Opioid Therapy Risk Report (OTRR) have the ability to determine whether a provider has prescribed opioids to a veteran without checking that veteran’s information in the OTRR. Available to providers through VHA’s electronic medical record system, OTRR is a clinical tool that provides information on any opioid and concurrent benzodiazepine prescriptions a veteran is receiving, the veteran’s current and prior health conditions, recent and upcoming appointments, and whether any opioid risk mitigation strategies have been employed (such as urine drug screening or PDMP query). However, we found that VHA Central Office cannot track the extent to which VHA providers use OTRR because this tool does not have this tracking capability. VHA officials said that adding tracking capabilities to OTRR is not a high priority for the agency due to limited resources and competing priorities. Instead, according to a draft memorandum, VHA Central Office is planning to address this CARA provision by requiring VHA providers to document the use of OTRR in a standardized way that VHA can monitor. However, as of March 2018, VHA has not established this requirement or outlined the process for monitoring providers’ use of OTRR. Without the ability to track the use of OTRR, VHA cannot sufficiently monitor whether providers are using the tool to help reduce the likelihood of opioid-related adverse events occurring among veterans receiving care through VHA. CARA also requires that VHA modify its electronic medical record system so that any provider who accesses the record of a veteran will be notified whether that veteran is receiving opioid therapy and has a history of substance abuse disorder or prior instances of overdose; has a history of opioid abuse; or is at risk of developing an opioid use disorder. However, we found that VHA does not plan to modify its electronic medical record system to implement this capability. When asked about this provision in CARA, VHA officials said that VHA’s medical record currently has real- time alerts to inform providers about veterans’ existing opioid prescriptions and that any patient exposed to an opioid could be at risk of developing an opioid use disorder. Additionally, they said that an alert regarding current or past history of opioid use disorder could have an unintended consequence of discouraging veterans from reporting their medical history due to the stigma surrounding drug use disorders. VHA Providers Do Not Always Adhere to Required Opioid Safety Risk Mitigation Strategies and Guideline Recommendations for Prescribing Opioids in a Safe Manner VHA Providers at Selected Medical Facilities Do Not Consistently Follow Opioid Risk Mitigation Strategies Our review of selected VHA medical facilities shows that providers do not always follow three key opioid risk mitigation strategies, two of which are required under VHA policy. Specifically, increasing the use of urine drug screening is an explicit goal of the OSI, and providers should generally ensure that an annual urine drug screening has been conducted. VHA policy requires providers to 1) query state PDMPs at least annually when prescribing opioids to determine if the veteran has obtained opioid medications or other controlled substances from a non-VA provider and 2) obtain written informed consent from patients about the risks of initiating long-term opioid therapy. These strategies are intended to help ensure that patients at VHA medical facilities are safely prescribed opioid medications. Overall, based on our review of 103 veterans at five selected facilities, we found that 75 percent of the veterans in our sample had an annual urine screening, 26 percent had their names queried in a PDMP, and 70 percent provided informed consent. Provider Adherence to Three Veterans Health Administration (VHA) Opioid Risk Mitigation Strategies at Five Selected Medical Facilities, March 2016 through March 2017 Increasing the use of urine drug screening is an explicit goal of VHA’s Opioid Safety Initiative (OSI), and providers should generally ensure that an annual urine drug screening has been conducted. VHA policy requires providers to (1) query state prescription drug monitoring programs (PDMP) at least annually when prescribing opioids to determine if the veteran has obtained opioid medications or other controlled substances from a non-VA provider, and (2) obtain written informed consent from patients about the risks of initiating long-term opioid therapy. However, our review of medical records for a random nongeneralizable selection of 103 veterans subject to these risk mitigation strategies found that Of the 53 veterans who received long-term opioid therapy: 32 veterans received an annual urine drug screening, which allows providers to monitor the types of medications in a veteran’s system, including controlled and illicit substances; 13 veterans had their names queried annually in a state PDMP to see if they had received prescriptions for controlled substances, including opioids from non-VHA prescribers; and 41 veterans had provided informed consent indicating that they had been educated on the risks and benefits of opioid use. Of the 25 veterans prescribed an opioid and benzodiazepine concurrently: 17 veterans received an annual urine drug screening; 8 veterans had an annual PDMP query; and 22 veterans had provided informed consent. Of the 25 veterans with the highest risk of an adverse event, such as a suicide, overdose, fall, or accident, based on their Stratified Tool for Opioid Risk Mitigation (STORM) risk score: 24 veterans received an annual urine drug screening; 5 veterans had an annual PDMP query; and 11 veterans had provided informed consent. We identified a number of factors that may have contributed to the inconsistent adherence to the three key opioid risk mitigation strategies at our selected VHA medical facilities. These factors may impede providers’ ability to consistently follow these strategies for all applicable patients at these facilities. To the extent that these factors are present across other facilities, VHA’s ability to ensure that all veterans are prescribed opioids in a safe and effective manner may be limited. PDMP access issues. Officials at four of the five selected medical facilities faced PDMP access issues. Officials at two facilities told us that not all facility staff can access state PDMPs due to state laws and regulations that do not allow access to all types of providers, such as nurses and pharmacists. Officials at one of these selected facilities explained that nurse practitioners in that state cannot access the state’s PDMP, so they must rely on other providers to obtain information from the PDMP about their patients. In addition, in some states, only providers licensed in the state may access the state’s PDMP. Because providers at VHA facilities may not be licensed in the state where the VHA facility is located but licensed in another state, these providers may be unable to access the state’s PDMP. Officials at two selected facilities also described difficulties accessing PDMPs in neighboring states that are part of the catchment area for the facility and where the veteran may reside. The low rates of adherence we identified may also be attributed to the fact that VHA did not require providers to query the PDMP until October 2016, 7 months into our review of patients from March 2016 to March 2017. CARA directed VA to ensure access by VHA providers to information on controlled substances prescribing through state PDMPs, including by seeking to enter into memoranda of understanding with states to allow shared access of such information between states and the VA. According to VHA officials, VHA Central Office has not taken steps to develop memoranda of understanding with states, nor has it developed any related guidance. Officials said this issue is likely being addressed by individual VISNs and medical facilities. In addition, VHA officials told us they have communicated with Members of Congress and the National Governors Association to address issues related to VHA provider access to the state PDMPs. Lack of required staff to support providers. We also found that not all of the selected medical facilities and their respective VISNs had filled required staff positions that can help ensure provider adherence to opioid risk mitigation strategies—specifically, academic detailers and pain champions. At the time of our review, not all facilities had access to VISN academic detailing services, which, according to VHA, can help ensure that providers follow opioid risk mitigation strategies. While VHA officials said that most VISNs across VHA had implemented an academic detailing program as required by VHA policy, two of the five VISNs for the selected facilities in our review had not. Nationally, as of March 2018, four VISNs had not implemented an academic detailing program. Additionally, 11 facilities across VHA had not received a visit from a detailer. At the time of our review, four of five selected facilities did not have a Pain Champion as required by VHA policy beginning in March 2015. Pain champions are generally primary care providers knowledgeable about pain care who can serve as a resource for other primary care providers by promoting safe and effective pain care. According to VHA officials, pain champions play a critical role in opioid safety and can help providers remedy gaps in pain care management for individual patients, such as incomplete opioid risk mitigation strategies. Lack of clinical opioid safety alerts. Another factor that may limit adherence to the opioid risk mitigation strategies is the fact that none of the selected facilities employ electronic reminders to help remind primary care nurses of strategies that have not been completed. Primary care nurses are typically responsible for ensuring adherence to these strategies, and VHA facilities often employ electronic alerts to notify providers when certain tasks need to be completed, such as regular screenings for depression and traumatic brain injury. Although VHA facilities are not required to develop these alerts, according to some primary care nurses we interviewed, it would be helpful to receive a reminder when a veteran is due for a PDMP query, urine drug screening, or has not given long-term opioid use informed consent. According to the nurses, such an alert could be issued through the electronic medical record system. Limited facility monitoring. We found that facilities’ monitoring of provider adherence to the opioid risk mitigation strategies was limited across the five selected facilities in our review, which could hinder identification of non-adherence to these strategies. Specifically, while we found that all five medical facilities and VISNs in our review have an active pain management committee, facility officials told us that three of five facility committees do not conduct regular medical record reviews, which VHA encourages under its pain management directive to improve pain management. The directive states that facility pain management committees should monitor the pain management practices at their facility. For example, the pain management committee could monitor providers’ care plans for individual veterans, which are to be documented in the veterans’ medical records. These types of medical record reviews could help identify providers who are not adhering to VHA’s opioid safety requirements. Some VHA Providers Do Not Consistently Follow Clinical Practice Guideline Recommendations Related to Opioid Safety We also found that some VHA providers at selected facilities do not consistently follow selected clinical practice guideline recommendations related to opioid safety. Our findings are based on our review of a random selection of medical records for 103 veterans prescribed opioids between March 2016 and March 2017. These guidelines recommend, for example, that providers consider using non-pharmacological treatments, such as acupuncture and yoga, for chronic pain and prescribe naloxone, a potentially lifesaving drug, as warranted. The guidelines provide evidence-based recommendations designed to assist in provider decision-making; however, they are not VHA requirements and variations in practice will occur based on provider discretion and the needs of individual patients. Overall, we found that, 20 percent of veterans in our sample were prescribed a non-pharmacological therapy, 23 percent of the veterans were prescribed naloxone, 54 percent had appropriate maintenance follow-up visits with a provider while prescribed opioids, and 17 percent had appropriate follow-up visits with a provider after a change in their opioid prescription. Provider Adherence to Selected Clinical Practice Guideline Recommendations for Management of Opioid Therapy for Chronic Pain at Five Selected Veterans Health Administration (VHA) Medical Facilities, March 2016 through March 2017 Clinical practice guidelines provide evidence-based recommendations designed to assist in provider decision-making; however, they are not VHA requirements and variations in practice will occur based on provider discretion and the needs of individual patients. Our review of medical records for a random, nongeneralizable selection of 103 veterans subject to the recommendations found that Of the 53 veterans who had been prescribed long-term opioids: 11 veterans were prescribed a non-pharmacological therapy, such as yoga, or cognitive behavioral therapy; 13 veterans were prescribed naloxone, which is a highly effective intervention for reversing an overdose; 29 veterans had a maintenance follow-up visit at least once every 30-180 days for the duration of the veteran’s opioid therapy; and 3 of 21 veterans who had a change in their opioid medication during the time of our review had a follow-up visit between 14 and 28 days following the change. Of the 25 veterans prescribed a concurrent opioid and benzodiazepine: 6 veterans were prescribed a non-pharmacological therapy; 4 veterans were prescribed naloxone; 11 veterans had a maintenance follow-up visit; and 0 of 9 veterans who had a change in their opioid medication during the time of our review had a follow-up visit between 14 and 28 days following the medication change. Of the 25 veterans with the highest risk of an adverse event, such as a suicide, overdose, fall, or opioid-induced respiratory depression, based on their Stratified Tool for Opioid Risk Mitigation (STORM) risk score: 4 veterans were prescribed a non-pharmacological therapy; 7 veterans were prescribed naloxone; 16 veterans had a maintenance follow-up visit; and 5 of 17 veterans who had a change in their opioid medication during the time of our review had a follow-up visit between 14 and 28 days following the medication change. There are a variety of reasons that VHA providers may not always follow clinical practice guideline recommendations. For example, the availability of these non-pharmacological therapies may be limited, according to officials at all five selected VHA medical facilities. Officials at some facilities noted that the availability of these therapies can be particularly challenging for facilities in rural areas. VHA officials explained that the biggest barrier to providing naloxone is educating providers, so that they consistently consider prescribing naloxone for their patients receiving opioid therapy. According to officials, an education course for providers on naloxone prescribing became available in December 2015, and naloxone education efforts are a key focus of academic detailing programs. According to VHA data, since fiscal year 2014, naloxone distribution has increased. Specifically, as of March 2018, the agency has dispensed almost 142,000 naloxone kits to veterans, an increase of about 58 percent since June 2017. Conclusions VHA is taking important steps under the OSI to help ensure that veterans receive safe care. For example, VHA has begun tracking and publicly reporting data on four key metrics related to opioid prescriptions, and these data show that opioid prescription rates have decreased since 2013. In addition, our review also found that VHA has made progress on most of its 2014 OSI goals. However, for two goals, VHA lacks documentation showing whether VISNs and medical facilities have completed required relevant actions, and in one case, VHA has not specified measurable outcomes, which makes it challenging to determine whether these goals have been accomplished. Without sufficient documentation and measurable outcomes, VHA cannot determine whether these OSI goals to help ensure safe and effective care for veterans prescribed opioids have been fully successful. Our review also shows that VHA needs to do more to ensure that its providers are following three key opioid risk mitigation strategies when prescribing an opioid medication to a veteran: conducting an annual urine drug screening, querying a PDMP, and obtaining written informed consent from the veteran on the benefits and risks of using opioid medications. VHA has several means at its disposal for improving adherence to these strategies—at a minimum it should ensure that each VISN has a fully staffed academic detailing program and that each facility has a designated primary care pain champion, as VHA policy requires. In addition to enforcing these requirements, VHA should direct its facilities to strengthen monitoring efforts to help ensure providers’ adhere to the opioid risk mitigation strategies. These efforts include regular reviews of veterans’ medical records and creating electronic alerts reminding providers when these risk mitigation strategies have not been completed. Academic detailers and pain champions would also help educate providers further about evidence-based clinical practice guideline recommendations, such as non-pharmacological alternatives to opioid therapy and prescribing naloxone. Without these efforts to improve adherence to key opioid risk mitigation strategies, VHA’s ability to ensure that all veterans are prescribed opioids in a safe and effective manner may be limited. Recommendations for Executive Action We are making the following five recommendations to VA: The Undersecretary for Health should ensure that Central Office, VISNs, and medical facilities document the actions they take towards achieving OSI goals. (Recommendation 1) The Undersecretary for Health should ensure that any OSI goals that have not been met have clearly defined, measurable outcomes, including milestones or numerical targets, as appropriate, and timeframes. (Recommendation 2) The Undersecretary for Health should track the use of the OTRR (or any subsequent tool) by providers prior to initiating opioid therapy. (Recommendation 3) The Undersecretary for Health should ensure that all VISNs have implemented an academic detailing program that supports all medical facilities in the VISN and that all VHA medical facilities have a designated primary care pain champion as required. (Recommendation 4) The Undersecretary for Health should require VHA medical facilities to take steps to ensure provider adherence to opioid risk mitigation strategies, including querying PDMPs, obtaining written informed consent, and conducting urine drug screening. For example, these steps could include creating alerts in the electronic medical record system to remind primary care teams when these actions should be completed or strengthening facility monitoring of providers. (Recommendation 5) Agency Comments We provided a draft of this report to VA for comment. While VA was reviewing a draft of this report, it requested further specificity in recommendation two; as a result, we revised the recommendation to be clearer. In its written comments, which are reproduced in appendix II, VA concurred with our recommendations and provided technical comments, which we have incorporated as appropriate. In its comments, VA agreed that clarifying ongoing priorities and plans and filling in gaps in implementation will help facilitate progress in its opioid safety efforts. VA stated that it will establish a workgroup to review all OSI goals and ensure that the goals have clearly defined measurable outcomes and timelines, and that documentation requirements are established. VA also informed us that in March 2018 it published a notice requiring VHA clinicians to conduct and document a data-based risk review using one of VHA’s clinical decision support tools for opioid management, such as STORM, prior to initiating opioid therapy. VA also stated that it will take actions to ensure that academic detailing programs are fully implemented and primary care pain champions are in place across the system. To improve its clinicians’ adherence to opioid risk mitigation strategies, VA stated that it will establish a workgroup to review and develop methods for increasing adherence. VA expects to complete all these actions by April 2019 or earlier. We are sending copies of this report to the appropriate congressional committees, the Secretary of Veterans Affairs, the Undersecretary for Health, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-7114 or clowersa@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix III. Appendix I: Known Efforts and Data Related to the Veterans Health Administration’s (VHA) Opioid Safety Initiative (OSI) 2014 Goals, July 2013-December 2017 Description of known related efforts and data According to VHA training system data, 13 of 21 VISNs developed at least one education course addressing the use of urine drug screening from January 1, 2015 through April 29, 2017. Ten of the 13 VISNs developed at least one course by the deadline (December 31, 2014) as required. Further, 6 of the 13 VISNs had at least one course with 30 or fewer providers completing it. Because we found the VISN course assignment data unreliable, we could not determine the completion rates for these courses. However, we found that, in response to a 2015 White House memorandum that required all federal employee opioid prescribers to complete training on the appropriate prescribing of opioids (which included discussion of urine drug screening) by April 15, 2017, of 20,231 prescribers who were identified and assigned the course, 19,242 completed it, for a completion rate of about 95 percent, according to VHA data from January 1, 2015 through April 29, 2017. Urine drug screening targets based on previous urine drug screening rates for all VHA facilities must be reached by 2nd quarter, fiscal year 2015 (March 31, 2015) According to VHA data, 151 of 157 facilities met or exceeded their urine drug screening target by the deadline (March 31, 2015) as required. In addition, VHA OSI metric data also show that there has been a 47 percentage point increase nationally in the percentage of patients on long-term opioid therapy who received a urine drug screen from the 4th quarter of fiscal year 2013 to the 1st quarter of fiscal year 2018. The increase in the percentage of patients receiving a urine drug screening was driven more by a reduction in the total number of patients on long-term opioid therapy rather than an increase in the number of patients receiving the screening. VHA Central Office lacks certifications from all VISNs that programs were established. Although officials at the five VISNs in our review told us they had supported training efforts on the use of PDMPs, only one VISN in our review provided documentation that it had established a program by the deadline (March 31, 2015) as required. However, in October 2016, VHA issued a directive requiring providers to query state PDMPs for patients prescribed opioids and also issued guidance on how PDMPs should be accessed and how these efforts should be documented in VA’s electronic health record system. In addition, according to VHA officials, querying the PDMP was added to the 2015 White House-required opioid safety training, as described earlier. According to VHA data, there has been a 22 percent increase in the querying of PDMPs by providers from 4th quarter, fiscal year 2016 to 3rd quarter, fiscal year 2017. Description of known related efforts and data VHA Central Office lacks documentation from all VISNs regarding VISN-specific protocols and implementation plans. Only one of the five VISNs in our review provided documentation regarding a VISN-specific protocol relating to patients on opioids and benzodiazepines, which was developed in 2013. Four of the five VISNs did not provide documentation of a VISN-specific protocol or implementation plans. However, VHA officials told us that they addressed this goal by issuing national tapering guidance including a provider reference guide in 2014, an opioid taper decision tool in 2016, and the VA/DOD clinical practice guideline in 2017. VHA officials said that the issuance of this guidance made the VISN-required action irrelevant. In addition, VHA OSI metric data show that there has been a 6.6 percentage point decrease nationally in the percentage of patients dispensed an opioid and benzodiazepine from the fourth quarter of fiscal year 2013 to the first quarter of fiscal year 2018. In 2016, VHA released its opioid risk stratification toolkit in the form of an opioid safety monitoring tool called the Stratification Tool for Opioid Risk Mitigation. In addition, VHA updated its pain management opioid safety education guide and quick reference guide for providers in July 2017. VHA Central Office lacks documentation from all VISNs regarding VISN-specific protocols and implementation plans. Only one of the five VISNs in our review provided documentation regarding a VISN-specific protocol relating to patients on opioids and benzodiazepines, which was developed in 2013. Four of the five VISNs did not provide documentation of a VISN-specific protocol or implementation plans. VHA-required action(s) Each VISN must certify that the treatment of all patients with a daily dose of greater than 200 morphine milligram equivalents has been reviewed by 2nd quarter, fiscal year 2015 (March 31, 2015) Description of known related efforts and data VHA Central Office lacks certifications from all VISNs that reviews were conducted. Although officials at the five VISNs in our review told us they had completed these reviews, only one VISN provided us with documentation to conclude that its facilities completed their review. However, VHA OSI metric data show that there has been a 2.1 percentage point decrease nationally in the percentage of patients dispensed greater than or equal to 100 morphine milligram equivalents per day from the fourth quarter of fiscal year 2013 to the first quarter of fiscal year 2018. Each facility must provide evidence that at least two evidence-based behavioral/psychological treatments or approved complementary or alternative modalities can be provided by 2nd quarter, fiscal year 2015 (March 31, 2015) According to VHA 2nd quarter, fiscal year 2015 data, all VHA medical facilities located in the United States provided at least one psychosocial service and at least one complementary and integrative health service. In February 2016, VHA began a pilot to implement a model of care known as the Collaborative Chronic Care Model into existing Behavioral Health Interdisciplinary Program teams at nine VHA medical facilities. According to a VHA document, the pilot will provide facilitation support to enhance existing Behavioral Health Interdisciplinary Program teams by incorporating evidence-based Collaborative Care Model elements, which can include a care manager to proactively monitor care and progress as well as other tools intended to improve communication between primary care and specialty care. In fiscal year 2017, efforts were expanded to 30 additional VHA medical facilities. According to one VHA official, the pilot is expected to be completed no earlier than August 2019. VISNs are regional networks that manage the VHA medical facilities located in their area. In October 2015, VHA began to implement a realignment of its VISN boundaries which resulted in the number of its VISNs decreasing from 21 to 18. One VISN provided evidence of one VHA medical facility’s tapering recommendations for patients on opioids and benzodiazepines. Based on information we obtained from VHA relative to this goal, “psychosocial” services refer to “behavioral/psychological” treatments, and “integrated health” is a term that may be used to refer to “complementary and alternative” modalities. According to a VHA official, the VHA facility in Manila, the Philippines did not offer at least one psychosocial service and at least one integrative health service for this time period. Appendix II: Comments from the Department of Veterans Affairs Appendix III: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Marcia A. Mann, Assistant Director; Stella Chiang, Analyst-in-Charge; Emily Binek, Krister Friday, Diona Martyn, and Michael Rose made key contributions to this report. Also contributing were Zhi Boon and Emily Wilson.

Why GAO Did This Study The Comprehensive Addiction and Recovery Act of 2016 and Senate Report 114-57 included provisions for GAO to report on VHA's OSI and the opioid prescribing practices of its health care providers. This report examines, among other issues, (1) the extent to which VHA has met OSI goals established in 2014 and (2) the extent to which VHA providers adhere to key opioid risk mitigation strategies. To do this work, GAO reviewed data and documents related to OSI efforts and goals and interviewed VHA officials. In addition, GAO reviewed a random, nongeneralizable selection of medical records for 103 veterans who were prescribed opioids at five selected VHA medical facilities from March 2016 through March 2017. GAO selected the facilities to obtain diversity in geography and rates of opioid prescribing. At the selected facilities, GAO reviewed facility data and documents related to opioid safety and interviewed officials. What GAO Found The Veterans Health Administration (VHA) has made progress improving opioid safety through its Opioid Safety Initiative (OSI). Launched in 2013, the OSI aims to help ensure that veterans are prescribed opioids in a safe and effective manner. Since the OSI began, VHA has seen reductions in opioid prescribing rates. For example, from the fourth quarter of fiscal year 2013 to the first quarter of fiscal year 2018, the percentage of patients dispensed an opioid decreased from about 17 percent to about 10 percent, or by about 267,000 veterans. Also, available evidence suggests VHA has accomplished six of nine OSI goals established in 2014; however, it is unclear whether the remaining three goals have been fully met. For example, in the case of OSI goal four (establishing safe and effective regional tapering programs for patients on opioids and benzodiazepines), GAO found that VHA lacked documentation that its regional networks established these programs. VHA also did not establish measures of safety or effectiveness under this goal. These limitations prevent VHA from fully evaluating progress and accurately determining the extent to which its efforts to help ensure safe and effective prescribing of opioids have been successful. In a review of a nongeneralizable sample of 103 veterans' medical records at five selected VHA medical facilities, GAO found that VHA providers did not always adhere to key opioid risk mitigation strategies, which are required by VHA policy or relevant to OSI goals. For example, among 53 veterans who were prescribed long-term opioid therapy (defined as a 90-day supply in the last 6 months), GAO found that 40 veterans did not have their names queried in a state-run prescription drug monitoring program database. The databases are used to identify patients who are receiving multiple prescriptions that may place them at greater risk for misusing opioids or overdosing; 21 veterans did not have a urine drug screening within the year prior to having their prescription filled. The screenings are used to determine whether veterans are taking their opioid medications as prescribed; and 12 veterans did not provide written informed consent. Informed consent is a formal acknowledgement that the veteran has been educated on the risks and benefits of opioid use prior to initiating long-term opioid therapy. GAO found several factors that may have contributed to inconsistent adherence to key opioid risk mitigation strategies at the selected VHA facilities. For example, four of the five selected facilities did not have a pain champion (a primary care position required by VHA that can help providers adhere to opioid risk mitigation strategies), and not all facilities had access to academic detailing, a program in which trained clinical pharmacists work one-on-one with providers to better inform them about evidence-based care related to the appropriate treatment of relevant medical conditions. In addition, three of the five facilities did not consistently review veterans' medical records to ensure provider adherence to these strategies. To the extent that these factors affect all VHA facilities, VHA will continue to face challenges ensuring that its providers prescribe opioids in a safe and effective manner. What GAO Recommends GAO is making five recommendations to VHA, including that it document actions and develop measurable outcomes related to its OSI goals, ensure that providers are adhering to opioid risk mitigation strategies, and ensure that all its regional networks have implemented academic detailing programs and that all VHA medical facilities have a designated primary care pain champion, as required. The Department of Veterans Affairs concurred with GAO's recommendations and described steps it will take to implement them.

gao_GAO-19-233

gao_GAO-19-233_0

Background Marine Corps Organizational Structure The Marine Corps, within the Department of the Navy, organizes itself into different Marine Air Ground Task Forces. Each Marine Air Ground Task Force consists of a command element that includes a ground combat element, air combat element, and logistics combat element that can conduct operations across a broad range of crisis and conflict situations. As shown in figure 1, there are four types of Marine Air Ground Task Forces: Marine Expeditionary Forces (MEFs), Marine Expeditionary Brigades, Marine Expeditionary Units, and Special Purpose Marine Air Ground Task Forces. The MEF is the principal warfighting organization for the Marine Corps and consists of one or more divisions, including subordinate units such as regiments and battalions. There are three MEFs in the active component of the Marine Corps: I MEF—Camp Pendleton, California; II MEF—Camp Lejeune, North Carolina; and III MEF—Okinawa, Japan. Headquarters Marine Corps consists of the Commandant of the Marine Corps (Commandant) and the staff organizations, which are responsible for advising and assisting the Commandant to carry out duties. For example, the Deputy Commandant for Programs and Resources is responsible for developing, defending, and overseeing the Marine Corps’ financial requirements and the Deputy Commandant for Plans, Policies, and Operations is responsible for establishing policy, procedures, training, and guidance on unit readiness reporting. Marine Corps Unit-Level Training Marine Corps units train to their core missions—the fundamental missions a unit is organized or designed to perform—and their assigned missions—those missions which an organization or unit is tasked to carry out. Units train to a list of Mission Essential Tasks that are assigned based on the unit’s required operational capabilities and projected operational environments. For example, the Mission Essential Tasks for a Marine Corps infantry battalion include amphibious operations, offensive operations, defensive operations, and stability operations. Marine Corps Training and Readiness manuals describe the training events, frequency of training required to sustain skills, and the conditions and standards that a unit must accomplish to be certified in a Mission Essential Task. Marine Corps Readiness Unit commanders are responsible for their units’ readiness, including assessing and reporting their units’ capabilities to accomplish Mission Essential Tasks to specified conditions and standards. Unit readiness assessments are tracked in the Defense Readiness Reporting System– Marine Corps. This information provides Marine Forces Command, Headquarters Marine Corps, the Office of the Secretary of Defense, Joint Staff, and Combatant Commands, among others, a means to assess ground combat forces’ readiness trends and to assist with strategic and operational planning. Marine Corps O&M Budget The Marine Corps’ O&M budget funds a wide range of activities, including the recruiting, organizing, training, sustaining, and equipping of the service. The Department of Defense (DOD) uses the Planning, Programming, Budgeting, and Execution (PPBE) process to allocate resources to provide capabilities necessary to accomplish the department’s missions. In this report, we refer to the PPBE process as the budget cycle. The budget cycle includes the following phases: The planning phase of the budget cycle examines the military role and defense posture of the United States and DOD in the world environment and considers enduring national security objectives, as well as the need for efficient management of defense resources. The programming phase of the budget cycle involves developing proposed programs consistent with planning, programming, and fiscal guidance, reflecting, among other things, the effective allocation of resources. The budgeting phase of the budget cycle refers to developing and submitting detailed budget estimates for programs. The execution phase of the budget cycle involves spending funds. The Marine Corps’ Office of Programs and Resources has multiple divisions that support Program Objective Memorandum (POM) development, strategy, independent analysis, budget justification and legislative coordination, among others. Two key divisions that have responsibilities regarding Marine Corps resources are: The Budget and Execution Division is responsible for leading development and submission of the POM, providing quality control over programmatic and financial data, and allocating funds to major commands. According to a Marine Corps official, the division also assists with defending the Marine Corps’ budget request to Congress and others. The Program Analysis and Evaluation Division is responsible for providing Marine Corps senior leaders with independent and objective analysis to inform resource allocation decisions and assessing institutional risk. The Program Budget Information System (PBIS) is the primary information system used by the Navy and Marine Corps in the programming and budgeting phases of the budget cycle to develop and submit financial plans (i.e., the POM and the budget) to the Office of the Secretary of Defense. Once appropriated, funds are passed via allocation and allotment to subordinate units and executed via the Standard Accounting, Budgeting, and Reporting System (SABRS). SABRS is used to (1) record and report financial information; (2) provide an accounting and reporting system for the execution of appropriations; and (3) record financial transactions that originate from source systems. The Marine Corps Cannot Fully Track All Unit-Level Training Funds for Ground Combat Forces through the Budget Cycle Our analysis of data from the three MEFs for fiscal year 2017 funds shows that the MEFs had some data available that could be used to track some training funds from budget request to obligation. According to the Marine Corps’ Financial Guidebook for Commanders, as part of the budget cycle, commanders should determine the cost involved in meeting requirements, among other things. To help develop a sound budget, commanders need to know what they were and were not able to accomplish as a result of funding in previous years. However, Marine Corps officials told us they faced limitations tracking training funds, as discussed below. Specifically, as shown in table 1, we found that I MEF and II MEF were able to provide data on their fiscal year 2017 budget request, allotment, and obligations for training exercises directed at the MEF and division level, but data on exercises at smaller unit levels, such as regiments and battalions, were not consistently available because officials at those levels do not always track funds for these exercises. We found that III MEF was able to provide obligations data for fiscal year 2017 training exercises at all unit levels, but was not able to provide data on funds requested and allotted by training exercise. Officials at III MEF stated that these data were not available because III MEF incurs several large one-time expenses that contribute to training, but allocating those costs across specific training exercises is difficult. One of the primary reasons that the Marine Corps cannot fully track all training funds through the budget cycle is that the Office of Programs and Resources has not established the consistent use of fiscal codes to provide greater detail about the use of funds across the budget cycle phases, and the accuracy of these fiscal codes is sometimes questionable. The Marine Corps uses a variety of fiscal codes to track funds in the programming and execution phases of the budget cycle in the PBIS and SABRS systems, respectively. Some of these codes are used across DOD, while others are specific to the Marine Corps. Two key fiscal codes that officials identified as relevant to efforts to track funds for unit-level training are the Marine Corps Programming Code (MCPC) and the Special Interest Code (SIC). However, we identified limitations with how these fiscal codes are applied, as detailed below. MCPCs are used to program funds for intended use, but are not clearly linked to executed funds. When the Marine Corps programs funds for intended use, it uses MCPCs to identify the funds; however, when it executes those funds, it uses a different set of fiscal codes to identify them. As a result, the Marine Corps cannot link the programmed intent of the funds to the execution of the funds, making it difficult to track funds through the budget cycle. In fiscal years 2011, 2012, and 2013, the Marine Corps found in a series of reports that it faced challenges tracking funds through the budget cycle, in part because MCPCs were used to program funds, but not to track them in the execution phase. According to the fiscal year 2012 report, such tracking would enable the Marine Corps to improve financial traceability and add consistently reliable program execution data that would promote an understanding of the current fiscal environment to Marine Corps financial managers, comptrollers and others. In 2014, the Marine Corps implemented a process to include MCPCs in the execution phase of the budget cycle. The process enabled SABRS to automatically generate MCPCs for executed funds, based on the fiscal codes already used in the execution phase of the budget cycle. According to officials in the Office of Programs and Resources, this process increased the amount of executed funding that could be linked to an MCPC. However, Marine Corps officials told us that the mapping of MCPCs used in the programming phase to those used in the execution phase were not cleanly aligned, causing uncertainty about their linkage. The MCPCs associated with executed funds are estimates based on subject matter expert and working group mapping of fiscal codes to an MCPC and require continuous manual validation to ensure their accuracy. Additionally, the data quality of the multiple execution fiscal codes that are used to generate MCPCs is questionable because the data quality of the various underlying systems that feed data into SABRS is poor, according to officials in the Office of Programs and Resources. Senior Marine Corps officials from the Office of Programs and Resources told us that due to these limitations, analysts cannot be certain that executed funds associated with an MCPC as reflected in SABRS correspond to the purpose for which the funds associated with the same MCPC were programmed in the Program Budget Information System. This limits the Marine Corps’ ability to assess the extent to which funds were executed consistent with their programmed intent and track funds through the budget cycle. SICs are not used consistently across units. The Marine Corps uses SICs to track funds associated with individual training exercises. However, units, including the MEF and its subordinate units, do not consistently use SICs in identifying funds associated with all training exercises. Specifically, officials at all three MEFs told us that units generate SICs for large-scale training exercises directed at the MEF or division level, but may not generate SICs to track expenses for small-scale exercises at lower unit levels such as the regiment and battalion, making it difficult to track those funds. Officials at I MEF and II MEF stated that tracking costs associated with small-scale exercises is less consistent because units are not required to use SICs to track funds associated with exercises at those levels, and SICs associated with each exercise may change from year to year. Further, officials at I MEF and II MEF stated that supply officers are responsible for financial management at units below the division level, and they may not prioritize use of SICs. Officials at III MEF stated that tracking costs associated with specific exercises was difficult because officials could not attribute several large one-time training expenses to specific training exercises. Officials at all three MEFs stated that there is currently no systematic way to ensure that SICs are used accurately to associate funds executed with training exercises, which means they do not have complete or consistent data on costs associated with individual training exercises. As a result, commanders may lack accurate data for making resource decisions about training exercises needed to complete Mission Essential Tasks and improve units’ training readiness. In 2014, the Marine Corps issued Marine Corps Order 5230.23, Performance Management Planning, with the mission of linking resources to readiness and requiring the Deputy Commandant for the Office of Programs and Resources to ensure visibility and traceability of funds through the budget cycle and accounting systems for all organizational units and programs. Officials in the Office of Programs and Resources cited one effort to align inconsistent fiscal codes, but this effort will not directly address the challenges we have identified. According to officials in the Office of Programs and Resources, the Marine Corps is currently conducting a fiscal code alignment effort to address inconsistent use of fiscal codes, but this effort is in its early stages, and the Marine Corps has not yet developed clear guidance for implementation of the effort. Further, while the Marine Corps uses a variety of fiscal codes to track funds in the programming and execution phases of the budget cycle, an official from the Budget and Execution division told us that this effort will focus on fiscal codes that are used across DOD due to manpower limitations. However, MCPCs are unique to the Marine Corps and not recognized in larger DOD budgeting systems. As a result, the fiscal code alignment effort will not include aligning MCPCs across the programming and execution phases of the budget cycle, even though the Marine Corps will continue to use MCPCs. Additionally, although an official told us that SIC codes will be a part of this effort, implementation guidance for the effort was still under development and as a result, it is unclear whether the effort will address the inconsistent use of SICs across unit-level training exercises. Without the ability to track unit-level training funds through the budget cycle, including aligning MCPCs and ensuring consistent use of SIC codes, the Marine Corps lacks data to assess the extent to which funds were obligated consistent with their programmed intent and to adequately forecast and defend budget requests for training. As a result, commanders may face challenges making informed resource decisions. The Marine Corps Has Made Limited Progress Establishing a Link between Training Funds for Ground Combat Forces and Readiness The Marine Corps Identified a Need to Link Training Funds to Readiness, but Did Not Designate Responsibility to Meet That Need Although internal Marine Corps assessments and guidance state that the Marine Corps needs an enterprise-wide process to link resources to readiness, the Marine Corps has made little progress fulfilling this need. The Marine Corps has been aware for years of the challenges it faces in explaining its resource needs in its budget estimates to Congress. As stated in its 2009 Financial Guidebook for Commanders, “Many of the congressional cuts the Marine Corps receives are because of an inability to explain why we spent the money the way we did.” From fiscal years 2009 through 2014, the Marine Corps Office of Programs and Resources issued a series of classified and unclassified reports—referred to as the Marine Corps Strategic Health Assessments—that evaluated the health of the Marine Corps. The reports cited a number of factors inhibiting the Marine Corps’ ability to link funding to readiness, including stove-piped efforts, lack of an analytical framework, limited data availability, and poor data quality. For example, the fiscal year 2013 and 2014 reports found that the lack of a comprehensive model to connect the output of institutional processes to readiness measures hindered the Marine Corps’ ability to link funding to readiness. Table 2 below summarizes some of the key related findings in the reports. In fiscal year 2014, the Marine Corps stopped issuing the Marine Corps Strategic Health Assessments, in part, because the person responsible for preparing the analyses moved to another position. A senior Marine Corps official also told us that the reports were discontinued because producing them was no longer a priority for Marine Corps leadership. However, the Marine Corps also issued guidance in August 2014 calling for an enterprise-wide effort to link institutional resources to readiness. Specifically, Marine Corps Order 5230.23 called for the development and implementation of an enterprise-wide performance management process that links resources to institutional readiness via a robust analytic framework. The order included requirements to, among other things, identify readiness goals, develop strategic performance indicators, and improve data and business processes to include ensuring the visibility and traceability of funds. While implementing this order could address a number of the findings in the Marine Corps Strategic Health Assessments, Marine Corps officials told us that the service had not prioritized implementation of this order. Specifically, the Marine Corps did not designate a single oversight entity with the authority to enforce the order and directly oversee and coordinate efforts to link training funds to readiness. For example, although the order directed the Deputy Commandant for Programs and Resources to organize a quarterly coordination event of key stakeholders to synchronize activities within each major line of effort, officials from this office told us that they have not been given the authority to direct the various efforts. As a result, problems identified in the Marine Corps Strategic Health Assessments have persisted, and the Marine Corps does not have a comprehensive model to connect the output of institutional processes to readiness measures, as called for in the fiscal year 2013 Marine Corps Strategic Health Assessment. According to Standards for Internal Control in the Federal Government, management should establish an organizational structure, assign responsibility, and delegate authority to achieve its objective. Marine Corps officials told us the benefits of having a single entity to oversee efforts to tie funds to readiness include having one authority responsible for ensuring a consistent data architecture—how data will be collected, stored and transferred across the Marine Corps—and data quality. Further, having a single entity would help ensure a unified approach that would help analysts better answer questions about how funds affect readiness. Marine Corps Has Not Assessed Its Current Initiative to Link Dollars to Readiness In the absence of a single entity responsible for overseeing the Marine Corps’ efforts to link training funds to readiness, two different organizations within the Marine Corps developed separate and overlapping initiatives. First, in 2012, the Commanding General of II MEF directed the development of C2RAM, a tool that attempts to link funding to readiness for ground combat forces by capturing and correlating resources and requirements associated with specific unit-level training exercises. C2RAM was developed in response to our recommendation that the Marine Corps develop results-oriented performance metrics that can be used to evaluate the effectiveness of its training management initiatives. The tool, a complex excel-based spreadsheet, is used to capture day-to-day operating costs for training exercises to meet a unit’s core and assigned Mission Essential Tasks for training readiness requirements. For example, unit operations and resource officials enter data on training exercise costs and the Mission Essential Tasks expected to be accomplished by each exercise, and the tool uses this data to project the unit’s expected training readiness levels. Further, commanders can use the tool to project the expected effect of decreases in funding on training readiness levels. According to Marine Corps officials, they spent approximately $11 million on the C2RAM initiative from fiscal years 2012 through 2017. Second, in 2015, the Headquarters Marine Corps Office of Programs and Resources adopted and made adjustments for Marine Corps purposes to the Air Force’s Predictive Readiness Assessment system and test-piloted it with Marine Corps units. The Marine Corps’ system was known as the Predictive Readiness Model (PRM). PRM was designed to evaluate the complex interactions between resources and readiness to help inform decisions about resource allocations and readiness outcomes. According to Headquarters Marine Corps officials, PRM attempted to map approximately 500 causal factors related to readiness ratings. The effort involved input from more than 70 subject matter experts from multiple Marine Corps organizations. In addition, data input into PRM was obtained from various authoritative sources, including readiness, financial, and training systems of record, as well as other unauthoritative sources, including C2RAM. According to Marine Corps officials, as of June 2018, the Corps had spent approximately $4 million to develop PRM. In March 2019, while responding to a draft of this report, the Marine Corps stated that it decided to discontinue development of PRM because the model did not meet its objectives. While these initiatives were both designed to help the Marine Corps link dollars to readiness, each had its own particular use and design. For example, unlike C2RAM, which focuses only on the training pillar of readiness for ground combat forces, PRM focused on all pillars of readiness tracked by the Marine Corps for ground combat forces and air combat forces. In addition, while PRM attempted to capture all training data, C2RAM does not. For example, it does not capture data on individual training. Moreover, while C2RAM is primarily used at the MEF level and below to help inform commanders’ decisions about how much training funding to request and identify the effect of funding on readiness, PRM was designed to help officials in Marine Corps Headquarters make service-wide decisions about budget development and resource allocation. During our review, we found data quality and classification challenges faced by both PRM and C2RAM, as discussed below. Data quality limitations. Some Headquarters Marine Corps officials questioned the accuracy and reliability of some of the data planned for use in PRM because the data had to be aggregated from multiple sources that have varying degrees of internal control. In addition, officials told us that existing data were insufficient or are not currently collected, so, in some cases, PRM had to rely on the opinion of subject matter experts to determine how causal factors affect readiness. According to Marine Corps officials at various levels, C2RAM data quality is questionable because data is manually input by various sources with varying degrees of expertise. This is exacerbated by weak processes for conducting quality checks of the data. Moreover, officials stated that cost data may be inaccurate because units may neglect to update cost estimates with actual costs after a training event is completed. Further, C2RAM is not consistently used across all three MEFs. For example, when we visited II MEF, we learned that their resource management officials do not use C2RAM to build their budgets because of concerns about data quality. Classification of Data. Another challenge that both efforts faced is the classification of aggregated data. Readiness data are classified; budget data are generally not. When these data are combined, the resulting data are classified, potentially making the tool less useful and available to officials seeking to make informed decisions about resource allocation. For example, C2RAM is currently an unclassified system that captures fiscal and training data, but not readiness data. However, officials at I Marine Expeditionary Force told us that if readiness data were incorporated into the tool, it could become classified, which would limit its availability and usefulness to lower unit levels. As the Marine Corps found in its Fiscal Year 2012 Strategic Health Assessment, its stove-piped processes often require integration at the senior leadership level to develop a comprehensive view of issues, including the effect of dollars on readiness. Development of C2RAM and PRM, however, was not integrated, resulting in two separate systems— each devoted to tackling the same problem, but in different ways and with similar weaknesses, such as data quality limitations. Moreover, there was some overlap between the two systems. For example, C2RAM was one of the many data sources for PRM. In addition, both PRM and C2RAM used some of the same data sources. For instance, both systems relied on information captured in the Marine Corps Training Information Management System as well as on data captured in SABRS. The Marine Corps assessed the feasibility of moving forward with the PRM tool and, in March 2019, while responding to a draft of this report, the Marine Corps stated that they decided to discontinue its development. However, the Marine Corps has not assessed C2RAM as part of an enterprise wide performance management process that links resources to readiness. For example, the Marine Corps could learn from the experience of commanders at the MEF level who find C2RAM useful and consider the extent to which those usability considerations could and should be brought into a service-wide model. Without conducting this analysis, the Marine Corps is unlikely to make headway in tackling the challenges posed by trying to link resources to readiness. Conclusions To meet the demands of its missions, the future security environment will require military forces to train across the full range of military operations, according to DOD. While the Marine Corps continues to ask for increased funding, according to a congressional report, the Marine Corps is unable to provide sufficient detail in its O&M budget estimates for training that would allow Congress to determine the benefits gained from additional funding. The Marine Corps has been aware for many years of the importance of providing accurate budget justifications to Congress. A number of factors have made it challenging for the Marine Corps to provide Congress the information it needs. First, the Marine Corps cannot fully track training funds through the budget cycle, making it difficult for the Marine Corps to, among other things, show that training funds were spent as planned. Second, the Marine Corps has not prioritized tackling the longstanding problem of how to link training resources to readiness. Although the Marine Corps has a standing order to develop an enterprise- wide performance management framework that links resources to readiness via a robust analytical framework, no single entity has been assigned the authority to enforce this order. In the absence of that leadership, certain components of the Marine Corps have developed their own, independent initiatives that were designed to achieve the same objective of linking funding to readiness, but had their own specific approaches and intended uses. Moreover, the Marine Corps has not assessed whether C2RAM provides an enterprise-wide performance management process linking resources to readiness. Until the Marine Corps assigns the authority needed to oversee development and implementation of a methodologically sound approach and assesses the degree to which C2RAM could be used, it will continue to face challenges making fully informed decisions about how much money it needs for training purposes and what it can reasonably expect to deliver for that money in terms of readiness gains. Recommendations for Executive Action We are making the following three recommendations: The Secretary of the Navy should ensure that the Deputy Commandant for the Office of Programs and Resources oversee development and implementation of an approach to enable tracking of unit-level training funds through the budget cycle. This approach should include aligning MCPCs across the Marine Corps and ensuring consistent use of SIC codes. (Recommendation 1) The Secretary of the Navy should ensure that the Commandant of the Marine Corps designates a single entity responsible for directing, overseeing, and coordinating efforts to achieve the objective of establishing an enterprise-wide performance management process that links resources to readiness. (Recommendation 2) The Secretary of the Navy should ensure that the Commandant of the Marine Corps assesses C2RAM to determine the extent to which this system, or elements of this system, should be adapted for use in an enterprise-wide performance management process linking resources to readiness. (Recommendation 3) Agency Comments and Our Evaluation We provided a draft of this report to DOD for review and comment. In written comments, DOD concurred with all three of the recommendations in the draft report and stated that the Marine Corps would take actions to track unit-level training funds, link resources to readiness, and examine C2RAM, as discussed below. DOD’s comments are reprinted in appendix II. DOD also provided technical comments, which we incorporated as appropriate. DOD concurred with the third recommendation in the draft report that the Secretary of the Navy should ensure that the Commandant of the Marine Corps assesses C2RAM and PRM and determine the extent to which these systems or elements of these systems could and should be adapted for use in the enterprise-wide performance management process linking resources to readiness. In its comments, the Marine Corps stated that work to develop PRM had been discontinued because the model did not satisfy the Marine Corps objectives. Given that the Marine Corps’ decision to stop development of PRM mitigates the potential for overlapping initiatives moving forward, we revised the report and recommendation to focus on the Marine Corps assessing C2RAM for use in the enterprise-wide performance management process linking resources to readiness. The Marine Corps stated in its written response that C2RAM has potential utility for supporting an understanding of resources to readiness and it will examine the system further. We are sending copies of this report to the appropriate congressional committees, the Acting Secretary of Defense, the Secretary of the Navy, and the Commandant of the Marine Corps. In addition, the report is available at no charge on our website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-2775 or FieldE1@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix III. Appendix I: Objectives, Scope, and Methodology This report evaluates the extent to which the Marine Corps (1) tracks unit- level Operations and Maintenance (O&M) training funds for ground combat forces through the budget cycle; and (2) links unit-level training funds for ground combat forces to readiness. This report focuses on ground combat forces which conduct a myriad of training at the Marine Expeditionary Forces (MEF). For our first objective, we requested and analyzed budget request, allocation, and obligations training exercise data for fiscal year 2017 from I Marine Expeditionary Force (MEF), II MEF, and III MEF. We collected data for this fiscal year because it was the most recently completed fiscal year for which actual obligated amounts could be obtained. We used this data request to determine the Marine Corps’ ability to provide the data as well as determine the source or sources they used to provide the data. We discussed the systems—Cost to Run a Marine Expeditionary Force (C2RAM) and Standard Accounting, Reporting, and Budgeting System (SABRS)—used to provide this data with knowledgeable Marine Corps officials, including discussion of the data reliability concerns with these systems which are identified in this report. We interviewed knowledgeable officials about the systems, reviewed the user guide for one of the systems, and observed how data was input and extracted to form reports. Although we found the data to be insufficient to consistently identify and fully track unit-level O&M training funding data though the budget cycle, we determined that the data we obtained were sufficiently reliable to provide information about the availability of fiscal year 2017 funding amounts requested, allotted, and obligated for unit-level training exercises, as discussed in this report. We also reviewed and analyzed data from a series of classified and unclassified reports that were issued by the Marine Corps from fiscal year 2009 through fiscal year 2014. These reports, known as the Marine Corps Strategic Health Assessment (MCSHA), evaluated the health of the Marine Corps, including its use of fiscal codes, through an enterprise- wide study of resource investments, organizational activities, and readiness outcomes. We also reviewed data about Marine Corps Programming Codes (MCPC) and Special Interest Codes (SIC) in Marine Corps documents such as the MCSHAs as well as the Standard Accounting, Budgeting, and Reporting System (SABRS) Customer Handbook. We assessed this information against Marine Corps Order 5230.23, Performance Management Planning, which requires the Deputy Commandant for Programs and Resources to ensure visibility and traceability of funds through the budget cycle and accounting systems for all organizational units and programs, as well as Standards for Internal Control in the Federal Government, which states that management should design an entity’s information system to ensure, among other things, that data is readily available to users when needed. For our second objective, we reviewed reports and supporting documentation on Marine Corps efforts to evaluate readiness levels achieved from O&M obligations for ground combat forces training and observed the operation of systems used to track training funds and readiness. Specifically, we reviewed and analyzed the MCSHAs to identify challenges that the Marine Corps reported facing in attempting to link training funds to readiness. As a part of our review of supporting documentation, we reviewed and analyzed the MCSHAs from fiscal years 2011 through 2014 issued by the Marine Corps Office of Program Analysis and Evaluation to summarize some of the key findings identified by the Marine Corps related to linking training funds to readiness. We reviewed these reports because they intended to provide a comprehensive overview of the health of the Marine Corps. From these reports, we identified and summarized key findings related to our review. Specifically, one GAO analyst reviewed the four reports to identify reported findings that prevent the Marine Corps from linking resources to readiness, such as stove-piped processes and inconsistent data management processes, while a second analyst confirmed the summary from this review. We shared our summary of key findings with Marine Corps officials and they concurred. In addition, we reviewed guidance and other related documents on the Predictive Readiness Model (PRM) and Cost to Run a Marine Expeditionary Force (C2RAM). We were briefed on and observed data being input into the C2RAM model and queries being run from that data. We were able observe the summary reports that resulted from the queries which helped to enhance our understanding of the Marine Corps efforts to link training funds to readiness. In addition, we reviewed previously issued GAO reports related to the issue. We assessed this information against Marine Corps Order 5230.23, Performance Management Planning, which calls for the development and implementation of an enterprise-wide performance management process that links resources to institutional readiness via a robust analytic framework, as well as Standards for Internal Control in the Federal Government, which states that management should establish an organizational structure, assign responsibility, and delegate authority to achieve its objective. To answer the two objectives for this review, we interviewed knowledgeable officials from the following offices: Office of the Secretary of Defense Cost Assessment and Program Evaluation Personnel and Readiness, Force Readiness Headquarters Marine Corps, Washington, D.C. Office of Programs and Resources Budget and Execution Division Program Analysis and Evaluation Division Command, Control, Communications, and Computers Marine Forces Command – Norfolk, Virginia Marine Corps Training and Education Command – Quantico, Virginia I Marine Expeditionary Force – Camp Pendleton, California II Marine Expeditionary Force – Camp Lejeune, North Carolina III Marine Expeditionary Force – Okinawa, Japan. We conducted this performance audit from August 2017 to April 2019 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Comments from the Department of Defense Appendix III: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact name above, Margaret Best, Assistant Director; William J. Cordrey; Pamela Davidson; Angela Kaylor; Amie Lesser; Tamiya Lunsford; Samuel Moore, III; Shahrzad Nikoo; Clarice Ransom; Cary Russell; Matthew Ullengren; and Sonja Ware made key contributions to this report.

Why GAO Did This Study Training is key to building readiness—the military's ability to fight and meet the demands of its missions. Through the Department of Defense (DOD) budget cycle, the Marine Corps estimates or programs its funding needs for training and spends funds to accomplish its training mission. Questions have been raised about whether the Marine Corps' training budget estimates are sufficiently detailed to determine training costs at the unit level or the expected readiness generated by those costs. House Report 115-200 included a provision for GAO to examine the military services' budgeting processes to build unit-level training readiness. This report examines the extent to which the Marine Corps (1) tracks unit-level training funds for ground combat forces through the budget cycle, and (2) links ground combat forces' unit-level training funds to readiness. GAO analyzed budget data and studies conducted by the Marine Corps and others, examined tools used by units to link training funds with readiness, and interviewed knowledgeable officials at various levels in the Marine Corps. What GAO Found The Marine Corps cannot fully track all unit-level training funds for ground combat forces through the budget cycle. According to GAO's analysis of data provided by the Marine Expeditionary Forces (MEFs), the principal warfighting organization for the Marine Corps, units can track some, but not all, funds for training exercises from the budget request through use of the funds. The Marine Corps cannot fully track all training funds through the budget cycle, in part, because it has not established the consistent use of fiscal codes. Two key fiscal codes that officials identified as relevant to track funds for unit-level training are the Marine Corps Programming Code (MCPC) and the Special Interest Code (SIC). The Marine Corps uses MCPCs to program funds, but GAO found that when the Marine Corps spends those funds, it uses a different set of fiscal codes. This makes it difficult to link the programmed intent of funds to the execution of those funds. The Marine Corps uses SICs to track funds associated with training exercises, but GAO found that units do not use SICs consistently. For example, officials at all three MEFs told GAO that units generate SICs for large-scale training exercises, but may not do so for small-scale exercises. The Marine Corps is taking steps to align fiscal codes across the budget cycle, but this effort is in its early stages and will not include MCPCs, and may not address the inconsistent use of SICs. Without the ability to track unit-level training funds through the budget cycle, the Marine Corps lacks readily available data to assess whether funds were obligated consistent with their programmed intent and to adequately forecast and defend budget requests for training. Although internal Marine Corps assessments and guidance state that the Marine Corps needs an enterprise-wide process to link resources to readiness, the Marine Corps has made little progress establishing a link between training funds for ground combat forces and readiness. The Marine Corps identified challenges with linking funds to readiness in a series of reports from fiscal years 2009 through 2014, citing factors such as stove-piped efforts and limited data availability and quality. Guidance directed that the Deputy Commandant for Programs and Resources organize quality coordination events with key stakeholders to synchronize activities within major lines of effort, but officials from this office stated that they have not been given the authority to direct the various efforts. Therefore, challenges have persisted, in part, because the Marine Corps has not designated a single entity with authority to oversee and coordinate efforts to link training funds to readiness. In the absence of a single oversight entity, two separate and overlapping tools were developed—the Cost to Run a MEF (C2RAM) tool and the Predictive Readiness Model (PRM). Although each tool had its own particular use and design, both were intended to link resources to readiness. Moreover, both faced similar challenges, such as data quality limitations, and relied on some of the same data sources. The Marine Corps recently assessed and discontinued development of PRM, however, it has not assessed C2RAM and how it could support an enterprise wide performance management process linking resources to readiness. Without dedicating a single entity with authority, and conducting an assessment of C2RAM, the Marine Corps is unlikely to make headway in addressing the challenges posed by trying to link resources to readiness. What GAO Recommends GAO recommends that the Marine Corps (1) tracks training funds through the budget cycle, (2) designates a single entity to oversee establishment of a process that links resources to readiness, and (3) conducts an assessment of C2RAM. DOD concurred, and based on its comments, GAO modified one recommendation.

gao_GAO-18-576T

gao_GAO-18-576T_0

Background NASA Acquisition Life Cycle for Space Flight Projects The life cycle for NASA space flight projects consists of two phases— formulation, which takes a project from concept to preliminary design, and implementation, which includes building, launching, and operating the system, among other activities. NASA further divides formulation and implementation into phases, phase A through phase F. Major projects must get approval from senior NASA officials at key decision points before they can enter each new phase. Formulation culminates in a review at key decision point C, known as project confirmation, where cost and schedule baselines are established and documented in a decision memorandum. Figure 1 depicts NASA’s life cycle for space flight projects. At the time of our review in May 2018, NASA had a portfolio of 26 major projects (see table 1). See appendix I for a brief description of each project. NASA Acquisition Management as a High- Risk Area NASA acquisition management is an area that we monitor on our high- risk list. Our high-risk series is a biennial report that keeps focused attention on government operations with greater vulnerabilities to fraud, waste, abuse, and mismanagement or that are in need of transformation to address economy, efficiency, or effectiveness challenges. In 1990, we first designated the area as high risk because there was little emphasis on end results, product performance, and cost control; the acquisition process itself was cumbersome and time-consuming; and NASA found itself procuring expensive hardware that did not work properly. For example, in April 1990, NASA deployed the $1.5 billion Hubble Space Telescope and soon after, the agency discovered that the primary mirror had been manufactured in the wrong shape, severely degrading some of the telescope’s scientific capabilities. Subsequently, we and other organizations, including the National Academy of Sciences and NASA’s Office of the Inspector General, found that NASA’s cost estimates were overly optimistic. Our reviews also found that NASA continued to experience significant cost and schedule growth due, in part, to not having a disciplined cost estimating process. In 1992, we reviewed the cost and schedule performance of 29 NASA programs and found that 25 of those programs experienced cost growth that ranged from 14 to 426 percent above their initial estimates. Further, the median estimate change for all programs was an increase of 77 percent. General reasons that NASA provided for the cost growth included insufficient definition studies, program and funding instability, overly optimistic assumptions by program officials, and unrealistic contractor estimates. The more specific reasons for the cost growth we found included program redesigns, technical complexities, budget constraints, and incomplete cost estimates. In 2004, we reviewed the cost and schedule performance of 27 NASA programs and found that 17 of the programs experienced cost growth. Cost growth for 10 of the 17 programs was over 25 percent. We found that considerable change in NASA’s program cost estimates—both increases and decreases—indicated that NASA lacked a clear understanding of how much its programs cost and how long they will take to achieve their objectives. Further, we found that NASA’s basic cost-estimating processes—an important tool for managing programs—lacked the discipline needed to ensure that program estimates are reasonable. In more recent years we have found that NASA’s leadership was focused on improving acquisition outcomes and had taken some steps to improve its management. In 2006, NASA established a management review process to enable NASA’s senior management to more effectively monitor a project’s performance, including cost, schedule, and cross-cutting technical and nontechnical issues. In 2009, NASA began requiring that NASA major programs and projects develop a joint cost and schedule confidence level (JCL) prior to project confirmation in order to ensure that cost and schedule estimates were realistic and projects thoroughly planned for anticipated risks. The JCL is a point-in-time estimate that, among other things, includes all cost and schedule elements, incorporates and quantifies known risks, assesses the impacts of cost and schedule to date, and addresses available annual resources. NASA policy generally requires that projects be baselined and budgeted at the 70 percent confidence level. In 2012, the agency established metrics to more consistently measure a project’s design progress and, in 2014, we found that most major projects in the portfolio were tracking and reporting those metrics. In addition, experts with whom we met confirmed that NASA’s metrics are valid measures to assess design maturity in space systems. Since 2015, we have observed a positive trend of higher numbers of projects maturing technologies prior to preliminary design review. Demonstrating that technologies will work as intended in a relevant environment serves as a fundamental element of a sound business case, and projects falling short of this standard often experience subsequent technical problems. Our best practices work has shown that maturing technologies prior to preliminary design review can minimize risks for projects entering development, which lowers the risk of subsequent cost growth and schedule delays. We believe that many of these steps NASA has taken contributed to the largely positive trend of cost and schedule performance for NASA’s portfolio of major projects between 2013 and 2017. In our May 2017 assessment of major projects, we found that out of 16 projects in development, 5 experienced cost growth and 4 experienced schedule delays over their development cost and schedule baselines. Both of these measures were at or near the lowest levels we have reported since we began our annual assessments in 2009. However, we also found in our February 2017 high risk update that NASA needed to do more with respect to anticipating and mitigating risks— especially with regard to large programs, estimating and forecasting costs for its largest projects, and implementing management tools. We highlighted several actions that would be critical to improving NASA’s acquisition outcomes, including the following: Ensuring that NASA conducted adequate and ongoing assessments of risks for larger programs because the impacts of any potential miscalculations will be felt across NASA’s portfolio. Ensuring that NASA understood long-term human exploration program costs. While the three major human exploration programs— Orion, SLS, and the Exploration Ground Systems (EGS)—have been baselined, none of the three programs has a baseline that covers activities beyond the second planned flight. Long-term estimates, which could be revised as potential mission paths are narrowed and selected, would provide decision makers with a more informed understanding of costs and schedules associated with potential agency development paths. Ensuring that program offices regularly and consistently updated their JCL across the portfolio. As a project reaches the later stages of development, especially integration and testing, its risk posture may change. An updated project JCL would provide both project and agency management with data on relevant risks that can guide project decisions. Ensuring that NASA continued its efforts to build capacity in areas such as cost and schedule estimating and measuring contractor performance. Further, in our 2016 and 2017 assessments of major projects, we found that while the cost and schedule performance of NASA’s portfolio was improving, a number of large, complex projects were in or would soon be entering the integration and test phase—the phase in development that often reveals unforeseen challenges that can lead to cost and schedule growth. In May 2017, projects in this phase included all three human spaceflight programs and the James Webb Space Telescope (JWST). Subsequently, we found that these programs experienced delays during this phase of development. For example, in December 2017, NASA announced a 13- to 19-month delay for the first integrated mission of Orion, SLS, and EGS. This mission is referred to as Exploration Mission 1 (EM-1) and will not have crew. In addition, in December 2017, we found that the JWST project continued to make progress towards launch, but the program was encountering technical challenges that required both time and money to fix and may lead to additional delays. Subsequently, the JWST project delayed its launch readiness date by at least 19 months from October 2018 to May 2020. Portfolio Cost and Schedule Deteriorated but Extent of Cost Growth Is Unknown The cost and schedule performance of NASA’s portfolio of major projects deteriorated between May 2017 and May 2018, but the extent of cost growth is unknown. NASA lacks a current cost estimate for its Orion crew capsule—one of the largest programs in the portfolio—but expects the program will exceed its cost baseline when NASA updates the program’s life-cycle cost estimate. Because the Orion program accounts for about 22 percent of all development costs, even a small percentage of cost growth for the Orion program could significantly affect portfolio cost performance. The known negative cost and schedule performance is largely driven by the cost and schedule growth of four projects—SLS, EGS, Space Network Ground Segment Sustainment (SGSS) and Mars 2020—that experienced technical problems compounded by programmatic challenges. Together, these projects experienced $638 million in cost growth and 59 months in aggregate schedule delays. Two projects—JWST and ICESat-2—experienced schedule delays due to technical challenges identified during integration and test. Another 3 projects—NASA Indian Space Research Organisation Synthetic Aperture Radar (NISAR), ICON, and GRACE-FO—experienced cost growth or delays largely due to factors outside of the projects’ control, such as launch vehicle delays. Portfolio Average Launch Delays Increased, but NASA Lacks a Current Orion Program Cost Estimate to Determine Extent of Cost Growth The average launch delay increased from 7 months in our May 2017 report to 12 months in our May 2018 report—the highest schedule delay we have reported to date. We were not able to determine the extent of portfolio cost growth this year because NASA does not have a current cost estimate for the Orion program—one of the largest programs in its portfolio—and officials expect the cost to increase. As of June 2017, the Orion program’s development cost was about $6.6 billion; based on that estimate, it accounts for 22 percent of the portfolio’s estimated $30.1 billion of development costs. As a result, a small percentage of cost growth for the Orion program could significantly affect cost performance. Even without including Orion cost growth, the overall development cost growth for the portfolio of 17 development projects increased to 18.8 percent, up from 15.6 percent in 2017 (see figure 2). Senior-level NASA officials told us they expect that the Human Exploration and Operations Mission Directorate and the Orion program will complete an updated life-cycle cost estimate in June 2018. This would be approximately 10 months after the program raised to senior-level officials’ attention that the program expects cost growth over its cost baseline during an August 2017 briefing concerning potential cost increases related to the launch delay for EM-1. In early June 2018, NASA officials said that they had not yet completed the updated life-cycle cost estimate. In our May 2018 report, we found that 7 of 17 NASA major projects had stayed within cost and schedule estimates since our 2017 annual assessment of major projects, but 9 projects experienced cost growth or schedule delays and cost growth is expected for the Orion program. Table 2 provides data on the cost and schedule performance between our May 2017 and 2018 reports for the 17 major projects in development that have cost and schedule baselines. The deteriorating cost and schedule performance of the portfolio in 2018 is the result of four projects—SLS, EGS, SGSS, and Mars 2020—addressing technical challenges that were compounded by risky programmatic decisions; two projects—JWST and ICESat-2—experiencing delays due to technical challenges identified during integration and test; and three projects—NISAR, ICON, and GRACE-FO—experiencing cost growth or delays largely due to factors outside of the projects’ control. We elaborate on these three scenarios below. Technical challenges compounded by risky programmatic decisions. Together, SLS, EGS, SGSS, and Mars 2020 experienced $638 million in cost growth and 59 months in aggregate schedule delays due to technical problems that were compounded by programmatic challenges since our May 2017 report. The SLS and EGS programs experienced cost growth and schedule delays associated with EM-1, their first combined mission along with the Orion program. We have found for several years that the human spaceflight programs—Orion, SLS, and EGS—are making progress maturing designs and building hardware, but also are experiencing some significant engineering and manufacturing challenges. For example, the SLS program ran into numerous challenges completing the welding of its core stage element in 2017. The program stopped welding on the core stage for months to identify and resolve low weld strength in the liquid oxygen and liquid hydrogen tanks due to low weld strength measurements found in the liquid oxygen tanks caused by a program and contractor decision to change the weld tool configuration during fabrication. The EGS program also experienced technical challenges, including with the design and installation of the ground support equipment and the 10 umbilicals that connect SLS and Orion to the Mobile Launcher—which supports the assembly, testing, and servicing of SLS and provides the platform on which SLS and Orion will launch. Finally, although the Orion program has not yet reported cost growth, it also experienced technical challenges. These challenges included software and hardware delays, and at least 14 months of delays with the European Service Module—which provides air, water, power, and propulsion to Orion during in-space flight—since the element’s critical design review in June 2016. In April 2017, we found that, according to program officials, the delays with the service module were largely due to NASA, the European Space Agency, and the European Space Agency contractor underestimating the time and effort necessary to address design issues for the first production service module and the availability of parts from suppliers and subcontractors. NASA expects the Orion program to experience cost growth over its cost baseline to the second combined mission, Exploration Mission 2 (EM-2). However, the extent of the growth is unknown because, as noted above, NASA is currently revising the program’s life-cycle cost estimate. Technical challenges such as these are not unusual for large-scale programs, especially human exploration programs that are inherently complex and difficult. However, we have found that NASA has made programmatic decisions—including establishing low cost and schedule reserves, managing to aggressive schedules, and not following best practices for earned value management or creating reliable cost and schedule baselines—that have compounded the technical challenges (see table 3). As a result, the three human spaceflight programs have been at risk of cost and schedule growth since NASA approved their baselines. In December 2017, NASA announced the new internal launch readiness date for EM-1 is now December 2019, and has allocated 6 months of schedule reserve available to extend the date to June 2020 for possible manufacturing and production schedule risks. This represents a delay of 13-19 months for EM-1. It is too soon to know if NASA has addressed the programmatic challenges identified above. We will continue to follow up through future reviews. Similarly, the SGSS project experienced new cost growth of $59.5 million and delayed its completion by 21 months. Project officials attributed the cost growth and delays to the contractor’s incomplete understanding of its requirements, which led to poor contractor plans and late design changes. But project management has been a challenge as well. The project has historically struggled to manage contractor performance and has faced both contractor and project staffing shortfalls, as we found in our prior reports starting in 2013. For example, NASA managers noted concerns with contractor plans and staffing estimates in 2013 during project confirmation. In March 2015, we found that the project was being rebaselined due to the contractor’s poor cost and schedule performance and in order to conform with limitations that NASA placed on the funding available to the contractor in fiscal years 2014 and 2015. The contractor was also operating with a limited number of staff at that time. In May 2017, we found that the project continued to experience contractor performance problems and had experienced cost growth and schedule delays over the 2015 rebaseline even as the project decreased its scope. In addition, the project experienced staff shortfalls in key areas, such as systems engineering and business management. The Mars 2020 project experienced $12.9 million in development cost growth, but no schedule delays. The cost growth was primarily due to technical challenges on a technology demonstration instrument and higher than anticipated integration costs for an entry, descent, and landing instrument. Both instruments are funded by the Human Exploration and Operations and Space Technology Mission Directorates. NASA officials attributed the cost growth of the technology demonstration instrument—which is designed to convert carbon dioxide to oxygen—to the complexity of the technology development for the effort. At the project’s preliminary design review in February 2016, a critical technology for the technology demonstration instrument did not meet the recommended level of maturity, which we have found can increase risk for systems entering product development. The project had matured the technology to this recommended level by its critical design review in February 2017. However, as a result of the focus on maturing this particular technology, other components of the instrument fell behind the planned schedule. Project costs for Mars 2020 also increased for an entry, descent, and landing instrument, due, in part, to cost increases for integration and to add additional staff to the instrument team to maintain schedule. Finally, the Radiation Budget Instrument project would have likely exceeded its cost baseline if NASA had not decided to cancel the project in January 2018. According to NASA’s cancellation memorandum, the project was canceled because of continued cost growth, technical issues, and poor contractor performance. In 2017, we found that the project was working to an aggressive schedule, and the prime contractor continued to experience cost overruns even after NASA added a deputy project manager and increased site visits and meetings with the contractor. Subsequently, the project—which was developing an instrument to be hosted on a National Oceanic and Atmospheric Administration satellite— determined that it would not be able to meet its delivery date for integration with the satellite without requiring additional funding in excess of the project’s cost baseline if other technical issues arose. In its cancellation memorandum, NASA stated continuing to fund the project from within the Earth Science Division budget would slow other important activities. Technical challenges identified during integration and test. The JWST and ICESat-2 projects experienced technical challenges during integration and test that delayed their schedules. Both projects were previously rebaselined before entering system-level integration and testing, and the current schedule delays are beyond the new schedules that NASA set for the projects in 2011 for JWST and in 2014 for ICESat- 2. The JWST project delayed its launch readiness date by at least 19 months from October 2018 to May 2020. NASA announced two delays for the project since our portfolio-wide review in May 2017. First, as we found in February 2018, the project delayed its launch readiness date by up to 8 months primarily due to the integration of the various spacecraft elements taking longer than expected. Specifically, execution of spacecraft integration and test tasks, due to complexity of work and cautious handling given the sensitivity of flight hardware, was slower than planned. In addition, before the delay, the project used all of its schedule reserves to its prior launch readiness date. This was the result of various contractor workmanship errors, particularly with respect to the spacecraft propulsion systems, as well as the resolution of various technical issues, including a test anomaly on the telescope and sunshield hardware challenges. Second, in March 2018, NASA announced that it had delayed the project’s launch readiness date by an additional 11 months to approximately May 2020 and planned to establish an external independent review board to analyze the project’s organizational and technical issues to inform a more specific launch time frame. The announcement also stated that after a new launch date is established, NASA would provide a new cost estimate that may exceed the $8 billion congressional cost cap that was established in 2011. NASA plans to finalize the project’s cost and schedule estimate by the end of June 2018. Because the additional delays were announced while a draft of our May 2018 report was with NASA for comment, we plan to follow up on the reasons for the additional delays and the results of the analysis in a future review. In our prior assessments of JWST, we have made recommendations with regard to improving cost and schedule estimating, updating risk assessments, and strengthening management oversight. NASA has generally agreed and taken steps to implement a number of our recommendations. For example, in December 2015, we recommended that the JWST project require contractors to identify, explain, and document anomalies in contractor-delivered monthly earned value management reports. NASA concurred with this recommendation and, in February 2016, directed the contractors to implement the actions stated in the recommendation. However, NASA did not implement some recommendations, which if implemented, may have provided insight into the challenges it now faces. For example, in December 2012, we recommended the JWST project update its JCL. Although NASA concurred with this recommendation, it did not take steps to implement it. An updated JCL may have portended the current schedule delays, which could have been proactively addressed by the project. The ICESat-2 project delayed its launch readiness date by 4 months from June to October 2018 due to technical issues with its only instrument, the Advanced Topographic Laser Altimeter System. A key part in the instrument’s lasers failed during instrument environmental testing, which delayed the project’s system integration review—the start of system-level integration and test. The manufacturer determined the primary cause of the anomaly was a flaw in the design of the mount that ensures a component of the optical module remains in a specific, precise position. The spare flight laser encountered the same problem during earlier testing, which indicated a systemic problem. The project redesigned and repaired the lasers and is proceeding through integration and test. External factors. External factors—including responding to requests for additional data collection and delays due to launch-vehicle related issues—contributed to cost increases or schedule delays for the NISAR, ICON, and GRACE-FO projects. The NISAR project experienced cost growth as the result of an increase in the scope of data collection in response to additional data needs being identified by an interagency working group. The additional data include soil moisture and natural hazard data that would be of value for other federal agencies and the science community. NASA officials said the additional funding for development would be used to upgrade the ground stations so that they can receive the additional data at a higher downlink data rate and volume. The ICON project missed its committed launch readiness date because of an accident involving its launch vehicle. In January 2017, two of the Pegasus launch vehicle’s three stages were involved in a transport accident. The stages were subsequently returned to the launch vehicle contractor facility for inspection and testing, and no damage was found. The project had been on track to launch early. Subsequently, in September 2017, an anomaly found in testing of the launch vehicle bolt cutter assemblies resulted in additional delays. NASA had planned to launch ICON in mid-June 2018, but recently announced a delay after off-nominal data was observed from the rocket during transit to the launch site. NASA announced a new launch date would be determined at a later date. The GRACE-FO project delayed its launch readiness date from February to May 2018 due to issues with its planned launch vehicle and launch site. The launch vehicle is the responsibility of NASA’s partner on the project—German Research Centre for Geosciences (GFZ). GRACE-FO had planned to launch at a Russian launch site. In February 2016, GFZ reported that it was notified by the Russian Federal Space Agency that the Dnepr launch vehicle was no longer available for GRACE-FO. GFZ, in June 2016, arranged to launch the two GRACE-FO spacecraft, along with commercial satellites, on a SpaceX Falcon 9. On May 22, 2018, GRACE-FO launched from Vandenberg Air Force Base in California. In addition, the Commercial Crew Program also experienced delays, which are not included above because the program does not have a schedule baseline. Since the award of the current Commercial Crew contracts in September 2014, the program, Boeing, SpaceX, and multiple independent review bodies have all identified the contractors’ delivery schedules as aggressive. In February 2017, we found that Boeing and SpaceX had determined that neither could meet their original 2017 dates for NASA to certify their systems for human spaceflight. In January 2018, we found that both contractors had notified NASA that final certification dates have slipped again and are now in the first quarter of calendar year 2019. The Commercial Crew Program’s schedule analysis indicates that certification may be further delayed to December 2019 for SpaceX and February 2020 for Boeing. NASA Is Likely to Encounter Additional Cost Growth and Schedule Delays The composition of the portfolio in the coming years is expected to include large and complex projects, putting NASA at risk of continued cost increases and schedule delays. Specifically, NASA plans to have complex projects enter the development portfolio in the next few years as it holds confirmation reviews and set cost and schedule baselines. This includes the Europa Clipper project and potentially the Wide-Field Infrared Survey Telescope (WFIRST) project. In February 2018, the President’s 2019 Budget Request proposed canceling the WFIRST project due to the project’s significant costs and higher priorities in the agency. However, the project may continue if funding is received. Together, preliminary estimates indicate that these two projects could cost as much as $7.8 billion. In addition, NASA expects to begin other large, complex projects like the Lunar Orbital Platform-Gateway— currently being discussed as a space station or outpost in lunar orbit— and a Europa Lander project in the coming years. A December 2017 space policy directive also instructed NASA to return astronauts to the moon for long-term exploration and to pursue human exploration of Mars and the broader solar system. To its credit, NASA recently took steps to put a process in place to control the costs of two projects while in formulation, which may prove useful if properly executed. The Europa Clipper project implemented a process whereby cost growth threats would be offset by descoping instruments in whole or in part. For example, if an instrument exceeds its development cost by 20 percent, the project would propose a descope option to NASA that brings instrument cost below that threshold. NASA had not descoped any instruments as of our May 2018 report. The WFIRST project is responding to findings from an independent review that was conducted to ensure the mission’s scope and required resources are well understood and executable. The review found that the mission scope is understood, but not aligned with the resources provided and concluded that the mission is not executable without adjustments and/or additional resources. For example, the study team found that NASA’s current forecasted funding profile for the WFIRST project would require the project to slow down activities starting in fiscal year 2020, which would result in an increase in development cost and schedule. NASA agreed with the study team’s results and directed the project to reduce the cost and complexity of the design in order to maintain costs within the $3.2 billion preliminary cost target. But even with these efforts, NASA’s cost and schedule performance may be further tested in upcoming years as some expensive, complex projects linger in the portfolio longer than expected. As previously discussed, the Orion program expects cost growth and faces other schedule and technical risks as it moves through the integration and test phase for EM-1 into at least 2019 and then through 2023 for EM-2. As of August 2017, NASA officials expected that new hardware and addressing development challenges would be the factors contributing to increased cost for the program. For example, there was a cost impact when the program moved from a single-piece, or monolithic, heatshield design to one that employs blocks in order to improve its structural strength. Program officials said they are also assessing schedule delays for EM-2, and noted that the EM-2 launch date depends on the outcome of the EM-1 launch date. The SLS and EGS programs continue to face cost, schedule, and technical risks as they move through the integration and test phase into at least 2019. For example, SLS will have to complete a “green run” test which requires multiple first-time efforts. Specifically, the test is the culmination of the development effort and includes the core stage integration with its four main engines, fully fueling with cryogenic hydrogen and oxygen, and then firing all four engines for about 500 seconds. NASA currently has no schedule reserve to its target December 2019 launch readiness date for two key areas in the core stage schedule. First, there is no reserve between the end of core stage production and the delivery of the core stage to the test facility. Second, there is no reserve between the end of the testing and delivery to Kennedy Space Center for final integration and testing prior to launch. As previously discussed, the JWST project is at risk of exceeding its congressional cost cap, and faces schedule risks as it completes its remaining integration and test work. These activities have taken considerably longer than planned due to a variety of challenges, including reach and access limitations on the flight hardware. Additionally, the project faces significant work ahead. For example, the project must complete integration of spacecraft element hardware and conduct deployment and environmental tests of the integrated sunshield and spacecraft. Further, it must integrate the telescope element with the spacecraft element to form the JWST observatory, and complete another set of challenging environmental tests on the full integrated observatory. At the same time, the project will need to mitigate dozens of remaining hardware and software risks to acceptable levels and address the project’s many potential single point failures to the extent possible. The SGSS project expects to experience additional cost growth through the final acceptance review because the full scope of the effort has not been included in the cost. NASA only approved its new cost estimate through the initial operational readiness review, currently planned for September 2019. A project official said NASA headquarters asked the project to determine if there are ways to reduce the cost between the operational readiness review and the final acceptance review. NASA plans to conduct an independent review of the project in mid-2018 to inform a decision on whether to continue the project past the operational readiness review. If NASA decides to continue the project past this review, additional cost growth is expected for SGSS when NASA revisits project costs through future budget cycles. In closing, NASA continues to make improvements to the acquisition management of its portfolio of major projects. However, the deterioration of the cost and schedule performance of NASA’s portfolio this year and the likelihood of additional cost growth and schedule delays demonstrate the need for NASA to continue to take actions to further reduce acquisition risk as we and others have recommended. Continuing to improve cost and schedule estimating tools and practices—such as by providing projects with sufficient cost and schedule reserves to address risks and unforeseen technical challenges and ensuring that program offices regularly and consistently update their JCLs across the portfolio— could help to better position NASA for improved outcomes. We look forward to continuing to work with NASA and this subcommittee in addressing these issues. Chairman Babin, Ranking Member Bera, and Members of the Subcommittee, this completes my prepared statement. I would be pleased to respond to any questions that you may have at this time. GAO Contact and Staff Acknowledgments If you or your staff have any questions about this testimony, please contact Cristina T. Chaplain, Director, Contracting and National Security Acquisitions at (202) 512-4841 or chaplainc@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this statement. GAO staff who made key contributions to this statement include Molly Traci, Assistant Director; Laura Greifner; Erin Kennedy; Miranda Riemer; Roxanna T. Sun; and Alyssa Weir. Appendix I: Descriptions of National Aeronautics and Space Administration Major Projects Reviewed in GAO’s 2018 Assessment Appendix I: Descriptions of National Aeronautics and Space Administration Major Projects Reviewed in GAO’s 2018 Assessment Project description The Commercial Crew Program facilitates and oversees the development of safe, reliable, and cost-effective crew transportation systems by commercial companies to carry NASA astronauts to and from the International Space Station. The program is a multi-phase effort that started in 2010. During the current phase, the program is working with two contractors— Boeing and SpaceX—that will design, develop, test, and operate the crew transportation systems. Once NASA determines the systems meet its standards for human spaceflight—a process called certification—the companies will fly up to six crewed missions to the space station. The DART project plans to travel to the near-Earth asteroid Didymos, a binary system, and impact the smaller of the two bodies. NASA will assess the deflection result of the impact for potential future use on other potentially hazardous near-Earth objects. The project responds to near-Earth object guidance by the Office of Science and Technology Policy to better understand our impact mitigation posture, and to recommendations by the National Research Council Committee to conduct a test of a kinetic impactor. The DART mission is part of the Asteroid Impact and Deflection Assessment, which is an international collaboration with the European Space Agency. The Europa Clipper mission aims to investigate whether the Jupiter moon could harbor conditions suitable for life. The project plans to launch a spacecraft in the 2020s, place it in orbit around Jupiter, and conduct a series of investigatory flybys of Europa. The mission’s planned objectives include characterizing Europa’s ice shell and any subsurface water, analyzing the composition and chemistry of its surface and ionosphere, understanding the formation of its surface features, and surveying sites for a potential landed mission. The EGS program is modernizing and upgrading infrastructure at the Kennedy Space Center and developing software needed to integrate, process, and launch the Space Launch System (SLS) and Orion Multi-Purpose Crew Vehicle (Orion). The EGS program consists of several major construction and facilities projects including the Mobile Launcher, Crawler Transporter, Vehicle Assembly Building, and launch pad, all of which need to be complete before the first uncrewed exploration mission using the SLS and Orion vehicles. The GRACE-FO mission will continue and expand upon the 2002 GRACE mission, which ended science operations in October 2017. The system, which consists of two spacecraft working together to obtain scientific measurements, will provide high-resolution models of Earth’s gravity field and insight into water movement on and beneath the Earth’s surface for up to 5 years. These models will provide rates of ground water depletion and polar ice melt and enable improved planning for droughts and floods. GRACE-FO is a collaborative effort with the German Research Centre for Geosciences. The ICESat-2 mission is a follow-on mission to ICESat that will measure changes in polar ice-sheet mass and elevation. The measurements will provide researchers a better understanding of the mechanisms that drive polar ice changes and their effect on global sea level. ICESat-2’s upgraded laser instrument will allow the satellite to make more frequent measurements and provide better elevation estimates over certain types of terrain than ICESat. InSight is a Mars lander with two primary objectives. It is intended to further understanding of the formation and evolution of terrestrial planets by determining Mars’s size, its composition, and the physical state of the core; the thickness of the crust; and the composition and structure of the mantle, as well as the thermal state of the interior. It will also determine the present level of tectonic activity and the meteorite impact rate on Mars. InSight is based on the Phoenix lander design. Phoenix successfully landed on Mars in 2008. Project description The ICON observatory will orbit Earth to explore its ionosphere—the boundary region between Earth and space where ionized plasma and neutral gas collide and react. Its four instruments will make direct measurements and use remote sensing to further researchers’ understanding of Earth’s upper atmosphere, the Earth-Sun connection, and the ways in which Earth weather drives space weather. JWST is a large, infrared-optimized space telescope designed to help understand the origin and destiny of the universe, the creation and evolution of the first stars and galaxies, and the formation of stars and planetary systems. It will also help further the search for Earth-like planets. JWST will have a large primary mirror composed of 18 smaller mirrors and a sunshield the size of a tennis court. Both the mirror and sunshield are folded for launch and open once JWST is in space. JWST will reside in an orbit about 1 million miles from the Earth. Landsat 9 is the next satellite in the Landsat series Program, which provides a continuous space-based record of land surface observations to study, predict, and understand the consequences of land surface dynamics, such as deforestation. The program is a collaborative, joint mission between NASA and the U.S. Geological Survey. The Landsat data archive constitutes the longest continuous moderate-resolution record of the global land surface as viewed from space and is used by many fields, such as agriculture, mapping, forestry, and geology. LCRD is a technology demonstration mission with the goal of advancing optical communication technology for use in deep space and near-Earth systems. LCRD will demonstrate bidirectional laser communications between a satellite and ground stations, develop operational procedures, and transfer the technology to industry for future use on commercial and government satellites. NASA anticipates using the technology as a next generation Earth relay as well as to support near-Earth and deep space science, such as the International Space Station and human spaceflight missions. The project is a mission partner and will be a payload on a U.S. Air Force Space Test Program satellite. Low Boom Flight Demonstrator (LBFD) LBFD is a flight demonstration project planned to demonstrate that noise from supersonic flight—sonic boom—can be reduced to acceptable levels, allowing for eventual commercial use of overland supersonic flight paths. Plans include multiple flights beyond fiscal year 2022 to gather community responses to the flights and to create a database to support development of international noise rules for supersonic flight. Lucy will be the first mission to investigate the Trojans, which are a population of never- explored asteroids orbiting in tandem with Jupiter. The project aims to understand the formation and evolution of planetary systems by conducting flybys of these remnants of giant planet formation. The Lucy spacecraft will first encounter a main belt asteroid—located between the orbits of Mars and Jupiter—and then will travel to the outer solar system where the spacecraft will encounter six Trojans over an 11-year mission. The mission’s planned measurements include asteroid surface color and composition, interior composition, and surface geology. Mars 2020 is part of the Mars Exploration Program, which seeks to further understand whether Mars was, is, or can be a habitable planet. Its rover and science instruments will explore Mars and conduct geological assessments, search for signs of ancient life, determine potential environmental habitability, and prepare soil and rock samples for potential future return to Earth. The rover will include a technology demonstration instrument designed to convert carbon dioxide into oxygen. Mars 2020 is based heavily on the Mars Science Laboratory, or Curiosity, which landed on Mars in 2012 and remains in operation. Project description NISAR is a joint project between NASA and Indian Space Research Organisation (ISRO) that will study the solid Earth, ice masses, and ecosystems. It aims to address questions related to global environmental change, Earth’s carbon cycle, and natural hazards, such as earthquakes and volcanoes. The project will include the first dual frequency synthetic aperture radar instrument, which will use advanced radar imaging to construct large-scale data sets of the Earth’s movements. NISAR represents the first major aerospace science partnership between NASA and ISRO. Orion is being developed to transport and support astronauts beyond low-Earth orbit, including traveling to Mars or an asteroid. The Orion program is continuing to advance development of the human safety features, designs, and systems started under the Constellation program, which was canceled in 2010. Orion is planned to launch atop NASA’s Space Launch System. The current design of Orion consists of a crew module, service module, and launch abort system. PSP will be the first NASA mission to visit a star. Using the gravity of Venus, the spacecraft will orbit the Sun 24 times and gather information to increase knowledge about the solar wind, including its origin, acceleration, and how it is heated. PSP instruments will observe the generation and flow of solar winds from very close range and sample and take measurements of the Sun’s outer atmosphere, where solar particles are energized. To achieve its mission, parts of the spacecraft must be able to withstand temperatures exceeding 2,500 degrees Fahrenheit and endure blasts of extreme radiation. The project was formerly named Solar Probe Plus, or SPP, and was renamed in May 2017. PACE is a polar-orbiting mission that will use advanced global remote sensing instruments to improve scientists’ understanding of ocean biology, biogeochemistry, ecology, aerosols, and cloud properties. PACE will extend climate-related observations begun under earlier NASA missions, which will enable researchers to study long-term trends on Earth’s oceans and atmosphere, and ocean-atmosphere interactions. PACE will also enable assessments of air and coastal water quality, such as the locations of harmful algae blooms. Psyche will be the first mission to visit a metal asteroid and aims to understand a previously unexplored component of the early building blocks of planets: iron cores. The project plans to orbit the Psyche asteroid to determine if it is a planetary core, characterize its topography, assess the elemental composition, and determine the relative ages of its surface regions. RBI is a scanning radiometer that NASA planned to launch on the National Oceanic and Atmospheric Administration’s (NOAA) Joint Polar Satellite System 2. RBI’s planned mission was to support global climate monitoring by continuing measurements of the Earth’s reflected sunlight and emitted thermal radiation made by NASA and NOAA satellites over the past 30 years. This data was intended to represent one of two key sets of measurements needed to determine whether the Earth is warming or cooling. The Restore-L project will demonstrate the capability to refuel on-orbit satellites for eventual use by commercial entities. Specifically, Restore-L plans to autonomously rendezvous with, inspect, capture, refuel, adjust the orbit of, safely release, and depart from the U.S. Geological Survey’s Landsat 7 satellite. Landsat 7 can extend operations if successfully refueled, but it is planned for retirement if the technology demonstration is unsuccessful. SLS is intended to be NASA’s first human-rated heavy-lift launch vehicle since the Saturn V was developed for the Apollo program. SLS is planned to launch NASA’s Orion spacecraft and other systems on missions between the Earth and Moon and to enable deep space missions, including Mars. NASA is designing SLS to provide an initial lift capacity of 70 metric tons to low-Earth orbit, and be evolvable to 130 metric tons, enabling deep space missions. The 70-metric-ton capability will include a core stage, powered by four RS-25 engines, and two five-segment boosters. The 130-metric-ton capability will use a new upper stage and evolved boosters. Project description The SGSS project plans to develop and deliver a new ground system for one Space Network site. The Space Network provides essential communications and tracking services to NASA and non-NASA missions. Existing systems, based on 1980s technology, are increasingly obsolete and unsustainable. The new ground system will include updated systems, software, and equipment that will allow the Space Network to continue to provide critical communications services for the next several decades. The Space Network is managed by the Space Communication and Navigation program. The SWOT mission will use its wide-swath radar altimetry technology to take repeated high- resolution measurements of the world’s oceans and freshwater bodies to develop a global survey. This survey will make it possible to estimate water discharge into rivers more accurately, and help improve flood prediction. It will also provide global measurements of ocean surface topography and variations in ocean currents, which will help improve weather and climate predictions. SWOT is a joint project between NASA and the French Space Agency—the Centre National d’Etudes Spatiales. TESS will use four identical, wide field-of-view cameras to conduct the first extensive survey of the sky from space for transiting exoplanets—or planets in other solar systems. The mission’s goal is to discover these exoplanets during transit, the time when the planet’s orbit carries it in front of its star as viewed from Earth. The project plans to discover rocky and potentially habitable Earth-sized and super-Earth planets orbiting nearby bright stars for further evaluation through ground- and space-based observations by other missions, such as JWST. WFIRST is an observatory designed to perform wide-field imaging and survey of the near- infrared sky to answer questions about the structure and evolution of the universe, and expand our knowledge of planets beyond our solar system. The project will use a telescope that was originally built and qualified by another federal agency. The project plans to launch WFIRST in the mid-2020s to an orbit about 1 million miles from the Earth. The project is also planning a guest observer program, in which the project may provide observation time to academic and other institutions. This is a work of the U.S. government and is not subject to copyright protection in the United States. The published product may be reproduced and distributed in its entirety without further permission from GAO. However, because this work may contain copyrighted images or other material, permission from the copyright holder may be necessary if you wish to reproduce this material separately.

Why GAO Did This Study GAO designated NASA's acquisition management as a high-risk area in 1990 after a history of persistent cost growth and schedule slippage in many of NASA's major projects. In more recent years, GAO found that NASA had taken some steps to improve its management, and, in May 2017, GAO found that projects were continuing a generally positive trend of limiting cost and schedule growth. But at the same time, GAO noted that many of these projects, including some of the most expensive ones, were approaching the phase in their life cycles when cost and schedule growth is most likely. This statement summarizes GAO's 2018 findings from its 10th annual snapshot of how well NASA is planning and executing its major acquisition projects, and describes (1) the cost and schedule performance of NASA's portfolio of major projects and (2) the extent to which NASA faces risks for further cost increases and schedule delays. To conduct its review for the 2018 report, GAO-18-280SP , GAO analyzed cost, schedule, and other data for NASA's 26 major projects, each with a life-cycle cost of over $250 million; reviewed monthly project status reports; and interviewed NASA officials. What GAO Found The cost and schedule performance of the National Aeronautics and Space Administration's (NASA) portfolio of major projects has deteriorated, but the extent of cost performance deterioration is unknown. NASA expects cost growth for the Orion crew capsule—one of the largest projects in the portfolio—but does not have a current cost estimate. In addition, the average launch delay for the portfolio was 12 months, the highest delay GAO has reported in its 10 years of assessing major NASA projects (see figure below). The deterioration in portfolio performance was the result of 9 of the 17 projects in development experiencing cost or schedule growth. Four projects encountered technical issues that were compounded by risky program management decisions. For example, the Space Launch System and Exploration Ground Systems programs are large-scale, technically complex human spaceflight programs, and NASA managed them to aggressive schedules and with insufficient levels of cost and schedule reserves. This made it more difficult for the programs to operate within their committed baseline cost and schedule estimates. Two projects ran into technical challenges that resulted in delays in the integration and test phase. For example, in December 2017, GAO found that the James Webb Space Telescope project encountered delays primarily due to the integration of the various spacecraft elements taking longer than expected, as well as the need to resolve technical issues during testing. GAO has previously found that integration and testing is when projects are most at risk of incurring cost and schedule growth. Three projects experienced cost growth or schedule delays due to factors outside of the projects' control, such as delays related to their launch vehicles. NASA continues to face increased risk of cost and schedule growth in future years due to new, large and complex projects that will enter the portfolio and expensive projects remaining in the portfolio longer than expected. What GAO Recommends GAO is not making any new recommendations in this statement. GAO has made recommendations in prior reports to strengthen NASA's acquisition management of its major projects. NASA generally agreed with these recommendations, but has not fully addressed some of them. GAO continues to believe they should be fully addressed.

gao_GAO-19-230

gao_GAO-19-230_0

Background Harm from Exposure of Personal Information Individuals’ sensitive personal information can be lost, stolen, or given away. Once exposed, individuals’ information can be misused to commit identity theft, fraud, or inflict other types of harm. Identity theft occurs when individuals’ information is used without authorization in an attempt to commit fraud or other crimes. In 2016, according to the Bureau of Justice Statistics, an estimated 26 million people—10 percent of U.S. residents aged 16 or older—reported that they had been victims of identity theft in the previous year. One potential source of identity theft is a data breach at an organization that maintains large amounts of sensitive personal information. Recent data breaches include the 2018 breach of Marriott International’s Starwood guest registration database, which may have exposed information of millions of individuals, and the 2017 data breach at Equifax, Inc., a nationwide consumer reporting agency, which exposed identifying information of at least 145.5 million people. The types of harm that can result from exposure of sensitive personal information include the following: Financial fraud from identity theft, which can include new-account fraud, in which thieves use identifying data, such as Social Security and driver’s license numbers, to open new financial accounts without that person’s knowledge; and, existing-account fraud, which is more common and entails the use or takeover of existing accounts, such as credit or debit card accounts, to make unauthorized charges or withdraw money. Tax refund fraud, which occurs when a Social Security number or other personally identifiable information is used to file a fraudulent tax return seeking a refund. Government benefits fraud, which occurs when thieves use stolen personal information to fraudulently obtain government benefits. For example, the Social Security Administration has reported that personal information of beneficiaries has been used to fraudulently redirect the beneficiary’s direct deposit benefits. Medical identity theft, which occurs when someone uses an individual’s name or personal identifying information to obtain medical services or prescription drugs fraudulently, including submitting fraudulent insurance claims. Synthetic identity theft, which involves the creation of a fictitious identity, typically by using a combination of real data and fabricated information. The federal government has identified synthetic identity theft as an emerging trend. Child identity theft, which occurs when a child’s Social Security number or other identifying information is stolen and used to commit fraudulent activity. Other types of fraud that occur when personal information is used; for example, to set up mobile phone or utility accounts, or to engage in activities such as applying for employment or renting a home. The harms caused by exposure of personal information or identity theft can extend beyond tangible financial loss, including the following: Lost time. Victims of identity theft or fraud may spend significant amounts of time working to restore their identities. In 2016, according to the Bureau of Justice Statistics survey of identity victims, most victims resolved issues in 1 day or less but about 1 percent of victims spent 6 months or more resolving their identity theft issues. Emotional distress and reputational harm. Exposed information also can cause emotional distress, a loss of privacy, or reputational injury. In 2016, according to the Bureau of Justice Statistics, about 10 percent of those who experienced identity theft reported suffering severe emotional distress. Harm from state-based actors. State-sponsored espionage can cause harm to individuals when nations use cyber tools as part of information-gathering, espionage, or other nefarious activities. Consumers’ Options to Address Risks or Harm Options available to consumers to help prevent or mitigate identity theft include actions they can take on their own (generally for free) or services they can purchase. Actions individual consumers can take themselves include the following: Placing a credit freeze. A credit or security freeze restricts potential creditors from accessing a credit report until the consumer asks the agency to remove or temporarily lift the freeze. Placing a fraud alert. A fraud alert on a credit report requires businesses to verify a consumer’s identity before they issue credit. Monitoring accounts and other information. Reviewing free annual credit reports. Individuals can request one copy of their credit report every 12 months (available for free at AnnualCreditReport.com).from each of the three nationwide consumer reporting agencies. Reviewing financial statements and other accounts. Individuals can review bank and other financial statements regularly for suspicious activity and make use of automatic transaction alerts and other free features that financial institutions offer to detect potential fraud. Individuals also can regularly review mobile phone or utility accounts for unusual activity. Reviewing health insurance benefits explanations and medical information. Individuals can review explanation-of- benefits statements from their health insurer to detect fraudulent insurance claims or monitor their files at their healthcare providers to detect unauthorized use of medical services. Consumers also can obtain various free or fee-based identity theft services, which are commercial products that generally offer tools intended to help consumers detect identity theft and restore their identity if it has been compromised. The private research firm IBISWorld estimated that the U.S. market for identity theft services was about $3 billion annually in 2015–2017. The services may be marketed directly to individuals for a monthly or annual fee. In addition, private- and public- sector entities that have experienced data breaches sometimes purchase these services and offer them to affected individuals at no cost. Identity theft services most often include credit monitoring, which tracks an individual’s credit reports and sends alerts about potentially suspicious activity; identity monitoring, which aims to monitor other sources such as public records and illicit websites (sometimes referred to as the “dark web”); identity restoration, which provides a range of services to recover from identity theft; and identity theft insurance, which reimburses individuals for certain costs related to the process of restoring identities. Other actions consumers can take to protect their identity include adoption of certain data security practices and early filing of tax returns. Data security practices can help protect sensitive information. For example, individuals can change or avoid sharing or re-using passwords, and make use of strong passwords and authentication options on online accounts; properly safeguard or shred sensitive paper documents; and limit access to their sensitive information on social media. Filing a tax return early reduces the risk of tax refund fraud, and some victims of tax refund fraud may be eligible for an Identity Protection Personal Identification Number (PIN)—issued by the Internal Revenue Service IRS)—to prevent future fraud. To protect their Social Security benefits, individuals can set up an online account at the Social Security Administration to monitor their benefits accounts. Limited Information Is Available on Effectiveness of Options after Data Breaches, but Credit Freezes Can Prevent New-Account Fraud We did not identify any studies that analyzed whether consumers who sign up for or purchase identity theft services encounter fewer instances of identity theft or detect instances of financial or other fraud more—or less—rapidly than consumers who take steps on their own. Views of experts varied, but most said identity theft services have limitations and would not address all data breach risks. Most experts also said that a credit freeze, which consumers place on their own for free, is a useful way to prevent one type of financial fraud—the illegal opening of new credit accounts in consumers’ names. Based on our review and discussions with experts, consumers can consider four factors when deciding on options to address risks after a data breach: the extent to which an option might prevent fraud; the cost of an option; its convenience; and the type of information that was exposed and may be at risk. No Independent Research Assesses Effectiveness of Consumer Options to Address Risks after Data Breaches Information that can help consumers assess their options for mitigating and addressing the risks of identity theft and other harm from data breaches is limited. Specifically, we did not identify any studies that analyzed whether consumers who sign up for free or purchase identity theft services encounter fewer instances of identity theft or detect instances of financial or other fraud more—or less—rapidly than consumers who take steps on their own for free—such as monitoring their credit reports or placing a credit freeze. For consumers who experienced identity theft, we did not find any studies that compared the effectiveness of free options to help consumers recover from identity theft with commercial identity restoration services. In addition to searching databases of scholarly publications and other sources, a range of academic, consumer, government, and industry experts we interviewed told us that they were unaware of any specific independent studies on the effectiveness of consumer options. We interviewed representatives of seven companies that provide identity theft services about how they assess the effectiveness of their services and found that what they measure does not directly address how effective these services would be in mitigating the risks of identity theft compared with options consumers can take on their own. For example, two company representatives said that their services focus on detection of fraudulent activity or assistance after identity theft has occurred, rather than on prevention of identity theft or other harms. The representatives of each of the providers said that their companies generally measure how customers use their products and services; customer satisfaction (for example, through surveys or other feedback); and whether the products work as intended (for example, whether alerts of fraudulent activity are successfully delivered to customers or customers can successfully access the company’s website when they need to). Companies that offer identity restoration services also measure the rate at which they complete the process of recovering stolen identities. While it is not possible to prevent identity fraud, four representatives said that early detection of fraud is important as it allows consumers to address potential fraud more quickly. FTC, a primary source for assistance to consumers on issues related to data breaches and identity theft, has advised consumers that the effectiveness of services that offer identity monitoring depends on factors such as the kinds of databases the service provider monitors, how well the databases collect information, and how often the service provider checks each database. For example, FTC suggests that consumers ask if service providers check databases that show payday loan applications or changes in addresses for misuse of their information as part of identity monitoring. In reviewing consumer education and promotional materials on the websites of five identity theft service companies we contacted that offer identity monitoring, we found that three providers included information about which types of databases they monitor; the other two did not. Government and commercial entities—such as federal agencies and retail stores—that decide to purchase identity theft services to offer to affected individuals after a breach of their data do not necessarily base their decision on how effective these services are. Rather, according to industry and some government representatives we interviewed, some base their decisions on federal or state legal requirements to offer such services and the expectations of affected customers or employees for some action on the breached entities’ part. Representatives of retail and banking associations we interviewed indicated that it has become the industry standard to offer 1 year of credit or identity monitoring services in the wake of a data breach. One industry representative said that in some cases the decision is not based on the effectiveness of the services. States such as California require companies to offer some type of identity theft service after a data breach. Moreover, Connecticut requires health insurers and certain health care-related companies to offer identity theft services following an actual or suspected data breach. In 2017, we reported that companies do not assess the effectiveness of an identity theft provider’s services when selecting a vendor to provide such services. Rather, they consider other selection factors, including price, reputation, capacity to respond quickly to large-scale breaches, and ability to provide comprehensive post-breach services, such as complying with statutory notification requirements. But companies that purchase identity theft services may be in a position to obtain more detailed information from potential providers than is publicly available to consumers. Views of Experts Varied, but Most Said Identity Theft Services Have Limitations and Would Not Address All Data Breach Risks In the absence of independent evidence of the effectiveness of identity theft mitigation options, we interviewed representatives and reviewed consumer education materials, working papers, and articles from academic, consumer, industry, and government entities. No one solution can protect against the full range of risks to individuals whose personal information was exposed in a data breach, based on our review of documentation and the views of academic, consumer, government, and industry experts. We obtained perspectives on the value of options available to consumers. The following summarizes key observations: Identity theft services. Representatives of 9 of the 10 consumer groups we interviewed generally viewed credit or identity monitoring (or both) to be of limited value. However, one consumer group representative noted that identity monitoring might be useful in circumstances in which Social Security numbers were compromised. In addition, a few consumer group representatives indicated that consumers could consider signing up for such services if they are offered for free. If identity theft services are not free, FTC and CFPB consumer education materials recommend that consumers consider the benefits and limitations of such services and compare them to free or low-cost options before signing up. A few consumer groups and one academic highlighted that consumers may not fully understand the limitations of signing up for identity theft services. A few consumer group representatives and one industry and state government representative cautioned that free services may be offered for only 1 or 2 years; exposed information can be used for identity theft or other harms over a much longer period. For example, in 2017, we reported that nation-state actors that steal consumer data as part of their espionage activities can wait much longer than a private identity thief to use compromised information (if at all), according to one identity theft service provider. In addition, CFPB consumer information and a few consumer group representatives noted that consumers should be aware that some services may try to charge consumers after the free period ends. Some consumer group and one industry representatives also said that the value of one feature of identity monitoring—dark web monitoring—is unclear. One representative said that there is nothing new that consumers can do once they learn their information was found on an illicit website. Rather, they must continue to monitor their accounts as they already should have been doing. In addition, one consumer group representative indicated that these services may provide consumers with a false sense of security. Experts we interviewed for our 2017 report said that identity restoration in particular could be helpful to consumers. FTC staff and one consumer group representative we interviewed said that one-on-one assistance can be helpful. Identity restoration typically is included with other identity theft services rather than offered as a stand-alone service. However, the level of service provided in identity restoration can vary substantially—some providers offer individualized hands-on assistance, while others largely provide self-help information that is of more limited value. In our 2017 report, we also found that another feature of identity theft services, identity theft insurance, may provide minimal benefits for consumers. More details about identity theft insurance appear later in this report. Options to prevent fraud or harm unrelated to credit accounts. Consumers have limited options to mitigate risks of other harms from data breaches, such as medical identity theft and identity theft tax refund fraud. Commercial identity theft services, credit freezes, and fraud alerts do not directly address these risks. Some consumer, government, and industry representatives cited self-monitoring as a way for consumers to be on the alert for these other types of fraud. Consistent with our 2017 report, identity theft service providers we interviewed generally indicated that their products and services do not directly monitor for these types of fraud. However, two noted that they would assist with any identity restoration involving medical identity theft, tax refund fraud, or government benefits fraud (such as fraudulently redirecting Social Security benefits). Identity theft services also may address these types of fraud indirectly—for example, detecting a fraudulent change of address can prevent sensitive health insurance information from being redirected to the fraudster. A few consumer groups said that consumers may not understand which risks commercial identity theft services address. Additionally, we reported in 2017 that identity theft services do not address non-financial harms, such as emotional distress, embarrassment, and harm to one’s reputation. For example, a House Committee report on the OPM data breaches noted that the information stolen from background investigations included some of the most intimate and potentially embarrassing aspects of a person’s life, such as mental health history, misuse of alcohol or drugs, or problems with gambling. Identity theft services also may be of limited value in cases of nation-state espionage. For example, in 2017, we reported that when the source of the data breach appears to be a nation state (as opposed to a private party), the risk of the information being sold for monetary purposes is likely to be lower, according to an FTC representative. Importance of data security. In the view of some experts, entities such as the federal government and private companies that hold consumer data have a responsibility to protect those data. A few experts said that the burden should not be on consumers to protect data they do not control. Except in certain circumstances, companies are generally not required to be transparent about the consumer data they hold or how they collect, maintain, use, and secure these data. Identity theft service providers may contract with third parties such as consumer reporting agencies or with third-party identity monitoring providers, such as dark web monitoring services. Moreover, one consumer group representative noted that identity monitoring services require consumers to provide additional personal information to enroll—which also could be compromised if the service provider’s information were breached. Finally, consumer group and government researchers we interviewed suggested other options that entities can (or already) use to address risks of harm. For example, one government researcher noted that financial institutions have started to use multifactor authentication and other technologies that can help institutions verify a consumer’s identity and thus help prevent fraud. Multifactor authentication involves first logging into an online account using the traditional username and password, and then the institution sending a verification code to a mobile phone or e-mail address that the consumer must enter as part of the log-in process. In addition, one researcher noted that some institutions have started to use facial recognition technology, or to ask an account holder to provide answers to questions such as the size of the account holder’s last deposit. Other biometric technologies such as fingerprint recognition on mobile phones, or one-time passcodes that are synced with financial institutions’ websites, also can help, according to one researcher and one consumer group representative. Other strategies can focus on reducing the riskiness of breaches by making information less useful for purposes of committing identity theft. For example, one researcher noted that organizations could encrypt data or use tokens so static account numbers could not be used on their own. There is no single solution to address all risks of harm, based on our review of documentation and the views of academic, consumer, government, and industry experts. Consumers Can Use Free Credit Freezes and Fraud Alerts to Effectively Prevent New-Account Fraud A credit freeze is the only consumer option that can prevent one type of identity theft-related fraud, and recent federal legislation made credit freezes free and easier to place or lift. This option is effective because it restricts potential creditors from accessing a consumer’s credit report to open a new account until the consumer asks the nationwide consumer reporting agency to remove or temporarily lift the freeze. In contrast, identity theft services and self-monitoring detect or remediate identity theft after it has occurred, but do not prevent the fraud from occurring in the first place. We interviewed representatives, or reviewed the consumer education or informational materials, of consumer, industry, and government entities and found that almost all of them included credit freezes on credit reports as a useful consumer option to protect against identity theft. More specifically, the Economic Growth, Regulatory Relief, and Consumer Protection Act, which took effect on September 21, 2018, required the three nationwide consumer reporting agencies (Equifax, Experian, and TransUnion) to make placing and lifting freezes free and specifies that the agencies must place a freeze within 1 business day, and lift it within 1 hour, of receiving a telephone or electronic request (see fig. 1). Consumers must contact each of the three agencies individually and request the freeze. Consumers obtain a PIN from each company, which enables them to lift or remove a freeze at a later date. Before the 2018 act, consumers typically had to pay $5-$10 per agency to place a credit freeze. Some experts had noted cost and inconvenience as some of the limitations to a credit freeze. The new law addresses these concerns to some degree by making credit freezes free and requiring these consumer reporting agencies to lift freezes expeditiously on request. While the new law removed some barriers to placing credit freezes, others still exist and the freezes have some limitations. For example, consumers still have to lift a freeze before applying for a loan or new credit account and need to place or remove a freeze at each consumer reporting agency separately, which could cause delays for consumers actively shopping for a home, car, or other purchase requiring the extension of credit. Two consumer groups said that there is confusion about how the law would affect minor children. (Under the new law, credit freezes only can be placed on behalf of children under age 16, but not minors ages 16 and 17—who must place freezes themselves). Moreover, as the new law only applies to the three nationwide consumer reporting agencies, credit freezes do not protect against new-account fraud resulting from the use of credit reports from other consumer reporting agencies. For example, one consumer group recommended that consumers place a fourth freeze with the National Consumer Telecom and Utilities Exchange—a consumer reporting agency that maintains credit reports that telecommunications or utilities companies may use to check the creditworthiness of consumers interested in opening phone or utility accounts. The law also permits insurance companies and employers to continue to access credit reports even after they are frozen, among other exceptions. One general limitation of credit freezes is that they do not protect against new-account fraud in cases in which credit reports are not used to verify a consumer’s creditworthiness. Furthermore, credit freezes do not protect against existing-account fraud, such as fraudulent credit card charges, or certain other types of fraud, such as identity theft tax refund fraud or synthetic identity fraud using elements of individuals’ identity information. While experts with whom we spoke across industry, government, and consumer groups generally believed credit freezes to be an effective tool in preventing new-account fraud, some consumer and industry experts indicated that fraud alerts also can be a good alternative for consumers. Unlike a credit freeze, a fraud alert still allows companies to access an individual’s credit report for the purpose of opening a loan or credit account. Fraud alerts notify companies requesting the reports that the individual may have been a victim of identity theft. The alerts require companies to verify consumers’ identities before they issue credit to a consumer. Fraud alerts therefore can make it harder for an identity thief to open accounts in a consumer’s name. Moreover, fraud alerts are easier to place than credit freezes, as consumers only need to contact one of the three nationwide consumer reporting agencies to place a fraud alert (that agency is then obligated to contact the other two on the individual’s behalf). The Economic Growth, Regulatory Relief, and Consumer Protection Act extended the period of an initial fraud alert from 90 days to 1 year. However, fraud alerts do not restrict access to consumers’ credit reports the way freezes do. Therefore, some consumer group and industry representatives noted that consumers should be aware that a fraud alert may not offer as strong a protection as a credit freeze does. We did not find any data or analysis on the effectiveness of fraud alerts compared to credit freezes or monitoring options. One consumer group told us that it recommends that after a data breach consumers first place a fraud alert, because it requires contacting only one of the three nationwide consumer reporting agencies, and then follow up by placing a credit freeze at the three agencies. The three nationwide consumer reporting agencies also offer a product called a credit lock that is functionally similar to a credit freeze in that it restricts access to an individual’s credit report. Credit locks do not require consumers to use a PIN and consumers can turn access to credit reports on or off through an application on their mobile phone. However, credit locks are not subject to the same federal requirements regarding the placement and removal of freezes and therefore do not offer the same degree of protection to consumers. Instead, credit locks are private products subject to the consumer reporting agencies’ terms and conditions, which could change. A credit lock is in place only as long as the individual subscribes to an agency’s service, but a credit freeze remains in place until the consumer chooses to remove it. Finally, consumers may be charged a fee to place a credit lock, whereas credit freezes can now be placed for free. Factors Consumers Can Consider When Assessing Options after Data Breaches Based on our interviews and review of consumer education materials and our 2017 report, we identified four factors that consumers can consider in deciding which options are best for them in responding to a breach of their personal information: Prevention. Consumers can consider the extent to which an option might prevent fraud. For example, because credit freezes block all access to an individual’s credit report, by definition they are effective in preventing new-account fraud where credit reports are used as part of the account-opening process. Identity theft services do not prevent fraud, but detect suspicious activity or help restore identities after identity theft. Cost. Consumers can consider the cost of a service. For instance, consumers can consider whether to pay for commercial identity theft services if they believe the value of the service outweighs the effort of monitoring their accounts on their own. In addition, they may consider that credit freezes now are available for free. Convenience. Consumers may consider the convenience of a service. For example, while consumers can monitor their own credit reports and accounts, some might prefer not to or may be limited in their ability to do so. In addition, technologies offered through financial institutions that automatically alert customers to any transactions involving their accounts can be a convenient, no-cost way for consumers to monitor their accounts. Type of information at risk. Finally, several experts from consumer and industry organizations indicated that the type of option that might be beneficial would depend on the type of information at risk. For example, one consumer group representative noted that if a credit card number were stolen, an identity monitoring service that monitored the dark web for Social Security numbers might not be needed. Furthermore, consumers should consider that credit monitoring will be of limited effectiveness in alerting them to misuse of an existing credit account—which is more common than fraud related to setting up new accounts. For more information on consumers’ options, see appendix II. Federal Agencies Provide Assistance to Consumers Affected by Data Breaches and Identity Theft Among federal agencies, FTC serves as a primary source for free assistance (including online resources, educational outreach, and customized assistance through IdentityTheft.gov) to consumers on ways to respond to data breaches, identity theft, and related harm. Approximately 13 percent of those affected by the 2015 OPM breaches used credit and identity monitoring and identity restoration services that OPM offered them and a fraction of a percent made identity theft insurance claims (the payouts for which averaged $1,800). Data we assessed for this report support a 2017 recommendation we made to the Office of Management and Budget (OMB) to revise guidance to federal agencies about responding to data breaches and one to Congress to consider permitting agencies to determine appropriate levels of identity theft insurance offered after data breaches. FTC Is Primary Provider of Federal Assistance to Consumers Affected by Data Breaches and Identity Theft Federal Trade Commission FTC, as a primary source for assistance to consumers on issues related to data breaches and identity theft, provides guidance and assistance through its website and through conferences and workshops. Online and printed resources. FTC’s home page includes links to identity theft-related resources, including information about key options consumers can consider to help them mitigate identity theft risks and other harms, and a link to IdentityTheft.gov (discussed later in this section). FTC updates the information regularly, such as after large-scale data breaches. Outreach. FTC maintains relationships with state government, law enforcement, and community and consumer organizations, through which it conducts outreach about how to respond to exposure or loss of personal information and identity theft mitigation. For example, FTC collaborated with the International Association of the Chiefs of Police to update the association’s model policy for identity theft to include referral information for IdentityTheft.gov. FTC also has held webinars, conferences, and workshops on topics related to data breaches and identity theft for groups including government officials, nonprofits, and the general public. Customized assistance (IdentityTheft.gov). FTC provides information and customized assistance through IdentityTheft.gov to individuals whose information was lost or stolen or who experienced identity theft or other harm, such as tax refund fraud. During fiscal year 2018, IdentityTheft.gov received almost 2 million unique visitors. The website in its current form has been in place since January 2016 and offers the following types of assistance: Steps to take after identity theft. IdentityTheft.gov provides individual victims with step-by-step instructions to resolve specific problems. From January 2016 (when FTC launched the current version of IdentityTheft.gov) through October 1, 2018, approximately 700,000 individuals set up and activated accounts on the website to help them recover from identity theft. Individuals who set up accounts can indicate what kind of information was stolen and what kind of adverse event they experienced. The site helps users generate pre- filled letters, affidavits, and forms to send to consumer reporting agencies, businesses, debt collectors, and IRS, as appropriate. For example, individuals who fill out an Identity Theft Report affidavit can use this report instead of filing a police report to request extended 7- year fraud alerts (available to identity theft victims) on their credit reports. In addition, individuals who experienced tax refund fraud can fill out a form on IdentityTheft.gov that is then submitted directly to IRS. An individual who experienced credit card fraud would be advised to take different steps than one who experienced fraud related to utility bills or medical insurance. Steps to take after data breaches or loss of personal information. IdentityTheft.gov/databreach provides checklists and suggestions for people whose personal information was lost or exposed but has not yet been misused. FTC also maintains an online chat function and telephone number for those who need additional assistance. For complex cases, FTC staff may refer individuals to the Identity Theft Resource Center, a nonprofit organization. We found that in developing and updating the website, FTC followed some key practices for consumer education planning. One key practice we identified was consulting with stakeholders. According to FTC staff we interviewed and documentation we reviewed, FTC obtained feedback from stakeholders such as law enforcement agencies and community organizations in developing IdentityTheft.gov. Another key practice we identified was assessing users’ needs. FTC conducted usability testing to ensure the site’s features were easy to use. FTC staff also told us that after receiving user feedback, they made it easier for users to set up an account. FTC also made changes to IdentityTheft.gov—such as incorporating the ability to auto-generate forms—to implement a 2014 Executive Order calling for federal agencies to centralize identity theft information at the website. Furthermore, in January 2018, FTC implemented a new function that allows users who report identity theft tax refund fraud to file reports directly with IRS. Since its launch in early 2018 through October 1, 2018, almost 22,000 IRS Identity Theft Affidavits (IRS Form 14039) were submitted to IRS through IdentityTheft.gov. In general, experts across consumer, government, and industry organizations and identity theft service providers we interviewed expressed the view that IdentityTheft.gov is a valuable or user-friendly resource. Other Federal Agency Resources Other federal agencies provide assistance to consumers on topics related to identity theft, including CFPB, the Department of Justice, IRS, and the Social Security Administration. CFPB. CFPB enforces, supervises for compliance with, and issues regulations to implement the federal consumer financial laws that address certain firms’ and financial institutions’ practices, which may include data security. A few of these laws and regulations contain provisions that can help protect the personal information of consumers. CFPB also offers consumer education resources. Similarly to FTC, CFPB included information about how consumers can address risks related to exposure of personal information and recover from identity theft in the bureau’s overall consumer education activities. CFPB provides consumer education materials related to data breaches and identity theft through its blog and its financial education resource, “Ask CFPB.” CFPB also maintains relationships with external groups, such as librarian networks. CFPB provides links to FTC resources about data breaches and identity theft topics on its website, so as not to duplicate efforts, according to CFPB staff. The two agencies also have coordinated some efforts. FTC and CFPB published a jointly produced blog post on September 21, 2018, the date the new free credit freeze and 1-year fraud alert provisions took effect. Such coordination is consistent with the 2014 Executive Order, which designated FTC as a centralized source of information about identity theft across the federal government. Staff of both agencies said that in developing new resources, they monitor information from a variety of sources, including consumer complaints, news and social media, and reports from other government entities, law enforcement, or nongovernmental stakeholders. Other federal and state agencies. IRS and the Social Security Administration provide some assistance to consumers for specific types of identity theft. For example, as noted previously, IRS provides some taxpayers with PINs if they are victims of identity theft tax refund fraud. In addition, states enforce laws and regulations and provide consumer education resources and assistance to consumers at risk of identity theft and other harms as a result of data breaches. For example, the Illinois Attorney General’s office maintains a call-in number for victims of identity theft, and the Colorado Bureau of Investigation can assist residents with identity theft issues. Few People Used Identity Theft Services OPM Provided, Very Few Made Insurance Claims, and Payouts Received Were Low OPM offered identity theft services to approximately 22.1 million individuals whose personal information was compromised during the 2015 data breaches at OPM. Personnel records or OPM systems containing information from the background investigations of current, former, and prospective federal employees and other individuals were breached. The services, offered at no cost to affected individuals, included credit monitoring, identity monitoring, identity restoration services, and identity theft insurance. To receive credit and identity monitoring services, affected people have to enroll with the identity theft service provider with which OPM contracted, but identity theft insurance and identity theft restoration services are available to the entire affected population whether or not they enroll. Few affected individuals have used the services. According to data from OPM, as of September 30, 2018, close to 3 million, or 13 percent, of individuals affected by the 2015 incidents had made use of the services. As seen in figure 2, the great majority of enrollments occurred in the months immediately following notification of the breach. OPM staff said that the spike in enrollments in July and August 2016 likely was due to the follow-up mailing that OPM sent to approximately 10 percent of affected individuals whose mailing addresses were incorrect in the original mailing of notifications. In addition, according to OPM-reported data we reviewed, of the 3 million individuals who used the services, about 1 percent made identity restoration requests and a fraction of 1 percent submitted insurance claims. According to data we reviewed, approximately 27,000 identity restoration cases had been resolved as of September 30, 2018. In addition, 61 insurance claims (of 81 submitted) had been paid, totaling $112,000, with an average payout of $1,800. Since 2015, OPM has obligated approximately $421 million for identity theft services and as of November 30, 2018, OPM paid out approximately $361 million of the obligated funds. OPM is required to provide identity theft services through September 2026. The contract to provide these services on behalf of OPM expired in December 2018; OPM re-competed and awarded a single contract that month to ID Experts, the company that had been providing these services. After the OPM breaches in 2015, OPM provided federal employees and other affected individuals with information and guidance about their options in mailed letters and on its website. On its website, OPM developed a Cybersecurity Resource Center and included background about the breaches and who was affected; instructions for how to enroll in identity theft services; and a Frequently Asked Questions webpage that included links to FTC resources, including IdentityTheft.gov. OMB’s 2017 policy guidance to federal agencies, including OPM, states that agencies should determine appropriate information to provide to affected individuals and review breach responses annually. Consistent with that guidance, OPM’s September 2017 Breach Response Plan calls for the agency to review its breach response plan annually, including to reinforce or improve training and awareness. In December 2018, OPM updated its website to incorporate changes in the cost of credit freezes and duration of fraud alerts resulting from new legislation we discussed earlier. OMB Has Not Revised Post-Data Breach Guidance to Agencies and Insurance Coverage Amount for Identity Theft Insurance Remains High Data we assessed for this report support a 2017 recommendation we made to OMB and a matter for congressional consideration, both of which have not yet been implemented. In our March 2017 report, we found that OMB policy guidance for federal agencies on how to prepare for and respond to data breaches did not address how agencies might assess the effectiveness of identity theft services relative to lower-cost alternatives. For example, the guidance did not discuss whether identity theft services would be preferable to alternatives (such as fraud alerts, credit freezes, or the agency conducting its own database monitoring). We concluded that the guidance might not fully reflect the most useful and cost-effective options agencies should consider in response to a breach—contrary to OMB’s risk-management and internal control guidance calling on federal leaders to improve effectiveness and efficiency. Therefore, we recommended that OMB conduct an analysis of the effectiveness of identity theft services relative to alternatives, and revise its guidance to federal agencies in light of the analysis. In oral comments on a draft of the 2017 report, staff from OMB’s Office of Information and Regulatory Affairs said that our draft recommendation to OMB on expanding OMB’s guidance to federal agencies would benefit from greater specificity, and we revised this recommendation to provide greater clarity. We contacted OMB several times between May 2018 and early March 2019 to update the status of this recommendation but as of March 2019, OMB had not responded with an update. In our current review, we found that information on the effectiveness of various consumer options continues to be limited. We also found that some free and low-cost alternatives to free or fee-based identity theft services can prevent or more directly address new account fraud and some options consumers can take on their own have become less burdensome. Therefore, we stand by this recommendation. In addition, as noted previously in this report, the identity theft insurance that OPM offered to affected individuals resulted in few insurance claims, and the amounts claimed have been small. These data are consistent with the findings of our 2017 report—which reported that the number and dollar amount of claims for identity theft generally were low. They also reinforce our conclusion that the $5 million per-person coverage limit mandated by Congress likely was unnecessary and might impose costs without providing a meaningful corresponding benefit. Specifically, we noted that $5 million in coverage would increase federal costs unnecessarily, likely mislead consumers about the benefit of the product, and create unwarranted escalation of coverage amounts in the marketplace. Therefore, we reiterate the matter for congressional consideration we made in our March 2017 report: in the event that Congress again requires an agency to provide individuals with identity theft insurance in response to a breach, it should consider permitting the agency to determine the appropriate level of that insurance. Agency Comments We provided a draft of this report to CFPB, FTC, and OPM. The agencies provided technical comments, which we incorporated as appropriate. We are sending copies of this report to the appropriate congressional committees, the Director of CFPB, the Chair of FTC, and the Acting Director of OPM. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions concerning this report, please contact me at (202) 512-8678 or ortiza@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made major contributions to this report are listed in appendix IV. Appendix I: Objectives, Scope, and Methodology This report examines (1) information and expert views about the effectiveness of options consumers can use to prevent or address the risks resulting from data breaches; and (2) federal assistance available to help consumers understand these options, including the status of one matter for congressional consideration and one recommendation relating to these issues in our 2017 report. To address the first objective, we conducted a literature review to identify any studies or independent research on the effectiveness of various options consumers have for mitigating data breach harms, consumer attitudes and behavior following data breaches, and identity theft and other harm to individuals from exposure of personal information. We searched databases of scholarly publications and other sources for work generally published within the last 5 years. Examples of databases searched include ProQuest, EconLit, Policy File Index, and SciTech Premium Collection. We searched for terms including “effective,” “data breach,” “identity theft,” “consumer attitudes,” and “consumer behavior” and options such as “credit freeze,” “fraud alert,” and “credit lock.” We also reviewed relevant academic literature to identify additional studies. From these searches, we did not identify any studies that assessed the extent to which commercial identity theft services were effective in preventing or mitigating harm from exposure of personal information. We identified and reviewed 54 studies that appeared in peer-reviewed journals or research institutions’ publications and were relevant to consumer attitudes and behavior related to privacy, data breaches, and identity theft. To ensure the selection of a range of perspectives on the effectiveness of options to mitigate harms, we reviewed the selection of experts and sources in our prior report and our literature review, and updated that selection through additional searches and recommendations from discussions with experts and identity theft service providers and review of relevant literature. We defined experts as those representing consumer and industry policy organizations that have conducted research or taken policy positions on consumers’ or entities’ options after data breaches; academics who conducted research on relevant topics; and federal and state government staff with specific positions of responsibility in consumer protection or education. We also contacted seven companies that provide identity theft services to consumers. We interviewed representatives of a nongeneralizable sample of 35 entities in the following categories: academic or independent research institution (4); consumer or privacy research and advocacy (10); industry association, identity theft service provider, or industry consultant (12); and federal or state government (9). We also reviewed relevant consumer education and other materials produced by consumer, government, industry, and other entities. We interviewed academics from Carnegie- Mellon University, RAND Corporation, the University of Maryland, and the University of Rochester. In addition, we interviewed representatives from the following organizations: Consumer or privacy groups: AARP, Consumer Action, Consumer Federation of America, Consumer Reports, Electronic Privacy Information Center, Identity Theft Resource Center, National Consumer Law Center, Privacy Rights Clearinghouse, U.S. PIRG, and World Privacy Forum. Industry associations or consultants: American Bankers Association, Consumer Data Industry Association, Property and Casualty Insurers Association of America, National Retail Federation, and Rational 360. Identity theft service providers: Credit Karma, Equifax, Experian, ID Experts, ID Shield, LifeLock, and TransUnion. Government agencies: Consumer Financial Protection Bureau (CFPB), Federal Reserve Bank of Philadelphia, Federal Trade Commission (FTC), Office of Personnel Management (OPM), and Offices of the Attorney General of California, Connecticut, Illinois, Massachusetts, and New York. Throughout this report, we use certain qualifiers when describing responses from interview participants and views of entities whose articles and written material we reviewed, such as “few,” “some,” and “most.” We define few as a small number such as two or three. The specific quantification of categories depends on the overall numbers of entities that addressed a specific topic. For example, we may refer to views shared by a proportion of the 10 consumer groups we interviewed, or those shared by identity theft service providers. We also reviewed provisions in the Economic Growth, Regulatory Relief, and Consumer Protection Act, enacted in May 2018, that address credit freezes and fraud alerts (two tools for preventing new-account fraud). To address the second objective, we reviewed and analyzed documentation and interviewed staff from FTC, CFPB, and OPM. We reviewed and analyzed FTC, CFPB, and OPM consumer education materials including blog posts, online fact sheets, and printed brochures and data on usage of the materials. For example, we analyzed FTC, CFPB, and OPM data and website analytics for their data breach- and identity theft-related web pages. We interviewed FTC and CFPB agency staff about their assistance to individuals and how they measure effectiveness of their efforts. We reviewed documentation and interviewed agency staff about the development, implementation, and assessment of consumer education materials and other resources and assistance. For example, we reviewed materials documenting FTC’s outreach to stakeholders and usability testing of IdentityTheft.gov. We compared the activities against a 2014 Executive Order on the security of consumer financial transactions, key practices for consumer education planning we identified in prior work, and federal standards for internal control. We analyzed data from the company with which OPM contracted to provide identity theft services to the approximately 22.1 million individuals whose information was exposed in the 2015 data breaches. We obtained data on the number of enrollments, the number and size of identity theft insurance claims submitted and paid, and number of identity restoration cases the companies handled. We assessed the reliability of the data by interviewing agency officials and reviewing documentation about the systems used to store the data. We found the data to be reliable for purposes of this reporting objective. We also reviewed the online guidance OPM provided to affected individuals and assessed the guidance against Office of Management and Budget guidance for agencies following data breaches and OPM’s 2017 Breach Response plan. In addition, for both objectives, we reviewed the evidence gathered and analyzed for the 2017 GAO report (GAO-17-254) and updated the status of the matter for congressional consideration and recommendations made in that report. We conducted this performance audit from November 2017 to March 2019 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: What Can Consumers Do After a Data Breach? Appendix II: What Can Consumers Do After a Data Breach? Figure 3 below provides information on actions consumers can take to monitor for identity theft or other forms of fraud, protect their personal information, and respond if they have been a victim of identity theft. This information summarizes prior GAO work and comments of academic, consumer organization, industry, and government experts. Appendix III: GAO Contacts and Staff Acknowledgements GAO Contact Staff Acknowledgments In addition to the contact named above, Kay Kuhlman (Assistant Director), Meghana Acharya, Carl Barden, Bethany Benitez, Catherine Gelb (Analyst in Charge), Danielle Koonce, Jill Lacey, Kathleen McQueeney, Barbara Roesmann, Jena Sinkfield, and Meg Tulloch made significant contributions to this report.

Why GAO Did This Study Recent large-scale data breaches of public and private entities have put hundreds of millions of people at risk of identity theft or other harm. GAO was asked to review issues related to consumers' options to address risks of harm from data breaches. This report, among other things, examines information and expert views on the effectiveness of consumer options to address data breach risks. GAO analyzed available data on options, collected and analyzed related documentation, conducted a literature review of studies, and interviewed a nongeneralizable sample of 35 experts (from academia, government entities, consumer and industry organizations) and identity theft service providers to reflect a range of views. What GAO Found No one solution can address the range of potential risks from a data breach, according to interviews with academic, consumer, government, and industry experts and documentation GAO reviewed. Perpetrators of fraud can use stolen personal information—such as account numbers, passwords, or Social Security numbers—to take out loans or seek medical care under someone else's name, or make unauthorized purchases on credit cards, among other crimes. Foreign state-based actors can use personal information to support espionage or other nefarious uses. Public and private entities that experience a breach sometimes provide complimentary commercial identity theft services to affected individuals to help monitor their credit accounts or restore their identities in cases of identity theft, among other features. Consumers also may purchase the services. As of November 30, 2018, the Office of Personnel Management (OPM) had obligated about $421 million for a suite of credit and identity monitoring, insurance, and identity restoration services to offer to the approximately 22 million individuals affected by its 2015 data breaches. As of September 30, 2018, about 3 million had used the services and approximately 61 individuals had received payouts from insurance claims, for an average of $1,800 per claim. OPM re-competed and awarded a contract to the previously contracted company in December 2018. GAO's review did not identify any studies that analyzed whether consumers who sign up for or purchase identity theft services were less subject to identity theft or detected financial or other fraud more or less quickly than those who monitored their own accounts for free. A few experts said consumers could sign up for such services if offered for free. Credit monitoring may be convenient for consumers and personalized restoration services may help identity theft victims recover their identities, but such services do not prevent fraud from happening in the first place. The services also do not prevent or directly address risks of nonfinancial harm such as medical identity theft. Consumer, government, and industry experts highlighted other free options, including a credit freeze, which prevents one type of fraud. A freeze restricts businesses from accessing a person's credit report—and can prevent the illicit opening of a new account or loan in the person's name. A provision of federal law that took effect in September 2018 made it free for consumers to place or lift credit freezes quickly at the three nationwide consumer reporting agencies (Equifax, Experian, and TransUnion). Consumers also can regularly monitor their accounts and review their credit reports for free every 12 months. In addition, they can take advantage of free federal assistance such as the guidance on the Federal Trade Commission's IdentityTheft.gov website. Finally, large amounts of personal information are outside of consumers' control and bad actors can use stolen information for years after a breach. Therefore, experts noted that data security at entities that hold such information—and efforts to make stolen information less useful for identity thieves, through use of new identity verification technologies, for example—are important ways to mitigate risks of harm for consumers. What GAO Recommends GAO reiterates a matter for congressional consideration and a recommendation from its 2017 report on identity theft services ( GAO-17-254 ). In that report, GAO found that legislation requiring federal agencies that experience data breaches, including OPM, to offer certain levels of identity theft insurance coverage to affected individuals requires coverage levels that are likely unnecessary. Therefore, Congress should consider permitting agencies to determine the appropriate coverage level for such insurance. GAO also recommended the Office of Management and Budget (OMB) update its guidance for agency responses to data breaches, after analyzing the effectiveness of identity theft services relative to lower-cost alternatives. OMB did not agree or disagree and had not taken action as of early March 2019.

gao_GAO-18-352

gao_GAO-18-352_0

Background VA’s Disability Compensation Claims Process VA’s process for deciding veterans’ eligibility for disability compensation begins when a veteran submits a claim to VA. The veteran submits his or her claim to one of VBA’s 56 regional offices, where staff members assist the veteran by gathering additional evidence, such as military and medical records, that is needed to evaluate the claim. Based on this evidence, VBA decides whether the veteran is entitled to compensation and, if so, how much. A veteran dissatisfied with the initial claim decision can generally appeal within 1 year from the date of the notification letter sent by VBA. Under the current appeals process (now referred to by VA as the legacy process), an appeal begins with the veteran filing a Notice of Disagreement. VBA then re-examines the case and generally issues a Statement of the Case that represents its decision. A veteran dissatisfied with VBA’s decision can file an appeal with the Board. In filing that appeal, the veteran can indicate whether a Board hearing is desired. Before the Board reviews the appeal, VBA prepares the file and certifies it as ready for Board review. If the veteran requests a hearing to present new evidence or arguments, the Board will hold a hearing by videoconference or at a local VBA regional office. The Board’s members, also known as Veterans Law Judges, review the evidence and either issue a decision to grant or deny the veteran’s appeal or refer (or remand) the appeal back to VBA for further work. New Appeals Process The 2017 Act made changes to VA’s legacy appeals process that will generally take effect no earlier than February 2019, which is approximately 18 months from the date of enactment. According to its appeals plan, VA intends to implement the Act by replacing the current appeals process with a process offering veterans who are dissatisfied with VBA’s decision on their claim one of five options: two of those options afford the veteran an opportunity for an additional review of VBA’s decision within VBA, and the other three options afford them the opportunity to bypass additional VBA review and appeal directly to the Board. Under the new appeals process, the two VBA options will be: 1. Request higher-level review: The veteran asks VBA to review its initial decision based on the same evidence but with a higher-level official reviewing and issuing a new decision. 2. File supplemental claim: The veteran provides additional evidence and files a supplemental claim with VBA for a new decision on the claim. The three Board options will be: 3. Request Board review of existing record: The veteran appeals to the Board and asks it to review only the existing record without a hearing. 4. Request Board review of additional evidence, without a hearing. 5. Request Board review of additional evidence, with a hearing. VA’s Appeals Plan The Act also requires VA to submit to the appropriate committees of Congress and GAO, within 90 days of the date of enactment, a comprehensive plan for (1) processing appeals under the legacy process until there are no more to process, (2) implementing the new appeals process, (3) processing of claims under the new appeals process in a timely manner, and (4) monitoring implementation of the new appeals process. In addition to these four broad elements, the Act lists 18 elements required to be included in the plan that relate to, among other things: staffing, IT, and other resources required to implement the plan; estimated timelines for hiring and training VA employees; and a description of risks associated with each element of the plan. The Act also includes a provision for GAO to assess the plan within 90 days after VA submits it. The Act also requires VA to provide progress reports to the appropriate committees of Congress and GAO at least once every 90 days (starting after VA submits its plan), until the date the Act’s legal changes to the appeals process generally go into effect and then at least once every 180 days after this date for 7 years. Rapid Appeals Modernization Program (RAMP) The Act also authorized VA to carry out a program to test any assumptions relied upon in developing its comprehensive plan and test the feasibility and advisability of any facet of the new appeals process. In its appeals plan, VA reported its decision to pilot test two of the five new options by allowing veterans with pending appeals in the legacy process (known as legacy appeals) to elect the VBA higher-level review or VBA supplemental claim options beginning in November 2017. This program, which VA refers to as RAMP, is intended to reduce legacy appeals by providing veterans with a chance for early resolution of their claims within VBA while the Board focuses on reducing its inventory of legacy appeals, according to VA. Participation in RAMP is voluntary, but veterans must withdraw their pending legacy appeal to participate, according to VA. Veterans dissatisfied with their RAMP decisions must wait until VA fully implements the new appeals process (in February 2019 at the earliest) before pursuing an appeal with the Board under the new process, according to VA officials. VA’s Plan Addresses Most of the Act’s Required Elements for the New and Legacy Disability Appeals Processes VA’s appeals plan addresses 17 of the Act’s 22 required elements, partially addresses 4 related to monitoring implementation and workforce planning, and does not address 1 element related to identifying total resources. For example, VA’s appeals plan addresses the required elements related to, among others, identifying legal authorities for hiring and removing employees, estimating timelines for hiring and training employees, and outlining the outreach VA expects to conduct. For the elements in the Act that VA’s appeals plan partially addresses or does not address, see table 1. For a detailed list of the 22 required elements in the Act, see appendix I. When we provided VA with our preliminary assessment, VA officials said they disagreed and that their appeals plan addresses all 22 of the required elements. In general, they said that data are not available, and VA cannot yet forecast the information required by the Act until aspects of the new appeals process are tested or implemented. However, in discussing our assessment at a January 2018 House Committee on Veterans’ Affairs hearing, the Deputy Secretary of Veterans Affairs stated that the agency agreed with our assessment and will work with us to address these gaps in VA’s appeals plan. Until VA’s appeals plan has complete information on all 22 of the required elements, Congress does not have the information it needs to fully conduct oversight of the plan and the agency’s efforts to implement and administer the new process while addressing legacy appeals. VA also needs information on resources, among other areas, to certify that the agency is prepared to carry out timely processing of appeals under the new and legacy appeals process. Further, as discussed below, addressing required elements through a more comprehensive plan and underlying analysis is consistent with sound planning practices and would better position VA to implement the new appeals process while attending to legacy appeals. For example, such an appeals plan would provide for carefully monitoring the new and legacy appeals processes against balanced goals and metrics, and clearly articulates resources, milestones and other information needed for effective program management. VA’s Appeals Plan Reflects Certain Sound Planning Practices, but Could Improve on Others VA’s appeals plan reflects certain sound planning practices, such as convening a working group on performance tracking; however, the plan could benefit from including important details related to three key planning areas: 1. articulating a balanced set of goals and related measures to monitor and assess the performance of the new appeals process, in conjunction with the legacy process; 2. developing a high-quality and reliable implementation schedule to manage key steps and activities of the project; and 3. assessing key risks in a comprehensive manner, including respective mitigation strategies, articulating clear criteria and an assessment plan for RAMP, and more fully testing or analyzing all appeals options. VA’s Appeals Plan Indicates Steps to Assess Process Changes, but Should Also Include Goals and Measures to Provide Full Picture of Success VA’s appeals plan reflects steps taken to track performance, but it could improve its planning practices related to monitoring and assessing performance on a range of key dimensions of success. Sound planning practices suggest that agencies develop overall goals tied to meaningful and balanced performance measures. These measures include a mix of outcome, output, and efficiency measures to ensure that an organization’s priorities—as well as government-wide priorities such as quality, timeliness, and cost of service—are addressed. VA’s appeals plan reports that the agency convened a working group to design a process for tracking timeliness of both the legacy appeals and appeals within the new process. In supporting documentation that we requested, VA officials stated they are also determining the best way to measure veterans’ satisfaction with the new appeals process. VA’s appeals plan and supporting documentation also identify timeliness goals for the two VBA-only options and one of the three Board options. Nevertheless, its appeals plan does not articulate a set of goals and measures that cover all aspects of its new appeals process, such as accuracy of decisions and cost. The plan also does not provide details on the metrics the agency will develop, how it will assess if the new appeals process is an improvement over the legacy appeals process, and how it will monitor the allocation of resources between legacy and new appeals claims. More specifically: VA’s reported timeliness measures are incomplete: VA’s appeals plan outlines timeliness goals for the two VBA options (average processing time of 125 days) and for the Board option that does not include new evidence or a hearing (average processing time of 365 days). However, VA’s plan does not establish timeliness goals for the other two Board options: Board review of additional evidence without a hearing and Board review of additional evidence with a hearing. In commenting on our findings, while VA officials indicated they expect the new process to be more efficient than the legacy process (and, therefore, more timely), they also said data to inform goal setting for all Board options will not be available until VA fully implements these options. Establishing timeliness goals for all options would provide a more complete picture of VA’s vision for the new appeals process, and help VA to develop concrete, objective, and observable performance measures to show progress in achieving that vision, as well as inform resource estimates. VA’s reported measures lack adequate balance: Other than including certain timeliness goals, VA’s appeals plan does not articulate additional aspects of performance important for managing appeals, such as accuracy of decisions, veteran satisfaction with the process, or cost. We previously reported that VA officials said they wanted to also use veteran survey results, wait times, and inventories as sources of information to measure progress under the new appeals process. Further, VA’s fiscal year 2018 annual performance plan includes an overall customer satisfaction score for veterans’ benefits. However, these and other potential measures of success are not specified in VA’s appeals plan for monitoring the new appeals process as compared with legacy appeals. By not articulating a set of comprehensive and balanced goals and measures in its appeals plan, VA could be inadvertently creating skewed incentives by focusing on one area of program performance to the detriment of other areas (e.g., processing claims quickly but inaccurately). In commenting on our findings, VA officials recognized the need to develop additional goals and measures. They indicated, for example, that they are developing and testing whether the existing quality assurance goal—92 percent accuracy—is appropriate for the new process. According to VA officials, once they have developed these other goals and measures, VA will communicate this information as part of the required progress reports to the appropriate committees of Congress and GAO. Moreover, at a January 2018 House Committee on Veterans’ Affairs hearing, the Deputy Secretary of Veterans Affairs acknowledged that their performance goals and measures are not yet complete and indicated that the agency will address these gaps in measuring performance. VA’s plan does not reflect how it will establish baseline data: VA’s approach for evaluating the efficiency and effectiveness of the implementation of the new appeals process falls short of sound practices for using baseline data to assess performance. Our prior work has demonstrated that by developing and tracking a performance baseline for all measures, including those that demonstrate the effectiveness of a program, agencies can better evaluate progress made and whether or not goals are being achieved. However, VA’s appeals plan did not provide important details about what aspects of the new appeals process’ performance will be compared to what aspects of the legacy process’ performance. In particular, section 5 of the Act lists a number of metrics VA is required to report periodically, including some that could be used as baseline measures. For example, VA is required to periodically publish on its website the average time that elapsed between the filing of an initial claim and the final resolution of the claim, for legacy appeals as well as appeals under the new system, which is consistent with our prior recommendation. However, VA’s appeals plan does not explain how or when the agency would collect and use these or other data about the legacy and new processes’ performance—such as accuracy, veteran satisfaction, and cost—to assess their relative performance. As we had previously reported, VA’s business case for reform in some instances relied on unproven assumptions and limited analyses of its legacy process to identify root causes of performance problems. Specifically, VA determined that the open-ended nature of its legacy appeals process, whereby a veteran can submit additional evidence numerous times at any point, can cause additional cycles of re- adjudication, a process VA refers to as “churning.” According to VA, this re-adjudication can occur multiple times and can add years to the time needed to reach a final decision on an appeal. Without fully articulating a plan for collecting and using baseline and trend data, VA cannot determine the extent to which the new appeals process, which also allows for multiple appeal opportunities, will achieve final resolution of veterans’ appeals sooner, on average, than the legacy process. In commenting on our assessment, VA indicated that it is working toward capturing the metrics listed in section 5 of the Act. VA officials also noted that reporting on the new appeals process will require IT system functionality that currently does not exist, but stated that efforts are underway to add this functionality. VA’s plan does not explain how the agency will monitor processing of legacy versus new appeals: In addition, VA’s appeals plan does not fully articulate how the agency will monitor whether resources are being appropriately devoted to both the new and legacy appeals process and how it will track both sets of workloads. An appeals plan that does not specifically articulate how VA will manage the two processes in parallel exposes the agency to risk that veterans with appeals in the legacy process may experience significant delays or otherwise poor results relative to those in the new appeals process or vice versa. In commenting on our findings, VA officials noted that VA was not required under section 3 of the Act to provide a description of its plans to capture metrics listed in section 5. Even if not required by the Act, developing an approach for carefully monitoring the management of new and legacy appeals would help VA track progress being made and achievement of goals. Until VA establishes complete and balanced goals and measures, identifies baseline data, and develops a plan for monitoring and assessing both the new and legacy processes, VA runs the risk of promoting skewed behaviors, or not fully understanding whether the new process is an improvement or whether veterans with appeals in the legacy process are experiencing poor results. VA’s Appeals Plan Needs a Reliable Implementation Schedule to Manage the Project VA’s appeals plan reflects certain aspects of sound planning practices related to managing the implementation of process change; however, other key components are not addressed. Sound planning practices for implementing process change suggest establishing a transition team. Consistent with such practices, VA’s appeals plan states that the agency convened an agency-wide governance structure to coordinate implementation of its new appeals process; it is comprised of senior-level employees with authority to make necessary decisions to keep the project on track. VA’s appeals plan also includes a copy of a master schedule. In its plan, VA asserts that the master schedule reflects timelines, interim goals and milestones, reporting requirements, and established deadlines, and that it will be used to guide implementation. VA’s appeals plan also reports that VA is consulting with project management professionals, who are using the master schedule, among other tools, to monitor implementation. In addition, VA made progress addressing some of the issues we previously identified by developing steps and timetables for updating training in anticipation of implementing the new appeals process. However, VA’s master schedule for implementing reform is missing elements of a high-quality and reliable implementation schedule for key activities. We have previously reported that having a well-planned schedule is a fundamental management tool. Generally recognized sound practices from the Project Management Institute (PMI) and GAO call for organizations to employ an integrated and reliable master schedule that defines when work activities will occur, who will complete the work, how long they will take, how they are related to one another, and the constraints affecting the start and completion of work elements, as well as whether resources will be available when they are needed. Such a project management schedule not only provides a road map for systematic project execution, but also provides the means by which to gauge progress, identify and address potential problems, and promote accountability. The master schedule VA provided in its appeals plan should have included other sound practices for project management related to a reliable schedule. Specifically: Key activities and their duration are not included: VA’s master schedule does not capture specific Board-related activities, such as efforts to develop metrics, and the schedule and other project plans we reviewed do not go beyond February 2019. For example, the schedule does not indicate the period of time when VA expects to no longer be processing legacy appeals. Also, VA’s master schedule submitted with its November 2017 appeals plan did not include the Rapid Appeals Modernization Program (RAMP) activities, even though this pilot test is occurring at the same time VA is preparing for full implementation of appeals options at VBA and the Board. However, VA’s updated schedule that accompanied its comments on our draft report was updated to include RAMP. When all key and necessary activities are not included, it raises questions about whether all activities are scheduled in the correct order, resources are properly allocated, or the estimated completion dates are reliable. In addition, if the schedule does not fully and accurately reflect VA’s efforts, it will not serve as an appropriate basis for analysis and may result in unreliable completion dates and delays. Sequencing and linkages among activities are not identified: For the high-level activities VA’s appeals plan identifies, VA’s master schedule does not indicate whether there were sequencing or linkages among them, which is not consistent with sound scheduling practices. Sequencing and linkages would show, for example, if any of these activities or sub-activities must finish prior to the start of other activities, or the amount of time an activity could be delayed before the delay affects VA’s estimated implementation date. For example, VA cannot train new employees until after it hires them. The activities VA identifies also do not appear supported by lower- level project schedules. Specifically, when we requested documentation to support VA’s high-level summary of activities and milestones, VA officials did not provide intermediate or more detailed schedules that reflected these practices. In particular, VA’s appeals plan lacks a complete schedule for IT modifications that clearly defines what is to be achieved and the time frames for achievement. We previously recommended that VA develop a schedule for IT updates that explicitly addresses when and how process reform will be integrated into new systems and when these systems will be ready to support the new appeals process at its onset. For example, VA’s appeals plan references several required IT modifications that do not appear in its master schedule. Schedules that are defined at too high a level may disguise risk that is inherent in lower-level activities. Interim goals are not reflected: VA officials stated that they have interim goals and milestones, though VA’s appeals plan and supporting documentation generally do not include this information. Sound planning and redesign practices suggest closely monitoring implementation and developing project goals that include a mix of intermediate goals to be met at various stages. We previously made a recommendation that VA develop a more robust plan for closely monitoring implementation of process reform, including metrics and interim goals to help track progress, evaluate efficiency and effectiveness, and identify trouble spots—all of which are consistent with sound planning practices. Without interim goals and milestones, VA lacks information to support sequencing of activities and to track and ensure accountability for steady progress. Resources are not assigned to all identified activities: The high- level summary schedule that VA provided us also lacks details regarding the assignment of resources for all activities. Specifically, while the plan identifies workgroups responsible for coordinating elements in the plan, such as regulations, training, and outreach, the schedule does not assign resources to the 40 listed activities. As discussed previously, VA’s appeals plan also does not provide information on the total resources required for this reform effort. Assigning resources to the listed activities, as well as providing other information, could provide a better indication of the estimated total resources required to implement the new appeals process and address legacy appeals. In commenting on our findings, VA officials stated that the agency is developing lower-level project schedules for key activities—such as RAMP and IT requirements—and will provide these schedules as part of the required progress reports to the appropriate committees of Congress and GAO. VA officials also noted that future updates will include additional dependencies and risks, which VBA and the Board are still developing. Further, in discussing our findings at a January 2018 House Committee on Veterans’ Affairs hearing, the Deputy Secretary of Veterans Affairs reiterated VA’s commitment to developing more robust project plans, particularly for IT. Until VA has a robust integrated master schedule, supported by detailed project plans that adhere to sound practices, VA’s appeals plan does not provide reasonable assurance that decision makers have the essential program management information needed for this complex and important effort. VA’s Plan Addresses Some but Not All Key Risks Related to the New Appeals Process VA’s appeals plan includes an assessment of risks involved in implementing the new appeals system, but could more comprehensively reflect key risks posed by such a significant reform effort. VA’s appeals plan and supplemental materials include a “risk register” that describes risks associated with many elements of its plan, and the remaining level of risk after its planned response to these risks. VA’s appeals plan also states that senior leaders will receive regular updates of risks and mitigation strategies. However, because VA has not yet articulated a balanced set of performance goals and measures in its appeals plan, it is hindered in its ability to identify and assess risks. Federal internal control standards state, and our previous work at VA and other agencies demonstrates, that establishing clear performance goals and objectives is a necessary pre-condition to effectively assessing risk. Having, for example, more complete timeliness goals, and goals and measures reflecting other areas of performance, would allow VA to better identify and target risks associated with managing two processes in parallel, including the potential that veterans with appeals in the legacy process may experience significant delays relative to those in the new appeals process. Importantly, VA is missing an opportunity to fully benefit from RAMP by not testing and assessing other aspects of the new appeals process. The Act authorizes VA to test the feasibility and advisability of any facet of the new appeals process, and VA is taking a positive step to mitigate some risks by testing the two review options available within VBA (review of a claim by a higher-level official based on the same evidence and review of a supplemental claim with additional evidence) through RAMP. In November 2017, VA began RAMP by inviting 500 veterans whose appeals have been pending the longest to participate. According to VA officials, VA plans to continue offering RAMP to additional eligible veterans with pending legacy appeals each month until January 2019—a month before VA anticipates fully implementing the new appeals system. However, as designed, RAMP does not include features—consistent with a well-developed and documented pilot test program—that would provide VA with an opportunity to evaluate fully the soundness of new processes and practices on a smaller scale. Specifically: VA’s plan does not clearly define success criteria for RAMP: VA’s appeals plan states that the agency will collect certain data from RAMP, such as the rate at which eligible veterans opt into the process, timeliness of claims processing, and individual employee productivity. VA also established an overall average processing time goal of 125 days for the two VBA options; however, the plan and supporting documentation did not clearly articulate whether RAMP reviews are expected to meet this timeliness goal. The plan also did not identify other success criteria for RAMP or the types of results expected before fully implementing the new appeals process. For example, VA’s plan does not articulate the expected number and type of subsequent appeals to the Board that result from RAMP. In commenting on our findings, VA noted that its intent in implementing RAMP was to collect data and test aspects of the new process, and that RAMP was not an initiative in and of itself. However, developing performance measures and data gathering procedures and defining success criteria for a pilot test before proceeding to full implementation are sound practices for process redesign and pilot testing. In addition, because RAMP was not included in VA’s risk assessment, we asked VA if it had identified any risks or mitigation strategies specific to RAMP. In its supplemental materials, VA stated that the greatest risk to RAMP is a low participation rate among eligible veterans with legacy claims. VA also indicated that it would need 10 percent of eligible veterans to opt into RAMP to yield meaningful results. However, this threshold is not articulated in VA’s appeals plan as an explicit success criterion or objective. According to data provided by VA, as of January 22, 2018, 238 veterans opted in. Of veterans with pending claims in RAMP, two-thirds chose the higher-level review option. VA also reported that 47 RAMP decisions have been made so far. As of yet, no appeals of RAMP decisions have been filed. VA’s plan does not articulate how it will assess RAMP before proceeding with full implementation: Although VA’s appeals plan describes a “close-out” phase in which VA intends to assess the results of RAMP, it does not detail the conditions that would have to be met (or not met) to trigger changes. For example, VA’s plan does not explain when or how it might respond to low opt-in rates for RAMP—other than stating it will increase outreach to eligible veterans—or to unexpectedly high appeal rates to the Board resulting from RAMP decisions. Sound redesign and change management practices both suggest that pilot tests be rigorously monitored and evaluated, and that further roll-out occur only after an agency’s transition team takes any needed corrective action and determines that the new process is achieving previously identified success criteria. Without fully articulating its plan for deciding how and when to roll out changes more broadly, it is not clear whether VA would be prepared to fully implement a new appeals process that achieves its aim of better serving veterans. RAMP does not test all aspects of the new appeals process: RAMP provides an opportunity to learn about experiences at VBA under the new system, such as the rate at which eligible veterans choose those options and the resources that will be required to process their appeals. However, RAMP was not designed to test how many veterans would choose to appeal directly to the Board and, therefore, it will not provide comparable information on the Board appeals options. Sound workforce planning practices suggest that agencies identify the total resources needed to manage the risk of implementing new processes and conduct scenario planning to determine those needs. In addition, although we previously recommended VA conduct additional sensitivity analyses to inform projections of future appeals inventories, VA’s appeals plan does not reflect VA’s use or intended use of sensitivity analyses when projecting staffing needs for new appeals options at the Board. In commenting on our findings, VA officials said they do not plan to conduct additional sensitivity analyses to project future workloads until they have more information from RAMP to inform their assumptions. As a result, VA will lack data on scenarios in which veterans may overwhelmingly choose options available at the Board over those at VBA when the appeals plan is fully implemented. This presents a risk that VA’s early production projections and initial resource allocations may not be properly balanced between the Board and VBA. This, in turn, may result in an unexpectedly large number of appeals pending with the Board, and corresponding lengthy average wait and decision times for some, if not all, Board options. Having information on the number of veterans who are likely to appeal to the Board is particularly critical, given that similar efforts to create additional review options at VBA did not achieve their goals of reducing the percentage of appeals that continue on to the Board. In 2001, VA established the Decision Review Officer (DRO) process—in which senior staff have the authority to overturn an initial disability claim decision without any new evidence—to resolve more appeals at the regional level and avoid long waits at the Board. However, we reported in 2011 that, although the DRO process helped some veterans get additional benefits at the regional office level, it did not accomplish the program’s primary goal of reducing the percentage of appeals continuing on to the Board. In responding to our findings, VA officials reiterated their plans to increase outreach in the event of low opt-in rates for RAMP and indicated they recently began to send follow-up RAMP invitation letters. With respect to assessing all appeals options, VA officials stated that, while no legal bar prevents testing of the Board options, the Board is focused on reducing its inventory of pending appeals while RAMP provides early resolution of appeals within the new VBA-only options. Officials conceded that this approach means they cannot collect data on the rate at which veterans opt to appeal directly to the Board (e.g., bypassing additional VBA review) until the new process is fully implemented. However, they noted that they can collect some data on the rate at which veterans whose appeals go through RAMP file subsequent appeals to the Board, even though the Board will not begin processing those appeals until full implementation. At a January 2018 House Committee on Veterans’ Affairs hearing, the Deputy Secretary of Veterans Affairs stated that given our assessment, VA will adjust its approach to identify and mitigate risks associated with implementing all of the appeals options, including those at the Board. Until VA pursues an approach that identifies and mitigates significant risks associated with implementing a new process, VA is taking a chance that untested aspects will not perform as desired. The Act provides VA authority to pilot aspects of the process and flexibility on the timing of implementing the new process, which could allow some additional time for VA to carefully measure performance under RAMP and determine whether any corrective actions are necessary. If VA does not take full advantage of this authority, it risks moving forward without knowing whether the new appeals process improves experiences for veterans, and potentially implementing a process that is more expensive or results in longer wait times than originally anticipated. Conclusions In implementing appeals reform after the enactment of the Veterans Appeals Improvement and Modernization Act of 2017, VA is undertaking a complex endeavor that has the potential to affect the lives of hundreds of thousands of veterans with service-connected disabilities. Such an endeavor demands a commensurate level of planning to be successful. While the Act required VA to submit its plan within 90 days of enactment, VA had proposed and began to plan for appeals reform much earlier, and had our March 2017 recommendations to guide its planning efforts from a foundation of sound practices. VA’s November 2017 appeals plan is a positive step forward. Certain elements of the plan—such as establishing an agency-wide governance structure to oversee implementation and testing aspects of reform prior to full implementation—are notable gains since our March 2017 report. At the same time, the plan partially addresses or does not address five of the required elements called for by the Act, such as delineating the total resources required by VBA and the Board to implement and administer the new appeals process and address legacy appeals. The plan also is not fully responsive to our past recommendations and does not reflect a number of sound planning practices that are essential for gauging progress, establishing accountability, and linking resources to results. One such key practice is articulating a desired “end state”—a vision for what successful implementation would look like for the new appeals process as well as the wind-down of the legacy process, such as accurate and timely processing of appeals while ensuring veteran satisfaction. Without establishing a complete and balanced set of goals and related performance measures to achieve this end state, and monitoring and assessing progress along the way, VA risks falling short of its overarching objective—to improve timeliness of appeals decisions for veterans overall. By not fully articulating how it plans to monitor workloads and devote resources to both the new and legacy processes, VA runs the risk of disadvantaging veterans with legacy appeals relative to those in the new process, or vice versa. Just as important is establishing a robust integrated master schedule— rather than a high-level timeline—that is built upon and clearly reflects extensive detailed planning and includes all activities necessary to execute the program and interdependencies between these activities. Without such a road map, VA’s appeals plan does not provide reasonable assurance that decision makers have the essential information needed to manage this complex and important program. We are encouraged that VA has taken some steps toward assessing risks, including establishing a risk register and implementing RAMP to collect information on the two VBA appeals options. However, unless VA assesses risks against a balanced set of goals and measures, VA may not be fully aware of risks that may impede successful implementation of appeals reform. Further, although VA will undoubtedly learn from the RAMP experience, it may not learn all that it should from its efforts without (1) establishing clear criteria for what success looks like (or the circumstances that would cause VA to consider making course corrections), and (2) building in time to take stock of the lessons learned before moving to full implementation. VA’s plan places a lot of weight on RAMP to, among other efforts, mitigate risk and generate estimates of the resources needed for successful implementation after fiscal year 2018, even though RAMP does not fully test options for appealing to the Board that will be available to veterans after full implementation. Unless VA addresses key risks associated with fully implementing appeals reform—by either testing or conducting sensitivity analyses for all five appeals options, to better understand potential workloads at the Board—VA runs the risk of fully implementing the process without knowing if it is improving the process for veterans. Recommendations for Executive Action We are making the following four recommendations to VA: The Secretary of Veterans Affairs should address all of the required elements in the Act in VA’s appeals plan to Congress—including delineating resources required for all VBA and Board appeals options— using sensitivity analyses and RAMP results, where appropriate and needed. (Recommendation 1) The Secretary of Veterans Affairs should clearly articulate in VA’s appeals plan how VA will monitor and assess the new appeals process compared to the legacy process, including specifying a balanced set of goals and measures—such as timeliness goals for all VBA appeals options and Board dockets, and measures of accuracy, veteran satisfaction, and cost—and related baseline data. (Recommendation 2) The Secretary of Veterans Affairs should augment the master schedule for VA’s appeals plan to reflect all activities—such as modifications to IT systems—as well as assigned responsibilities, interdependencies, start and end dates for key activities for each workgroup, and resources, to establish accountability and reduce overall risk of implementation failures. (Recommendation 3) The Secretary of Veterans Affairs should ensure that the appeals plan more fully addresses risk associated with appeals reform—for example, by assessing risks against a balanced set of goals and measures, articulating success criteria and an assessment plan for RAMP, and testing or conducting sensitivity analyses of all appeals options—prior to fully implementing the new appeals process. (Recommendation 4) Agency Comments and Our Evaluation We provided a draft of this report to the Department of Veterans Affairs (VA) for comment. In its comments, reproduced in appendix II, VA outlined its planned actions to address the recommendations. VA clarified in a subsequent communication that the agency agreed with all our recommendations. VA did not provide technical comments. With respect to our first recommendation that VA address all the required elements in the Act in VA’s appeals plan (including delineating total resources), VA stated that both VBA and the Board will use existing resources to implement the new appeals process. VA also stated it plans to take additional steps to determine resource requirements for addressing workloads under both the legacy and new appeals process. For VBA, VA stated that it will continue to rely on RAMP to project resource requirements, while acknowledging the need to augment its analysis of RAMP data by adopting additional strategies to project resource requirements. VA did not describe these strategies, but stated that it will share them with Congress and GAO in the near future. Meanwhile, VA noted that its 2019 budget request includes 605 additional FTEs for VBA to process appeals, but did not indicate how it developed this budget request. For the Board, VA stated that it plans to develop better predictions regarding resource allocations among dockets by leveraging project management and other support within the agency. We will continue to monitor VA’s efforts to delineate needed resources, including how it uses the results of pilot tests and prediction analysis. With respect to our second recommendation to articulate how it will monitor and assess the new appeals process compared to the legacy process, VA stated that it is working to develop a complete and balanced set of measures for the new appeals process, and timeliness goals for all appeals options. Further, VA indicated it will track performance for Board appeal options using an existing process. We are encouraged by VA’s proposed actions, which will provide a more complete picture of VA’s vision for the new appeals process. However, VA does not detail whether or how the agency will develop a baseline or compare performance of the new appeals process to the legacy process. Until VA develops a baseline and a plan for monitoring and assessing both the new and legacy processes—using a complete and balanced set of goals and measures— VA risks not fully understanding whether the new process is an improvement. With respect to our third recommendation to augment its master implementation schedule to manage the project, VA provided an updated schedule with additional key activities and responsibilities, such as RAMP. Moreover, VA restated its plans to use an agency-wide governance structure to coordinate and track implementation of its new appeals process. We are encouraged by VA’s efforts to develop a more detailed implementation schedule. However, the updated schedule VA provided does not include major activities, such as integrated IT system testing, and completion dates for major activities, for example, adding functionality to VA’s primary claims processing system. In addition, VA provided an updated calendar for six major IT activities through the end of calendar year 2018. However, we continue to believe that VA will need to develop a longer term schedule that projects when processes will be integrated into new systems and when new systems will be ready to support the new appeals process. The schedule also does not indicate whether activities are interrelated, such that a delay in one activity could affect other activities and thereby affect VA’s estimated implementation date. This sound planning practice is especially important because VA stated the agency is concurrently executing many of the activities. With respect of our fourth recommendation to more fully assess risks associated with appeals reform prior to its full implementation, VA stated that it will assess risks against a balanced set of goals it plans to select. Moreover, using its existing risk management process, VA stated it has identified additional risks and mitigation strategies after submitting its November 2017 plan. For example, VA states that it is addressing the continued low opt-in rate for RAMP, which is testing the new VBA-only options. VA is also acknowledging that delays in the development of IT required to implement the appeals process may prevent the agency from certifying readiness in January 2019. Importantly, VA states that the Board is exploring the development of a pilot program to identify needs and concerns related to full implementation—including all Board appeals options—and to make predictions about timeliness and productivity under the new appeals process. However, VA did not define success criteria for its current pilot test, RAMP, or clearly articulate how the agency will assess results of either RAMP or a new test of Board appeals options before proceeding to full implementation. Implementing our recommendation in a complete and timely manner is important because it would improve VA’s ability to identify and mitigate significant risks associated with implementing a new process. We will continue to monitor the status of VA’s actions to address our recommendations and how they are implemented, to help ensure that VA is undertaking a level of planning appropriate to implementing a complex endeavor, and thereby improving VA’s chance of success. We are sending copies of this report to the appropriate congressional committees, the Secretary of Veterans Affairs, and other interested parties. In addition, the report is available at no charge on GAO’s website at http://www.gao.gov. If you or your staff have any questions about this report, please contact Elizabeth Curda at (202) 512-7215 or curdae@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who contributed to this report are listed in appendix III. Appendix I: Our Assessment of VA’s Appeals Plan Against Required Elements in the Act To assess the extent to which VA’s appeals plan addresses the required elements in the Veterans Appeals Improvement and Modernization Act of 2017 (Act), we first identified and developed a checklist reflecting each required element for VA’s appeals plan (including sub-parts) under section 3(a) and (b) of the Act. To compare the required elements and their sub-parts against VA’s appeals plan and supplemental materials provided by VA, we developed decision rules for determining whether VA’s appeals plan addressed, partially addressed, or did not address each required element (see table 2). Specifically, we concluded that VA’s plan addressed (or partially addressed) a required element if the plan included information related to all (or some) subparts of the requirement. We focused on the plan as presented, rather than auditing the information VA relied on in developing the plan. For example, the Act’s section 3(b)(10) required VA’s plan to include a description of the modifications to the information technology (IT) systems required to carry out the new appeals system, including cost estimates and a timeline. We concluded that VA’s plan addressed all sub-parts of this element because it provided a description of required IT modifications, a reference to costs included in the Appeals Modernization IT budget, and a timeline. However, our determination that VA addressed this element should not be construed to necessarily mean that VA fully identified or described all IT requirements, or provided complete estimated costs and timelines associated with those requirements, or that the information in VA’s appeals plan comported with sound planning practices. This type of assessment was outside the scope of this objective. Appendix II: Comments from the Department of Veterans Affairs Appendix III: GAO Contact and Staff Acknowledgments GAO Contact Elizabeth H. Curda, (202) 512-7215 or curdae@gao.gov. Staff Acknowledgments In addition to the contact above, the following staff members made significant contributions to this report: Michele Grgich (Assistant Director), James Whitcomb (Analyst in Charge), and Rachael Chamberlin. In addition, key support was provided by Susan Aschoff, Mark Bird, David Chrisinger, Daniel Concepcion, Clifton Douglas, Alex Galuten, Nisha Hazra, Melissa Jaynes, Benjamin Licht, Patricia McClure, Sheila McCoy, Lorin Obler, Gloria Proa, Almeta Spencer, James Sweetman, Walter Vance, and Greg Whitney.

Why GAO Did This Study VA's disability compensation program pays cash benefits to veterans with disabilities connected to their military service. In recent years, the number of appeals of VA's benefit decisions has been rising. For decisions made on appeal in fiscal year 2017, veterans waited an average of 3 years for resolution by either VBA or the Board, and 7 years for resolution by the Board. The Veterans Appeals Improvement and Modernization Act of 2017 makes changes to VA's current (legacy) appeals process, giving veterans new options to have their claims further reviewed by VBA or appeal directly to the Board. The Act requires VA to submit to Congress and GAO a plan for implementing a new appeals process, and includes a provision for GAO to assess VA's plan. This report examines the extent to which VA's plan (1) addresses the required elements in the Act, and (2) reflects sound planning practices identified in prior GAO work. GAO reviewed and assessed VA's appeals plan and related documents against sound planning practices, and solicited VA's views on its assessments. What GAO Found The Department of Veterans Affairs' (VA) plan for implementing a new disability appeals process while attending to appeals in the current process addresses most, but not all, elements required by the Veterans Appeals Improvement and Modernization Act of 2017 (Act). VA's appeals plan addresses 17 of 22 required elements, partially addresses 4, and does not address 1. For example, not addressed is the required element to include the resources needed by the Veterans Benefits Administration (VBA) and the Board of Veterans' Appeals (Board) to implement the new appeals process and address legacy appeals under the current process. VA needs this information to certify, as specified under the Act, that it has sufficient resources to implement appeals reform and make timely appeals decisions under the new and legacy processes. VA's appeals plan reflects certain sound planning practices, but it could benefit from including important details in several key planning areas: Performance measurement : VA's plan reflects steps taken to track performance, but could articulate a more complete and balanced set of goals and measures for monitoring and assessing performance on a range of dimensions of success. Specifically, the plan reports that VA is developing a process to track timeliness of the new and legacy processes. However, contrary to sound planning practices, the plan does not include timeliness goals for all five appeals options available to veterans, does not include goals or measures for additional aspects of performance (such as accuracy or cost), and does not explain how VA will monitor or assess the new process compared to the legacy process. Unless VA clearly articulates a complete and balanced set of goals and measures, it could inadvertently incentivize staff to focus on certain aspects of appeals performance over others or fail to improve overall service to veterans. Project management : VA's plan includes a master schedule for implementing the new appeals plan. However, this schedule falls short of other sound practices for guiding implementation and establishing accountability, such as articulating interim goals and needed resources for, and interdependencies among, activities. Unless VA augments its master schedule to include all key activities and reflect sound practices, VA may be unable to provide reasonable assurance that it has the essential program management information needed for this complex and important effort. Risk assessment : VA has taken steps to assess and mitigate some risks related to appeals reform by, for example, pilot testing two of the five appeals options through its Rapid Appeals Modernization Program (RAMP). However, as designed, RAMP does not include key features of a well-developed and documented pilot test. For example, VA has not articulated how it will assess RAMP before proceeding with full implementation. In addition, RAMP is not pilot testing three options and, as a result, VA will not have data on the extent to which veterans will appeal directly to the Board when given the option. Unless VA identifies and mitigates key risks associated with implementing a new process, VA is taking a chance that untested aspects will not perform as desired. What GAO Recommends GAO recommends that VA (1) fully address all legally required elements in its appeals plan, (2) articulate how it will monitor and assess the new appeals process as compared to the legacy process, (3) augment its master schedule for implementation, and (4) address risk more fully. VA agreed with GAO's recommendations and outlined its planned actions to address them.

gao_GAO-18-298

gao_GAO-18-298_0

Background The mission of the IRS, an agency within the Department of the Treasury (Treasury), is to provide America’s taxpayers with top quality service by helping them understand and meet their tax responsibilities; and to enforce the tax law with integrity and fairness to all. In carrying out its mission, IRS annually collects over $3 trillion in taxes from millions of individual taxpayers and numerous other types of taxpayers. It also manages the distribution of over $400 billion in refunds. To guide its future direction, the agency has six strategic goals: (1) empower and enable all taxpayers to meet their tax obligations; (2) protect the integrity of the tax system by encouraging compliance through administering and enforcing the tax code; (3) collaborate with external partners proactively to improve tax administration; (4) cultivate a well-equipped, diverse, flexible and engaged workforce; (5) advance data access, usability and analytics to inform decision making and improve operational outcomes; and (6) drive increased agility, efficiency, effectiveness and security in IRS operations. The mission of IRS’s Information Technology organization is to deliver IT services and solutions that drive effective tax administration to ensure public confidence. It is led by the Chief Information Officer (CIO), who oversees several subordinate offices. Figure 1 shows the structure of IRS’s Information Technology organization. IRS Relies on Major IT Investments for Tax Processing For fiscal year 2016, IRS’s IT portfolio contained 137 investments, of which 23 were classified as major. According to the agency, it spent approximately $2.7 billion on its IT investments during fiscal year 2016. Of the $2.7 billion, approximately $1.9 billion (70 percent) was spent for operations and maintenance activities, and approximately $800 million (30 percent) was spent for development, modernization, and enhancement. Among the agency’s investments that we selected for our review, the following four were primarily in development during fiscal year 2016: The Affordable Care Act (ACA) Administration investment encompasses the planning, development, and implementation of IT systems needed to support IRS’s tax administration responsibilities associated with the Patient Protection and Affordable Care Act. The agency reported spending $253 million on this investment in fiscal year 2016. Customer Account Data Engine 2 (CADE 2) is to, among other things, provide daily processing of taxpayer accounts, address a financial material weakness, and maintain a clean audit opinion. It is expected to replace the nearly 50 year old IMF system that IRS is using to process individual taxpayer accounts. A key project supporting CADE 2 is the Individual Tax Processing Engine project, which, according to the agency, is a complex effort to, among other things, convert approximately 200,000 lines of IMF’s legacy assembly language code to Java. According to IRS, the agency has completed an initial phase of converting the assembly language code for core IMF components to Java; however, significant work remains to complete the conversion. Specifically, in October 2017, IRS’s CIO stated that the agency could deliver a system to replace the core IMF components in 5 years if the agency was provided with 50 to 60 employees and the associated funding, direct hire authority to hire employees with the right skills, and approximately $85 million each year. The agency reported spending $182.6 million on CADE 2 in fiscal year 2016. The Return Review Program (RRP) is IRS’s primary system for fraud detection. As such, it supports the agency’s capabilities to detect, resolve, and prevent criminal and civil tax noncompliance. According to IRS, as of May 2017, the system had helped protect over $4.5 billion in revenue. The agency reported spending $100.2 million on this investment in fiscal year 2016. Enterprise Case Management (ECM) is to provide an enterprise solution for performing case management across IRS’s business units. According to the agency, its current systems provide limited visibility into case management practices between programs, process redundancies, and multiple handoffs that can lead to, among other things, increased risks; and ECM is expected to address these limitations. The agency reported spending $38.1 million on the investment in fiscal year 2016. Five other investments that we selected for our review were in the operations and maintenance phase during fiscal year 2016: MSSS represents approximately 73 percent of IRS’s IT infrastructure. Specifically, this investment encompasses the design, development, and deployment of servers, middleware and large systems, and enterprise storage infrastructures, including systems software products, databases, and operating systems. The MSSS investment began in 1970. The agency reported spending $499.4 million on this investment in fiscal year 2016. Telecommunications Systems and Support (TSS) provides the voice and data network infrastructure services, video services, and engineering throughout IRS. The TSS investment began in 2001. The agency reported spending $336.4 million on this investment in fiscal year 2016. End User Systems and Services (EUSS) provides desktops, laptops, mobile devices, software, incident management services, and asset management services to end users in IRS. The EUSS investment began in 2002. The agency reported spending $238.0 million on this investment in fiscal year 2016. IDRS is used by IRS employees to review tax information, issue notices to taxpayers, and update taxpayer records. The IDRS investment began in 1973. The agency reported spending $15.8 million on this investment in fiscal year 2016. IMF is IRS’s system for processing individual taxpayer account data. Using this system, accounts are updated, taxes are assessed, and refunds are generated as required during each tax filing period. Virtually all IRS information system applications and processes depend on output, directly or indirectly, from this data source. As previously noted, the agency uses assembly language code to program this system, which began in the late 1960s. The agency intends to decommission IMF once CADE 2 is fully implemented; however, as we recently reported, the agency has not provided a target date for decommissioning IMF. The agency reported spending $14.3 million on this investment in fiscal year 2016. GAO Has Made Recommendations to Improve IRS’s Major IT Investments, Government- Wide Legacy Systems, and IT Workforce Planning For several years, we have reported on the performance of IRS’s IT investments and identified opportunities for improving the management of these investments. In February 2015, we reported that the agency had provided summary-level Chief Technology Officer risk assessment ratings for the majority of IT investments in quarterly reporting to Congress. However, the agency did not provide such ratings for selected investments for which Congress required detailed reporting, including CADE 2 and RRP, which are the subject of this review. We noted that summary-level risk assessment ratings would improve the visibility into risks faced by the investments and provide Congress with the information to more easily determine the investments requiring greater attention. Consequently, we recommended that IRS provide summary-level risk assessment ratings for all major investments in its quarterly reporting to Congress. In response to our recommendation, IRS began providing summary- level risk information for all major investments in its fiscal year 2015 second quarter report to Congress. In its report for the fourth quarter of fiscal year 2016, the agency reported that selected aging systems were facing increased risks. In this regard, the agency CIO provided a risk assessment rating for major IT investments, which incorporated risks associated with people, infrastructure, deferred scope, and delivery of agreed-upon scope. IRS reported that two investments had risk ratings that went from yellow to red. Specifically, IMF received a red risk rating for the people factor and CADE 2 received a red risk rating for the delivery of agreed upon scope factor. In a report in June 2016, we noted, among other things, that CADE 2 and ACA did not report information on planned versus actual delivery of functionality in accordance with best practices. In addition, ACA did not report timely information on planned versus actual costs. Accordingly, we recommended that IRS report, at least quarterly, scope and cost performance for CADE 2 and ACA, consistent with best practices. In response to our recommendation, the agency began reporting on planned versus actual delivery of functionality for CADE 2 starting in fiscal year 2016. However, the agency has not reported on planned versus actual functionality for ACA. In March 2017, officials responsible for managing the investment told us that the agency had not implemented the recommendation because it did not see the benefit in doing so given that the remaining development work was minimal. IRS subsequently completed the development work for ACA in September 2017, at which point the investment transitioned to operations and maintenance. Given the status of the investment, we agree that the recommendation is no longer applicable. In addition, our prior work has emphasized the importance of IRS more effectively managing its legacy systems. As part of a government-wide review in November 2013, we reported on the extent to which 10 agencies’ large investments had undergone operational analyses—a key performance evaluation and oversight mechanism required by OMB to ensure investments in the operations and maintenance phase continue to meet agency needs. We noted that MSSS had not had an operational analysis for fiscal year 2012. As a result, we recommended that Treasury perform an operational analysis for the investment. The department did not comment on our recommendation but subsequently performed an operational analysis for the MSSS investment. In May 2016, we reported on legacy IT systems across the federal government, noting that these systems were becoming increasingly obsolete and that many of them used outdated software languages and hardware parts that were unsupported by the vendor. As part of that work, we highlighted Treasury’s use of assembly language code and Common Business Oriented Language (COBOL)—a programming language developed in the late 1950s and early 1960s—to program its legacy systems. We noted the need for agencies to move to more modern, maintainable languages, as appropriate and feasible. Further, we noted that a leading IT research and advisory company had reported that organizations using COBOL should consider replacing the language, and that there should be a shift in focus to using more modern languages for new products. We also pointed out that the use of COBOL presents challenges for agencies given that procurement and operating costs associated with this language will steadily rise, and because fewer people with the proper skill sets are available to support the language. Further, we reported that IMF was over 50 years old and, although IRS was working to modernize it, the agency did not have a time frame for completing the modernization or replacement. In addition, we noted that IRS did have not have specific activities or timelines for updating MSSS and EUSS. Thus, we recommended that Treasury direct the CIO to identify and plan to modernize or replace IRS’s legacy systems. The department had no comments on our recommendation. We have also previously reported on agencies’ IT workforce planning efforts. Specifically, in November 2016 we identified eight key IT workforce planning activities based on relevant laws and guidance and noted that the five federal departments in our review, including Treasury, had mixed progress in addressing the activities. We made one recommendation to Treasury and the department agreed with our recommendation. Performance of Selected IRS IT Investments Varied The performance of selected IRS IT investments has varied. In this regard, we found that the four selected investments in development had spent less than planned, and that most were behind schedule and had delivered less scope than planned. In addition most of these investments had significant variances, meaning that actual cost, schedule, or scope varied from plans by more than 10 percent. For the five selected investments in the operations and maintenance phase, we found that most had met all of their operational performance targets and all performed operational analyses required by OMB. However, none of the analyses addressed all key factors specified in OMB’s guidance. The Selected Investments in Development Spent Less than Planned, but Most Were Behind Schedule and Had Delivered Less Scope than Planned Best practices highlight the importance of monitoring the performance of projects in development by comparing actual cost, schedule, and scope to plans in order to allow appropriate corrective actions if actual performance deviates significantly from planned performance. With regard to the four selected investments in development that we reviewed, IRS reported cost, schedule, and scope performance information for ECM, CADE 2, and RRP, but the agency reported only cost and schedule information for ACA. Table 1 provides details reported by the agency on the performance of these IT investments. Regarding ECM, the agency reported that it spent $1.5 million less than budgeted, had an approximately 9 percent schedule overrun for the three projects it worked on during the time frame of our review, and delivered about 90 percent of planned scope. However, after 18 months of working with a contractor, the agency paused all development activities for the investment because the product that was being delivered did not meet the agency’s needs. Specifically, according to agency officials, including the CIO, the contractor’s solution was not sufficiently automated to be scalable across the agency. Thus, IRS subsequently established a new effort to acquire a product that would be aligned with its business needs. The officials stated that the strategy for acquiring the new product includes collaboration with other agencies on experiences in implementing enterprise case management systems and requesting information on potential solutions from commercial vendors. Regarding CADE 2, IRS reported that it spent $4 million less than it budgeted, had a 54 percent schedule overrun for the 15 projects it worked on during the time frame of our review, and delivered 46 percent of the planned scope. Officials responsible for managing CADE 2 stated that the cost, schedule, and scope variances from planned performance were due to human resource and funding shortages. Specifically, the agency reported that it does not have an adequate number of staff with expertise in assembly language code and tax processing to perform development work on both its core tax processing system (IMF) and its tax processing system modernization effort (CADE 2), or enough Java programmers to develop and maintain new code. As a result, the agency paused 7 of the 15 projects. (IRS’s efforts to address its human resources constraints are discussed later in this report.) For RRP, the agency reported that it spent $29.5 million less than budgeted, had a 19 percent schedule overrun for the 4 projects it worked on during the time frame of our review, and delivered about 80 percent of planned scope. According to the agency, these variances were due to, among other things, overestimation of planned costs, deferral of planned scope to a future release, and additional time needed to address a development defect. For ACA, the agency reported that it spent $41.6 million less than planned and was on time for the four projects it worked on. According to the agency, the cost variance was due to, among other things, an initial overestimation of the costs to complete planned work. IRS did not track scope delivery for ACA and, as a result, we could not determine the scope performance for the investment. As previously mentioned, in June 2016, we recommended that IRS report on actual scope information for ACA at least quarterly. In its March 2017 response to the recommendation, the agency stated that it did not see the benefits of implementing the recommendation for the investment given the minimal development work remaining. IRS has since completed this development work and transitioned ACA to the operations and maintenance phase. We agree that the recommendation is no longer applicable given the status of the investment. The Majority of Selected Investments in the Operations and Maintenance Phase Met Performance Targets According to OMB’s fiscal year 2016 capital planning guidance, ongoing performance of operational investments should be monitored to ensure the investments are meeting the needs of the agency, are delivering expected value, and/or are consistent with the agency’s enterprise architecture. To achieve these goals, agencies are required to establish and publically report on five operational metrics for major IT investments, as well as planned and actual performance against these metrics. According to OMB, these metrics seek to answer more subjective questions about areas such as customer satisfaction and financial performance. IRS reported operational performance metrics, as required, for the five selected investments we reviewed that were in the operations and maintenance phase. Further, three of these investments—IMF, MSSS, and TSS—met all of their operational performance targets, and the remaining two investments—IDRS and EUSS—met four of five operational performance targets during the time frame that we reviewed. Table 2 lists the operational performance metrics for each of the five investments, the metrics’ areas of focus, as well as the extent to which IRS met planned performance targets. With regard to the investments that did not meet their performance targets: IDRS did not meet its target for IRS employees’ usage of the investment for 9 out of the 18 months we reviewed. Officials responsible for managing IDRS stated that this was likely due to a reduction in the number of staff at the agency who access taxpayer accounts and to a lag experienced early in the months before notices are sent out for the filing year. EUSS did not meet its target for the average amount of time IRS employees wait to receive telephone support for 6 out of the 18 months we reviewed. According to officials responsible for managing EUSS, this target was missed due to the attrition of telephone support staff and the agency’s inability to hire additional support staff. Operational Analyses for Selected Investments in the Operations and Maintenance Phase Addressed Most, but Not All Key Factors OMB’s fiscal year 2016 capital programming guidance highlights the importance of operational analyses in examining the ongoing performance of operational investments. The guidance further notes that such analyses should be conducted at least annually and should address, among other things, the following: the extent to which the investment supports customer processes as designed, and how well the investment is delivering the goods or services it was designed to deliver; how well the investment contributes to achieving the organizations a comparison of current performance with a pre-established cost alternative methods of achieving the same mission needs and greater utilization of technology or consolidation of investments to better meet organizational goals. The five selected investments that we reviewed in the operations and maintenance phase performed operational analyses that addressed most, but not all of these key factors identified in OMB guidance. Specifically, four of the investments addressed five of the six factors, and one investment addressed four of the factors. Table 3 provides our assessment of the investments’ operational analyses. With regard to the investments that did not address all key factors identified in OMB’s guidance: The IMF operational analysis did not address the factor associated with greater utilization of technology or consolidation of investments to better meet organizational goals. Specifically, the analysis stated that the agency is researching the validity of converting legacy assembly language code to a modern programming language. However, the analysis did not more broadly address greater utilization of technology or consolidation to better meet organizational goals, consistent with the key factor in OMB’s guidance. In addition, the analysis did not reflect IRS’s progress to date in modernizing IMF and the associated challenges. This omission is concerning given the risk exposure from the agency’s continued use of the legacy assembly language code. (Such risk is further discussed later in this report.) With respect to IDRS, while the investment is intended to, among other things, provide for systemic review of tax information, issue notices to taxpayers, and update taxpayer records, IRS’s performance metrics generally focused on system availability and usage and did not address the extent to which the intended functionality was being provided. Agency officials agreed that IDRS’s metrics could be improved to address the extent to which intended functionality is being provided. For TSS, while the investment provides, among other things, video conferencing and enterprise voice and fax services, the operational analysis did not address how well these service offerings were being delivered. IRS officials stated that they had instead evaluated these service offerings in a post-implementation review, which is a one-time effort conducted after an investment has completed development. However, by not addressing the factor in the operational analysis, which is an annual exercise, IRS risks not being continually informed of the extent to which the investment is meeting the needs of the agency. In addition, the operational analysis for the TSS investment did not appropriately include a comparison of current performance with a pre-established cost baseline. Specifically, while the analysis included planned and actual cost figures for fiscal year 2016, the planned cost figure was not complete as it did not account for reimbursable costs and user fees. Regarding MSSS, the operational analysis did not address alternative methods of achieving the same mission needs and strategic goals. Senior officials in IRS’s Information Technology organization stated that the agency had performed analyses of alternative methods for achieving mission needs and strategic goals, but these analyses were not included in the operational analysis for the investment. The operational analysis for EUSS did not appropriately include a comparison of current performance with a pre-established cost baseline. Specifically, while the analysis included planned and actual cost figures for fiscal year 2016, the planned cost figure was not complete as it did not include multi-year funding and user fees. A Branch Chief for the IDRS investment stated that IRS has used the same operational performance metrics for the investment for 10 to 15 years, and the agency has not revisited them to justify their validity over time or to modify them. The Branch Chief further noted that the operational performance metrics are usage-based and do not provide a qualitative measure of how well the investment is delivering intended services. IRS officials did not identify the causes for the deficiencies we noted with the other selected investments’ operational analyses. Until IRS addresses the shortcomings noted for the selected operational investments, the agency risks not having critical information needed to determine whether the investments fully meet intended objectives and whether there are alternative ways to efficiently meet the agency’s mission. Selected Legacy Investments Face Significant Risks and IRS Has Not Implemented Steps Needed to Effectively Manage These Risks Three selected investments we reviewed—IMF, IDRS, and MSSS—are facing significant risks due to their reliance on legacy programming languages, outdated hardware, and a shortage of human resources with critical skills. However, IRS has not implemented steps needed to effectively manage these risks—and thus, the agency’s ability to carry out its tax processing and modernization efforts may be impacted. Two of the three selected investments—IMF and IDRS—rely on legacy programming languages, resulting in increased risk to continuing operation of these investments. Specifically, IRS reported that IMF is written in assembly language code and COBOL, and IDRS is written in COBOL. As we previously reported, reliance on assembly language code and COBOL has risks, such as a rise in procurement and operating costs, and a decrease in the availability of individuals with the proper skill sets. In addition, one investment in our review—MSSS—relies on a significant amount of outdated hardware. Specifically, at the start of fiscal year 2017, the agency reported an inventory of approximately $684.2 million in hardware associated with this investment. Of this amount, approximately $430.3 million, or 63 percent, was for outdated hardware, with about 21 percent of that amount directly supporting tax processing. The $430.3 million is broken down as follows: $112.6 million in communications equipment, which includes devices such as network switches and telephone systems; $171.5 million in systems supporting IRS employees, which includes desktop and laptop computers, scanners, and printers; $88.9 million in equipment directly supporting tax processing, which includes servers and UNISYS mainframes; and $57.3 million in storage equipment, which includes automated tape libraries and disk arrays. Figure 2 illustrates the categories of hardware associated with MSSS, including outdated hardware. IRS officials stated that the outdated hardware associated with MSSS is expensive to maintain because it is often past the warranty. Specifically, after a warranty for hardware ends, the maintenance fees for this hardware commonly increase by approximately 25 percent per year. In addition, the officials stated that, relying on this hardware has the potential to expose IRS to equipment failures that could preclude its systems from supporting the annual tax filing season and expanding the systems and tools for enforcement approaches, among other things. The three selected investments—IMF, MSSS, and IDRS—are also facing risks due to the attrition of key personnel. For example, IMF program officials noted that developers are responsible for maintaining taxpayer accounts and applying business rules associated with the tax process for a given situation or tax year, and thus require skills beyond creating or updating lines of code. However, according to an internal staffing report for IMF for 2011 to 2017, the agency experienced attrition of developers skilled in legacy programming languages and tax processing, exposing the investment to increased risks of not being able to successfully process tax information. For example: According to the report, from 2011 to 2017, 24 developers responsible for performing work on the IMF investment retired or were transferred to other positions. In addition, as a result of this attrition, 32 developers were available to perform IMF system updates for the 2017 tax filing season, which was about 4 developers (5,840 work hours) less than needed to perform the work. Further, as of July 2017, IMF projected a shortage of 3 developers (4,042 work hours) needed for the 2018 tax filing season. In an internal document identifying options to address the loss of knowledge caused by the attrition of staff for IMF, IRS reported that it has taken various actions as a result of the ongoing attrition of developers. Among others, these actions include: (1) cancelation of planned system enhancements; (2) training and transfer of developers from other projects to perform work on IMF; and (3) reduction in the amount of development work being completed for CADE 2 to address a financial material weakness. According to IMF risk logs, the investment also reported potential impacts on tax processing as a result of the attrition. These impacts include (1) the agency’s delay in implementing modifications to IMF for the filing season to reflect changes in the tax law, (2) tax processing delays due to the lack of adequate institutional knowledge to resolve complex issues, and (3) a lack of necessary data from IMF, which the agency uses as input for other tax processing systems. Further, according to the agency’s CIO, it takes 4 to 5 years to train developers performing work on the IMF investment. The agency, however, is facing challenges with such training and development. For example, IMF program staff stated that the agency has historically recruited and trained future developers from within the agency, where staff had an understanding of IRS business processes and concepts. However, according to the program staff, budgetary reductions limiting travel, moving costs, or stipends, have prevented the agency from continuing such efforts. According to our analysis of an IRS report showing staffing allocations for MSSS, as of April 2017, IRS reported that there were 12 COBOL developers supporting the MSSS investment. Agency officials noted impacts as a result of attrition among its developers, such as the loss of historical knowledge and expertise required to ensure proper maintenance of systems and prevent disruptions during the tax filing season. With regard to IDRS, IRS officials reported the attrition of 30 developers from January 2012 through January 2017. In addition, these officials noted that, as of March 2017, the attrition had resulted in a shortage of 20 developers required to complete work on the investment. In addition, the agency identified 10 “single points of failure” for this investment, meaning that only one staff is available to support a function. Further, the officials noted that attrition of staff may result in (1) a delay in updating systems to reflect tax law changes and (2) IRS’s inability to complete critical IDRS project activities on time. The Three Selected Legacy Investments Did Not Fully Implement Key Practices for Managing Risks We established an evaluation framework based on leading practices from The Software Engineering Institute and OMB guidance. The framework consists of 6 practices and 22 associated activities for managing IT investment risks. Table 4 identifies these practices and activities. IRS has not fully implemented all of the key practices for managing risks for any of the three selected legacy investments that we reviewed. Specifically, based on our analysis, the agency fully implemented one key practice and partially implemented the remaining five for IMF; fully implemented two key practices, partially implemented two, and did not implement the remaining two for IDRS; fully implemented one key practice, partially implemented three, and did not implement the remaining two for MSSS. Table 5 provides our assessment of the extent to which IRS implemented key risk management practices for the selected three legacy investments. IRS fully implemented one key risk management practice for IMF. Specifically, IRS officials responsible for managing IMF risks stated that the agency continuously identified risks through, among other things, monthly meetings. Further, these risks were documented using the Item Tracking Reporting and Control tool, IRS’s risk and issue repository. In addition, we determined that IRS partially implemented the remaining five key risk management practices for IMF. Specifically, the agency prepared for risk management by using IRS’s Application Development organization risk management strategy along with the Item Tracking Reporting and Control tool, which describe how projects are to identify, analyze, prioritize, mitigate, and monitor risks and issues. However, the risk management strategy did not address risk constraints or risk assumptions. In addition, IRS’s risk analysis for IMF included criteria for evaluating and quantifying risk likelihood and severity, but it did not address residual risk, which is the exposure remaining after action has been taken to manage a risk. Further, the agency’s prioritization of risks for IMF included consideration of risk criticality, but did not include the creation of a risk profile, which documents the highest priority risks. With respect to risk mitigation, IRS developed a Stabilization Plan in December 2016 for IMF, and used the Item Tracking Reporting and Control tool. The Stabilization Plan and tool addressed risk mitigation plans, which included specific actions to be taken, as well as an assignment of responsibility and commitment of resources. Further, the agency documented the rationale for accepted IMF risks and established a schedule or period of performance for risk handling activities. However, IRS did not meet the activities for developing alternative courses of action for all critical risks, and establishing threshold values for acceptability of risks, or threshold values for each risk category. Finally, regarding monitoring, IRS reviewed risks at least annually, but did not implement a strategy to escalate and monitor unresolved risks, even though the Application Development risk management strategy outlined a process for doing so. Further, the agency did not compare risks status to acceptability thresholds to determine the need for implementing a risk mitigation plan. Officials responsible for overseeing risk management activities for the investment also did not review its risk management process at least annually to ensure that the process remains appropriate and effective. IRS fully implemented two of the key risk management practices for IDRS. Specifically, the agency continuously identified IDRS risks via bi- weekly meetings and documented risks using the Item Tracking Reporting and Control tool. In addition, the agency prioritized risks based on a documented risk inventory and risk profile. In addition, IRS partially implemented two key practices. Specifically, similar to IMF, the agency prepared the IDRS investment for risk management by using IRS’s Application Development organization risk management strategy; however, it did not address risk constraints or risk assumptions. Further, the agency partially analyzed IDRS risks by including criteria for evaluating and quantifying risk likelihood and severity; however, it did not include residual risks and, thus, may not be aware of additional risk mitigation actions that are needed. Further, IRS did not implement the remaining two practices. With respect to risk mitigation, as of March 2017, the agency did not include risk mitigation plans, as required, for 15 of the 20 risks identified for IDRS. Further, the agency did not maintain dates for risk handling activities for the investment, as the majority of completion dates and projected impact dates for identified risks had passed and these dates were not updated. The agency also did not meet any of the key activities for monitoring identified risks for the investment. For example, it did not compare risk status to acceptability thresholds to determine the need for implementing a risk mitigation plan. Further, the agency did not provide evidence that executive leadership is monitoring all top risks for IDRS. For example, while IRS officials closed 19 of 20 identified risks noting that these risks were tracked by the Applications Development Risk Review Board, the meeting minutes that we received from this board did not show that the risks were being monitored. Lastly, IRS did not review the IDRS risk management process at least annually. IRS fully addressed one key risk management practice for MSSS. Specifically, the agency prioritized risks in the MSSS risk log by assigning “red,” “yellow,” and “green” indicators to identified risks. Further, IRS identified the most significant risks for the MSSS investment in a weekly status report which is intended to address the agency’s readiness to support the tax filing season. IRS partially implemented three practices for MSSS. The agency continuously identified the investment’s risks via several processes. For example, the agency uses its Sustaining Infrastructure Program to address infrastructure components in need of replacement. According to IRS officials and documentation on the Sustaining Infrastructure Program, the program includes (1) an identification of aging infrastructure components; and (2) risk scoring and ranking of the components based on age, the extent to which the asset is associated with critical IRS processes, and the asset’s impact on operations. IRS officials stated that this process results in a prioritized list of assets that are candidates for replacement. In addition, the agency documents MSSS risks via a risk log; however, the risk log does not include all risks for the investment. For example, while officials responsible for managing the MSSS investment told us about human resource risks, these risks were not included in the risk log. The agency also partially mitigated the investment’s risks by developing risk mitigation plans and specific actions for the items identified in its risk log; however, the actions did not include a schedule or period of performance. IRS also did not establish threshold values for MSSS risk categories, or alternative courses of action for critical risks. The agency partially monitored MSSS risks by reviewing risks regularly, providing executive monitoring of top risks, and implementing a strategy to escalate unresolved risks. However, the agency did not compare risk status to acceptability thresholds to determine the need for implementing a risk mitigation plan. In addition, an IRS official responsible for managing risk and issue data for the MSSS investment stated that IRS did not review the risk management process for MSSS annually. Instead, the officials only updated the date and version number on the document that captures this process. In addition, IRS did not implement two key practices. First, the agency did not prepare the MSSS investment for risk management. Specifically, while the agency provided the risk management plan intended to document how it prepares the investment for risk management, we found—and IRS’s Director for Demand Management and Project Governance, and approver of this plan confirmed—that the plan did not describe the risk management activities that the agency was carrying out for the investment. In addition, the agency did not meet key activities for analyzing MSSS risks. For example, the agency used criteria for evaluating and quantifying risk likelihood but it did not document criteria to analyze the severity of all of its risks or consider inherent and residual risks. IRS cited various reasons for the inconsistent implementation of risk management key practices for the three selected investments. For example, IDRS officials responsible for risk management stated that their current guidance did not clearly address some of the key practices. Further, MSSS officials responsible for risk management stated that the majority of the risk management activities for the investment are not documented. In addition, MSSS officials stated that selected risks were not documented due to their perception that a reduced budget and hiring freeze would not allow the agency to mitigate the risks. However, documenting these risks would ensure that IRS appropriately forms a baseline for initiating risk management activities. Until IRS fully implements all of the key practices for managing risks, it will be challenged to successfully identify and mitigate risks before they adversely impact the agency’s ability to carry out its mission. IRS Has Not Implemented Key IT Workforce Planning Practices As we have previously reported, implementing effective IT workforce planning practices can better position agencies to address human capital risks. Accordingly, our prior work has highlighted four key IT workforce planning practices and supporting activities identified in various laws enacted and guidance issued over the past 20 years that call for agencies to perform workforce planning activities. These key practices are (1) setting the strategic direction for workforce planning, (2) analyzing the workforce to identify skill gaps, (3) developing strategies to address skill gaps, and (4) monitoring and reporting on progress in addressing skill gaps. The key IT workforce planning practices and supporting activities are identified below. While IRS has initiated IT workforce planning efforts, the agency has not yet implemented any of the four IT workforce planning practices. Specifically, we found that the Human Capital Office and IT organization have collaboratively developed a tool to automate the IT workforce planning process but the tool is in the initial stages of implementation and IRS has not yet performed any of the activities associated with setting the strategic direction for workforce planning. In addition, the agency has developed an inventory of its current IT workforce but it has not yet developed the competency and staffing requirements nor conducted any of the activities associated with analyzing the workforce to identify skill gaps, developing strategies to address skill gaps, or monitoring and reporting on progress in addressing skill gaps for the agency. While IRS has not implemented key practices for IT workforce planning at the agency level, staff for IMF and CADE 2—two of the four investments selected for our review—provided evidence of efforts they had taken to address their workforce needs. For example, For IMF, in 2016, IRS established a process for continuously matching the current workforce capacity, in terms of skills and staffing, with a projected level of work. In addition, IMF staff identified competencies and staffing requirements for the investment, and assessed the gaps in competencies and staffing by assessing net available staff hours with needed staff hours for particular skill types. Lastly, IMF staff developed strategies and implemented activities in an effort to address IT skill gaps by creating a Stabilization Plan, which includes short and long-term activities for training and realignment of resources. For CADE 2, IRS conducted an assessment in 2015 to identify government and contractor resource needs and utilization. IRS also identified skill gaps and developed strategies and implemented activities such as knowledge transfer sessions to begin addressing these skill gaps. The CADE 2 program manager stated that the program is waiting for additional guidance and direction from the human capital office, as the work in this area was a rudimentary one- time effort. Staff for the remaining two investments we reviewed—IDRS and RRP— stated that they were awaiting further implementation of the agency-wide workforce planning tool to address their IT workforce planning needs. IRS officials attributed the limited progress in implementing IT workforce planning practices to resource constraints and competing priorities. Nevertheless, until the agency implements these practices, it will continue to face challenges in assessing and addressing the gaps in knowledge and skills that are critical to the success of its key investments, some of which we identified earlier in the report. Conclusions IRS has performed operational analyses to examine the ongoing performance of some IT investments, but it has not fully addressed key factors specified in OMB guidance. Until IRS fully addresses all key factors for performing operational analyses, the agency risks not having the information it needs to determine whether the investments fully meet intended objectives, or if there are alternative or more efficient ways to do so. In addition, IRS faces significant risks that could impact key tax processing investments. Specifically, IMF, IDRS, and MSSS are reliant on legacy programming languages and outdated hardware, and the agency is experiencing shortages of staff with the skills to support these investments. However, the agency has not implemented key risk management practices, placing the tax processing and modernization efforts at risk. By fully implementing key risk management practices, IRS will have better assurance that it is proactively addressing risks before they can impact the agency’s ability to carry out its mission. Further, although human capital risks have in part led to significant cost, schedule, and scope variances for CADE 2, a key modernization system, IRS has not implemented key IT workforce planning practices. Specifically, while the agency has initiated efforts to address workforce planning agency- wide, which it plans to continue, the efforts have not yet been implemented for all of the agency’s IT investments. Until IRS implements effective key workforce planning practices, it will not be best positioned to address the human capital risks it faces and ensure the timely and effective delivery of its investments. Recommendations for Executive Action We are making the following 21 recommendations to IRS: The Commissioner of the IRS should ensure the operational analysis for IMF fully addresses greater utilization of technology or consolidation of investments to better meet organizational goals. (Recommendation 1) The Commissioner of the IRS should ensure the operational analysis for IDRS addresses the extent to which the investments support customer processes as designed, and how well the investments are delivering the goods or services they were designed to deliver. (Recommendation 2) The Commissioner of the IRS should ensure the operational analysis for TSS addresses the extent to which the investments support customer processes as designed, and how well the investments are delivering the goods or services they were designed to deliver. (Recommendation 3) The Commissioner of the IRS should ensure the operational analysis for TSS includes a comparison of current performance with a pre-established cost baseline. (Recommendation 4) The Commissioner of the IRS should ensure the operational analysis for EUSS includes a comparison of current performance with a pre- established cost baseline. (Recommendation 5) The Commissioner of the IRS should ensure the operational analysis for MSSS addresses alternative methods of achieving the same mission needs and strategic goals. (Recommendation 6) The Commissioner of the IRS should fully implement the risk management key practice associated with preparing for risk management for the IMF investment. (Recommendation 7) The Commissioner of the IRS should fully implement the risk management key practice associated with analyzing risk for the IMF investment. (Recommendation 8) The Commissioner of the IRS should fully implement the risk management key practice for prioritizing risk for the IMF investment. (Recommendation 9) The Commissioner of the IRS should fully implement the risk management key practice associated with mitigating risk for the IMF investment. (Recommendation 10) The Commissioner of the IRS should fully implement the risk management key practice associated with monitoring, reporting, and controlling risk for the IMF investment. (Recommendation 11) The Commissioner of the IRS should fully implement the risk management key practice associated with preparing for risk management for the IDRS investment. (Recommendation 12) The Commissioner of the IRS should fully implement the risk management key practice associated with analyzing risk for the IDRS investment. (Recommendation 13) The Commissioner of the IRS should fully implement the risk management key practice associated with mitigating risk for the IDRS investment. (Recommendation 14) The Commissioner of the IRS should fully implement the risk management key practice associated with monitoring, reporting, and controlling risk for the IDRS investment. (Recommendation 15) The Commissioner of the IRS should fully implement the risk management key practice associated with preparing for risk management for the MSSS investment. (Recommendation 16) The Commissioner of the IRS should fully implement the risk management key practice associated with identifying risk for the MSSS investment. (Recommendation 17) The Commissioner of the IRS should fully implement the risk management key practice associated with analyzing risk for the MSSS investment. (Recommendation 18) The Commissioner of the IRS should fully implement the risk management key practice associated with mitigating risk for the MSSS investment. (Recommendation 19) The Commissioner of the IRS should fully implement the risk management key practice associated with monitoring, reporting, and controlling risk for the MSSS investment. (Recommendation 20) The Commissioner of the IRS should fully implement IT workforce planning practices, including the following actions (1) setting the strategic direction for workforce planning; (2) analyzing the workforce to identify skill gaps; (3) developing strategies and implementing activities to address skill gaps; and (4) monitoring and reporting on progress in addressing skill gaps. (Recommendation 21) Agency Comments and Our Evaluation We received written comments on a draft of this report from IRS. In its comments, which are reproduced in appendix II, IRS did not state whether it agreed or disagreed with our recommendations. However, the agency acknowledged the importance of strengthening its risk management process by implementing the key leading practices we identified, and described actions underway which confirm the significance of the risks described in our report. The agency also reported actions it has taken since the end of our review to address the IT workforce planning recommendation and stated that it would provide a detailed corrective action plan addressing each of our recommendations. Further, IRS provided technical comments, which we incorporated, as appropriate. We are sending copies of this report to interested congressional committees, the Commissioner of IRS, and other interested parties. This report will also be available at no charge on our website at http://www.gao.gov. If you or your staffs have any questions on matters discussed in this report, please contact me at (202) 512-9286 or pownerd@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made major contributions to this report are listed in appendix III. Appendix I: Objectives, Scope, and Methodology Our objectives were to (1) evaluate the performance of selected Internal Revenue Service (IRS) information technology (IT) investments, (2) summarize any risks associated with selected legacy systems and evaluate the steps IRS has taken to manage them, and (3) determine the extent to which IRS has implemented key IT workforce planning practices. To select investments for our review, we first considered investments identified by IRS as essential to tax processing. We then considered the following factors: (1) investments impacting the greatest number of IRS business areas and associated services based on our review of IRS’s 2016 Technology Roadmap; (2) investments with the highest levels of funding for fiscal year 2016, as reported on the Federal IT Dashboard; (3) investments that IRS’s Chief Information Officer rated as having significant risk with respect to human capital or infrastructure; and (4) investments with planned future system migration efforts as outlined in IRS’s Enterprise Transition Plan. In addition, we considered a mix of investments in development and operations and maintenance. We selected the following nine investments (presented in order of those considered mission critical, followed by those most prominently meeting the above selection factors): (1) Individual Master File (IMF), (2) Integrated Data Retrieval System (IDRS), (3) Telecommunications Systems and Support (TSS), (4) Mainframes and Servers Services and Support (MSSS), (5) End User Systems and Services (EUSS), (6) Enterprise Case Management (ECM), (7) Customer Account Data Engine 2 (CADE 2), (8) Return Review Program (RRP), and (9) Affordable Care Act Administration (ACA). To address our first objective, we analyzed IRS’s reporting on the performance of the nine investments in our selection. The investments included four that were primarily in development—CADE 2, RRP, ECM, and ACA—and five that were primarily in the operations and maintenance phase—IMF, IDRS, TSS, MSSS, and EUSS. For the three investments which were using IRS’s Investment Performance Tool—CADE 2, RRP, and ECM—we compiled and analyzed quarterly reports showing planned versus actual cost, schedule, and scope for work IRS was performing on these investments during fiscal year 2016 through the first 2 quarters of 2017. For the fourth investment—ACA—we compared reported planned and actual costs, as well as planned and actual completion dates for development activities for fiscal year 2016 and the first 2 quarters of 2017. IRS did not report information on ACA’s performance in meeting scope goals. In addition, for the five operational investments, we compiled and analyzed operational performance information reported for the selected investments for fiscal year 2016 and the first 2 quarters of 2017; this information included, where reported, the performance target and actual results for each metric. Further, we determined the extent to which an operational analysis was performed for each of the investments in accordance with best practices. To do so, we obtained operational analyses for fiscal year 2016 and analyzed the analyses against relevant practices outlined in the Office of Management and Budget’s (OMB) fiscal year 2016 capital programming guidance. To assess the reliability of the data for the investments in development used for this objective, we interviewed officials responsible for overseeing the use of the Investment Performance Tool to confirm the completeness of the data generated from the tool, as well as our understanding of what these data represent. We also followed up with these officials to discuss detected anomalies we found in the performance data. In addition, we relied on data reliability assessments we previously completed on IRS’s financial management system because it is a source of the actual cost data found in the Investment Performance Tool. Finally, we followed up on IRS’s actions to address recommendations we previously made to improve the reliability of the cost, schedule, and scope performance data. While we found additional actions are needed to address our recommendations, we determined the investments’ data are sufficiently reliable for our purposes. With respect to the reported operational performance data, we reviewed documentation describing the performance metrics and interviewed IRS officials regarding the process for reporting such metrics. We determined these data were sufficiently reliable for purposes of reporting on operational performance. For our second objective, we selected three investments from our initial selection of nine—IMF, IDRS, and MSSS—because they were placed into operation in the late 1960s and early 1970s, and are therefore considered legacy investments. To summarize the risks associated with these investments, we reviewed, among other things, risk logs captured in IRS’s Item Tracking Reporting and Control tool and risk detail reports. In addition, we obtained resource assessment documentation, where available, and documentation from IRS identifying staff availability and the legacy programming languages supporting these investments. We also identified aged hardware components supporting the selected investments by obtaining reports from the Knowledge, Incident/Problem, Service Asset Management system. This system is used for tracking and managing IRS assets, to include recording and tracking asset acquisitions, transfers, and disposals. We supplemented our review of documentation with interviews of IRS officials responsible for software and infrastructure maintenance. To evaluate the steps IRS has taken to mitigate risks, we analyzed documentation such as risk management plans; risk logs captured in IRS’s Item Tracking Reporting and Control tool and through other means; risk detail reports that included the probability, impact, and overall status for identified risks; risk mitigation plans; and meeting minutes from IRS’s Applications Development Risk Review Board. We also interviewed IRS officials involved in the risk management process, including software developers responsible for maintaining aging programming languages, system administrators, and risk coordinators. We selected officials based on (1) the median number of years the officials had worked to support the investment; (2) their position as an investment Branch Chief or Section Chief in order to fairly represent the management’s perspective on matters discussed; (3) consideration of employees who serve in more than one role in the risk management process in order to obtain diverse perspectives on the process; and (4) their fair representation of all programming language types and infrastructure supporting the investments. We evaluated IRS’s risk management efforts by comparing information obtained to key practices from the Software Engineering Institute’s Capability Maturity Model® Integration for Development, as well as OMB guidance. To assess the reliability of the data used for our second objective, we interviewed officials responsible for overseeing the use of IRS’s Item Tracking Reporting and Control tool and users of this tool to determine if the tool was being consistently implemented. In addition, as part of our interviews with software developers responsible for maintaining aging programming languages, we determined the extent to which risks shared by these officials were consistent with formally documented risks. Lastly, we corroborated IRS’s identification of legacy programming languages, and infrastructure components with our staff possessing expert knowledge of IRS’s IT environment. We determined these data were sufficiently reliable for purposes of describing risks faced by selected investments, as well as for evaluating IRS’s risk management efforts. For our third objective, we obtained documentation describing a tool that IRS is planning to implement agency-wide to address IT workforce planning. Further, we obtained a demonstration of the functionality provided by this tool, and interviewed officials in IRS’s human capital office, as well as investment staff, to determine the extent to which this tool has been implemented across the agency. We also obtained and reviewed information relative to cross-functional acquisition training, efforts intended to strengthen IT program management, and results of IT skills assessments. We compared IRS’s efforts to key practices for IT workforce planning derived from sources, including the Clinger-Cohen Act of 1996, Office of Personnel Management workforce planning guidance, and OMB Circular A-130, and identified in our report on IT workforce planning efforts. In addition, we selected four investments from our initial selection of nine—IMF, IDRS, CADE 2, and RRP—to identify efforts they had taken to address their workforce needs. We selected the four investments based on one or more of the following factors: (1) mission critical designation by IRS, (2) exposure to human capital risk, or (3) status as a key development effort at IRS. Officials responsible for managing two of the investments—IMF and CADE 2—provided information on their efforts. For IMF, we reviewed workforce capacity planning documentation as well as short and long- term workforce plans to assess IMF implementation of workforce planning efforts such as skills gap analysis, development of strategies and implementation of activities to address IT skills gaps, and monitoring and reporting progress in addressing IT skills gaps. For CADE 2, we reviewed documentation from its resource assessment conducted in 2015, which included information relative to resource needs and skills gaps. In addition, we reviewed documentation of efforts to address skills gaps, including training and knowledge transfer programs. We conducted this performance audit from November 2016 to June 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Comments from the Internal Revenue Service Appendix III: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the individual named above, the following staff made key contributions to this report: Sabine Paul (Assistant Director), Bradley Roach (Analyst in Charge), Andrew Banister, Mark Canter, Vern Cumarasegaran, Rebecca Eyler, Paul Middleton, and Martin Skorczynski.

Why GAO Did This Study IRS relies extensively on IT investments to annually collect more than $3 trillion in taxes, distribute more than $400 billion in refunds, and carry out its mission of providing service to America's taxpayers in meeting their tax obligations. For fiscal years 2016 and 2017, the agency reported spending approximately $2.7 billion and $2.6 billion, respectively, for IT investments. GAO was asked to review IRS's IT operations. GAO's specific objectives were to (1) evaluate the performance of selected IRS IT investments, (2) summarize any risks associated with selected legacy systems and evaluate the steps the agency has taken to manage such risks, and (3) determine the extent to which IRS has implemented key IT workforce planning practices. GAO analyzed planned versus actual performance information for nine selected investments for fiscal year 2016 and the first 2 quarters of fiscal year 2017—four in development and five in the operations and maintenance phase; identified risks facing three legacy investments and analyzed IRS's efforts to manage these risks against key practices; and analyzed IRS's IT workforce planning efforts against best practices. What GAO Found The performance of the Internal Revenue Service's (IRS) selected information technology (IT) investments that GAO reviewed varied. Specifically, the four selected investments in the development phase that GAO reviewed spent less than planned, but most were behind schedule and delivered less scope than planned (see table below). In addition, the five selected investments in the operations and maintenance phase that GAO reviewed had performed internal qualitative assessments of performance as required by the Office of Management and Budget (OMB); however, none of the analyses addressed all key factors specified in OMB guidance. Three investments GAO reviewed in the operations and maintenance phase that are legacy investments—Individual Master File (IMF), Integrated Data Retrieval System (IDRS), and Mainframes and Servers Services and Support (MSSS)— are facing significant risks due to their reliance on legacy programming languages, outdated hardware, and a shortage of human resources with critical skills. For example, IRS reported that it used assembly language code and Common Business Oriented Language (both developed in the 1950s) for IMF and IDRS, which exposes these investments to a rise in procurement and operating costs, and a decrease in staff available with the proper skill sets. Further, MSSS relies on a significant amount of outdated hardware exposing the investment to rising warranty and maintenance fees, as well as equipment failures. Despite these risks, the agency has not fully implemented key risk management practices and may be challenged in mitigating risks effectively so that they do not impact the agency's ability to carry out its mission. IRS has not yet fully implemented any of the key IT workforce planning practices GAO has previously identified. Specifically, the agency has developed a tool to automate the IT workforce planning process, but the tool is in the initial stages of implementation. IRS officials attributed the limited progress in implementing IT workforce planning practices to resource constraints and competing priorities. Nevertheless, until the agency fully implements these practices, it will continue to face challenges in assessing and addressing the gaps in knowledge and skills that are critical to the success of its key IT investments. What GAO Recommends GAO recommends that IRS perform operational analyses consistent with guidance, implement key risk management practices, and fully implement key IT workforce planning practices. IRS did not agree or disagree with the recommendations, but said it would provide a plan for addressing each recommendation.

gao_GAO-18-679

gao_GAO-18-679_0

Background As of January 2018, there were 22 TCS programs authorized across 18 states (see fig. 1). All TCS programs are state programs; there are no federal TCS programs. Decisions about whether to develop and operate a TCS program (and how to structure the program) are completely at the discretion of each state; there is no federal role in establishing these programs. Most TCS programs began within the last 10 years; the first TCS program awarded scholarships in Arizona in 1998 and Florida created the newest program in 2018, according to state program documents and officials. Scholarships are funded through donations from private individuals and businesses, and the financial impact to states from TCS programs primarily occurs through forgone revenue resulting from the associated tax credits. In all 22 programs, state agencies and nonprofit organizations both play a role in administering the programs, with the specific responsibilities varying by program: State departments or agencies responsible for tax administration, education, or both, generally administer these programs. For example, they may approve schools or nonprofit scholarship granting organizations or disseminate program information or guidance to potential donors, scholarship students, or the public. Nonprofit scholarship granting organizations (SGO) are generally responsible for managing some aspects of the donation process— such as collecting donations—as well as awarding scholarships to students. Most Programs Use Income to Determine Student Eligibility and Consider Various Factors When Setting Scholarship Award Amounts States’ TCS programs often determine student eligibility for scholarships based on household income and use a range of factors to determine scholarship award amounts. Eligibility Requirements Income requirements: Seventeen of the 22 TCS programs have income limits (i.e., the maximum amount of household income a student can have and still be eligible for a scholarship). As shown in figure 2, income limits varied widely among programs, ranging from just under $32,000 to about $136,500 per year for students from a four-person household in SY 2017-2018. For context, we compared these income limits to the 2012- 2016 5-year ACS estimates of state median household income for four- person households. Six of the 17 programs had household income limits in SY 2017-2018 above their state’s median income. This included two programs each in Arizona and Pennsylvania which collectively accounted for about one-third of all TCS scholarships awarded to students in SY 2016-2017, according to state-reported data. Of the 17 programs that have household income limits, 6 also require SGOs to further consider income when selecting scholarship recipients among eligible students. Such requirements include giving preference to scholarship applicants from lower-income households or ensuring that a certain percentage of scholarship recipients come from lower-income households. Of the remaining five TCS programs that do not use income to determine eligibility, three use one or more other types of eligibility criteria, such as whether the student has a disability, and two—Montana’s TCS program and Arizona’s Original Individual Income Tax Credit program—are open to all school-aged residents. TCS programs collected limited information on the household incomes of scholarship recipients. The 11 programs that had income information on recipient families collected and reported it in different ways. For example, the Alabama program requires SGOs to report the total number and amount of scholarships awarded to students qualifying for the federal free and reduced-price lunch program and makes the information publicly available. Arizona makes an annual report publicly available on the state’s four TCS programs, including breakdowns of the number of students receiving scholarships from various income levels. Other eligibility requirements: Some TCS programs’ eligibility criteria for student scholarship recipients include other factors, such as students’ disability status or previous schooling. Specifically, 7 of 22 programs are limited to students with disabilities or allow students with disabilities to qualify for a scholarship even if they do not meet some requirements for students without disabilities. For example, to be eligible for Virginia’s program, all students must have a household income below a certain amount, but that amount is higher for students with disabilities. South Carolina’s program is limited to students with disabilities. In addition, some programs may require students to have previously attended a public school (9 of 22) or live in the attendance area of a public school with performance challenges (5 of 22). See appendix II for more information on the eligibility criteria of TCS programs. Scholarship Uses and Amounts TCS programs have different requirements for how students can use their scholarships and different methods for calculating scholarship amounts. More than half of the programs (13 of 22) allow students to use their scholarship money for costs like transportation and books in addition to tuition, whereas the remaining programs (9 of 22) require scholarships funds to be used for tuition only. Four programs allow donors to recommend that their donations fund scholarships for specific students. Average scholarship awards in SY 2016-2017 ranged from $500 to $5,468 per student among the 16 programs that published such information or provided it to us. (See appendix II for more information). Most programs require SGOs to consider one or more factors related to student or school characteristics when determining scholarship award amounts. As shown in table 1, these factors may include the cost of private school tuition or the state funding amounts for public school students, among other factors. See appendix II for more information on program requirements related to scholarship amounts. Effect of TCS Donations on Taxes Owed Depends on Factors Such as Tax Credit Percentages and Individual Circumstances The extent to which TCS program donations affect the amount that donors owe in state and federal taxes depends on program characteristics—such as the percentage of the donation that the rules of the program allow donors to claim as a state tax credit (referred to in this report as “tax credit percentages”) and limits on donation amounts—along with donors’ financial circumstances. Almost all of the TCS programs (20 of 22) offer tax credits to businesses for income or other types of taxes, while more than half offer tax credits to individuals for their income taxes (13 of 22). More than half of programs (13 of 22) offer tax credits for cash donations only, while the remaining 9 programs also allow for at least one type of “in kind” donation, such as a property donation. Tax Credit Percentages Eleven of the 22 programs allow eligible donors (either individuals, businesses, or both) to claim 100 percent of their donations as state tax credits, meaning that, for each dollar donated, the amount of state taxes owed (i.e., the donor’s tax liability) is reduced by a dollar, up to any maximum donation limits set by the program. The other 11 programs offer tax credits of 50 percent to 85 percent of donations (see table 2). For example, Indiana and Oklahoma offer tax credits of 50 percent of the donation value, meaning that donors can reduce their state tax liability by 50 cents for every dollar donated. All but one of the programs prohibit donors from receiving a tax credit greater than their tax liability in a given year, although two thirds of the programs allow donors to carry forward portions of the credits to use in future years. Limits on Tax Credit Amounts Sixteen of the 22 programs limit the amount of tax credits each donor may claim per year and programs vary in how they structure these limits. The programs that set annual limits for donors generally do so in one or both of the following ways: Dollar amount limits: Thirteen programs limit the dollar amount of TCS program tax credits that donors can claim in a given year. These limits ranged from a maximum tax credit of $150 for either individuals or businesses in Montana’s program, to a maximum tax credit of $1 million for either individuals or businesses in Illinois’ program in CY 2018. Limits based on percentage of tax liability: Four programs limit the amount of the TCS program tax credits a donor can claim to a percentage of the donor’s total income tax liability. These limits ranged from 50 to 90 percent of a donor’s income tax liability in CY 2018. For example, in South Carolina donors could receive a tax credit up to 60 percent of their total income tax liability for the year of the donation. All but three programs specify a maximum total amount, or cap, of TCS program tax credits that may be claimed each year for the program as a whole (see table 3). Programs’ procedures vary if the cap is reached in a given year. For example, in Rhode Island, potential donors may apply for credits on a “first come, first served” basis once the application period starts until all credits are taken. In 2018, all of the credits were claimed on the first day of the application period and a drawing was held to determine who would receive credits among those who applied on that first day. Georgia’s TCS program offers a maximum tax credit percentage of 100 percent when total donations do not exceed the donation cap. However, if total donations exceed the program cap, the allowable tax credit percentage is prorated among donors who apply on the day the program- wide cap on tax credits is reached. Twenty programs published or provided us with information on donation amounts, such as total donations and average donations. Among these programs, total program-wide donation amounts in CY 2016 ranged from $43,865 to $553 million. (See appendix III for more information about donation amounts.) TCS Donations and Federal Taxes for Individual Donors In addition to reducing their state tax liabilities, some individuals who make TCS program donations may also be able to reduce their federal income tax liabilities through the federal tax deduction for charitable contributions. In August 2018, IRS and Treasury published proposed regulations that, if finalized without modification, would change the extent to which individuals who make TCS program donations can reduce their federal tax liability. However, the proposed regulations were not final at the time this report was published and are therefore subject to change. As a result, the information we present below does not address the proposed regulations. Currently, the extent to which individuals may reduce their federal income tax liabilities as a result of their TCS donation depends on their specific circumstances, such as whether they itemize their deductions (versus taking the standard deduction), the federal rates at which their income is taxed, and the amount of federal deductions they take for state and local taxes. More specifically, the effect of a TCS donation on an individual donor’s federal tax liability depends on the following: Itemizing federal deductions and taking the deduction for charitable contributions: Taxpayers benefit from itemizing deductions—such as those for state and local taxes, mortgage interest, and charitable contributions—if they exceed the standard deduction. Taxpayers, including TCS donors, may only claim a federal deduction for charitable contributions if they itemize. Federal tax rate: The reduction in federal taxes owed as a result of the federal deduction for charitable contributions depends on the donor’s applicable federal tax rate. Given the same deduction amount, taxpayers subject to higher tax rates will generally reduce their tax liabilities by larger amounts than taxpayers subject to lower tax rates. Deduction for state and local taxes: When filing federal taxes, taxpayers who itemize may take a deduction for state and local taxes they have paid during the tax year. Beginning in tax year 2018, individual taxpayers may deduct no more than $10,000 in state and local taxes on their federal tax returns. Taxpayers who claim state tax credits for TCS program donations reduce their state tax liability, which may in turn reduce the amount they may deduct on their federal tax return for state and local taxes paid. Interaction between the federal deduction for charitable contributions and the federal deduction for state and local taxes: Generally, if a donor pays $10,000 or less in state and local taxes, the amount of the deduction for charitable contributions may be fully or partially offset (i.e., canceled out) by a decrease in the deduction for state and local taxes paid as a result of the TCS program tax credit. Conversely, taxpayers who pay more than $10,000 in state and local taxes cannot deduct the full amount of state and local taxes they paid. Therefore, the reduced state and local taxes paid as a result of the tax credit generally may not offset the amount of the deduction for charitable contributions for these taxpayers. See figure 3 for a description of how individuals’ TCS program donations could affect their federal tax liabilities. Combined Reduction in State and Federal Tax Liability TCS program donations can lead to a range of possible changes to an individual’s state and federal income tax liabilities, including some scenarios where donors could reduce their combined state and federal tax liability by an amount that is greater than the amount of their donation (see for example, Donor A in figure 4). Figure 4 shows four examples of how state and federal income taxes may be reduced for hypothetical individual donors in states with 100 percent and 50 percent tax credit scholarship programs. Agency Comments and Our Evaluation We provided a draft of this report to Education and IRS for review and comment. While the draft was under review at these agencies, IRS and Treasury issued proposed regulations related to state tax credits and the federal deduction for charitable contributions. We updated the report to include information about these proposed regulations but did not alter our analysis to reflect the proposed regulations because they were not final at the time this report was published and are therefore subject to change. We provided a revised draft to IRS as the revisions directly relate to IRS’s areas of responsibility, and informed Education about our approach to addressing the proposed regulations. IRS did not provide formal comments on the draft report. Education’s comments are reproduced in appendix I. Education also provided technical comments, which we incorporated as appropriate. In its comments, Education noted that it has no role in developing, operating, or overseeing TCS programs, and provided a variety of comments and observations on the draft report. For example, Education suggested that we add additional details about certain TCS program requirements, such as more information about state tax rules and permissible uses of scholarship funds. We incorporated these comments as appropriate. Education also suggested that we delay publication of this report until the IRS regulations are finalized, as Education thought that the report could be more helpful at that time. GAO policy is to communicate audit and evaluation results in a timely manner to decision makers and others who either requested the work or may need the information to bring about needed changes. Therefore, we are issuing the report as planned. We are sending copies of this report to the appropriate congressional committees, the Secretary of Education, and the Commissioner of Internal Revenue. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (617) 788-0580 or nowickij@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix IV. Appendix I: Comments from the Department of Education Appendix II: State Tax Credit Scholarship Program Eligibility and Scholarship Award Characteristics Appendix III: Information About Tax Credit Scholarship Program Tax Provisions Program name Programs available for both individual and business donors Alabama Educational Scholarship Program $50,000 or 50% of tax liability, whichever is lower (individual) 50% of business tax liability $1,000 (individual) 75% of business tax liability $1 million (individual and business) $350,000 (Individual and business) $150 (Individual and business) $1,000 (individual) $100,000 (business) 60% of tax liability (Individual and business) Virginia Education Improvement Scholarships Tax Credits Program Programs available for individual donors only Arizona Original Individual Income Tax Credit Program $81,250 (individual) No limit for business donors $555 (individual) $552 (individual) Percent of donation that may be claimed as a credit Maximum donation amounts vary from 50% to 100% of tax liability $510,000 or no more than 10% of program credits 2 percent of wages paid no In Oklahoma, Pennsylvania, and Rhode Island, the percentage of donations that can be claimed as a tax credit increases if donors commit to donating for 2 years. In Oklahoma, that percentage increases from 50 percent to 75 percent. In the two Pennsylvania programs and the Rhode Island program, the percentage increases from 75 percent to 90 percent. Program name Programs available for both individual and business donors Alabama Educational Scholarship Program Total donations in CY 2016 (rounded) Appendix IV: GAO Contact and Staff Acknowledgments GAO Contact: Staff Acknowledgments: In addition to the individual named above, Nagla’a El-Hodiri (Assistant Director), Barbara Steel-Lowney (Analyst-in-Charge), Jeff Arkin, and Jessica L. Yutzy made key contributions to this report. Also contributing to this report were Deborah Bland, Lilia Chaidez, Sarah Cornetto, Caitlin Cusati, Paulissa Earl, Alison Grantham, Kirsten Lauber, Sheila R. McCoy, Mimi Nguyen, Jessica Orr, Michelle Philpott, Paul Schearf, and Andrew J. Stephens.

Why GAO Did This Study TCS programs offer state tax credits to individuals or businesses that donate to scholarship funds for students to attend private elementary and secondary schools. Through these credits, donors may reduce the amount they owe in state taxes by the full or a partial amount of their donation, depending on each program's rules. Designing a TCS program requires that many decisions be made, such as which students will be eligible to receive scholarships and the effect donations will have on donors' state taxes. GAO was asked to review key characteristics of TCS programs. This report examines (1) state TCS programs' policies regarding student eligibility and scholarship awards, and (2) how donating to a TCS program could affect the amount of state and federal taxes owed by donors. For both objectives, GAO reviewed publicly-available documents about student eligibility and tax provisions for all 22 programs authorized as of January 2018 and verified the accuracy of the information with state program officials. GAO did not conduct an independent legal review of state laws and regulations. GAO also interviewed federal officials and reviewed relevant federal guidance and policy documents. What GAO Found In 2018, there were 22 tax credit scholarship (TCS) programs authorized across 18 states, which provide state tax credits for individual and business donations that fund scholarships for students to attend elementary and secondary private schools (see figure). To determine the eligibility of students for these scholarships, most TCS programs use household income and have various approaches to determine scholarship award amounts. Income limits vary widely among programs, ranging from approximately $32,000 to $136,500 per year for students from a four-person household in school year 2017-2018. Programs have different requirements for how students can use their scholarships and different methods for calculating scholarship amounts. More than half of the programs (13 of 22) allow students to use their scholarship money for costs like transportation and books in addition to tuition, whereas the remaining programs (9 of 22) require scholarships funds to be used for tuition only. Average scholarship awards in school year 2016-2017 ranged from $500 to $5,468 per student among the 16 programs that published or provided GAO with such information. The effect of TCS donations on donors' tax liability depends on program characteristics and donors' financial circumstances. Specifically, half of the 22 programs allow eligible donors to claim 100 percent of their donations as state tax credits, meaning that for each dollar donated, state taxes owed are reduced by a dollar, up to any maximum set by the state. The remaining 11 programs allow donors to claim from 50 to 85 percent of their donations as state tax credits. Programs often specify a maximum tax credit that may be claimed each year by a donor, by all donors combined, or both. Individual donors may also reduce their federal tax liabilities through the federal deduction for charitable contributions, depending on their financial circumstances and applicable tax provisions.

gao_GAO-18-232

gao_GAO-18-232_0

Background Federal Housing Programs That Assist Older Adults Older adults may receive federal housing assistance through a number of programs, but only a few programs specifically target older households. To be eligible for those programs, the head of household or spouse must be 62 or older. Programs Targeting Older Households HUD has two programs specifically targeted to older households—the Supportive Housing for the Elderly (Section 202) program for renters and the Home Equity Conversion Mortgage program for homeowners. HUD’s Section 202 program is the only program that provides supportive housing targeted to very low-income older adults. Supportive housing is defined as non-institutionalized housing that connects residents with the services they need to live independently, such as in-home care, meal delivery, and transportation. Until fiscal year 2012, the Section 202 program funded the development of new units. The Home Equity Conversion Mortgage program, administered by the Federal Housing Administration (FHA), a component of HUD, allows older homeowners to access FHA-insured mortgages to convert some of the equity in their homes into monthly streams of income or lines of credit (reverse mortgage). In addition to being at least 62 years old, borrowers must occupy the property as a principal residence, and any existing lien on the property must be small enough to be paid off at settlement. In a reverse mortgage, the loan balance increases and home equity decreases over time. As the borrower receives payments from the lender, the lender adds the principal and interest to the loan balance, reducing the homeowner’s equity. The homeowner remains responsible for paying insurance and property taxes. The Rural Housing Service has one homeownership program that specifically serves households aged 62 and older. The Section 504 Rural Home Repair and Rehabilitation Grant program finances the removal of health and safety hazards or remodels dwellings to make them accessible for disabled household members. According to USDA officials, grants often are made in conjunction with Section 504 loans that have a 1 percent interest rate. Other Programs That Can Serve Older Households HUD has rental assistance programs that do not target, but serve a significant number of older households. According to HUD officials, HUD’s primary rental assistance programs, including the Housing Choice Voucher, Public Housing, and Project-Based Section 8 programs, serve nearly 1.5 million such households. In addition, FHA’s Section 221(d)(3) and Section 221(d)(4) Multifamily Rental Housing for Moderate-Income Families provide mortgage insurance to finance multifamily properties, some of which may be designated for the elderly. The HOME Investment Partnerships Program provides formula grants to states and localities that communities use to fund a range of activities including building, purchasing, and rehabilitating affordable housing for rent or homeownership or providing direct rental assistance to low-income households. HUD Public Housing Designated for Older Households Local public housing agencies can apply to HUD for approval to designate public housing developments (or portions of developments, such as buildings or floors) for occupancy only by elderly families, by disabled families, or both. HUD refers to this as “designated public housing”. 42 U.S.C. § 1437e. However, HUD officials stated that HUD had fewer than 40,000 units of designated public housing and the majority of elderly persons were not assisted through such housing. Similarly, the Rural Housing Service administers a number of homeownership and rental programs not targeted to older rural households, but that can serve them. They include the Section 502 Rural Direct Home Loan; Section 502 Home Loan Guarantee; Section 504 Rural Housing Repair and Rehabilitation Loan; and the Section 521 Rural Rental Assistance programs. The Section 521 program provides rental subsidies to low-income, elderly, or disabled households living in properties funded by the Section 515 Rural Rental Housing program and others. Finally, the Low-Income Housing Tax Credit, established under the Tax Reform Act of 1986 and administered by the Internal Revenue Service, is the largest source of federal assistance for developing affordable rental housing for low-income households, including elderly households, and as of 2017, had financed about 2.9 million rental units. It provides tax credits to encourage private-equity investment in affordable housing development. Supply of Federally Assisted Housing Worst-Case Needs for Rental Housing HUD tracks older adults in its biennial Worst Case Housing Needs report. HUD defines households with worst-case needs as very low-income renters who do not receive government housing assistance and paid more than half of their income for rent, lived in severely inadequate conditions, or both. The latest report (2017) indicated that severe housing problems were on the rise among unassisted renter households in 2015. This included older renters, for whom the number and proportion of households with worst-case needs increased from 2013, by 382,000 and 2.6 percentage points, respectively. In 2015, 1.85 million such households experienced worst-case housing needs, including unaffordable rents. The report noted that low- income older households that rely on fixed incomes rather than wages may be less likely to benefit from economic recovery trends that raised incomes for others in recent years. In recent years, HUD and USDA have lost subsidized housing stock. Losses can happen under several scenarios, including when federal rental assistance contracts expire, federally subsidized mortgages reach maturity or are paid off and owners convert the units to market-rate rentals, or units fall into disrepair. In its fiscal year 2014–2018 strategic plan, HUD reported that its public housing stock faced a capital needs backlog, estimated at $26 billion, which would be difficult to meet given federal fiscal constraints. The decreasing supply of adequate affordable housing may affect older low-income renters (see sidebar), who were well represented in HUD- and USDA-assisted housing. In addition to losing housing stock, HUD and USDA have programs that previously added to the supply of affordable housing but no longer do so. One example is HUD’s Section 202 program, which specifically serves very low-income older adults. Except for $10 million in its fiscal year 2017 appropriation that could contribute to capital advances, since fiscal year 2012, the program has primarily funded contract renewals for existing rental assistance and for service coordinators. According to USDA officials, the Section 515 program (direct loans for multifamily housing) had no funds for new construction. They told us that Section 515 funds almost exclusively were being used for unit rehabilitation, and that only the Section 538 loan guarantee program was funding new construction. According to USDA officials, many of the rehabilitated Section 515 properties are described in loan applications as properties serving the elderly, as are some newly constructed properties funded by the Section 538 program. Reports on Integration of Housing and Health Services Federal and other entities have reported on a lack of integration among housing and health programs and services and the benefits of closing gaps. In 2002, the congressionally mandated Commission on Affordable Housing and Health Facility Needs for Seniors in the 21st Century reported that a lack of integration between housing and health care for older adults resulted in inefficiencies, noting that the basis was partly historical differences in policies, funding systems, and regulatory structures. The Commission found that, with few exceptions, older adults obtained their housing from one source and health care and supportive services from a different source. In July 2015, the White House Conference on Aging highlighted the importance of collaboration across sectors and the need to better integrate housing, transportation, health care, and long-term services and supports to encourage healthy aging. It noted that opportunities existed to leverage approaches taken by states and localities to consider how best to serve older adults. In 2016, the Bipartisan Policy Center’s Health and Housing Task Force reported that bridging the current policy gap between housing and health had the potential to provide a number of benefits, including improving health outcomes for older adults, reducing costs incurred by the health care system, and enabling many older adults to age in their own homes and communities. Senior leadership in HUD’s Office of Policy Development and Research recently amplified the discussion about housing and health, noting that the Housing Act of 1937 recognized the linkage between the two. That office also has observed that efforts to better link housing and health services hold promise to improve the ability of older adults to safely, comfortably, and more affordably age in place. Collaboration on Housing Assistance and Health Services for Older Adults Did Not Include USDA or Define Common Outcomes Collaborations that were focused on coordinating housing and health services for older adults involved HUD and HHS. While these collaborations demonstrated some leading practices we identified, they did not include USDA (a relevant participant) and did not define common outcomes for these efforts. USDA, HUD, and HHS Collaborated on Some Housing and Health Service Efforts That Did Not Focus on Older Adults USDA has collaborated with HUD on two efforts related to housing, and with HUD and HHS on one effort that addressed both housing and health services, although these efforts were not focused specifically on older adults. Officials at HUD and USDA told us that they primarily collaborated on administrative initiatives through the Rental Policy Working Group. For example, in 2016, HUD, USDA, and the Department of the Treasury entered into a memorandum of understanding to formalize the activities of the Rental Policy Working Group, such as reducing duplicative physical unit inspections at properties assisted by one or more of the agencies. While older renter households may benefit from administrative improvements made through the Rental Policy Working Group, its efforts were not intended to focus on any particular household type served by HUD and USDA. Under a 1991 memorandum of understanding, HUD and USDA also are required to coordinate where both agencies’ rental programs could overlap in serving rural families. In 2016, USDA started participating in interagency training intended to help coordinate the provision of home and community-based services and supports to vulnerable populations, including older adults. The training was initiated by HHS, and since 2011, HHS and HUD have conducted it for program staff as part of their efforts to strengthen cross-agency collaboration. In 2017, the 1-day training session focused on housing as a platform for improving the quality of life of persons with disabilities, older adults, and other at-risk populations. USDA participated in a discussion on housing and health collaboration between federal, state, local, and community partners. Representatives of both HUD and HHS said that USDA’s increased participation had been positive. HUD officials also told us that in 2018, the agenda for the 1-day training would focus on housing and health supports for older and disabled persons living in rural areas. Additionally, in 2016, HUD and USDA began coordinating on developing topics for HUD’s research agenda. Specifically, HUD’s Office of Policy Development and Research develops a research agenda with input from external stakeholders, and both HUD and USDA officials told us of efforts to develop rural housing topics for inclusion in HUD’s research. HUD Policy Development and Research officials stated that HUD conducted extensive research in rural areas and was interested in coordinating more with USDA, though this research has not been specific to older adult issues. In response to HUD’s outreach, USDA submitted topics to HUD, some of which were incorporated into HUD’s 2017 research agenda update. They include the relationship between housing, food, and health and the impact of home equity loss on rural homeowners, particularly those who are aging. HUD and HHS Have Had Several Collaborations on Services and Data Sharing for Older Adults HUD and HHS have undertaken multiple collaborative efforts that link data on housing and health services and often have focused on older adults. The Support and Services at Home demonstration was launched in 2011 to connect older residents of affordable housing properties with home and community-based supportive services and promote health care coordination. The latest evaluation report was published in January 2016 and described the program’s ongoing implementation and impact from 2011 through 2014. It found lower rates of growth in Medicare expenditures among program participants than among a comparison group. As of August 2017, HUD representatives told us that the departments were still collaborating on evaluating the demonstration. Data sharing between HUD and HHS on another demonstration program—the Supportive Services Demonstration for Elderly Households in HUD-Assisted Multifamily Housing—began in 2014. Like the Support and Services at Home demonstration, the Supportive Services Demonstration is intended to test aging-in-place models that show potential for delaying or avoiding the need for nursing home care. HUD has been conducting a 4-year, two-part evaluation, which includes an evaluation of the implementation process and an impact evaluation that will match Medicare claims data from HHS’s Centers for Medicare and Medicaid Services and HUD administrative data. HUD and HHS completed a pilot of the data match in 2014, and HUD has submitted two semi-annual reports to Congress on program implementation, the latest in May 2017. HUD officials stated that in October 2017, HUD secured a contract to conduct a full evaluation of the Supportive Services Demonstration, as described above. According to the officials, the demonstration implementation team conducted an in-person, 2-day training event for care coordination teams in November 2017. The National Center for Health Statistics at the Centers for Disease Control and Prevention, also a component of HHS, in 2014 began to link national health survey participant data to HUD administrative data covering HUD’s largest housing assistance programs through 2014. The data linkage is intended to help those agencies and other federal entities and researchers complete independent projects for statistical and research purposes. The first linkage included two household surveys (National Health Interview Survey and National Health and Nutrition Examination Survey) that covered 1999–2012 and it was completed in July 2016. HUD officials told us that HUD and HHS were working on the second data linkage project. HHS confirmed that the second linkage was underway as of February 2018, with an expected completion date of summer 2018. It will add survey years 2013–2016 and administrative data years through 2016. While the project was not focused specifically on older adults, it could include a significant proportion of older adults living in non-institutionalized settings who receive housing assistance from HUD. HHS’s Office of the Assistant Secretary for Planning and Evaluation and HUD completed a study on health care utilization among HUD- assisted older adults in 12 jurisdictions, and published a final report in August 2016. The study, which began in 2009, explored the feasibility of matching administrative data from HUD and HHS’s Centers for Medicare and Medicaid Services to determine if doing so could help track housing and health outcomes, and reliably support future research and policy analysis. More specifically, it linked HUD individual tenant-level data to Medicare and Medicaid beneficiary enrollment, payment, and claims data. In the 2016 report, HHS and HUD found that, controlling for previously identified factors, HUD- assisted dual Medicare-Medicaid beneficiaries were less likely to use some Medicare-covered services such as acute hospital stays, but more likely to use Medicaid-covered home and community-based supportive services. According to the report, the study demonstrated that linking the agencies’ data could inform decisions about future program investment. Select Collaborations Did Not Include USDA or Define Common Outcomes We found that the collaborative efforts between HUD and HHS (pilot programs and data sharing related to older adults) demonstrated some but not all relevant leading practices we previously identified for effective interagency collaboration. As we reported in 2012, interagency collaborative mechanisms can be enhanced by leading practices, including written guidance and documenting agreements on how the participating agencies will be collaborating, clarifying roles and responsibilities, leveraging resources (such as funding and staffing), including all relevant participants, and clearly defining outcomes. Written Guidance and Agreements HUD and HHS established interagency agreements or memorandums of understanding for collaborative efforts that focused on or included older adults. For example, HUD and HHS entered into a memorandum of understanding for the National Center for Health Statistics data linkage effort and had contracts for the Support and Services at Home demonstration and other collaborations. Our leading collaboration practices state that agencies articulating their agreements in formal documents can strengthen their commitment to working collaboratively, as long as they are continually updated and monitored. Roles and Responsibilities The roles and responsibilities of agencies participating in a collaborative effort may be defined in a number of ways, including through laws, policies, memorandums of understanding, or other requirements. Clarity of roles and responsibilities allows participating agencies to understand and agree on accountability for the joint effort, and a process for making and enforcing decisions. In accordance with their memorandum of understanding, HUD and HHS had clear roles and responsibilities for the National Center for Health Statistics data linkage effort. For example, the agreement documented specific data that HUD was to provide to the National Center for Health Statistics and how the latter would attempt to link HUD’s data to its survey participant data. It also specified both agencies’ roles in data access, storage, and disposition. Leveraging Resources Collaborating agencies should identify human, information technology, physical, and financial resources needed to initiate or sustain their collaborative effort. And it is important that the agencies leverage sufficient resources to accomplish their objectives. HUD’s and HHS’s written agreements generally included a description of how they would leverage each agency’s resources, including staffing, funding, and data. For the Support and Services at Home demonstration, HUD provided resources to help evaluate the program model, while HHS funded a wellness nurse who worked with a service coordinator to perform such tasks as assessing residents’ needs; identifying and coordinating service delivery; monitoring receipt and follow-through of services; and building and sustaining partnerships with providers. Relevant Participants HUD and HHS, including their component agencies and offices, were the main participants in these collaborative efforts. HUD officials stated that USDA was not included because these collaborative efforts were pilots originally intended to better understand the health of households assisted by HUD. But no current plan exists to include USDA in such collaborative efforts in the future. As these collaborative efforts mature, USDA, which provides assistance to low-income older adults in rural areas, may benefit from inclusion. Effective collaborative efforts benefit from having participants with the necessary knowledge, skills, and abilities to contribute to the outcomes of the collaboration. Officials at HUD and HHS said they were open to greater collaboration with USDA in the future. And while HUD has a presence in rural communities, USDA’s participation would allow the Rural Housing Service to provide input on and help address challenges that may be unique to rural older adults. By not including USDA in future collaborative efforts on older adult housing and health services, HUD and HHS may miss opportunities to identify and respond to the changing needs of some older adults living in federally assisted housing—such as by drawing on the experience and resources of the Rural Housing Service in serving rural populations. In addition, USDA may miss opportunities to benefit from lessons learned or programmatic improvements that HUD and HHS might undertake as a result of their collaborations. Outcomes Although HUD and HHS have established specific objectives for their various evaluations and data-sharing initiatives, they have not defined common outcomes for these collaborative efforts. For example, while some of these efforts were expected to explore the housing and health relationship and inform more evidence-based program decisions, the agencies have not defined common outcomes for their interagency efforts as a whole. We reported that collaborating entities should determine whether they have clearly defined short-term and long-term outcomes, can track their progress, and whether they each have collaboration- related performance standards against which to evaluate individual performance. Such common outcomes could include both quantitative and qualitative information. For example, a set of measures against which to track and monitor their collaborations (including demonstrations, data matching, and studies) might include the extent to which hospital stays were reduced, as noted for the Support and Services at Home and Supportive Services demonstrations, or program costs saved. Monitoring and reporting such measures would provide greater transparency to agency and congressional decision makers about how these collaborative efforts have resulted in potential cost savings and other benefits across agency lines. A more long-term common outcome could include developing proposals for programmatic improvements that would leverage the lessons learned from the collaborative efforts. Senior leadership at HUD has said that HUD considers the nexus of housing and health to be a priority for future work, and that federal agencies needed to continue finding ways to move beyond their programmatic scope to engage in more comprehensive, cross-cutting efforts. The official pointed to data-matching efforts with HHS as low-cost initiatives that could enhance HUD’s knowledge about the health status of assisted households and potentially inform cost-saving policies. Because many of their collaborative efforts began in the last 5 years and some remain ongoing, HUD and HHS may not have prioritized developing common outcomes that relate both to older adult housing and health services. In contrast, HUD and HHS have developed broader goals that relate to how they serve older adults.However, without defined common outcomes to help guide ongoing and future efforts, HUD and HHS (and potentially USDA) lack measures against which to monitor, evaluate, and report the results of their collaborative efforts. Conclusions Federal agencies, particularly HUD and HHS, have found opportunities to collaborate in meaningful ways on services provided to older adults. But collaborative federal efforts to address the housing needs of older adults and tie into health services would benefit from consistent USDA involvement and from defining common outcomes. Greater USDA participation would result in a better nationwide assessment of the housing and health needs of older Americans who live in all federally assisted housing and leverage USDA’s expertise and resources in serving rural populations. Outcome information would help the agencies articulate to stakeholders and Congress the results the collaborations achieved; activities, strategies, or areas on which to focus in the future; and how scarce federal resources were leveraged and managed. Recommendations for Executive Action We are making a total of three recommendations (one recommendation each to HUD, HHS, and USDA): The Secretary of Housing and Urban Development should work with HHS and USDA’s Rural Development to define common outcomes and identify opportunities to include USDA in future collaborative efforts on older adult housing and health services. (Recommendation 1) The Secretary of Health and Human Services should work with HUD and USDA’s Rural Development to define common outcomes and identify opportunities to include USDA in future collaborative efforts on older adult housing and health services. (Recommendation 2) The Assistant to the Secretary for Rural Development should work with HUD and HHS to define common outcomes and identify opportunities to include USDA in future collaborative efforts on older adult housing and health services. (Recommendation 3) Agency Comments and Our Evaluation We provided a draft of this report for review and comment to HHS, HUD, and USDA. HHS and HUD provided written comments that are reproduced in appendixes I and II, respectively. USDA provided technical comments, which we incorporated as appropriate. All three agencies concurred with our recommendations. The Deputy Administrator of Multifamily Housing Programs provided USDA’s concurrence in an e-mail dated April 11, 2018. In its comment letter, HUD stated that our recommendation to include an additional agency in housing and health demonstrations is consistent with direction provided by the Senate Committee on Appropriations in fiscal year 2016. The committee directed HUD to partner with other federal agencies to pursue a service coordination demonstration in non- metropolitan areas. HUD noted that the committee advised it not to delay existing demonstration efforts while a non-metropolitan component was being designed. HUD also stated that it convened an expert panel to better understand challenges to service coordination for low-income older adults in rural areas, and summarized the findings in a report. HUD said further action on the panel report was contingent on funding availability and direction from appropriations committees. HUD stated that, in regard to our recommendation that additional federal agencies might benefit from data-linkage projects similar to its project with the National Center for Health Statistics, it would be pleased to offer assistance, guidance, and insights to other agencies. Our recommendation is that HUD work with HHS and USDA’s Rural Development to define common outcomes and identify opportunities to include USDA in future collaborative efforts on older adult housing and health services. This would include data sharing, service demonstrations, research, and other collaborations. Ongoing housing and health collaborations among HUD, HHS, and USDA would benefit from greater USDA involvement, which also would serve to enhance assessments of the housing and health needs of older Americans in federally assisted housing. We are sending copies of this report to the appropriate congressional committees, the Secretary of Agriculture, the Secretary of Health and Human Services, the Secretary of Housing and Urban Development, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-8678 or garciadiazd@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix III. Appendix I: Comments from the Department of Health and Human Services Appendix II: Comments from the Department of Housing and Urban Development Appendix III: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Paul Schmidt (Assistant Director), Bernice Benta-Jackson (Analyst in Charge), Abigail Brown, Stephen Brown, William Chatlos, Charles Culverwell, Kirsten Lauber, John McGrail, Marc Molino, Dae Park, Nadine Garrick Raidbard, Barbara Roesmann, Joseph Silvestri, and Jeff Tessin made key contributions to this report.

Why GAO Did This Study According to the Census Bureau, by 2030, about 1 in 5 Americans will be 65 and older. This aging of the population presents challenges and opportunities for policymakers and service providers in helping ensure that the older population's needs—including housing and health services—are met. Federal agencies with programs that provide housing assistance to low-income older households include HUD and USDA. Several HHS programs provide those households with health services. This report assesses the extent to which the three agencies collaborated to address the housing and health service needs of older adults living in federally assisted housing. GAO compared agency efforts to leading collaboration practices it has identified (including written agreements; roles and responsibilities; leveraged resources; relevant participants; and defined outcomes) and interviewed HUD, HHS, and USDA officials. What GAO Found The Departments of Housing and Urban Development (HUD) and Health and Human Services (HHS) have collaborated on older adult housing and health issues, but these efforts did not fully demonstrate leading practices GAO identified for effective collaboration. The HUD-HHS efforts demonstrated some leading practices. For example, the agencies have written agreements for data-sharing projects and have leveraged resources to conduct research on older adults. The Department of Agriculture (USDA) was not included in the efforts although it provides housing assistance to older rural households. GAO identified the inclusion of all relevant participants as a leading practice. According to HUD, the efforts were intended to explore the health of HUD-assisted households. However, by not including USDA in future collaborations, HUD and HHS may miss opportunities to leverage expertise and USDA may not be able to benefit from any resulting insights and improvements. The HUD-HHS collaborative efforts also did not define common outcomes, another leading practice GAO identified, likely because their collaboration is relatively new. Without common outcomes (for instance, focused on recipient impact or cost savings), the agencies lack measures against which to monitor, evaluate, and report the results of any collaborations. Future collaborations would benefit from consistent USDA involvement. And by defining common outcomes, the agencies would help inform Congress and stakeholders of results achieved and strategies or areas on which to focus. What GAO Recommends GAO is making three recommendations (one each to HUD, HHS, and USDA). They focus on including USDA in collaborations on older adult housing and health services and defining outcomes for the efforts. The three agencies concurred with GAO's recommendations. HUD stated that it had begun examining challenges relating to services for low-income rural older adults.

gao_GAO-19-73

gao_GAO-19-73_0

Background Roles and Responsibilities for DOD Real Property The Under Secretary of Defense for Acquisition and Sustainment has overall responsibility and oversight for DOD’s real property and provides overarching guidance and procedures for real property management. The Assistant Secretary of Defense for Energy, Installations, and Environment, assists with developing policy and guidance for real property inventory and serves as the focal point for all matters related to the inventory of real property assets. The military services and WHS are responsible for implementing policies, programs, and procedures in accordance with OSD’s guidance to maintain an accurate and complete real property inventory. They are also responsible for ensuring that real property requirements are being met when other DOD components, such as defense agencies and DOD field activities, utilize real property under their jurisdiction. The defense agencies and DOD field activities are responsible for confirming that all real property assets that they occupy, operate, or maintain are contained within the real property inventory and for reconciling any real property data, when needed, with their supporting military service or WHS. Real Property Inventory Data Real property inventory data are used at the installation, military service, and OSD levels for the recording, planning, managing, and reporting of DOD real property assets, as shown in figure 1. Installation level. Real property officials are to record transactions to document new acquisitions, changes to existing facilities, and disposals and to collect information—including physical characteristics, space usage, and facility condition—on the real property at each installation. Officials are to enter this information into corresponding military service or WHS real property data systems. Installation officials stated they use real property information for a variety of purposes such as prioritizing facilities for sustainment and restoration projects, preparing installation master plans, and conducting fire and safety planning. Service level. Military service headquarters and WHS use inventory information to oversee and manage their real property needs across their installations. For example, according to officials, these data inform how they use property to support their missions and to budget for required sustainment, restoration, or construction of real property. In addition, this information is used to account for real property asset holdings that are included in financial statements prepared to meet federal financial reporting requirements. OSD level. OSD requires that the military services and WHS submit their real property inventories to be compiled into a department-wide data set—RPAD. The OSD focal point is responsible for providing information from the RPAD to assist various OSD offices with responsibilities for budget and mission planning. For example, the information is used in budgeting for sustainment of facilities. Additionally, OSD offices use the information in mission planning for certain DOD components—defense agencies, DOD field activities, and U.S. Special Operations Command—and for certain types of facilities, including sustainable buildings, historic property, and ranges. Moreover, OSD uses this information to meet reporting requirements outside of DOD. These include reports to Congress on the utilization of DOD’s facilities. All executive branch federal agencies are required to annually submit real property data to the General Services Administration to compile into the Federal Real Property Profile. DOD also reports information to the Office of Management and Budget on disposals and square footage of certain types of purposes to meet report requirements for the National Strategy for the Efficient Use of Real Property. OSD provides annual guidance that gives specific requirements for content and format for the military services’ RPAD submissions, including data elements and any associated business rules. For fiscal year 2016, OSD required 216 data elements to be maintained in RPAD and provided a data dictionary, called the Real Property Information Model, which defines these elements. OSD also has a process to verify and validate the data the military services and WHS submitted annually to the RPAD that includes OSD using a verification and validation tool to determine whether each data element has an entry that is in the correct format and complies with established business rules. When data anomalies are discovered with the data, OSD provides the data back to the submitting organization for review and correction as necessary. The military services and WHS certify annually that the real property information submitted to OSD accurately reflects each of their inventories. Some key real property data elements are significant for planning and reporting on real property assets: Operational status. A code used to identify the current operational status of the real property asset, such as whether the site location of the asset is active, the existence of the asset, and the usage of the asset. Asset review. A date used to document any type of review of an asset. DOD requires that each facility be physically inventoried on a cycle—every 5 years for non-historic facilities and every 3 years for historic facilities. Plant replacement value. A calculation of the cost to replace the current physical plant (facilities and supporting infrastructure) using today’s construction costs (labor and materials) and standards (methodologies and codes). Utilization rate. A percentage (on a scale 0 to 150) used to represent the extent to which a real property asset is used by the primary user for the current program based on its design purpose. DOD has not established cutoff points to determine unutilized, underutilized, and utilized real property. However, according to OSD officials, DOD considers a utilization rate of 101 to 150 as over utilized, meaning an asset’s available space is not sufficient to meet the primary user’s space requirement. Facility condition. A measure of a facility’s physical condition that is expressed as a percentage (on a scale of 0 to 100). Factors used to calculate the facility condition include the facility’s estimated deferred maintenance and repair costs and the facility’s plant replacement value. DOD guidance states a condition of 0 to 59 is failing; 60 to 79 is poor; 80 to 89 is fair; and 90 to 100 is good. Figure 2 displays these real property data elements. Real Property and Financial Management DOD has undertaken several financial management improvement initiatives over the years to address deficiencies in business systems, processes, and controls through its Financial Improvement and Audit Readiness (FIAR) Plan. The FIAR Plan guidance includes 40 of the data elements required to be reported to OSD and maintained in RPAD within the scope of the effort. As part of the department’s FIAR effort, each of the military services developed individual plans to prepare their management processes, such as their accountability systems and procedures for real property, which would be tested during financial audit. The military services’ real property efforts to prepare for financial audit have included developing manuals, monitoring activities such as testing of the implementation of real property procedures, and implementation of corrective actions to address identified deficiencies in the processes and procedures. DOD’s Real Property Asset Database Contained Inaccurate Data and Lacked Completeness, Although Certain Data We Reviewed Improved DOD’s RPAD has data quality issues specific to accuracy of certain data elements and completeness of the dataset, although certain data we reviewed improved since fiscal year 2014. Accuracy of data elements and completeness of RPAD are important to OSD, other federal agencies, and Congress because they use this information to determine facility sustainment funding and to understand DOD’s utilization of its real property as a means to identify potential excess property for disposal, among other things. Accuracy of Certain Data in RPAD Improved but Other Data Did Not Comply with Information Requirements We found that accuracy of certain data in the selected set of key data elements we reviewed improved while other data contained discrepancies that resulted in inaccuracies in RPAD for fiscal years 2014 through 2016. For some data we reviewed, the magnitude of such discrepancies decreased while others increased from fiscal year 2014 to fiscal year 2016. Specifically, we found: Operational status. For operational status codes that are not an active status, such as an asset that was determined to be excess or surplus, or disposed, OSD’s business rules require a corresponding date that documents when the status was determined or when a disposal was completed. If the corresponding date is not provided, then the operational status cannot be verified as correct. Our analysis of operational status from fiscal year 2014 through fiscal year 2016 found improvements in data on surplus and disposed facilities. The percentage of surplus and disposed facilities without a valid date improved from 37.5 percent to 0 percent and 3.3 percent to 0.3 percent, respectively. However, the percentage of excess facilities without a valid date increased from 22.7 percent to 47.9 percent. Asset review date. All facilities are required to have a date that documents a physical inventory; these reviews are to be conducted at least every 5 years, unless a historic asset. The percentage of facilities with a review date older than 5 years improved from 34.1 percent in fiscal year 2014 to 22.1 percent in fiscal year 2016. RPAD in fiscal year 2016 indicated that 143,420 facilities had a physical inventory date that was older than 5 years, which suggests that the information for these facilities may not be accurate because the information has not been updated within the required time frame. According to real property installation officials, overdue dates can occur because the physical inventory was either not conducted or the information from the physical inventory was not entered into the military services’ data systems. The percentage of facilities with a missing review date increased from 3.4 percent in fiscal year 2014 to 7.2 percent in fiscal year 2016. Plant replacement value. All facilities are required to have a plant replacement value not less than zero, meaning it cannot be a negative number. For all 3 fiscal years, none of the facilities had a negative plant replacement value and missing entries were an insignificant number. The business rules allow for values of zero though these entries may potentially create problems for other data elements that use plant replacement value as part of their calculation. For example, plant replacement value is a denominator in the formula used to calculate facility condition index. If a plant replacement value is zero, the facility condition index cannot be determined. The percentage of facilities with a plant replacement value of zero declined from 3.4 percent in fiscal year 2014 to 2.3 percent in fiscal year 2016. Utilization rate. All facilities are required to have a utilization rate from 0 to 150. The percentage of facilities missing a utilization rate improved from 23.3 percent in fiscal year 2014 to 2.4 percent in fiscal year 2015 before increasing to 14.4 percent in fiscal year 2016. As such, in fiscal year 2016, about 93,600 facilities did not have an indication of the utilization and this information was not available to users of RPAD. Facility condition index. All facilities are required to have a facility condition from 0 to 100. The percentage of facilities that had missing facility condition entries increased from 0.5 percent in fiscal year 2014 to 5.6 percent in fiscal year 2016. Figure 3 displays our analysis of discrepancies between the information requirements and data entries in RPAD. RPAD Was Incomplete as It Did Not Include All of DOD’s Existing Real Property Assets RPAD did not include all of DOD’s existing real property assets in fiscal years 2014 through 2016, resulting in an incomplete data set. Specifically, we found (1) the military services have not recorded all assets that existed and reflected previously disposed facilities that no longer existed as active in their respective data systems, (2) the military services did not report all assets in the RPAD submission to OSD that were recorded in each military service’s data system, and (3) OSD did not include all assets reported by the military services in RPAD, as shown in figure 4. We and others found instances of facilities that existed that the military services did not record in their data systems and of disposed facilities that no longer existed but were still reflected as active in RPAD. During our 12 site visits, officials at two installations stated that there were real property assets on their installations that were not recorded in their real property data system at the time of our visit. For example, real property officials at an Army installation identified over 2,000 existing assets—primarily linear structures—that were not in the inventory. Real property officials at a Marine Corps installation acknowledged that they were aware of assets that were not recorded in the data system but did not know the quantity of these. The officials stated they were in the process of reconciling the real property inventory with the assets in existence on the installation. In May 2018, Marine Corps Headquarters officials stated they plan to send real property officials to this location from other installations to assist with entering identified assets into the inventory. With the additional support, the officials expect the reconciliation to be completed in fiscal year 2019, 3 years earlier than initially planned. Moreover, in our review of 120 facilities during site visits, we found that 6 of the facilities had been disposed of but were recorded as active in the fiscal year 2015 RPAD data of the Air Force and Army. For example, all four of the Army’s disposals occurred previous to fiscal year 2015 but were not entered into the data system until fiscal year 2016. The changes were made and reflected in the inventory submission for fiscal year 2016. Also, one of the Air Force’s assets, fencing, had been disposed of years ago with the housing project that it enclosed, but was not included in the original disposal documentation. The real property installation officials had identified this omission when reviewing the list of assets that we selected for our review and began documenting the disposal prior to our site visit. DOD reported in its 2017 Agency Financial Report that material weaknesses in its internal controls over real property resulted in, among other things, that the department could not substantiate that all existing assets were recorded in the military services data systems. Similar to our site visit results, the Navy Office of Financial Operations also reported in June 2017 that 15 of 650 real property assets tested from a non- generalizable sample were reported to have been disposed of, but were not recorded as disposed of or removed from the Navy’s data system. Additionally, we found RPAD did not include some facilities that were in the military services’ data systems. The number and total plant replacement value across these three data sets should be identical, but were not in each of the 3 years that we reviewed. This means information on the excluded real property was not available to users of RPAD or to the Federal Real Property Profile. Specifically, The military services did not report all facilities in their data systems to OSD for inclusion into RPAD. Our analysis found the Army, Navy, and Marine Corps did not report to OSD between approximately 40,900 facilities (6.1 percent) and 103,600 facilities (15.9 percent) of the facilities included in their data systems in fiscal years 2014 through 2016. If all of these facilities still existed during those years, these unreported facilities had a total plant replacement value that ranged from $12.8 billion to $56.5 billion during the 3 fiscal years. We could not include the Air Force in this analysis because it was not able to provide its end-of-year real property inventory for fiscal years 2014 through 2016. Air Force officials stated that their contractor did not archive copies of the end-of-year real property inventory for these years but would begin to do so for fiscal year 2017. OSD did not include all facilities reported by the military services and WHS in RPAD. Additionally, our analysis showed that the number of facilities OSD did not include in RPAD ranged from about 3,300 facilities (0.5 percent) to 19,400 facilities (2.6 percent) of the facilities reported by the military services and WHS in fiscal years 2014 through 2016. If all of these facilities still existed during those years, the total plant replacement value of the unreported facilities ranged from $3.4 billion to $21.6 billion. OSD and military service officials agreed that accuracy and completeness issues with real property have been a long standing issue, but stated recent audit efforts associated with FIAR should result in some improvements of the data. For example, military service installations officials stated that they are working to reconcile differences between existing real property and information in their data systems to include adding existing assets that are not in the data system and correcting information on disposed assets. Moreover, military service officials stated that they have emphasized conducting timely physical inventories and require installations to report on the currency of their physical inventories. According to officials, when reporting real property to OSD and when OSD consolidates this information into RPAD, assets with significant errors in their records are excluded to improve the accuracy of the information in the data set. The officials explained as the accuracy of the data improves through physical inventories, fewer assets will be excluded in the reporting process, which will improve completeness of RPAD. However, as we describe further in this report, the audit efforts will not correct all identified accuracy and completeness issues. Deficiencies Exist in DOD’s Processes for Recording and Reporting Real Property Data DOD’s processes for recording and reporting real property data have deficiencies that contribute to inaccuracies and incompleteness in the RPAD data. Specifically, we identified inconsistencies in the military services’ recording of real property transactions and physical inventories of assets. In addition, we found the military services have not corrected identified discrepancies in their real property data reported to OSD in the annual RPAD submissions. DOD Has Processes for Recording and Reporting Real Property Data According to a DOD instruction, OSD must establish, issue, and maintain data requirements for DOD’s real property inventory. As such, DOD requires that the military services maintain an accurate and complete record of their real property, regardless of the organization using or funding the real property. The real property accountable officers at each installation must implement processes to ensure that all real property transactions are auditable and that information recorded, including physical inspections, is accurate, complete and retained in accordance with applicable laws and regulations. OSD also requires that the military services report their real property data for RPAD following OSD requirements and that they utilize OSD’s verification and validation tool to identify discrepancies between data entries and DOD’s real property information requirements. OSD and the military services have developed some procedures to implement these policies. For example, OSD established an annual reporting process, to include defining the specific content and format for the submission of information. Moreover, the military services have developed written procedures that clarify how specific transactions should be conducted. For example, the Marine Corps has developed detailed guidance on control processes for appropriately documenting disposed assets. The Navy has developed procedures for conducting physical inventories. The Army has defined roles and responsibilities for accounting for real property, including changes to facility function (i.e., category code). Lastly, the Air Force has developed overall policies and procedures for accounting for real property that defines the roles and responsibilities of accountable officials. Military Services Did Not Consistently Record Real Property Data The processes for recording real property information include documenting and entering into the data system when transactions— acquisition of, change to, and disposal of a real property asset—or physical inventories occur. To document a transaction or physical inventory, real property installation officials are expected to complete the required supporting records. According to Standards for Internal Control in the Federal Government, appropriately designed control activities could include requiring documentation should be completed within a reasonable time frame after the event occurs. Then, the officials are to promptly enter the updated information into the real property data system. DOD also requires a review of each real property asset record, including a physical inventory of each real property asset every 5 years for non- historic assets or every 3 years for historic assets. Physical inventories help ensure current and accurate information on assets are reflected in the military services’ data systems. Furthermore, the Standards for Internal Control in the Federal Government require agencies to design control activities to achieve objectives, to monitor activities, and to remediate identified deficiencies on a timely basis. Such activities could include appropriately documenting and accurately and timely recording transactions, and implementing procedures to help ensure that processes are monitored and evaluated for deficiencies on an ongoing basis, corrective actions are determined for any identified deficiencies, and these actions are completed and documented to correct deficiencies on a timely basis. Real Property Transactions We and the military services identified that transactions were not being consistently documented with required supporting records or entered into the military services’ data systems within reasonable time frames. Specifically, during our site visits to 12 military services’ installations, officials at 5 installations stated that they were experiencing delays with documenting and entering into the data system some transactions. According to the officials, this occurred due to challenges with obtaining required information from contractors, heavy workloads, and staff shortages. Moreover, the military services found through testing in 2017 that they did not consistently document transactions with required supporting records or enter real property transactions into the data system. The military services conducted these tests as part of their preparation for financial statement audits to identify deficiencies in the recording of real property transactions. The military services were then to develop corrective action plans and remedy any identified deficiencies prior to the department’s audit of the fiscal year 2018 financial statements. Specifically, The Air Force conducted tests in March 2017 and reported that of 271 assets tested, 171 did not have appropriate supporting records. The Air Force also reported in a separate test of 27 assets that 17 of these were not timely or accurately recorded. The Army conducted tests in October 2017 and reported that more than half of the assets selected did not pass its testing for one or more of the 9 key data elements associated with plant replacement value. Lack of adequate supporting records was the most common reason for test failure. The Navy conducted tests in October 2017 and identified documentation issues or key elements that were not timely or accurately entered into its data system for 11 out of 58 assets tested. The Marine Corps conducted tests in July 2017 and identified documentation issues or key elements that were not timely or accurately entered into its data system for 20 of 55 assets tested. Physical Inventories We and the military services have identified that real property installation officials do not consistently document or enter physical inventory information into the military services’ data systems. We found during our site visits to 12 military service installations that for 21 facilities out of 106 facilities tested, real property installation officials had not entered physical inventory information in the military services’ data system within the last 5 years. The 21 facilities we reviewed included 16 from the Air Force (with 2 reflected as being last inventoried in January 1934 or October 1992), 4 from the Army, and 1 from the Marine Corps (which showed as being been last inventoried in November 2003). The military services also identified similar inconsistencies with recording physical inventories in testing of their real property assets as part of their preparation for financial statement audits: The Air Force conducted tests in March 2017 and reported that installation officials, for 89 out of 281 assets tested, did not have complete supporting records or did not timely provide the most recent physical inventory checklist that reconciled with the Air Force’s data system. The Army conducted tests in September 2016 and reported 1 of the 5 installations tested did not have adequate supporting records for asset changes identified in physical inventories. The Navy conducted tests in June 2017, and reported 5,918 of the 34,104 assets tested had not had a physical inventory for more than 5 years. Furthermore, in October 2017, the Navy reported that 9 of 41 assets it tested did not have supporting records that the inventory was performed per DOD requirements for timeliness. The Marine Corps had an external auditor conduct tests in September 2017 and reported that installation officials could not support the last physical inventory performed for 83 of 998 assets tested. The military services did not fully monitor recording processes on an ongoing basis, including evaluating whether or the extent to which activities are being carried out and remediating any identified deficiencies. We found that this occurred in part due to the military services not being required to conduct ongoing monitoring of the processes used for recording real property transactions and physical inventories. According to military service officials, they conduct monitoring of recording and have begun developing corrective action plans as part of the recent audit readiness effort that are based on the Financial Improvement and Audit Readiness Guidance. However, this guidance aimed principally at improving financial reporting addresses 40 of the 216 data elements required to be maintained in RPAD. DOD has not determined to what extent the remaining data elements are a priority for other management purposes beyond financial reporting. Accordingly, the remaining 176 data elements, or approximately 80 percent, are not required to be monitored. For example, the recording of RPAD-required data elements for dates that support an operational status of excess, surplus, and disposed or document when a facility was built are not included in current monitoring efforts. The monitoring of the recording of only about 20 percent of the required data elements in RPAD results in inaccurate and incomplete data not being systemically evaluated and corrective actions not being taken to resolve the issues. Unless the military services are required to monitor on an ongoing basis the processes used for recording all required real property information, DOD will continue to have data quality issues related to accuracy and completeness in the military services’ data systems that will be reflected in RPAD. Military Services Have Not Corrected Identified Discrepancies Reported in Their RPAD Submissions The military services have not corrected identified discrepancies in their real property data reported to OSD in the annual RPAD submissions. OSD provided the military services with a verification and validation tool to identify data that does not comply with information requirements. Specifically, from fiscal years 2014 through 2016, the military services used OSD’s verification and validation tool to identify discrepancies and submitted reports summarizing the results to OSD, but have not corrected all discrepancies identified by the tool. According to real property installation officials, they have not been directed by headquarters to correct discrepancies in their data systems that were identified in their annual RPAD submission. Our review of 120 assets during the 12 installation site visits confirmed that 61 assets with discrepancies in five key data elements tested in the fiscal year 2015 RPAD data set continued to have these discrepancies in 2017. Based on our analysis, the military services have not corrected identified discrepancies in part because OSD’s guidance for annual RPAD reporting does not define which data elements were most significant to the department’s decision making and should be a priority for correction. Furthermore, we found that the guidance does not require the military services to develop and implement corrective action plans to remediate discrepancies in significant data elements in their data systems that are identified by OSD’s verification and validation tool. According to OSD and military service officials, identifying significant data elements could assist with streamlining and prioritization of efforts to improve data quality. In addition, OSD officials agreed that requiring the services to develop and implement corrective action plans would benefit data quality, but stated there are challenges with the verification and validation tool that would need to be addressed to leverage its full potential. By OSD not defining significant data elements and coordinating corrective action plans to remediate discrepancies, the military services may continue to submit information with discrepancies from year to year in some data elements and will miss an opportunity to improve the accuracy of inventory data. DOD Has Not Addressed Risks to Its Ability to Use Information to Manage Its Real Property DOD has not addressed three risks—unfilled real property positions to manage its data, lack of a department-wide approach to improving data quality, and a limited plan for the implementation of its expanded data platform—that diminish its ability to use real property information to manage its real property. Unfilled Real Property Positions We found that DOD has not addressed how it will overcome unfilled real property positions throughout the department, which poses a risk to data quality. For example, real property installation officials at 10 of 12 installations we visited told us that they had unfilled real property positions, including real property accountable officers, engineers, realty specialists, planners, and space management analysts. Real property installation officials told us that their unfilled real property positions contributed to workload backlogs and prevented them from sufficiently maintaining their real property data. The Army, Marine Corps, and Navy completed various workforce plans that found they did not have a sufficiently sized workforce to adequately maintain their real property data: Army: In March 2015, the Army completed a workforce analysis that found current authorized manning documents are short 223 real property positions of the total 495 positions required to perform these functions, which include real property accountable officers and realty specialists. Marine Corps: In August 2016, the Marine Corps identified that it had an immediate need for 20 real property accountable officer positions to effectively maintain its real property data. According to Marine Corps officials, they have since filled 19 of the 20 positions. Navy: For its fiscal year 2018 planning, the Navy identified a need for 63 real property positions—ranging from real property accountable officers to geospatial specialists—to meet real property requirements. The Air Force has not identified the workforce it needs to maintain quality data on its real property, but Air Force officials told us in May 2018, that they are beginning efforts to better understand their workforce needs. OSD and military service officials stated that they continually face challenges due to unfilled real property positions. However, they do not expect to fill all of their authorized positions because senior leadership has prioritized staffing at other offices and military service officials stated that they face challenges in finding qualified applicants for open positions. Despite the recognized needs, DOD has not outlined how it plans to overcome challenges related to its unfilled real property positions. According to an OSD official, OSD cannot direct the military services to fill their real property positions; however, OSD has not coordinated with the military services to identify opportunities to overcome unfilled positions. Potential opportunities may include using available staff more efficiently or evaluating opportunities to better address how they will manage unfilled positions. Lack of a Department-wide Approach to Improving Data Quality Absent a department-wide approach to improving data at various levels within DOD, military service headquarters have individually initiated actions to improve data quality for certain data elements. These efforts are largely uncoordinated and result in inconsistent approaches to address similar data quality risks and may contribute to inefficient use of resources and accuracy issues in the real property data. For example, military service headquarters officials told us they have taken action to improve data quality when they do not receive specific guidance from OSD, including communicating priorities to installations and developing contracts to improve select data elements. We found instances where the military services took different actions to improve their information on utilization rates prior to OSD issuing a memorandum to have a standardized approach to determine this information. For example, the Army developed a database to record space authorization information for each asset. The Marine Corps used a contract to obtain space utilization information at certain installations. Moreover, a Marine Corps headquarters official stated in some cases that after Marine Corps headquarters implemented its own policy and provided guidance to the installations to fill a gap that OSD issued guidance with a different approach. The official stated that the Marine Corps had spent financial resources on a contract to improve a data element that they later had to categorize in a different way due to OSD guidance. Also, real property installation officials at a Marine Corps installation stated that their headquarters had made large-scale changes to the records of their housing assets due to a new approach to determine specific data elements for those assets, which resulted in inaccuracies. Officials noted that headquarters later retracted that approach and restored the records. In addition, we observed in our review of real property records during the site visits that real property installation officials did not apply the same criteria for determining an asset’s operational status for the codes of disposed, closed, and nonfunctional that resulted in inaccuracies. Navy regional command officials provided written guidance and a decision support tool for determining appropriate codes for operational status to help improve accuracy within this data element. However, according to real property installation officials, the Air Force and Army did not have similar guidance. OSD and military service officials agreed that better coordination among OSD and the military services would assist their effort to improve data quality. Limited Plan for Implementation of OSD’s Expanded Data Platform OSD has not fully identified how it will complete implementation of a new module for real property within its expanded data platform, known as the Data Analytics and Integration Support platform, and DOD faces a risk to information accessibility as it may not fully realize the anticipated benefits of the effort. OSD currently uses the platform for generating unique identification numbers for its real property assets and as a dashboard for tools related to military construction planning. However, OSD has neither outlined how it will accomplish its stated objectives and goals for expansion of this platform as OSD’s new data system for real property, nor has it set time frames for the expansion. In September 2017, OSD modified its contract for updating the Data Analytics and Integration Support platform, but that contract does not specify when full implementation of the expansion to include a new module for real property will occur. OSD is planning to expand the use of the Data Analytics and Integration Support platform to make it a near real-time, department-wide information source of required real property information accessible to a greater number of users who manage real property. If implemented, this expanded platform would replace DOD’s annual data call to the military services for end-of-year real property information to compile into the RPAD. Further, the expanded platform would interface daily with the military service data systems. This would provide near real-time information to users for the department-wide management of DOD’s real property. According to OSD officials, users could also access real property information themselves and run their own data analyses when OSD expands this platform to replace the annual data call to the military services. Figure 5 displays a comparison of RPAD to the proposed expansion of the Data Analytics and Integration Support platform. OSD officials told us that the military services will need to ensure their data systems can fully interface with the Data Analytics and Implementation Support platform for full integration to occur. Specifically, the officials stated that the Army’s data system can fully interface with OSD’s expanded platform, but the Navy wants to test how its data system would interface with the platform before it can fully connect. In addition, officials noted that the Air Force’s current data system is the least compatible with OSD’s expanded data platform because it is currently working to design and implement a new data system for real property. OSD and Air Force, Marine Corps, and Navy officials noted that they are aware the military service data systems are not fully integrated with OSD’s expanded data platform. Guidance from DOD and the Office of Management and Budget note that risk management is integral to effective program management. The Standards for Internal Control in the Federal Government states management should define objectives clearly such as through specific and measurable terms that allow for the assessment of performance toward achieving those objectives and that management should identify, analyze, and respond to risks related to achieving defined objectives. The office of the Assistant Secretary of Defense for Energy, Installations, and Environment is responsible for providing the guidance and procedures for implementing real property management policy, including ensuring the information is available to determine if an asset is used effectively. One way an organization can manage risk is by developing a risk management strategy that identifies risks to program objectives, and includes time frames and performance metrics for addressing those risks. DOD has taken some actions that when fully implemented should result in some improvements to select data elements and the potential to enhance information accessibility. However, in part, DOD’s weaknesses with quality information on real property and accessibility to this information continue to exist because DOD has not developed a strategy that identifies and addresses risks, such as those previously described, and includes time frames and performance metrics. OSD and military service officials agreed that a strategy for addressing risks would help the department to further its effort to improve the quality and accessibility of the information. Developing and implementing such a strategy would allow the department to take key steps toward improving its information for managing its real property. Without a strategy for improving the quality of the data and information used to manage its real property, DOD, Congress, the Office of Management and Budget, and the General Services Administration will not have information needed for effective decision making and do not have reasonable assurance that risks to data quality and information accessibility are being managed appropriately. Specifically, information would be limited in decision making related to improving space management at installations, to adequately sustaining DOD’s real property assets, and to accurately generating financial statements. Conclusions DOD’s efforts to reform its real property management is complicated by not having quality data on its large inventory of assets—over 568,000 facilities with an estimated combined plant replacement value of about $1 trillion. An accurate and complete inventory of its assets is essential for DOD to make informed management decisions about its real property. The department has taken action to improve data quality of some data elements through financial improvement and audit readiness efforts. However, deficiencies in the processes for recording and reporting real property data continue to lead to inaccurate and incomplete information. The military services do not require monitoring of the recording of all required real property information, to include evaluating on an ongoing basis whether or to what extent these activities are carried out and remediating any identified deficiencies. In addition, OSD has not defined which data elements were significant to the department’s decision making and which should be a priority for correction. Also, the military services do not have plans to correct the discrepancies in significant data elements in their data systems that are identified by OSD’s verification and validation tool. Without taking actions to address these deficiencies, DOD will continue to have inaccurate and incomplete real property data and unreliable information in RPAD. DOD also has not developed a strategy that establishes time frames and performance metrics to address risks to data quality and information accessibility. Specifically, DOD faces risks related to unfilled real property positions, a lack of a department-wide approach to improving data, and a limited plan for implementation of OSD’s expanded data platform. Without a strategy to address these risks, DOD is missing an opportunity to ensure that the information needed for effective decision making, such as budget decisions and oversight by Congress, is available to meet real property accountability and reporting objectives and to avoid inefficient and potentially costly workarounds, such as additional data calls to installations. Recommendations for Executive Action We are making a total of 6 recommendations to the Department of Defense: The Secretary of the Army should require monitoring of its processes used for recording all required real property information—to include evaluating on an ongoing basis whether or to what extent these activities are being carried out—and remediating any identified deficiencies. (Recommendation 1) The Secretary of the Navy should require monitoring of Navy and Marine Corps processes used for recording all required real property information—to include evaluating on an ongoing basis whether or to what extent these activities are being carried out—and remediating any identified deficiencies. (Recommendation 2) The Secretary of the Air Force should require monitoring of its processes used for recording all required real property information—to include evaluating on an ongoing basis whether or to what extent these activities are being carried out—and remediating any identified deficiencies. (Recommendation 3) The Secretary of Defense should ensure that the Undersecretary of Defense for Acquisition and Sustainment, in collaboration with the military services, defines and documents which data elements within the RPAD submissions are most significant for decision-making. (Recommendation 4) The Secretary of Defense should ensure that the Undersecretary of Defense for Acquisition and Sustainment, in collaboration with the military services, coordinates on corrective action plans to remediate discrepancies in significant data elements in its real property data system that are identified by OSD’s verification and validation tool. (Recommendation 5) The Secretary of Defense should ensure that the Undersecretary of Defense for Acquisition and Sustainment, in collaboration with the military services, develops a strategy that identifies and addresses risks to data quality and information accessibility. At a minimum, this strategy should establish time frames and performance metrics for addressing risks related to (1) unfilled real property positions, (2) a lack of a department- wide approach to improving its data, and (3) implementation of OSD’s expanded data platform. (Recommendation 6) Agency Comments and Our Evaluation We provided a draft of this report to DOD for comment. In written comments, DOD concurred with four recommendations and partially concurred with three recommendations. DOD’s comments are summarized below. DOD also provided technical comments, which we incorporated as appropriate. In written comments, DOD stated that Recommendations 2 and 3 should be combined to more appropriately align with authority and responsibility of the U.S. Navy and U.S. Marine Corps as a single Military Department and DOD concurred with the combination of the two recommendations. Based on these comments, we combined the draft recommendations for separate actions by the Secretary of the Navy and the Commandant of the Marine Corps into one recommendation. In our final report, the action is addressed to the Secretary of the Navy in Recommendation 2 and our total number of recommendations is decreased to six. DOD partially concurred with our recommendation that the Undersecretary of Defense for Acquisition and Sustainment collaborate with the military services to develop a strategy that identifies and addresses risks to data quality and information accessibility (Recommendation 6). We recommended that the strategy, at a minimum, include timeframes and performance metrics for addressing risks and include other specific information. However, DOD stated that it plans to collaborate with the military services on separate service strategies that reflect each military service’s operating environment. We continue to believe that DOD would benefit from one department-wide strategy to improve data quality and information accessibility. For example, we found that the military services’ efforts to improve data quality have been largely uncoordinated and had led to inconsistent approaches, which may have contributed to data inaccuracies. Further, we found that OSD has not fully identified how it will complete implementation of a data platform expansion to include real property information and may not realize the anticipated benefits of the effort. The platform is an effort managed by OSD and would benefit from a single DOD strategy addressing key points noted in our recommendation. Accordingly, we believe our recommendation remains warranted. DOD’s comments are reprinted in their entirety in appendix II. We are sending copies of this report to the appropriate congressional committees and to the Secretary of Defense; the Under Secretary of Defense for Acquisition and Sustainment; the Under Secretary of Defense (Comptroller); and Secretaries of the Departments of Air Force, Army, and Navy, the Commandant of the Marine Corps, and the Director of Washington Headquarters Services. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact Brian J. Lepore at (202) 512-4523 or leporeb@gao.gov or William J. Cordrey at (404) 679-1873 or cordreyw@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs are listed on the last page of this report. GAO staff who made key contributions to this report are listed in appendix III. Appendix I: Physical Inventory For the military services and Washington Headquarters Services’ real property inventories, DOD requires that the data elements shown below in table 1 be validated through a physical inventory of each real property asset. Physical inventories are to be performed every 5 years or every 3 years for historic assets. Appendix II: Comments from the Department of Defense Appendix III: GAO Contacts and Staff Acknowledgments GAO Contacts Staff Acknowledgments In addition to the contacts named above, Gina Hoffman (Assistant Director), Paul Kinney (Assistant Director), Susan Langley (Analyst-in- Charge), Scott Bruckner, Vincent Buquicchio, Josh Edelman, Chad Hinsch, Brad Johnson, Amie Lesser, Carol Petersen, Sam Portnow, Richard Powelson, Michael Silver, and John Yee made key contributions to this report. Related GAO Products DOD Financial Management: The Navy Needs to Improve Internal Control Over Its Buildings, GAO-18-289. Washington, D.C.: May 10, 2018. High-Risk Series: Progress on Many High-Risk Areas, While Substantial Efforts Needed on Others, GAO-17-317. Washington, D.C.: February 15, 2017. Defense Facility Condition: Revised Guidance Needed to Improve Oversight of Assessments and Ratings, GAO-16-662. Washington, D.C.: June 23, 2016. DOD Financial Management: Greater Visibility Needed to Better Assess Audit Readiness for Property, Plant, and Equipment, GAO-16-383. Washington, D.C.: May 26, 2016. Defense Infrastructure: More Accurate Data Would Allow DOD to Improve the Tracking, Management, and Security of Its Leased Facilities, GAO-16-101. Washington, D.C.: March 15, 2016. Underutilized Facilities: DOD and GSA Information Sharing May Enhance Opportunities to Use Space at Military Installations, GAO-15-346. Washington, D.C.: June 18, 2015. Defense Infrastructure: DOD Needs to Improve Its Efforts to Identify Unutilized and Underutilized Facilities, GAO-14-538. Washington, D.C.: September 8, 2014. Defense Infrastructure: Army Brigade Combat Team Inactivations Informed by Analyses, but Actions Needed to Improve Stationing Process, GAO-14-76. Washington, D.C.: December 11, 2013. Federal Real Property: Greater Transparency and Strategic Focus Needed for High-Value GSA Leases, GAO-13-744. Washington, D.C.: September 19, 2013. Military Bases: Opportunities Exist to Improve Future Base Realignment and Closure Rounds, GAO-13-149. Washington, D.C.: March 7, 2013. Military Base Realignments and Closures: Updated Costs and Savings Estimates from BRAC 2005, GAO-12-709R. Washington, D.C.: June 29, 2012. Excess Facilities: DOD Needs More Complete Information and a Strategy to Guide Its Future Disposal Efforts, GAO-11-814. Washington, D.C.: September 19, 2011. Defense Infrastructure: The Enhanced Use Lease Program Requires Management Attention, GAO-11-574. Washington, D.C.: June 30, 2011. Federal Real Property: Progress Made on Planning and Data, but Unneeded Owned and Leased Facilities Remain, GAO-11-520T. Washington, D.C.: April 6, 2011. Military Base Realignment and Closures: DOD Is Taking Steps to Mitigate Challenges but Is Not Fully Reporting Some Additional Costs, GAO-10-725R. Washington, D.C.: July 21, 2010. Defense Infrastructure: Continued Management Attention Is Needed to Support Installation Facilities and Operations, GAO-08-502. Washington, D.C.: April 24, 2008. Federal Real Property: Progress Made Toward Addressing Problems, but Underlying Obstacles Continue to Hamper Reform, GAO-07-349. Washington, D.C.: April 13, 2007. Defense Infrastructure: Issues Need to Be Addressed in Managing and Funding Base Operations and Facilities Support, GAO-05-556. Washington, D.C.: June 15, 2005. Defense Infrastructure: Changes in Funding Priorities and Strategic Planning Needed to Improve the Condition of Military Facilities, GAO-03-274. Washington, D.C.: February 19, 2003. Defense Infrastructure: Military Services Lack Reliable Data on Historic Properties, GAO-01-437. Washington, D.C.: April 6, 2001. Military Infrastructure: Real Property Management Needs Improvement, GAO/NSIAD-99-100. Washington, D.C.: September 7, 1999.

Why GAO Did This Study DOD manages a portfolio of real property assets that as of fiscal year 2016 reportedly included about 568,000 facilities with a combined plant replacement value of about $1 trillion and 27.2 million acres of land. DOD requires the military services and Washington Headquarters Services to collect and maintain information about each of the assets in their inventories to assist the department with management decision making. In May 2017, the House Armed Services Committee, Subcommittee on Readiness, asked GAO to review DOD's management and use of its real property data. This report evaluates (1) how accurately and completely RPAD reflects DOD's real property assets, (2) DOD's processes to ensure accuracy and completeness in recording and reporting real property data, and (3) DOD's actions to ensure it has addressed risks that may affect the use of real property information for managing its assets. GAO analyzed the RPAD and military services' data for fiscal years 2014-2016; reviewed documentation; conducted site visits; and interviewed DOD officials. What GAO Found GAO found that the Department of Defense's (DOD) Real Property Assets Database (RPAD) contained inaccurate data and lacked completeness, although certain data that GAO reviewed had improved their accuracy since fiscal year 2014. RPAD is a department-wide database of real property data annually compiled by the Office of the Secretary of Defense from the inventories of the military services and DOD's Washington Headquarters Services, which manages real property in the National Capital region. DOD uses RPAD to report on DOD's real property to Congress and other federal agencies, such as the Office of Management and Budget and the General Services Administration to assist in managing federal real property. DOD has weaknesses in its processes for recording and reporting real property data that have led to inaccurate and incomplete information. GAO and others found military services have not consistently recorded real property transactions (i.e., acquisition of, change to, and disposal of a real property asset) and physical inventories of assets. GAO also found that the military services have not corrected identified discrepancies in their data systems, such as missing entries for utilization and facility condition and overdue asset reviews. GAO reviewed records of 120 facilities with identified discrepancies in fiscal year 2015 RPAD data and compared them to the records in the respective data system in 2017 and found that 61 discrepancies remained. The military services had corrected the data in the remaining 59 reviewed facilities in their data systems. DOD's efforts to prepare for an upcoming financial audit have helped identify issues and improve accuracy of some data. However, if DOD does not require the military services to fully monitor recording processes and implement corrective actions to resolve data discrepancies, the department will continue to have incomplete and inaccurate real property data and unreliable RPAD information. DOD has not addressed three risks that can adversely affect its ability to use its information to manage its real property. Specifically, DOD (1) has unfilled real property positions limiting its capacity to manage its data, (2) lacks a department-wide approach to improving its data quality, and (3) has not identified how it will complete implementation of an effort to improve access to data. These risks exist, in part, because DOD has not developed a strategy that identifies and addresses risks with accompanying time frames and performance metrics. If DOD does not develop a strategy that identifies and addresses risks to data quality and information accessibility, DOD may miss the opportunity to reasonably ensure that the information needed for effective decision making by DOD, Congress, and other federal agencies is available to meet real property accountability and reporting objectives. What GAO Recommends GAO is making six recommendations to improve DOD's real property data, including fully monitoring recording processes; developing and implementing corrective actions for identified data discrepancies; and developing a strategy to address risks associated with data quality and information accessibility. DOD concurred or partially concurred with all draft recommendations. In response, GAO agreed to combine two recommendations.

gao_GAO-18-224

gao_GAO-18-224_0

Background IDT Refund Fraud Viewed broadly, IDT refund fraud is composed of two crimes: (1) the theft or compromise of PII, and (2) the use of stolen (or otherwise compromised) PII to file a fraudulent tax return and collect a fraudulent refund. Figure 1 presents an example of how fraudsters may use stolen PII and other information, real or fictitious (e.g., sources and amounts of income), to complete and file a fraudulent tax return and successfully receive a refund. In this example, a taxpayer may alert IRS of IDT refund fraud. Alternatively, IRS can detect IDT refund fraud through its automated filters that search for specific characteristics as well as through other reviews of taxpayer returns. IRS reported that, through September 2017, the number of taxpayers reporting that they were a victim of IDT refund fraud had decreased by about 40 percent compared to the same period in 2016 (from 348,650 to 208,503). IRS officials attribute this decline to improved fraud filters. Pre-refund Compliance Checks We have long highlighted the importance of pre-refund compliance checks as a means to improve compliance while minimizing taxpayer burden. As we testified in 2011, pre-refund compliance checks help IRS to confirm taxpayers’ identity, quickly and efficiently correct some errors with virtual certainty, and identify and audit some returns before refunds are issued. They also have the potential to deter billions of dollars in erroneous refunds, especially for refundable tax credits. These credits have complex eligibility requirements and are often overclaimed. IRS’s ability to match tax returns to information provided by third parties, including from financial institutions, can help enforce compliance with the tax laws. Pre-refund checks benefit taxpayers directly when IRS identifies underclaimed benefits. Pre-refund compliance checks can reduce the tax gap created when taxpayers file returns that, for example, underreport their tax liability. In 2016, IRS estimated that the average annual gross tax gap was $458 billion for tax years 2008 to 2010. IRS estimated that through late payments and enforcement actions, it would collect an additional $52 billion annually for those tax years, resulting in an average net tax gap of $406 billion. Because of the importance of improving voluntary compliance and addressing the tax gap, we continued to include Enforcement of Tax Laws as a high-risk area in our 2017 High-Risk Report. Systemic Verification As noted previously, beginning in 2017 the Protecting Americans from Tax Hikes Act of 2015 requires employers to submit W-2s to SSA by January 31 (about 1 to 2 months earlier than in prior years, depending on the method of filing). It also requires IRS to hold refunds for all taxpayers claiming EITC or ACTC until February 15. In October 2017, IRS reported that, among the 13.4 million refunds subjected to this hold, it had completed processing 10.3 million refunds totaling $51.2 billion. Although IRS has authority to hold additional refunds until it receives more W-2 data, IRS, in consultation with Treasury, decided not to exercise this authority in 2017. IRS officials explained that they did not do more than required by the law because it would be a major shift in refund issuance causing a strain on the economy, industry partners, taxpayers, and IRS telephone and other operations. Officials said that they expect to learn from their experience during the 2017 filing season and will continue to consider changes for future filing seasons as they have for 2018. However, all returns—with EITC or ACTC and without EITC or ACTC— were subject to systemic verification as well as other fraud filters. Systemic verification is one element of IRS’s Return Review Program (RRP), its primary system to detect fraud and noncompliance. RRP is a platform that runs individual tax returns through a comprehensive set of rules and models to detect potential taxpayer fraud and other noncompliance, then selects returns for various treatment options. Systemic verification categorized taxpayer returns in one of three outcomes to detect potentially fraudulent or noncompliant returns (see figure 2): 1. Wage information verified: Income and withholding on the return matches W-2 data within the allowed threshold. 2. False or incorrect income: Information on the return is not valid when compared to W-2 data. This mismatch can include income, withholding, employer identification number, or other characteristics. 3. Unable to verify: Unable to verify income or withholding on the return because W-2 data are unavailable or the taxpayer did not report wage income but had other types of income such as Social Security or self- employment. IRS reprocessed (looped) all returns that reported wage income through RRP when new third-party data became available. For EITC or ACTC returns that IRS was required to hold until February 15, IRS had additional time to reprocess these returns before releasing the refund. After systemic verification is completed, IRS either continues processing the refunds for release or holds the refunds for additional review. For returns where IRS either verified or was unable to verify the wage information, the refunds were processed (beginning February 15 for returns with EITC or ACTC) unless selected by the fraud filters for review. However, IRS does not have the authority to correct a taxpayer’s return based on W-2 data, so it must initiate a correspondence audit which, as we have reported, is more costly to IRS, more burdensome on the taxpayer, and more time consuming for both. Therefore, for returns with false or incorrect income, IRS froze the refund and directed it to various units for review depending on the results of systemic verification and fraud filters. For example, if IRS suspected that the return was IDT refund fraud, it directed it to the Taxpayer Protection Program to verify the taxpayer. For returns where IRS suspected potential noncompliance, it directed the return to the Integrity and Verification Operations group. Systemic Verification Shows Promise but Legacy Information Technology and Issues with Employer W-2 Filing Limit Its Success Early W-2 Data and the Refund Hold Helped IRS Prevent Fraud and Noncompliance for a Limited Number of Tax Returns By mid-February 2017, 2 weeks following the new W-2 filing deadline and about when the refund hold expired, IRS had received more than twice as many (over 214 million) W-2s than SSA provided at a similar time in 2016 (see figure 3). Nevertheless, IRS did not have all W-2 data in time to conduct pre-refund checks of wages, withholding, and other information before issuing refunds, especially early in the filing season. Despite not having all W-2 data, IRS was able to identify and prevent some fraud and noncompliance before issuing refunds. IRS received and initially processed through systemic verification a total of about 35.1 million individual tax returns through February 14, representing nearly $200 billion in refunds. As shown in table 1, nearly 13.4 million (38 percent) of those returns claiming about $115 billion in refunds were filed by taxpayers who claimed EITC, ACTC, or both, and were subject to the refund hold. Using systemic verification, as of February 14, 2017, IRS determined that nearly 150,000 of these 13.4 million returns (1 percent) were potentially fraudulent because they included false or incorrect income. The returns represented approximately $800 million in refunds. IRS also verified wage and other information for approximately 4.72 million (35 percent) of those returns filed and processed through February 14, representing $73.5 billion in refunds. However, IRS was unable to verify 7.79 million (58 percent) of these returns before it released refunds because W-2 data were unavailable, as described later in this report. Finally, table 1 also notes that, as of October 2017, IRS reported that, among those returns filed and processed through February 14, 10.3 million had completed processing and $51.2 billion in refunds had been issued. As the February 15 refund hold expiration approached, IRS continued to reprocess (loop) returns through systemic verification as more W-2 data became available. In doing so, IRS staff identified 12,000 more returns, in addition to the 150,000 initially identified, that they suspected to be fraudulent. This brought the total number of potentially fraudulent or noncompliant returns to about 162,000 with nearly $863 million in refunds. IRS manually held these refunds and referred the suspicious returns for further screening to the Integrity and Verification Operations group. IRS later cleared approximately 150,000 (93 percent) of these returns and released about $797 million in refunds. IRS confirmed that approximately 12,000 (7 percent) of the returns that it had not cleared were fraudulent, eventually protecting $65 million, which included $51 million in EITC or ACTC claims. To reduce false positives (when legitimate tax returns are erroneously selected for review), an IRS working group made several changes to how IRS’s fraud filters make selections based on W-2 data and other information. For 2018, IRS plans to automatically select returns that it had held manually in 2017. However, officials noted that while verifying wage information is important, the complexity of determining EITC and ACTC eligibility remains a challenge. We reviewed IRS’s systemic verification results and found that IRS improved its selections of potentially fraudulent returns with W-2 data contributing to its fraud filters. As of February 15, returns selected for review by systemic verification comprised 14,618 (6 percent) of all paper and electronic returns selected as potential identity theft by the fraud filters. By September 15, selections from systemic verification increased to nearly 78,369 (about 10 percent) of all returns selected as potential identity theft. Moreover, we found that if more W-2 data were available earlier, IRS could have excluded more returns from review, thereby reducing or eliminating work and reducing taxpayer burden by not delaying legitimate taxpayers’ returns. For example, systemic verification allowed IRS to exclude about 321,000 electronically-filed tax returns out of more than 700,000 that had been selected for review by the fraud filters. IRS’s Ability to Process W-2s Was Limited by Its Information Technology Systems and Issues with Employer W-2 Filing We found that IRS’s ability to verify information on tax returns early in the filing season was limited because of its Information Technology (IT) systems and issues with employers filing W-2s on paper or after the filing deadline. IT systems. IRS receives and maintains validated taxpayer data, including W-2 and 1099-MISC forms, through the Information Return Master File (IRMF) system. IRS received W-2 data from SSA daily but only loads the data onto IRMF weekly due to the legacy design of this system. This contributed to IRS’s inability to verify more than half (7.79 million or 58 percent) of tax returns with EITC or ACTC claiming $38.4 billion in refunds when the February 15 refund hold expired. IRS officials stated that due to the system’s legacy design, adding new or updating existing information return documents requires the agency to reload its entire file, which contains billions of information returns. Officials reported that this process can take up to 3 days or more to complete, depending on the file size of the incoming and existing data, and has prevented IRMF from processing and making the W-2 data available for use, as it is received from SSA. Consequently, while IRS had received a total of about 210.9 million W-2s by February 13, it received an additional 3.9 million W-2s between February 13 and 20 that IRS was unable to use in systemic verification before the February 15 refund hold expired. In October 2017, IRS officials told us several reasons why they were not addressing IT limitations. At that time, they said they had discussed various options to make W-2 data available faster, but they had not assessed whether IRMF processing could occur more than once weekly. Further, these officials said IRS developed a plan to modernize IRMF, which would allow for faster processing, but officials told us that this effort is on hold because of competing priorities and funding shortages. These officials also said they had not considered the potential financial benefits of either modifying existing procedures or continuing to pursue modernizing IRMF to process W-2 data more frequently for use in systemic verification. However, in response to our discussions, in November 2017, IRS officials reported they had started to assess the possibility of processing W-2 data on IRMF daily. Specifically, IRS is planning to assess daily processing for the months of January and February during the 2019 filing season when the number of information returns is lower and the file is less time consuming to load. They noted they would not have time to assess their options and make necessary changes to process W-2s daily for the 2018 filing season. As we reported in October 2017, IRS faces challenges with managing its aging legacy systems, and with establishing a process for prioritizing its modernization efforts. IRS’s planned action is consistent with its strategic plan, which includes objectives to strengthen refund fraud prevention by using third-party data and analytics for timely, informed decision making, and to innovate technology systems to support IRS’s business needs. It is also consistent with Standards for Internal Control in the Federal Government, which calls for management to design and implement internal controls within programs based on the related benefits and costs. By taking its planned action to assess processing W-2 data more frequently, IRS would be in a better position to make informed decisions about the future of IRMF and its modernization efforts. Paper W-2 processing. Of the 253 million W-2s that SSA received by December 1, 2017, about 23 million (9 percent) were paper. SSA receives and processes paper W-2s at the Wilkes-Barre Direct Operations Center (WBDOC) in Wilkes-Barre, Pennsylvania. Beginning in October or November of each year, WBDOC programs and tests its systems for transmitting transcribed paper W-2 data. The majority of W- 2s that WBDOC receives are in optical character recognition (OCR) format, which SSA can scan into its systems instead of manually entering the data. Officials stated that W-2s that are not in OCR format require more time and effort to process. This process of developing, testing, scanning, or entering data manually occurs between October and March before WBDOC begins transmitting the paper W-2 data to SSA’s Baltimore facility. Beginning in March, SSA continually transmits all paper and electronic W-2 data to IRS. By law, employers who file 250 or more W-2s are required to file W-2s electronically, while those who file fewer than 250 W-2s may opt to file on paper or electronically. This requirement has not changed since 1989 when employers filed electronically using magnetic media or other machine-readable forms. Since then, technological advancements allow employers to file for free using SSA’s website or other software packages. Consequently, even though not required, by July 28, 2017, SSA had received approximately 69 million electronically filed W-2s from about 4.4 million employers who filed fewer than 250 W-2s. In August 2014, we reported that lowering the electronic W-2 filing requirement would not only contribute to IRS’s ability to verify employment information on tax returns, but it could reduce administrative costs for SSA. According to SSA estimates, the cost to transcribe and process a total of 24.2 million paper W-2s in 2016 was about $13.3 million, or $0.55 per paper W-2. In addition to the cost savings from lowering the electronic filing requirement, as we reported in August 2014, there would be fewer transcription errors and fewer W-2s subject to the longer paper W-2 processing time. In that report, we suggested that Congress should consider providing the Secretary of Treasury with the regulatory authority to lower the requirement for electronic filing of W-2s from 250 returns annually to between 5 to 10 returns, as appropriate. In August 2017, SSA officials estimated that SSA can save between $9.7 and $11.3 million per year if the W-2 paper filing requirement is lowered to 10 or fewer W-2s. These officials reported that this estimate is based on a projected increase of 17.6 million to 20.6 million in electronically filed W-2s and a decrease of paper W-2s by more than two-thirds. Late W-2 filing. IRS began publicizing the change in the W-2 deadline in June 2016. Nevertheless, about 260,000 employers missed the January 31 filing deadline, accounting for late filing of about 7.9 million W-2s in 2017. IRS officials stated that, of the 27,764 employers who had requested an extension for time to file W-2s, as discussed below, IRS approved approximately 6,500 (23.4 percent), which account for approximately 1.1 million W-2s (13.9 percent) of the 7.9 million late filed W-2s. Because IRS has not yet started to assess penalties, it does not yet know how many of these will be subjected to a penalty. Generally, an employer must pay a penalty for failing to file an information return timely or correctly unless an exception applies, such as being granted an extension. However, IRS has changed how it enforces late filing penalties by not mailing some proposed penalty notices to employers who fail to file W-2s timely. For example, IRS mailed all penalty notices to employers who failed to timely file in 2014. However, it did not mail all penalty notices for 2015 and 2016 to employers who failed to file W-2s timely. IRS officials told us that, due to a lack of resources to manage all the penalty cases, they began applying a risk-based selection process to prioritize compliance efforts. Moreover, officials told us they did not collect data to track how many penalty notices IRS did or did not mail for late-filed W-2s, nor the associated penalties IRS proposed to assess for 2015 and 2016. By law, there are some exceptions to the enforcement of penalties on those who fail to file correct information on or before the required filing date, and who fail to include all of the information required to be shown on the return, or include incorrect information, without correction. However, by not mailing all penalty notices, as it did in the past, IRS is not using a tool to collect, at the least cost, the proper amount of tax revenue, is not enhancing or promoting voluntary compliance, and it is missing an opportunity to educate and help the employer understand his or her legal obligations and rights. Additionally, without timely W-2 data to complete pre-refund checks against filed returns, IRS risks releasing fraudulent and noncompliant refunds or burdening legitimate taxpayers whose returns could be cleared with the W-2 data. IRS officials told us that they are monitoring the effect of not mailing all notices on the number of late filings. However, as of November 2017, IRS did not have plans to track and evaluate the extent to which the late W-2s are associated with fraudulent or noncompliant refunds. In addition, IRS does not mail penalty notices until up to a year and a half after the missed deadline. For example, IRS will not assess and mail penalty notices for the approximately 260,000 employers who filed W-2s or other information returns late in 2017 until summer 2018. In part, this is because IRS waits to compile all late-filed information returns, not just W- 2s, some of which are not due until April. Further, late-filing penalty amounts increase incrementally until August 1 for employers who file or correct information returns after the filing deadline. Finally, for 2017, IRS did not finish transcribing and processing the more than 40 million paper- filed information returns until about late September. However, IRS officials have not assessed the options for mailing penalty notices for late W-2s earlier or communicating with the employers earlier in the process. These officials told us that the penalty notice process is consistent with IRS’s enforcement procedures. They further added that mailing multiple penalty notices could increase burden and cost for both the taxpayer and IRS. However, quickly responding to employers that filed late increases the potential for compliance, thereby increasing the availability of W-2 data for systemic verification to detect and prevent fraud and noncompliance. Finally, because it takes up to a year and a half for IRS to identify the late filing and mail the penalty notice, it is possible that the employer could have filed W-2s late two years in a row without IRS notifying him/her of the first late filing. W-2 extensions. In 2017, IRS received 27,764 employer requests for an extension of time to file W-2s, which is substantially higher compared to prior years. IRS officials attribute the increase in extension requests to the new early filing deadline. IRS also began requiring employers to provide reasonable cause and only file their requests on paper. Prior to 2017, employers could file for an automatic 30-day request for extension, electronically or on paper. Because IRS manually processed all requests to determine if the cause was reasonable, IRS did not complete its processing until November 2017. Consequently, employers would not know until after the extended deadline whether IRS granted them the extension. IRS officials told us that, for 2017, they notified about 10,000 employers who requested but were not granted an extension that they would not be penalized this year. Officials also notified these employers that they would be penalized next year under the same conditions. In November 2017, IRS officials said that they are reviewing the extent to which extension requests made in 2017 affected systemic verification. For 2018, IRS plans to continue requiring employers to file extension requests on paper. IRS’s Analyses of the Refund Hold Are Limited, Hampering Their Usefulness for Decision Making and IRS Has Not Explored Additional Uses of Systemic Verification IRS Completed Its Analyses of Potential Changes to the Refund Hold but Did Not Fully Assess the Benefits and Costs and Taxpayer Burden IRS officials examined the effectiveness of the February 15 refund hold by analyzing how systemic verification results differ under several hypothetical scenarios. For example, IRS could extend the refund hold date beyond February 15 when more W-2 data are available for systemic verification before issuing refunds. While the law states that IRS cannot release refunds with EITC or ACTC before February 15, IRS has discretion to continue to hold all refunds until it can verify W-2 data, and has the authority to expand the refund hold to all taxpayers, not just those who claimed EITC or ACTC. Further, the law does not preclude IRS from releasing refunds with EITC or ACTC on a rolling basis after February 15, or in conjunction with an extension of the refund hold. In October 2017, the National Taxpayer Advocate (NTA) told us that she supports potential modifications to the refund hold. In addition, in a June 2017 annual report to Congress, the NTA stated that holding the refunds for all taxpayers longer so that IRS can verify W-2 data could help IRS prevent tax refund fraud before refunds are issued. The NTA also recommended that IRS research the benefits and costs of delaying refund payments. During the 2017 filing season, IRS reviewed limited preliminary systemic verification data to assess potential changes to the February 15 refund hold. In October 2017, IRS completed its final analysis, which included more data on taxpayers who filed after the February 15 refund hold and estimated potential amounts of protected refunds. However, both analyses have limitations. We assessed IRS’s preliminary analysis of the 35.7 million returns filed by all taxpayers (those who claimed EITC or ACTC and those who did not) before February 15 and which were subjected to systemic verification. IRS’s analysis included actual results from systemic verification for these tax returns for each week between February 15 and March 15 after reprocessing the returns when new W-2 data became available. Our assessment of IRS’s preliminary analysis showed that by both extending the refund hold date beyond February 15 and expanding the refund hold to all returns: IRS could have verified more than twice as many returns. By March 15, IRS could have verified wage information for more than twice as many returns before issuing refunds—30.5 million compared to 14.3 million verified by February 15. By only holding returns until February 15, IRS would be unable to verify W-2 data for 20.2 million (56 percent) tax returns, representing $66.6 billion in refunds, before releasing the refunds. IRS could have detected about $3 billion—twice as much–in potential fraud and noncompliance. If IRS had held all taxpayers’ refunds until late February or early March, it could have detected about twice as much potential fraud or noncompliance before issuing refunds, as shown in figure 4, because it had more W-2 data available at that time compared to February 15. For example, if IRS held all taxpayers’ refunds until March 1, it could have identified $2.87 billion compared to $1.47 billion as of February 15, about a 95 percent increase. If IRS held all taxpayers’ refunds until March 8, it could have identified even more in potentially fraudulent or noncompliant refunds before issuing them ($3.18 billion compared to $1.47 billion as of February 15, an increase of about 116 percent). However, these potential fraudulent or noncompliant refunds do not represent potential refunds that IRS could protect. This is because IRS limits the number of cases it selects for review due to the large volume of work this represents and limited staff available. Further, some returns that IRS selects for review are false positives—legitimate tax returns erroneously selected for review. Our Fraud Risk Framework provides a comprehensive set of overarching concepts of fraud risk management and leading practices that serve as a guide for agency managers to use when developing efforts to combat fraud in a strategic, risk-based way. For example, a leading practice in the Fraud Risk Framework emphasizes risk-based preventive activities for strategically managing fraud risk to help avoid a costly and inefficient “pay and chase” model. Additional leading practices call for federal agencies to continuously monitor and evaluate the effectiveness of preventive activities and to consider the benefits and costs of its control activities. Further, key concepts in the Fraud Risk Framework highlight the importance of measuring outcomes to adapt fraud detection and prevention activities. Additionally, we have reported that program evaluation provides agencies with objective information on program effectiveness and efficiency. Program evaluation is necessary to inform and improve IRS’s fraud risk management activities. However, when we compared IRS’s preliminary analysis to the Fraud Risk Framework and program evaluation standards, we found that it was limited in several areas: IRS has not documented an evaluation plan, goals, or strategy related to the refund hold. To ensure an evaluation’s credibility, agencies should develop evaluation plans with clearly defined program goals and researchable evaluation questions. However, IRS did not have documentation detailing an evaluation plan or program goals that includes the purpose of the analysis and the research questions it is assessing. Moreover, IRS plans to continue assessing the effectiveness of the refund hold on systemic verification. In May 2016, we recommended that IRS develop an overall compliance strategy that includes refundable credits, such as EITC and ACTC. In February 2017, IRS reported that it is taking steps to implement this recommendation. However, it is unclear how IRS plans to incorporate the results of its analysis of systemic verification into its overall compliance and fraud risk management strategy. IRS did not determine how many potentially fraudulent or noncompliant refunds it issued before verifying against W-2 data. A key benefit of obtaining W-2 data early in the filing season is to verify that the information matches before issuing the refund. In its preliminary results, IRS reported the number and amount of refunds it identified as potentially fraudulent or noncompliant before issuance only for taxpayers that claimed EITC or ACTC and whose refunds IRS held until February 15. Of the 22.1 million taxpayers who filed before February 15 and did not claim these credits, IRS identified approximately 196,000 returns filed by taxpayers claiming nearly $580 million in refunds as potentially fraudulent or noncompliant. IRS did not report the number of refunds that were issued before IRS had identified them as potentially fraudulent. IRS has not fully assessed the burden on the taxpayers who were subjected to the refund hold date. We have reported that a key concept in tax administration is reducing unnecessary taxpayer burden, which is the direct time and money that taxpayers spend to comply with tax laws, including costs for paid tax preparation. Three economic experts we interviewed cited key factors that IRS could consider in assessing the burden to taxpayers as a result of the refund hold. For example, experts told us that IRS could examine changes in taxpayer behavior, such as waiting to file a return later, or shifting from using Free File to paid preparation that can offer refund- related financial products such as an advance on their refund. These experts also indicated that IRS could compare the amount of fraud or noncompliance that IRS prevented among taxpayers claiming EITC or ACTC against taxpayers who do not claim these credits. All experts we interviewed agreed that more than 1 year of data might be needed to assess short-term and long-term effects of the refund hold on taxpayer behavior and patterns of fraud and noncompliance. IRS officials told us that they have added a question to IRS’s customer satisfaction survey to determine how taxpayers got their information about the refund hold. They have also indicated they are analyzing taxpayer behaviors related to the timing of filing and taxpayers’ use of refund-related financial products. However, IRS has not provided us with the revised survey or its results, nor provided documentation of what is included in the analyses. IRS officials told us that they have limited resources to conduct research and have not completed the work because they are prioritizing other research efforts. The limitations of IRS’s preliminary analysis prevent IRS from fully understanding the effectiveness of systemic verification and refund hold, and hampers IRS’s broader fraud risk management and compliance efforts. IRS officials stated that they did not document an evaluation plan, include key data, determine how many refunds were issued before detecting potential fraud and noncompliance, nor assess taxpayer burden. Without a documented evaluation plan that includes key data to assess the success of preventing fraud and noncompliance before issuing refunds, IRS risks relying on insufficient information to make decisions on potential changes to the refund hold date and those subjected to it. For example, by not assessing taxpayer burden, IRS does not understand how taxpayers are affected by the current hold date or whether extending the hold or expanding it to all taxpayers would increase taxpayer burden. IRS completed its final analysis of the refund hold in October 2017 and provided us with a draft. Based on our initial review, IRS’s findings correspond with those in the preliminary analysis discussed above in that IRS could detect much more potential fraud and noncompliance if it held refunds longer. However, there were key differences between the preliminary analysis and IRS’s final analysis. First, IRS assessed two potential refund hold dates after February 15—February 28 and March 1, when IRS receives the majority of W-2s. Second, IRS included all returns that would be affected by the two extended refund hold dates rather than only those that filed before February 15. Third, IRS estimated the total amount of fraud and noncompliance that it could protect under these two extended refund hold dates. Finally, IRS based its estimates on returns that had completed final processing rather than returns that had not completed processing. In its final analysis, IRS estimated that it could detect about $7.1 billion in potential fraud and noncompliance if it held refunds with EITC or ACTC until March 1, of which it could protect about $533 million. This is about $468 million more than what IRS protected by holding refunds with EITC or ACTC until February 15. Further, IRS estimated that it could have protected $100 million in fraud and noncompliance had it held all taxpayer refunds until February 15—$35 million more than it protected with the current hold and verification process. IRS further estimated that by holding all refunds until March 1, it could protect about $895 million. Various factors account for the differences between what IRS could detect as potential fraud and noncompliance and what it estimated that it could protect. First, IRS limits the number of cases it selects for review due to the large volume of work required to review all returns flagged by systemic verification and other fraud filters and limited staff available. Second, some returns that IRS selects for review are false positives— legitimate tax returns erroneously selected for review—so not all the returns will be confirmed as fraud or noncompliant. In its final analysis, IRS had not addressed the limitations noted above for the preliminary analysis. However, IRS expects to further explore the possibility of holding refunds beyond February 15. IRS also plans to complete additional analyses, including the effect of W-2 extension requests on systemic verification and taxpayers’ use of refund-related financial products. As IRS continues analyzing the effectiveness of the refund hold date on systemic verification, the limitations we outlined above will continue to prevent IRS from fully understanding the effectiveness of systemic verification and refund hold, and hamper IRS’s broader fraud risk management and compliance efforts. IRS Has Not Analyzed the Benefits and Costs of Additional Uses of Early W-2 Data and Systemic Verification As noted, the Fraud Risk Framework emphasizes the use of fraud prevention activities to help federal agencies avoid the costly and inefficient “pay-and-chase” model. However, IRS has not assessed the benefits and costs of additional uses of early W-2 data to prevent fraud and noncompliance before issuing refunds. For example, IRS has not determined the value of using W-2 data to address employment fraud or underreporting prior to issuing refunds. Employment fraud is a type of identity theft refund fraud that occurs when an identity thief uses a taxpayer’s name and Social Security number to obtain a job and claims a refund. Underreporting occurs when a taxpayer underreports income or claims unwarranted deductions or tax credits. With its Automated Underreporter program, which is utilized after the filing season and after refunds have been issued, IRS electronically matches income information reported to IRS by third parties, such as banks and employers, against the information that taxpayers report on their tax returns. With earlier W-2 information, IRS can detect more possible employment fraud or underreporting before issuing refunds. For example, if IRS has two W-2s reporting wage income for a taxpayer, but that taxpayer did not report both on his or her tax return, the taxpayer may have underreported his or her income or could be a victim of employment fraud. IRS officials stated that they are not using systemic verification to review such instances before issuing a refund because it would require them to follow the deficiency process. IRS typically begins this process when it has completed all of its compliance checks later in the filing season when it has most third-party data available for verification. IRS then sends the taxpayer a notice that informs him or her that IRS has proposed an adjustment to taxes owed because the third-party data IRS received does not match what the taxpayer reported on his or her tax return. The notice also informs the taxpayer of his or her right to challenge any resultant tax increase with the U.S. Tax Court. IRS officials told us they do not want to issue the notice earlier because that could encourage taxpayers to file in Tax Court before IRS has completed its review. IRS officials stated that they did not see the potential benefits of taking intermediate steps before sending a notice of deficiency, such as holding the refunds and corresponding with the taxpayer to resolve the discrepancy. However, while IRS had not explored this or other potential uses of W-2 data, IRS officials acknowledged that it would be worthwhile to consider additional opportunities of earlier W-2 data. Earlier availability of W-2s and other information returns can help IRS identify and prevent fraud and noncompliance before issuing refunds. However, without assessing the benefits and costs, IRS does not know the extent to which it can use earlier W-2 data for other pre-refund compliance checks. Conclusions During the 2017 filing season, IRS’s ability to detect and prevent fraud and noncompliance improved because it received significantly more W-2 data earlier and utilized it to verify wage, withholding, and other information on millions of tax refunds. Based on these results, systemic verification shows promise for preventing fraudulent refunds and reducing noncompliance. Nevertheless, the agency faced challenges that limited its success in implementing systemic verification. Similar to taking action to assess the potential for processing more W-2s early in the filing season, IRS can take additional steps to increase the availability of more W-2 data. By not collecting data to track late W-2 filings, IRS could not measure the extent to which late W-2 filings are associated with fraud and noncompliance. Further, by not taking earlier action to improve enforcement of penalties for late W-2 filings, IRS is missing an opportunity to encourage compliance with the W-2 filing deadline and verify more wage information before releasing refunds. As a result, IRS risks releasing fraudulent and noncompliant refunds. We have also previously identified action Congress could take to increase the availability of W-2 data to IRS early in the filing season. In August 2014, we suggested that Congress provide the Secretary of the Treasury with the authority to lower the electronic filing requirement from 250 W-2s to 5 to 10 W-2s. This action would also have the benefit of reducing SSA’s W- 2 paper processing costs by millions of dollars each year. In addition, the February 15 refund hold for EITC and ACTC claims afforded IRS an opportunity to verify return information with early W-2 data before issuing refunds. IRS took steps to collect and assess preliminary data on systemic verification and the refund hold during the filing season. In addition, IRS completed its final analysis that considers different scenarios for holding refunds longer and the potential revenue it could protect. However, IRS’s efforts are not guided by an evaluation plan to assess the results of systemic verification in preventing fraud and noncompliance before issuing refunds. Developing and implementing an evaluation plan that fully assesses the benefits and costs of that hold date would help IRS determine the effectiveness of systemic verification, its fraud risks, and the effect of the refund hold on taxpayer burden. IRS would then be in a better position to modify the refund hold under its existing authority and balance detecting and preventing fraud and noncompliance with taxpayer burden. Further, it is not clear how the analysis informs IRS’s broader fraud risk management efforts and other compliance strategies. Finally, with these data, IRS has the potential to improve tax enforcement in other areas such as for underreporting or employment fraud. While IRS has measures in place to address these issues after paying refunds, taking action before issuing refunds can prevent fraud and noncompliance and save IRS time and resources. Recommendations for Executive Action We are making the following six recommendations to IRS. The Acting Commissioner of Internal Revenue should collect data to track late W-2 filing penalty notices and the extent to which they are associated with fraud and noncompliant returns. (Recommendation 1) The Acting Commissioner of Internal Revenue should assess options for improving enforcement of late W-2 filing penalties, for example, by mailing notices before the next filing deadline. (Recommendation 2) The Acting Commissioner of Internal Revenue should develop an evaluation plan to fully assess the benefits and costs, including taxpayer burden, of modifying the February 15 refund hold, and determine how this effort informs IRS’s overall compliance strategy for refundable tax credits and fraud risk management. (Recommendation 3) Based on the benefits and costs assessment in Recommendation 3, the Acting Commissioner of Internal Revenue should use IRS’s existing authority to modify the refund hold such that it minimizes the risk of releasing fraudulent or noncompliant refunds. (Recommendation 4) The Acting Commissioner of Internal Revenue should assess the benefits and costs of additional uses and applications of W-2 data for pre-refund compliance checks, such as addressing underreporting, employment fraud, and other fraud or noncompliance before issuing refunds. (Recommendation 5) Based on the assessment in Recommendation 5, the Acting Commissioner of Internal Revenue should implement any identified changes to improve pre-refund compliance checks. (Recommendation 6) Agency Comments and Our Evaluation We provided a draft of this product to Treasury and SSA for review and comment. IRS provided written comments, which are summarized below and reproduced in appendix II. SSA responded in writing with no comments (see appendix III). SSA and Treasury provided technical comments, which we incorporated as appropriate. In its written comments, IRS did not state whether it agreed or disagreed with our recommendations, but outlined planned steps to address five of the six recommendations. If implemented as planned, IRS’s proposed actions for recommendations 1, 4, 5 and 6, could meet the intent of the recommendations. However, for the third recommendation to develop an evaluation plan to fully assess the costs and benefits of modifying the February 15 refund hold, it is not clear whether IRS’s planned actions will fully satisfy the recommendation. IRS stated that it would assess and evaluate options for improvements to its refundable tax credits and fraud risk management strategies. However, IRS did not specify whether this evaluation would fully assess the benefits and costs, including taxpayer burden, of modifying the February 15 refund hold. As we reported, a documented evaluation plan that includes key data to assess the success of preventing fraud and noncompliance before issuing refunds will help IRS make better-informed decisions on potential changes to the refund hold date and those subjected to it. This includes, for example, assessing taxpayer burden to understand how taxpayers are affected by the current hold date and whether extending the hold or expanding it to all taxpayers would increase taxpayer burden. Regarding the second recommendation in our draft report to assess options for improving enforcement of late W-2 filing penalties by mailing notices before the next filing deadline, IRS stated that the timing of the receipt of a W-2 account file from SSA and the overall complexity of the process precludes notices from being issued prior to the start of the next filing season. Specifically, IRS noted that it assesses penalties on approximately 40 different types of information returns—in addition to W- 2s—and that the penalty calculation is complex. For W-2s, IRS explained that it receives a reconciled file from SSA in December that identifies those employers that should not be penalized. Finally, IRS noted that issuing penalty notices on a piecemeal basis would burden the taxpayer and potentially lead to erroneous notices. We recognize that there are challenges to issuing penalty notices, or other communications, before the next filing season. However, there are also benefits. As we reported, earlier communication with the employer, whether it includes a penalty assessment or not, increases the potential for compliance, helps taxpayers avoid filing late in the subsequent year, and increases the availability of W-2 data for systemic verification to detect and prevent fraud and noncompliance. However, we continue to believe that assessing the options for improving enforcement of late W-2 filing penalties, such as through earlier communication, would help IRS identify potential opportunities to encourage compliance with the W-2 filing deadline and verify more wage information before releasing refunds. We intended the recommendation to be inclusive of other options beyond mailing notices earlier. As a result, we clarified the recommendation to make mailing notices before the next filing deadline an example. We are sending copies of this report to the appropriate congressional committees, the Acting Commissioner of Internal Revenue, the Acting Commissioner of Social Security, the Secretary of the Treasury, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-9110 or lucasjudyj@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. Key contributors to this report are listed in appendix IV. Appendix I: Objectives, Scope, and Methodology Our objectives in this report were to assess the Internal Revenue Service’s (IRS) performance in detecting fraud and noncompliance using systemic verification, and the Social Security Administration’s (SSA) performance providing timely Form W-2, Wage and Tax Statement (W-2) data to IRS; and the extent to which IRS analyzed the effectiveness of the refund hold on systemic verification as well as opportunities for IRS to apply systemic verification to other efforts to detect fraud and noncompliance. To answer the first objective, we obtained and analyzed IRS documents and data, including documents describing the implementation of IRS’s systemic verification of W-2 data and preliminary systemic verification data on the 2017 filing season, and used this information to determine how IRS used early W-2 data; reviewed the Protecting Americans from Tax Hikes Act of 2015 (the Act) and related tax laws and regulations to understand IRS’s systemic verification matching W-2 data against individual income tax returns affected by the Act (taxpayers claiming the Earned Income Tax Credit (EITC) or the Additional Child Tax Credit (ACTC)), as well as other returns not affected by the Act (those not claiming EITC or ACTC), and statutory requirements for penalties, electronic filing, and authority to hold refunds; reviewed IRS laws, regulations, and policies on penalty assessments for filing W-2s late, IRS data on late W-2s for 2017, and interviewed IRS officials to understand the process for assessing penalties; interviewed officials from IRS’s Wage and Investment division (which is responsible for managing filing season operations) on the challenges in implementing systemic verification, as well as planned improvements; interviewed officials from IRS’s Information Technology Applications Development unit to understand the technological capabilities of IRS’s Information Return Master File and related systems and identify system limitations and improvements. We compared IRS’s actions to IRS’s Strategic Plan, which includes objectives to strengthen refund fraud prevention by using third-party data and analytics for timely, informed decision making, and to innovate technology systems to support IRS’s business needs. We also compared IRS’s actions to the Standards for Internal Control in the Federal Government, which call for management to design and implement internal controls within programs based on the related benefits and costs; observed and interviewed SSA employees processing and transcribing paper W-2s at the SSA’s Wilkes-Barre Direct Operations Center in Wilkes-Barre, Pennsylvania to understand how this work is performed and the time required for completing it; reviewed our prior reports, including reports on the filing season, tax credits, and identity theft, and evaluated IRS’s actions to implement selected prior recommendations; and interviewed SSA managers and staff who oversee and process paper Form W-2 data and transmit the data to IRS, and obtained and analyzed SSA goals, documents, and data, including data on costs for processing paper W-2s. To answer the second objective, we reviewed IRS documents that included internal working group meeting minutes, planning documents, and management reports; assessed IRS’s data for 2017 and its preliminary and final analyses on the systemic verification results and outcomes under different scenarios for the 2017 filing season; compared IRS’s efforts to detect and prevent fraudulent and noncompliant refund payments with leading practices in our A Framework for Managing Fraud Risks in Federal Programs (Fraud Risk Framework) and program evaluation; interviewed officials from IRS’s Wage and Investment division on the benefits and costs of systemic verification and to determine whether IRS conducted an economic analysis of the effects on taxpayers burden as a result of holding all taxpayer’s refunds until February 15; interviewed three economic experts to identify factors that IRS should consider or study following implementation of the Act. We selected economists based on their expertise in the field of tax policy and refundable tax credits, and to ensure variation in perspectives on tax issues. We asked similar questions of each economist and analyzed their comments to identify commonalities. We used these interviews to identify factors that IRS should consider in evaluating the refund hold date and any potential changes to it. The views of the economic experts are not generalizable; and interviewed officials from the Department of the Treasury’s (Treasury) Bureau of the Fiscal Service and Office of Tax Policy about their actions to prepare for releasing a large volume of refunds on February 15 and to determine what analyses, if any, Treasury had conducted on taxpayer burden related to the holding all taxpayer’s refunds until February 15. To assess the reliability of the data we used for this report, we reviewed IRS and SSA reports on W-2 data and IRS reports on systemic verification and its results. We also reviewed IRS reports on the performance of its fraud filters. We examined systemic verification data to identify obvious errors or outliers and assessed potential data limitations that would affect use of the data for assessing performance. We also interviewed IRS officials about their data quality procedures and the reliability and limitations of these data. We determined that the data presented in this report are sufficiently reliable for the purposes of our reporting objectives. We conducted this performance audit from March 2017 to January 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Comments from the Internal Revenue Service Appendix III: Comments from the Social Security Administration Appendix IV: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Joanna Stamatiades, Assistant Director; Erin Saunders Rath, Analyst-in-Charge; Jessica Ard; Mark Canter; Jacqueline Chapin; Jehan Chase; Felisa Garmon; Robert Gebhart; Tom Gilbert; Andrew Howard; Kirsten B. Lauber; Japheth McGee; Paul Middleton; Ed Nannenhorn; Sabine Paul; Bradley Roach; and Robert Robinson made key contributions to this report.

Why GAO Did This Study IRS continues to confront the ongoing problems of identity theft (IDT) refund fraud. The agency estimates that at least $1.68 billion was paid in IDT refund fraud in 2016. To help address this issue, consistent with GAO's prior reporting, the Protecting Americans from Tax Hikes Act of 2015 advanced the deadline for employers to file W-2s to SSA to January 31 (about 1 to 2 months earlier than in prior years). This change allows IRS more time to match wage information to tax returns through systemic verification, and identify any discrepancies before issuing refunds. GAO was asked to assess how well IRS implemented systemic verification. GAO assessed IRS's performance using systemic verification and the extent to which IRS analyzed the effectiveness of the refund hold on this process. GAO analyzed IRS and SSA data and documents, observed SSA's paper W-2 process, and interviewed IRS and SSA officials. GAO compared IRS actions to laws; IRS policies; and standards for internal control, fraud risk management, and program evaluation. What GAO Found Beginning in 2017, as required by law, the Internal Revenue Service (IRS) held all refunds for taxpayers claiming the Earned Income Tax Credit (EITC) or Additional Child Tax Credit (ACTC) until February 15. IRS also took actions to verify wage and other information reported on tax returns before issuing refunds, referred to as systemic verification, but several factors limited its success. IRS received over twice as many (over 214 million) Forms W-2, Wage and Tax Statement (W-2) by February 15 compared to the same time in 2016, and reported that W-2 data were responsible for improving fraud detection and reducing taxpayer burden. However, IRS was unable to verify over half of the returns it held until February 15 before issuing the refunds. For example, IRS received W-2s daily but its information technology systems processed them weekly. In response to GAO's review, IRS reported it is planning to assess options for processing W-2s daily. Also, some employers submit W-2s late, but IRS did not track the extent to which late W-2s are associated with fraud or noncompliance. Further, IRS has not assessed options for enforcing late W-2 penalties earlier. Additionally, about 9 percent (about 23 million) of W-2s were filed on paper, which IRS does not begin to receive from the Social Security Administration (SSA) until March. By law, employers who file 250 or more W-2s are required to file W-2s electronically, while those who file fewer than 250 W-2s may opt to file on paper or electronically. In August 2014, GAO suggested that Congress provide the Secretary of the Treasury with the authority to lower the electronic filing requirement from 250 W-2s to 5 to 10. This action could also have the benefit of reducing SSA's W-2 paper processing costs by $9.7 to $11.3 million per year. These issues reduce IRS's access to timely W-2 data, limiting its ability to prevent fraud and reduce noncompliance before issuing refunds. IRS's preliminary and final analyses of the February 15 refund hold both showed that IRS could have detected significantly more in potential fraud and noncompliance if it held all refunds until late February, when it had more W-2 data available. There are differences between these analyses. For example, the final analysis included more returns and estimated total revenue IRS could protect by extending the refund hold and expanding it to all taxpayers. In that analysis, IRS estimated that it could have protected $100 million in fraud and noncompliance had it held all taxpayer refunds until February 15—$35 million more than it protected by holding refunds with EITC or ACTC. IRS further estimated that moving the refund hold to March 1 for all taxpayers could protect $895 million compared to $533 million if it only held refunds with EITC or ACTC until that date. However, GAO found limitations to IRS's analyses. For example, while IRS has plans to further explore holding refunds longer, it does not have an evaluation plan to assess the effectiveness of the refund hold on systemic verification. Also, IRS did not fully assess the benefits and costs, including taxpayer burden, of the refund hold, nor how its analysis informs its broader fraud risk management or compliance efforts. As a result, IRS does not have sufficient information to inform a decision on potential changes to the refund hold date and those subjected to it. Finally, IRS has not assessed the benefits and costs of expanding systemic verification to use for pre-refund compliance checks in other areas such as income underreporting and employment fraud. Therefore, IRS may be missing opportunities to maximize use of early W-2 data. What GAO Recommends GAO recommends IRS collect data to track late W-2 filing penalties and assess options for earlier enforcement; assess the benefits and costs of using existing authority to hold refunds longer, hold all refunds, or both, and expanding systemic verification to other areas; and take actions based on the assessments. IRS listed steps to respond to 5 of 6 recommendations, but said it could not enforce penalties earlier. GAO recognizes the challenges but clarified that assessing other options would provide benefits, as discussed in the report.

gao_GAO-18-34

gao_GAO-18-34_0

Background VA serves veterans of the U.S. armed forces and provides health, pension, burial, and other benefits. The department’s three operational administrations—VHA, Veterans Benefits Administration, and National Cemetery Administration—operate largely independently from one another. Each has its own contracting authority, though all three also work with national contracting organizations under the Office of Acquisition, Logistics, and Construction for certain types of purchases, such as medical equipment and information technology. VHA, which provides medical care to about 7 million veterans at 170 medical centers, is by far the largest of the three administrations. These medical centers are organized into 18 VISNs, organizations that manage medical centers and associated clinics across a given geographic area. Each VISN is served by a corresponding Network Contracting Office. Figure 1 shows the organizational structure of the procurement function at VA. MSPV-NG Program For over a decade, each of VA’s 170 medical centers used VHA’s legacy MSPV program to order medical supplies, such as bandages and scalpels. Many of those items were purchased using the Federal Supply Schedules, which provided medical centers with a great deal of flexibility. As we reported in 2016, this legacy program, however, prevented VHA from standardizing items used across its medical centers and affected its ability to leverage its buying power to achieve greater cost avoidance. Standardization is a process of narrowing the range of items purchased to meet a given need in order to improve buying power, simplify supply chain management, and provide clinical consistency. For example, a hospital network might find that it purchases 100 varieties of bandages, but might ultimately determine—with input from clinicians—that it can narrow those choices down to 10 varieties to fill most needs, which would provide greater consistency and allow the hospital to negotiate lower prices. In part because the legacy MSPV program limited standardization, VHA decided to transition to a new iteration, called MSPV-NG. VHA launched the MSPV-NG program in December 2016 but allowed a 4-month transition period. After April 2017, medical centers could no longer use the legacy program. MSPV-NG now restricts ordering to a narrow “formulary”—a list of specific items that medical centers are allowed to purchase. VA has had a formulary in place for pharmaceuticals since 1997, and many leading hospital networks rely on a similar formulary approach when it comes to purchasing their own medical supplies. VHA policy requires medical centers to use MSPV-NG—as opposed to other means such as open market purchase card transactions—when purchasing items that are available in the formulary. Figure 2 illustrates the program structure and key participants involved in the transition to MSPV-NG. VA’s primary MSPV-NG program goals are to: Standardize requirements for supply items for greater clinical consistency. Achieve cost avoidance by leveraging VA’s substantial buying power when making competitive awards; VA set a goal of achieving $150 million in cost avoidance in 2016 through a supply chain transformation effort, of which MSPV-NG is a primary part. Achieve greater efficiency in ordering and supply chain management, including a metric of ordering 40 percent of medical centers’ supplies from the MSPV-NG formulary. Involve clinicians in requirements development to ensure uniform clinical review of medical supplies. VHA gave responsibility for developing and implementing MSPV-NG to its Healthcare Commodity Program Executive Office (program office), an organization within VHA’s Procurement and Logistics Office. According to documentation, the program office and SAC, a VA-wide contracting organization, identified several steps to allow for a successful transition to MSPV-NG. These steps included the following: 1. Identify and develop requirements – Determine which types of medical supplies should be made available to medical centers via the MSPV-NG formulary and their key characteristics. The program office was responsible for this aspect of the transition. 2. Award contracts and establish agreements – SAC was responsible for awarding distribution contracts to a select number of prime vendors within certain geographic areas to deliver supplies to medical centers. SAC was also responsible for awarding contracts and establishing agreements with suppliers that provide the products themselves, which set prices for individual items. 3. Implement MSPV-NG at medical centers – MSPV-NG orders are placed by ordering officers—members of the logistics staff at each medical center that are delegated authority by SAC contracting officers to place orders for medical supplies. Each medical center’s most frequently purchased items—referred to as their core list—vary based on the type of care provided, local preferences, and other factors. Leading Practices for Organizational Transformation Efforts We have previously reported that organizational transformations (such as MSPV-NG) require careful planning and implementation to be successful. For instance, one leading practice is for leadership to set clear implementation goals and a timeline to achieve them. Likewise, communicating a strategy and progress to stakeholders—as well as seeking feedback—is a hallmark of successful organizational transformations. We have reported that at the center of any serious change management initiative are the people. Thus, to facilitate success, is to recognize the “people” element and implement strategies to help individuals maximize their full potential in the organization, while simultaneously managing the risk of reduced productivity and effectiveness that often occurs as a result of the changes. Building on the lessons learned from the experiences of large private and public sector organizations, the key practices and implementation steps that we identified in our prior work can help agencies transform their cultures so that they can be more results oriented, customer focused, and collaborative in nature. Standards for Internal Control in the Federal Government also identify related principles, such as the importance of the tone from the top and ensuring that data used in decision-making are reliable. Supply Chain Practices Identified by Selected Leading Hospital Networks Leading hospital networks we spoke with have similar goals to VA in managing their supply chains, including clinical standardization and reduced costs. In managing their supply chain efforts, the leading hospital networks we identified take consistent approaches to drive change and achieve savings. These hospitals reported they analyze their spending to identify items purchased most frequently, and which ones would be the best candidates to standardize first to yield cost savings. These hospitals also acknowledge that this is an iterative process and do not attempt to standardize all categories of medical supplies at a single time, but instead prioritize categories of supplies based on the potential for standardization. The hospitals’ supply chain managers establish consensus with clinicians through early and frequent collaboration on supply chain standardization. These hospitals also continually involve clinicians in determining key supply characteristics and evaluating potential items, understanding that clinician involvement is critical to the success of any effort to standardize their medical supply chain. For example, a supply chain official from one large hospital we spoke with stated that selecting an item that does not meet clinician needs could damage clinician buy-in for future efforts, so they take great care to be thorough in taking clinician input into account. Supply chain officials from these leading hospitals have reported positive results from these efforts, such as increased cost savings and the potential for improved patient care. By tackling a few specific categories at a time and communicating with clinicians on an ongoing basis about the outcomes of these processes and the decisions taken, these hospitals are able to achieve efficiencies, including significant cost savings in some cases, while maintaining buy-in from their clinicians. Figure 3 depicts the key steps that selected hospitals’ supply chain managers reported following when standardizing their medical supply chains, including the critical role of clinicians throughout the process. Emergency Procurements The Federal Acquisition Regulation (FAR) generally requires agencies to contract using full and open competition, but permits contracting without full and open competition in specified circumstances, such as when the agency’s need for supplies or services is of unusual and compelling urgency. The VHA Procurement Manual describes an emergency as a situation—such as response to fires or floods—where delay in award of a contract would result in financial or physical injury to the VA or a veteran. The manual also states that neither a lack of advance planning nor concerns about a need to obligate funds before the end of the fiscal year are valid justifications for an urgent or emergency procurement request. For needs that cannot be met through MSPV-NG, medical centers submit purchase requests to their local VHA contracting office—the Network Contracting Office. The contracting office provides medical centers with expected lead times for various types of procurements, which can be from days to months, depending on the complexity of the requested item. However, if a medical center has an urgent need that must be met more quickly than the expected lead times, the customer submitting the request can identify it as an emergency. The purchase request is entered into two VA data systems, the Integrated Funds Distribution Control Point Activity, Accounting and Procurement and VA’s Electronic Contract Management System (eCMS). The medical center designates the priority level of the request as: 1. Emergency: life threatening cases, emergency physical plant repair, and requires acquisition action within 24 hours; 2. Special: urgent, non-life threatening, and requires acquisition action within 72 hours; and 3. Standard: all other cases and requires acquisition action within 40 days. Incoming requests are screened by Network Contracting Office managers and assigned to individual contracting officers, who must prioritize emergency requests over other pending contract actions. Figure 4 illustrates the typical process for submitting and awarding an emergency procurement. VHA’s Implementation of MSPV-NG Program Has Not Yet Achieved Its Goals VHA’s implementation of the MSPV-NG program—from its initial work to identify a list of supply requirements in 2015, through its roll-out of the formulary to medical centers in December 2016—was not executed in line with leading practices. Despite changes aimed at improving implementation, the agency continues to face challenges that have precluded achievement of program goals. Specifically, VHA lacked a documented program strategy, leadership stability, and workforce capacity for the transition that—if in place—could have facilitated buy-in for the change throughout the organization. Furthermore, the initial requirements development process and tight time frames contributed to ineffective contracting processes. As a result, VHA developed an initial formulary that did not meet the needs of the medical centers. VA made some changes in the second phase of requirements development to address deficiencies identified in the initial roll out, namely by increasing the level of clinical involvement. However, VHA has not yet achieved its goals for utilization and cost avoidance. VA Lacked an Overarching Strategy for Implementing MSPV-NG VA did not document a clear overall strategy for the MSPV-NG program at the start and has not done so to date. According to program office and SAC officials responsible for developing and executing the program, no document existed at the outset of the MSPV-NG program that outlined the overall strategy. About 6 months after our initial requests for a strategy or plan, an official provided us with an October 2015 plan focusing on the mechanics of establishing the MSPV-NG formulary. However, this plan was used only within the VHA Procurement and Logistics Office and had not been approved by VHA or VA leadership. Leading practices for organizational transformation state that agencies must have well-documented plans and strategies for major initiatives (such as MSPV-NG) and communicate them clearly and consistently to all involved—which included VHA headquarters, the SAC, and all 170 medical centers. Without such a strategy, VA could not ensure that all stakeholders understood VHA’s approach for MSPV-NG and worked together in a coordinated manner to achieve program goals. This is also in contrast to the practices of several leading hospital networks we met with, which placed an emphasis on designing and communicating a strategy and governance structure for their medical supply standardization efforts before making any changes to purchasing. If VA continues to move forward with MSPV-NG without an overarching strategy that it communicates to all stakeholders to ensure they understand VHA’s approach for MSPV-NG, VA will continue to face challenges in meeting program goals. Leadership Instability and Staffing Shortages Were Obstacles to Effective Implementation of MSPV- NG Leadership instability and workforce challenges also made it difficult for VA to execute its transition to MSPV-NG. Due to a combination of budget and hiring constraints, and lack of prioritization within VA, the program office, which has primary responsibility for implementing MSPV-NG, has never been fully staffed and has experienced instability in leadership. As of January 2017, 24 of the office’s 40 positions were filled, and program office officials stated that this lack of staff affected their ability to implement certain aspects of the program within the planned time frames. Our work has shown that leadership buy-in is necessary to ensure that major programs like MSPV-NG have the resources and support they need to execute their missions. We have also previously found that leadership must set a tone at the top and demonstrate strong commitment to improve and address key issues. However, leadership of VHA’s Procurement and Logistics Office changed frequently during the implementation of MSPV-NG, and two of its leaders, the Chief Procurement and Logistics Officer and the Deputy Chief Logistics Officer, were serving in an acting capacity. A similar instability in leadership affected the program office itself. Since the inception of MSPV-NG, the program office has had four directors, two of whom were acting and two of whom were fulfilling the director position while performing other collateral duties. For instance, one of the acting MSPV-NG program office directors was on detail from a VISN office to fulfill the position but had to abruptly leave and return to her VISN position due to a federal hiring freeze. Without prioritizing the hiring of the program director position on a permanent basis, this lack of stability could continue to affect execution of MSPV-NG. Moreover, VA’s Chief Acquisition Officer (CAO), whose responsibilities include oversight of VA acquisition programs such as MSPV-NG, is serving in an acting capacity and is not a “non-career employee.” By statute, VA is required to appoint or designate a non-career employee as the agency’s CAO. VA provided information to show that since 2009, VA has designated career employees as “acting” CAOs rather than appointing or designating non-career employees to the CAO position. As we reported in 2012, clear, strong, and effective leadership, including a CAO, is key to an effective acquisition function that can execute complicated procurements like MSPV-NG. By appointing a CAO in a non-acting capacity, VA could improve the effectiveness of its acquisition function. During our 2012 review, VA indicated that it sought to establish an Assistant Secretary for Acquisition, Logistics, and Construction, who would serve as VA’s CAO. In connection with the current review, VA’s Office of General Counsel cited a statutory limitation on the number of assistant secretaries that may be established within VA as the reason it has not established that additional assistant secretary position. VA’s Office of General Counsel indicated that the agency was considering requesting, in the reform plan that VA was required to submit to the Office of Management and Budget in September 2017, a change to the statute that limits the number of VA assistant secretaries. However, subsequently, VA’s Office of General Counsel indicated that the plan will not include such a request. By not appointing or designating a non- career employee as CAO, VA will continue to be noncompliant with the statute. Figure 5 summarizes the history of leadership changes in these positions, which are all currently filled in an acting capacity. Further, according to officials, leadership vacancies at medical centers and competing demands on logistics staff time made implementation of MSPV-NG more challenging at the selected VISNs and medical centers we visited. For instance, longstanding vacancies in the Chief Supply Chain Officer positions existed at one of the VISNs and its medical center that we visited. The VISN-level position was vacant for about 4 years, with Chief Supply Chain Officers from individual medical centers filling in for periods of time, according to the current Chief Supply Chain Officer, who took the position in January 2017. In one medical center within that VISN, the local position was also vacant for several years, according to the current Chief Supply Chain Officer, who took the position in 2016. He stated that he found that the staffing of the office had suffered in the absence of a leader, leaving it poorly-equipped to execute the transition to MSPV-NG. Medical center logistics staff also had several other major transformation efforts to manage alongside the MSPV-NG transition, such as implementing a new system for managing equipment. Several Chief Supply Chain Officers we interviewed stated that these additional demands made it challenging for their staff to implement the MSPV-NG program. The MSPV-NG Initial Requirements Development Process Had Limited Clinician Involvement and Did Not Prioritize Categories of Supplies The MSPV-NG program office initially developed requirements for medical and surgical supply categories—identifying items to include in the formulary—based almost exclusively on prior supply purchases, with limited clinician involvement. The program office concluded in its October 2015 formulary plan that relying on data on previous clinician purchases would be sufficient and that clinician input would not be required for identifying which items to include in the initial formulary. Further, rather than standardizing purchases of specific categories of supplies—such as bandages or scalpels—program officials told us they identified medical and surgical items on which VA had spent $16,000 or more annually and ordered at least 12 times per year, and made this the basis for the formulary. Officials said this analysis initially yielded a list of about 18,000 items, which the program office further refined to about 6,000 items by removing duplicate items or those that were not considered consumable commodities, such as medical equipment. In 2015, the program office also took the lead in developing requirements for these 6,000 items. In documentation, and as confirmed by agency officials, we found that the program office did not solicit input from clinicians for most items and did not prioritize categories of supplies. Instead, the program office relied on historical purchase data to set requirements across medical and surgical categories because officials said they thought this would provide a good representation of medical centers’ needs. This approach to requirement development stood in sharp contrast to those of the leading hospital networks we met with, which relied heavily on clinicians to help drive the standardization process and focused on individual categories of supplies rather than addressing all categories simultaneously. Initial Requirements Development and Tight Time Frames Contributed to Ineffective Contracting Practices for Initial Formulary Based on the requirements developed by the program office, SAC began to issue solicitations for the 6,000 items on the initial formulary in June 2015. From June 2015 to January 2016, medical supply companies responded to only about 30 percent of the solicitations. As a result, according to SAC officials, they conducted outreach and some of these companies told SAC that VHA’s requirements did not appear to be based on clinical input and instead consisted of manufacturer-specific requirements that favored particular products instead of broader descriptions. Furthermore, SAC did not solicit large groups of related items, but rather issued separate solicitations for small groups— consisting of 3 or fewer items—of supply items. This is contrary to industry practices of soliciting large groups of related supplies together. Therefore, according to SAC officials, some medical supply companies told them that submitting responses to SAC’s solicitations required more time and resources than they were willing to commit. By its April 2016 deadline for having 6,000 items on the formulary, SAC had been working on the effort for over a year and had competitively awarded contracts for about 200 items, representing about 3 percent of the items. Without contracts for the items on the formulary in place, VA delayed the launch of the MSPV-NG program until December 2016. To continue the legacy MSPV program through the new launch date, SAC awarded bridge contracts—short-term sole-source contracts—to its legacy prime vendor contractors for a second year. We previously reported that bridge contracts had resulted in higher costs to the government. In part because of these costs, SAC officials stated that VA leadership did not view a third set of bridge contracts for the legacy MSPV program as a viable option. As a result of the pressure not to miss the revised December 2016 deadline, which VA documents we reviewed stated would have been “catastrophic,” SAC abandoned its original goal of using competitive procedures and relied instead on a non-competitive strategy for placing most of the items on the MSPV-NG initial formulary. Starting in August 2016, SAC established 175 limited source blanket purchase agreements with Federal Supply Schedule vendors to complete the initial Phase 1 formulary. While this approach enabled the MSPV- NG program office to establish the formulary more quickly, it did so at the expense of one of the primary goals of the MSPV-NG program— leveraging VA’s buying power to obtain cost avoidance through competition. We previously reported that a senior VA procurement official said VA could save 30 percent, on average, on the prices available under the Federal Supply Schedules when awarding competitive contracts that leveraged VA’s buying power under the legacy MSPV program. The discounts VA obtained from these limited source agreements were generally much less. We reviewed a non-generalizable sample of 10 randomly-selected limited source blanket purchase agreements and found that most items (332 of the 376 items covered by these agreements) were discounted 5 percent or less. Competition is the cornerstone of the acquisition system; its benefits are well established, including saving the taxpayer money. As shown in figure 6, the non- competitive agreements awarded in the last few months before the launch of MSPV-NG accounted for approximately 79 percent of the items on the January 2017 version of the formulary. Initial Formulary and Unclear Communication Contributed to a Lack of Medical Center Buy-In for MSPV-NG Implementation Once VA’s MSPV-NG initial formulary was established in December 2016, each medical center was charged with implementing it. Previously, medical centers had hundreds of thousands of items they could obtain through the legacy MSPV program. In order to transition to the new formulary—consisting of around 6,000 items at launch—the program office directed medical centers to determine if items they had ordered in the past could be fulfilled by the formulary. To do this, each medical center’s Chief Supply Chain Officer—the head of the logistics office—was to review their center’s core list of previously ordered items to try to identify matches on the MSPV-NG formulary in three different categories: 1. Direct matches – For some items, the exact same item a medical center had been purchasing was available in the formulary. Identifying these matches may not necessarily be simple, as the names and identification numbers were not typically the same. 2. Potential clinical equivalents – Many items that were no longer available under the MSPV-NG formulary had close matches on the formulary. However, because these were not exactly the same, work was required to ensure that they were clinically equivalent—in nearly all cases, this required clinician input. Clinical Product Review Committees at each medical center, which are comprised of clinicians and others, are responsible for approving new supplies before they are introduced to a medical center. 3. Items without matches – Finally, there were some items that medical centers had been purchasing for which logistics staff were not able to identify a clinical equivalent in the MSPV-NG formulary. In these cases, logistics staff sought non-MSPV methods of obtaining the same items they had previously purchased—usually via purchase card transactions and, in a few cases, via requests to their local contracting office to award new contracts for the items. Figure 7 shows the typical process for identifying MSPV-NG matches for core list items at individual VA medical centers, as described by logistics officials at the selected medical centers. According to logistics officials we spoke with, the MSPV-NG formulary matching process was challenging for the selected medical centers, and they had varying levels of success, in part, due to incomplete guidance from the program office. The MSPV-NG program office provided some guidance, including a tool for identifying direct matches, but three of the Chief Supply Chain Officers at the selected medical centers stated that they did not find it very helpful, in part, because it only included matches for the highest-volume items. Based on our discussions with the MSPV- NG program office and selected medical centers, as well as our review of communications provided to medical centers, the program office provided various emails and held conference calls, but did not provide complete guidance to summarize the steps medical centers should take to execute the matching process. Without complete guidance, each selected VISN and medical center approached the process somewhat differently. One medical center devoted a great deal of effort to matching items early on, had completed its review, and determined its purchasing strategy for nearly all core list items before the transition period was complete. Others devoted less attention to this and planned instead to rely on purchase cards to continue buying the same items they had purchased under the legacy MSPV program, which works against VA’s goal of leveraging buying power through MSPV-NG. The amount of clinician input on the matching process varied among medical centers in our review, in part, because the various communications from the program office did not provide complete information on how to involve clinicians and Clinical Product Review Committees at medical centers. While the program office asked medical centers to involve clinicians, it did not specify a process for how to do so, and centers were left to develop their own approaches. For example, in one selected VISN, the Deputy Chief Medical Officer became involved with the logistics office coordination effort and obtained active participation from clinicians at each medical center, who formed working groups to review potential clinical equivalent matches. In other VISNs and medical centers, there was little concerted effort to involve clinicians at this stage of the process, and only a few clinical equivalent items were reviewed and matched with clinical input. Without effective matching to the formulary, VA cannot achieve the MSPV-NG utilization rates it needs to meet the program’s goals. Without complete guidance, these centers may be unable to effectively match their core lists to the MSPV-NG formulary and, thus, increase their utilization of it. The MSPV-NG formulary also continued to change while the medical centers were working to match their core list items, which made the process more challenging. Several clinicians and logistics staff at the medical centers we visited expressed frustration about the frequency by which items were being added and deleted on the formulary and the impact it had on their purchasing strategies. Our analysis found that in April 2017, 690 items were added to the formulary, but, in June, 628 items were deleted. These medical center officials also noted that they had not received any communications from the program office or SAC regarding why items were being added and deleted, and were unsure why the changes were taking place. SAC and MSPV-NG program office officials stated that these continuing changes stemmed from several factors, including elimination of duplicate items from multiple vendors and addition of other items identified as necessary by VHA or medical centers. In some cases, medical center officials told us that that they were less willing to expend effort on the matching process because the formulary was a moving target. Without visibility into or an understanding of the criteria used by the program office on its process for adding or removing items on the formulary, medical centers will likely continue to face challenges in matching their items to the formulary. See Table 1 for the number of items added and deleted from the formulary from January to July 2017. Many medical centers were unable to find direct matches or substitutes for a substantial number of items on their core lists, which negatively impacted utilization rates for the initial formulary. In October 2015, the program office estimated that the items on the initial formulary would meet 80 percent or more of the medical centers’ needs. However, according to SAC, as of June 2017, only about a third of the items on the initial version of the formulary were being ordered in any significant quantity by medical centers, indicating that many items on the formulary may not be those that are needed by medical centers. Senior VHA acquisition officials attributed this mismatch to shortcomings in their initial requirements development process as well as with VA’s purchase data. VA set out a target that medical centers would order 40 percent of their supplies from the MSPV-NG formulary, but utilization rates are below this target with a nationwide average utilization rate across medical centers of about 24 percent as of May 2017. Instead of fully using MSPV-NG, the selected medical centers are purchasing many items through other means, such as purchase cards or new contracts awarded by their local contracting office, in part, because they said the formulary does not meet their needs. These approaches run counter to the goals of the MSPV-NG program and result in VA not making the best use of taxpayer dollars. Specifically, Chief Supply Chain Officers—who are responsible for managing the ordering and stocking of medical supplies at the six selected medical centers—told us that many items they needed were not included in the MSPV-NG formulary. As discussed above, the difficult transition process also created a lack of clinician desire to find substitutes on the formulary. As such, we found that these six medical centers generally fell below VA’s stated utilization target that medical centers order 40 percent of their items from the MSPV-NG formulary. As shown in figure 8, among the six selected medical centers we reviewed, one met the target, while the remaining five were below 25 percent utilization. The one facility that met the target, Hampton VA Medical Center, is categorized by VA as a smaller, less complex facility, and had fewer items to match, which could contribute to its higher utilization. The utilization rate is VA’s primary metric for the success of MSPV-NG— broad usage of the formulary is necessary for VA to meet its goals of more efficient supply purchasing, standardization, and cost avoidance. Utilization is calculated by dividing the purchases made via MSPV-NG by the total purchases under the medical supply budget category. This is the same metric used under the legacy MSPV program, and most medical centers were meeting the 40 percent target prior to the transition to MSPV-NG. Officials stated that the current metric does not provide enough information and, as a result, VHA is in the process of preparing a new metric to more precisely assess MSPV-NG use and effectiveness, and has begun conducting routine surveys of its medical centers to obtain their feedback on MSPV-NG. Greater utilization of MSPV-NG is essential to VA achieving the cost avoidance goal of $150 million for its supply chain transformation effort. Under the legacy MSPV program, the National Acquisition Center tracked cost avoidance achieved by comparing prices for competitively-awarded MSPV supply contracts with prices available elsewhere. However, VHA officials stated that they are not currently tracking cost avoidance related specifically to MSPV-NG. VHA officials told us they plan to use a new cost avoidance metric that compares total supply spending for VHA as a whole across fiscal years. This new metric, however, does not measure whether cost savings are being achieved specifically through MSPV-NG. Officials stated the broader metric was more useful than measuring cost avoidance specific to MSPV-NG. VA’s practices are in contrast with those of the leading hospitals we met with, which maintain detailed, item-level data on cost avoidance and use them to inform future supply requirements and contracting. These hospitals we interviewed reported substantial cost savings from their standardization efforts. For example, the director of supply chain management at one leading hospital network stated that it achieved a goal of $100 million in cost savings on medical supplies in the first 2 years of their standardization effort, and an additional $35 million annually in the several years since. This hospital achieved these results despite its purchasing power being less than VA’s. Without calculating cost avoidance attributable to MSPV-NG, VHA cannot assess whether the program is meeting its goals, nor can it use cost avoidance data to guide future MSPV-NG requirement development and contracting strategy efforts. VA Encouraged Greater Clinical Involvement in the Second Phase of Requirements Development, but Faced Further Staffing and Schedule Constraints In Phase 2 of MSPV-NG, the program office has taken some steps to incorporate greater clinical involvement in subsequent requirement development, but both its requirements development and SAC’s contracting efforts have been hampered by staffing and schedule constraints. Work on Phase 2 began while medical centers were implementing Phase 1 and beginning to order from the MSPV-NG formulary. Figure 9 shows key dates in the concurrent requirements development, contracting, and implementation processes for Phases 1 and 2. In the fall of 2016, the program office began to establish panels of clinicians—including physicians, surgeons, and nurses working in the medical centers—to serve on MSPV-NG integrated product teams (IPT) assigned to the task of developing updated requirements for the second phase of the formulary. The IPTs were to review categories of medical supplies such as operating room surgical supplies and patient exam room instruments and supplies. According to VA officials and our analysis, this revised approach was based on a recognition that more robust mechanisms were needed for incorporating clinician input, in part, because VA had sought information on best practices from leading hospital networks, and because of shortcomings with the Phase 1 requirements that became apparent in the contracting process. Similar to the analysis performed in support of the initial formulary, the MSPV-NG program office analyzed updated data on medical center supply purchases to generate a list that had grown from the 6,000 items established for the initial formulary to a new total of about 9,900 items for these new IPTs to review. The program office set a March 2017 deadline to complete this second, IPT-based phase of requirements development—VHA ultimately met this compressed timeline, but in a rushed manner that limited the impact of the clinical involvement. Program officials said they had difficulty recruiting clinicians to participate, and the program office’s first IPTs were not established until the fall of 2016. In December 2016, slightly more than half (20 of the 38) of the IPTs had begun their work to review items and develop updated requirements. Many of the remaining IPTs were still looking for additional clinicians to participate. Program officials said they received assistance from the Assistant Deputy Under Secretary for Health for Administrative Operations in December 2016. According to program officials, this involvement proved critical in successfully recruiting staff to participate in some of the remaining IPTs, which were then able to make progress in reviewing each item in the formulary. However, the program office did not provide training for the IPTs on how to carry out their work until late January 2017, about 2 months before the IPTs were scheduled to complete the development of all medical and surgical requirements. Further, staff on the IPTs had to complete their responsibilities while simultaneously managing their regular workload as physicians, surgeons, or nurses. By early March 2017, the IPTs still had about 4,200 of the 9,900 items to review. Faced with meeting this unrealistic time frame, the MSPV-NG program office had 9 IPT members travel to one location—with an additional 10 members participating virtually—to meet for 5 days to review the remaining items. Members told us that this time pressure limited the extent to which they were able to pursue the goal of standardizing supplies, and that their review ended up being more of a data validation exercise than a standardization review. In addition, the program office attempted to pursue standardization across all supply categories rather than those with the greatest potential for standardization and cost avoidance and continues to lack a strategy for doing so going forward. Standards for Internal Control in the Federal Government state that management should define what is to be achieved and who is to achieve it, how it will be achieved, and the time frames for achievement. In addition, this approach runs counter to how leading hospitals standardize their supply chains by tackling individual categories one at a time and obtaining deep clinician involvement. Without a strategy for how best to prioritize these items by category for future phases of the requirement development process, these IPTs will be limited in fully contributing to VHA’s goals of more efficient supply purchasing, standardization, and cost avoidance. VA Plans to Replace Limited Source Blanket Purchase Agreements, but Faces Challenges Doing So Before They Expire SAC’s ongoing Phase 2 contracting effort also faces an unrealistic schedule. The SAC plans to replace the existing Phase 1 limited source agreements with competitive awards based on the Phase 2 requirements generated by the IPTs, but it may not be able to keep up with expiring agreements. Because they were made on a non-competitive basis, the Phase 1 limited source blanket purchase agreements were established for a period of one year. In order to keep the full formulary available, the SAC director said his staff must award several hundred contracts before the Phase 1 limited source agreements expire later this year. However, the SAC director stated that doing so will be difficult because his staff must award between 200 to 250 contracts in a 3-month period from the end of September 2017 through December 2017. To adhere to this ambitious schedule, each of the 15 contracting staff on the MSPV-NG team would need to award between 13 to 17 contracts within 3 months, equaling one contract per staff member every 5 to 6 days, which is significantly faster than SAC’s typical pace. SAC officials acknowledged that it is unlikely that they will be able to award the 200 to 250 contracts by the time the existing limited source agreements expire. According to SAC officials, they are in the process of hiring more staff to deal with the increased workload. Further, the SAC division director told us that they cancelled all outstanding Phase 2 solicitations in September 2017 due to low response rates, protests from service-disabled veteran-owned small businesses, and changes in overall MSPV-NG strategy. SAC is still assessing alternative approaches, which poses additional challenges for replacing expiring agreements by December 2017. For cases where limited source agreements expire without new contracts in place, SAC officials said they intend to use a different type of agreement called a distribution and pricing agreement as a stopgap. They stated that the use of these agreements with suppliers who have existing limited source agreements would prevent items from falling off the formulary. However, like BPAs, the agreements are not contracts—the supplier informally agrees to continue to sell its products to VA at the same price and terms. SAC officials stated that VA has not used these types of agreements previously, and they pose a risk in that the supplier is not required to perform and VA has no remedy if the supplier opts to end the agreement or raise the price. These agreements also do not allow VA to achieve its goal of achieving greater cost avoidance through supply standardization and competitive contracts. Despite the unrealistic time frames and the risks of the stopgap approach, VA has not developed a plan for how to mitigate these risks, established an achievable schedule for making the competitive Phase 2 contract awards, or prioritized the various categories of supplies. Establishing such a plan would help ensure that VA is better positioned to mitigate risks and prioritize supply categories that are most likely to yield cost avoidance. VA Is Revising Its Approach to MSPV-NG VA is currently revising its approach to MSPV-NG requirement development to adopt a model that focuses on clinician-driven sourcing, a key step that leading hospital networks reported following in standardizing their medical supply chains. The MSPV-NG program office continues to refine its strategy for requirement development and is seeking greater clinician involvement in future requirement development efforts, which it refers to as clinician-driven sourcing. For example, program officials said they plan to involve VHA’s national clinical program offices—groups of clinicians at VHA that provide national policy and leadership within their clinical specialty—to obtain greater buy-in from senior clinical leaders and to implement a more structured approach for identifying clinicians willing to serve on integrated product teams. This approach, if implemented effectively, could mitigate some of the prior challenges in recruiting clinicians to participate. However, these efforts are in their early stages, and the MSPV-NG program office has not outlined whether or how it will use input from these clinical groups to prioritize its requirements development and standardization efforts. Without input from these national clinical program offices, VA will continue to be challenged to focus on supply categories that offer the best opportunity for standardization and cost avoidance. Senior VHA and MSPV-NG program officials also told us each VA medical center was expected to use a standing committee, known as the Clinical Product Review Committee, to review new items to include on the formulary and to evaluate opportunities to streamline the formulary through standardization. This approach will likely require additional effort on the part of the MSPV-NG program office to implement, as some centers’ clinicians said the Clinical Product Review Committees were not operating as intended. VA is also exploring major changes in its contracting strategy for MSPV- NG. Specifically, MSPV-NG program office and SAC officials plan to replace the current contract and formulary process with a new contract where the vendor would not only provide distribution services, but also develop the formulary. In October 2017, VA sought information from industry on their capabilities to support such a program. VA stated that its target completion date for this new MSPV-NG contracting strategy is December 2018. To date, VA has provided only limited details on this potential new approach, thus, we cannot assess whether it has the potential to address the shortcomings with the current MSPV-NG approach described in this report. Frequent Use of Emergency Procurements Impacts VHA’s Ability to Strategically Manage Its Acquisitions Some emergencies are to be expected, as VHA operates one of the largest health care systems in the country. However, VHA designated a substantial number of its procurements in fiscal year 2016 as emergencies, and we found that it frequently uses emergency procurements to buy routine supplies and on-going services that do not warrant the emergency designation defined in VHA guidance. Among the 18 contract actions we reviewed from three VISNs, we found instances of emergency procurements caused by shortcomings in planning, funding, and communication. These emergency procurements strain the capacity of VA’s acquisition workforce and put the government at risk of paying more than it should for goods and services. Emergency Procurements Account for Almost Two Billion Dollars’ Worth of VHA Contract Obligations Based on our analysis of VA data, we found that emergency procurements accounted for approximately 20 percent of VHA’s overall contract actions in fiscal year 2016, with obligations totaling about $1.9 billion. As shown in figure 10, we found that the percentage of requests designated as emergencies varied across the 19 VISNs. Some Selected Procurements Identified as Emergencies Stemmed from Lack of Planning, Communication Problems, and Other Factors We selected a non-generalizable sample of 18 contract actions designated by customers as emergencies. Most of these contracts were not awarded on a competitive basis, and half cited the unusual and compelling urgency exception to full and open competition. Table 2 shows instances in which the 18 contract actions were awarded without competition, those that cited unusual and compelling urgency as the basis for use of non-competitive procedures, and our observations on the main contributing factor to designating these procurements as emergencies. Additional information on each of the contributing factors follows. Planning Challenges VHA guidance specifies that neither a lack of acquisition advance planning nor concerns about a need to obligate funds before the end of the fiscal year are valid justifications for an urgent or emergency procurement request. However, among our selected contract actions, lack of planning by customers was a principal contributing cause, leading to 7 of the 18 contract actions being procured as emergencies, resulting in some non-competitive awards to the incumbent vendor for the same requirement. For instance, one medical center procured medical gas on an emergency basis through consecutive non-competitive contracts. The initial contract was terminated because the company was not licensed by the state where services were being provided, which led to a 3-month emergency contract being awarded to a different vendor. This was followed by a series of non-competitive bridge contracts to that incumbent vendor over a 3-year period. In another case, a medical center routinely procured custom surgical packs through consecutive emergency sole- source purchase orders. The contracting officer’s representative told us the medical center may be paying more for custom surgical packs ordered on an emergency basis than it would under a competitive, long- term contract. Funding Issues Funding uncertainty also contributed to three awards being designated as emergencies. For example, one medical center submitted an emergency request to outsource patient laundry due to funding uncertainties for repairs of on-site, VA-owned and operated laundry equipment. The contracting officer’s representative stated that the VISN could not provide funds to repair the equipment, leading to a series of last-minute emergency requests, a few months at a time, for contracted patient laundry services to prevent a gap in service. At another VISN, a large amount of funding became available late in the fiscal year, which led to an emergency request to purchase postage to ensure the funding was spent before it expired at the end of that fiscal year. The contracting office issued an order for $890,000 worth of metered mail postage, which medical center staff told us would cover 1 to 2 years of usage. Lack of Communication We found that shortcomings in communication between customers and contracting offices also contributed to eight awards made on an emergency basis for routine items. For one of the contracts in our review, a medical center resubmitted a request in January 2016 to purchase equipment for a new operating room that had previously been submitted as a standard request months earlier. However, the contracting officer’s representative at the medical center told us that no action was taken by the contracting office, and he did not receive a response for 6 months. The medical center then upgraded the request to an emergency since the operating room was scheduled to open in June 2016. The contracting officer’s representative noted that delays procuring the equipment past the scheduled opening date would delay the opening of the new operating room and possibly result in the rescheduling or cancelling of procedures, affecting patient care. After the order was upgraded to an emergency, the equipment was ultimately delivered before the operating room was opened. In another case, an inventory manager routinely submitted emergency purchase requests for cardiac catheters as a strategy to manage stock levels. The reason he cited was that he was uncertain how long it would take the contracting office to fulfill standard requests. He stated that the contracting office’s time frames for standard orders are unpredictable, and more consistent communication about the expected delivery date of any given order would reduce his need to place emergency orders. He noted that being able to plan around delivery dates was important for maintaining stock at designated levels for the various types of catheters used in the cardiology department. Figure 11 shows a medical center stock room and designated stock levels for one type of catheter. The “L” indicates the standard stock level, and “R” indicates the level of stock at which refill is needed. Ordering officers use these levels to inform when they should place orders. Overreliance on Emergency Procurements Can Drive Up Costs and Overtax the Workforce In addition to being contrary to VHA guidance, overuse of emergency procurement requests has negative effects on the overall operation of VA’s procurement system. In reviewing the 18 selected contracts, we identified two primary effects—the potential for increased costs and increased burden on the contracting workforce that could take resources away from other important efforts. Increased Costs As noted above, half of the contract actions we reviewed (9 out of 18) cited unusual and compelling urgency as the basis for the use of non- competitive procedures. When unusual and compelling urgency exists, an agency may limit competition to the firms it reasonably believes can perform the work in the time available. In all nine cases, however, there was no competition at all, which puts the government at risk of paying more than it should for goods and services. Promoting competition— even in a limited form—increases the potential for quality goods and services at a lower price. We have previously reported that competition in contracting is a critical tool for achieving the best return on investment and that it can improve contractor performance and promote accountability for results. Burden on the Contracting Workforce Emergency procurement requests must be processed quickly, and contracting officers have limited ability to question the validity of an emergency request. Nevertheless, many of the contracting officials we spoke with that had responsibility for our 18 selected contracts told us they generally communicate directly with the requestor to clarify the requirement and assess the nature of the request. As stated in the VHA procurement manual, contracting officers generally must process emergencies within 5 days or less. However, the manual acknowledges that different Network Contracting Offices assign different time frames to priority categories. For instance, officials from all three selected Network Contracting Offices told us they generally process emergencies immediately. Several contracting officials we interviewed stated that, because they do not have clinical expertise, they infrequently question the medical center staff customer about whether their request is truly an emergency. Even if they work with customers to reach a compromise, such as purchasing a smaller quantity to fill just the immediate need, emergencies still require immediate attention and result in deprioritizing other tasks. The impact on the contracting officer workload can be exacerbated by low staffing levels. For example, none of the three Network Contracting Offices we visited were staffed to their authorized levels. Table 3 shows the number of emergency actions processed by each selected Network Contracting Office in fiscal year 2016, along with staff levels. We have previously reported that when contracting officers process frequent and emergency small-dollar transactions, it reduces their ability to plan ahead and take a strategic view of procurement needs. Several of the VA contracting officials we spoke with noted that regularly processing emergency contracts and extensions affects their ability to work on bigger-picture efforts, some of which would reduce workload. For instance, one contracting officer stated that awarding emergency contract extensions has prevented him from competitively awarding more than 40 lab contracts. In these cases, the contracting officer stated that he instead extended the period of performance of the non-competitive contracts to the incumbent vendors. In addition, emergency contracts are generally awarded for short periods of time—often 1 year or less—while competitive contracts often have terms of 5 years. According to some contracting officers we spoke with, this can result in contracting officers spending much of their time tending to a large number of short-term contracts, instead of a smaller number of fully-competed contracts with longer periods of performance. Better Planning and Management of Contracting Strategies May Help Reduce Emergencies We found that greater planning and coordination between medical center and contracting staff can help to leverage VA’s buying power by employing principles of strategic sourcing—a process that moves away from numerous individual procurements to a broader aggregate approach—and thereby reducing the need for emergencies. For example, inventory managers responsible for two of the selected cardiac catheter contracts in our sample stated that managing catheter inventory was difficult because of the unpredictability of the needs, the high cost of the items, and the long turnaround times from their respective contracting offices. As a result, they had to place frequent emergency orders to keep stock at safe levels. One inventory manager noted, however, that there is no longer a need to place emergency orders for catheters because the SAC has since put in place a purchasing agreement that enabled her to place orders directly, without requiring involvement from the contracting office. In addition to reducing contracting office workload, the supply technician said this agreement greatly reduced the amount of work required to place an order and allowed her to more effectively maintain her inventory with short and predictable turnaround times. She also stated that the agreement protected against the frequent price increases she experienced when purchasing the catheters on the open market through the contracting office. The agreement also reduced workload for the local VISN contracting office. In analyzing eCMS data on awards from fiscal years 2014 through 2016, we identified several types of goods and services that were repeatedly purchased on an emergency basis through stand-alone contract actions. This suggests there may be additional opportunities, at both the VISN and national levels, to reduce emergencies by making supplies and services available through more efficient, competitively-awarded contract vehicles. In addition to reducing burden on logistics and contracting staff, reviewing existing spending to find opportunities to leverage buying power is also in line with strategic sourcing best practices. MSPV-NG is one such contracting mechanism for procuring routine supplies, and a more strategic approach to developing requirements for the formulary could help avoid some emergency procurements. Our analysis of VA eCMS data found that many awards designated as emergencies were for medical-surgical items, some of which could likely be purchased through MSPV-NG. Figure 12 shows the number of medical-surgical procurements designated as emergencies within each VISN in fiscal year 2016. Within our sample of 18 contract actions, we found several instances of reoccurring emergency procurements for medical-surgical supplies, such as custom surgical packs and catheters. Procuring routine supplies on an emergency basis defeats the objectives of MSPV-NG to leverage VA’s large buying power and make the process of ordering supplies more efficient and transparent. However, while data on emergency procurements are available, VHA’s Procurement and Logistics Office does not currently analyze this data to identify items frequently purchased on an emergency basis to determine whether such items could be referred to SAC to be added to the MSPV-NG formulary. In addition, local VISN Network Contracting Offices have also not used available data on emergency purchases to identify items frequently purchased on an emergency basis. Steps by VHA’s Procurement and Logistics Office and individual VISN contracting offices to review such data and identify opportunities for leveraging MSPV-NG or other national contracts could help VA achieve greater efficiency. Purchasing medical supplies through individual emergency contract actions is much less efficient than using MSPV-NG; moreover, by making numerous individual procurements at the local level and not leveraging its aggregate buying power, VA is paying more for items than it needs to. Conclusions Any major organizational change requires a solid strategic plan that is communicated with stakeholders, stable leadership, and stakeholder involvement and buy-in. VHA was missing all of these elements when it rolled out the MSPV-NG program, which presented obstacles to effective implementation and buy-in and affected the program’s ability to meet its goals. Moving forward, without an overall strategy that is communicated to all stakeholders and enhanced leadership stability, VHA will likely continue to face these challenges. In addition, in the initial requirements development process, VHA relied on prior purchase data—rather than clinician input—and did not prioritize categories of medical supplies, both of which veered from practices employed by leading hospital networks. Once the initial formulary was established, medical centers faced challenges matching supply items to the formulary and took varying approaches, in part, due to incomplete guidance on key aspects of the process and frequent changes in the items on the formulary. Providing complete guidance and communicating the criteria and processes for adding or removing items from the formulary would help centers more effectively match items to the formulary, thereby increasing utilization, which as of May 2017 was below VA’s established target. Further, because it does not calculate cost avoidance attributable to MSPV-NG, VA cannot accurately measure the extent to which the program is contributing to its overall cost avoidance goal. VA made changes during the second phase of requirements development, in particular to encourage greater clinician involvement. However, the program faces an unrealistic contracting schedule and has not yet developed a plan for how to manage or mitigate the associated risks. Establishing such a plan is essential for risk mitigation, and supply category prioritization could help VA target those categories most likely to yield cost avoidance. In addition, while the program is planning to involve national clinical program offices to obtain greater clinician buy-in, it has not outlined whether or how it will use input from these groups to prioritize its requirements development efforts. Without such input, VA will continue to face challenges focusing on those supply categories that offer the best opportunity for standardization and cost avoidance. Further, VA is considering another major change in its MSPV program in which the prime vendor may absorb some of the work currently conducted by SAC. However, VA may face challenges in this new approach until it addresses the existing shortcomings in the MSPV-NG program, such as the absence of a documented overall strategy, insufficient clinician involvement in the requirements development process, and lack of medical center buy-in. Meanwhile, among the 18 contract actions we reviewed, we found shortcomings in planning and communication that led to medical centers’ overreliance on emergency procurements to obtain routine goods and services—some of which could be made available via MSPV-NG— bypassing effective contracting practices like competition. These emergency procurements can be a particular drain on resources, especially those of contracting officers who must respond immediately to fulfill emergency orders. Identifying opportunities to more strategically purchase frequently purchased goods and services—both at the local levels and nationwide through the MSPV-NG program—could help minimize these workforce challenges and minimize costs. Recommendations for Executive Action We are making 10 recommendations to VA. The Director of the MSPV-NG program office should, with input from the Strategic Acquisition Center (SAC), develop, document, and communicate to stakeholders an overarching strategy for the program, including how the program office will prioritize categories of supplies for future phases of requirement development and contracting. (Recommendation 1) The VHA Chief Procurement and Logistics Officer should take steps to prioritize the hiring of the MSPV-NG program office’s director position on a permanent basis. (Recommendation 2) The Secretary of Veterans Affairs should assign the role of Chief Acquisition Officer to a non-career employee, in line with statute. (Recommendation 3) The Director of the MSPV-NG program office should provide complete guidance to medical centers for matching equivalent supply items, which could include defining the roles of clinicians and local Clinical Product Review Committees. (Recommendation 4) The Director of the MSPV-NG program office should, with input from SAC, communicate to medical centers the criteria and processes for adding or removing items from the formulary. (Recommendation 5) The VHA Chief Procurement and Logistics Officer, in coordination with SAC, should calculate cost avoidance achieved by MSPV-NG on an ongoing basis. (Recommendation 6) The MSPV-NG program office and SAC should establish a plan for how to mitigate the potential risk of gaps in contract coverage while SAC is still working to make competitive Phase 2 awards, which could include prioritizing supply categories that are most likely to yield cost avoidance. (Recommendation 7) The VHA Chief Procurement and Logistics Officer should use input from national clinical program offices to prioritize its MSPV-NG requirements development and standardization efforts beyond Phase 2 to focus on supply categories that offer the best opportunity for standardization and cost avoidance. (Recommendation 8) The VHA Chief Procurement and Logistics Officer should direct VISN Network Contracting Offices to work with medical centers to identify any opportunities to more strategically purchase goods and services frequently purchased on an emergency basis. For example, offices could do this by analyzing existing data. (Recommendation 9) VHA Chief Procurement and Logistics Officer should analyze data on items that are frequently purchased on an emergency basis, determine whether such items are suitable to be added to the MSPV-NG formulary, and work with SAC to make any suitable items available via MSPV-NG. (Recommendation 10) Agency Comments and Our Evaluation We provided a draft of this report to the Department of Veterans Affairs for review and comment. VA provided written comments on a draft of this report. In its written comments, reprinted in appendix II, VA concurred with all of our 10 recommendations. In its response to our recommendation that VA assign the role of Chief Acquisition Officer to a non-career employee, as required by statute, VA stated that it is unable to implement the recommendation without congressional action and requested closure of the recommendation. We asked VA officials what congressional action they believe is necessary to follow the recommendation. The officials told us they believe the CAO position should be assigned to an assistant secretary, but that the number of assistant secretaries within VA is limited by statute. We decline to close this recommendation. VA should assign the role of CAO to a non-career employee, as required by statute. If VA maintains its view that it cannot meet this requirement without congressional action, then VA should request the specific congressional action that VA believes is necessary. VA provided technical comments on the draft report, which we incorporated as appropriate. As agreed with your office, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies to the appropriate congressional committees, the Secretary of Veterans Affairs, and other interested parties. In addition, the report will be available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions concerning this report, please contact me at (202) 512-4841 or by email at oakleys@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix III. Appendix I: Objectives, Scope, and Methodology You requested that we examine the Department of Veterans Affairs’ (VA) transition to the Medical Surgical Prime Vendor-Next Generation (MSPV- NG) program and the extent to which the department contracts for good and services on an emergency basis. This report addresses the extent to which: (1) VA’s implementation of MSPV-NG was effective in meeting program goals, and (2) Veterans Health Administration (VHA) awards contracts on an emergency basis for routine supplies and ongoing services, and what impact, if any, these awards have on VHA’s acquisition function. To review the extent to which implementation of MSPV-NG was effective, we reviewed policy and guidance related to the program. We obtained and analyzed the MSPV-NG program’s formulary development plan, which explained the program’s rationale for pursuing its initial requirements development approach. We also obtained and reviewed additional program documentation, including communications to medical centers and other stakeholders, briefings, and training and tools provided to medical centers. We interviewed leaders in the VHA Procurement and Logistics Office and Healthcare Commodity Program Executive Office (the program office for MSPV-NG), as well as other staff involved in planning and executing aspects of MSPV-NG. We also interviewed VA’s Chief Acquisition Officer during the development of MSPV-NG, cognizant Office of General Counsel staff, and others regarding the program. We also interviewed supply chain managers from four leading hospital networks regarding their medical supply management practices. We selected the hospital networks because they were identified by an industry study as having leading supply chain practices. During interviews, we asked each of these supply chain managers a standard set of questions about processes followed to standardize their hospital networks’ supply chain. VA had also identified two of these hospital networks as having leading supply chain practices and used the industry study to identify these hospital networks. We used this information from the leading hospital networks to compare the key steps—identified by each of the four hospital networks—followed in standardizing their medical supply chains to those steps that VA followed when implementing the MSPV-NG program. We also confirmed these key steps with the leading hospital networks. We conducted site visits at a non-generalizable selection of three Veterans Integrated Service Networks (VISNs), and two medical centers within each selected VISN: VISN 6: Durham, North Carolina Durham, North Carolina VA Medical Center Hampton, Virginia VA Medical Center VISN 8: St. Petersburg, Florida Tampa, Florida VA Medical Center Gainesville, Florida VA Medical Center VISN 22: Long Beach, California Long Beach, California VA Medical Center San Diego, California VA Medical Center The VISNs were selected primarily based on highest total contract obligations in fiscal years 2014 through 2016 and representation of multiple geographic areas and prime vendor contractors. The first site visit to VISN 22 was also chosen based on the rollout schedule for the graphical user interface, an IT system related to MSPV-NG. The final site visit to VISN 6 was also chosen as the VISN with the highest percentage of contract actions designated as emergencies over the fiscal year 2014 through 2016 period. The selected medical centers in each VISN were chosen based on our review of VA Electronic Contract Management System (eCMS) data on emergency procurements within each VISN (see below) and geographic proximity to the VISN office location. At each selected VISN, we interviewed the Chief Supply Chain Officer and other members of leadership. At medical centers in each selected VISN, we met with the Chief Supply Chain Officer, ordering officers, other logistics staff, clinicians involved in the MSPV-NG transition, and on-site representatives of the prime vendor contractors. We evaluated MSPV-NG program office status briefings and integrated product team training briefings, which documented the planned role of clinicians in the Phase 2 requirements development process. We interviewed VHA Procurement and Logistics Office leadership, other MSPV-NG program office staff, and integrated product team managers and clinicians about the evolution of the program office’s requirements development approach, including the role of clinicians in preparing item descriptions and evaluating items. Three integrated product teams were selected for interviews based on those that covered the greatest number of items, as well as for diversity of types of medical supplies. We also met with members of additional integrated product teams during site visits to the selected medical centers. We obtained and analyzed the Strategic Acquisition Center’s acquisition strategy for MSPV-NG supply contracts and discussed its evolution with the Center’s acquisition staff. We analyzed the MSPV-NG formulary as of January 2017 to determine what acquisition instrument was used to add a particular item to the formulary, how the cumulative total of items by award type changed from fiscal year 2014 to fiscal year 2017, and when certain MSPV-NG items would be removed from the formulary because the underlying acquisition instrument had expired. We also analyzed the contents of the formulary monthly from January to July 2017 to determine the number of items added and deleted each month. We determined that the MSPV-NG formulary data were sufficiently reliable for the purposes of our reporting objectives. For the formulary data, we corroborated the supplier’s name, award number, award type, and the award’s effective and expiration dates with data in the Federal Procurement Data System- Next Generation. We were also able to corroborate the total number of items on the January 2017 MSPV-NG formulary through other documentation, such as program briefings. To determine the level of discounts obtained by the MSPV-NG program office, we randomly selected 10 limited source blanket purchase agreements. We reviewed each agreement and compared the price for each item on the supplier’s price list with the item’s Federal Supply Schedule price. We obtained and analyzed the current MSPV-NG indefinite delivery, indefinite quantity solicitations and the Defense Logistics Agency’s documentation on distribution and pricing agreements. We also reviewed related prior GAO reports and relevant parts of the Federal Acquisition Regulation. We obtained information on the metrics used by VA to assess the performance of MSPV-NG, primarily the utilization metric, which is calculated by VA based on budget object code spending data from the financial system and MSPV-NG spending data. We obtained data on the performance of the six selected medical centers for May 2017 and July 2017. We also interviewed officials responsible for maintaining this data to gather information on processes, accuracy, and completeness, as well as on planned changes in the metric. We found the utilization metric data to be sufficiently reliable for our purposes. To assess the extent to which VA has awarded contracts on an emergency basis for routine supplies and ongoing services, and the effect on VA’s acquisition function, we obtained and analyzed VA and VHA policy and guidance documents, reviewed relevant parts of the Federal Acquisition Regulation, and reviewed prior GAO reports. We obtained eCMS data for fiscal years 2014 through 2016, and analyzed these data to determine the number of actions designated by customers as emergencies, the percentage of actions designated as emergencies in each VISN, and the total obligations attributed to these actions. We also calculated the number and value of all actions designated as emergencies in selected Product and Service Codes related to medical supplies and services for fiscal year 2016. We determined that these eCMS data were sufficiently reliable for the purposes of determining the extent of emergency procurements by reviewing information on system controls and conducting validation of data, including tracing selected information to source documents for the contracts that we selected. We selected a non-generalizable sample of 18 contracts from the three selected VISNs. The selection was based primarily on: contracts designated by the customer as emergencies in eCMS data; use of the term “emergency” or “urgent” in the description field; high dollar value; and Product and Service Codes for services and medical supplies. We obtained and reviewed the contract files for each of the selected contracts, which are also stored in eCMS, including signed awards, limited competition justifications, work statements, and other documents. We compared key information, such as extent of competition, against data reported in eCMS. We interviewed the requesters—in most cases the contracting officer’s representative—for all selected contracts. We also visited Network Contracting Offices for each of the three selected VISNs and interviewed leadership at each location, as well as the contracting officials responsible for each selected contract. Finally, we met with a Strategic Acquisition Center contracting officer to discuss a related contract award. We conducted this performance audit from November 2016 to November 2017 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Comments from the Department of Veterans Affairs Appendix III: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the individual named above, Lisa Gardner, Assistant Director; Emily Bond; Matthew T. Crosby; Lorraine Ettaro; Michael Grogan; Jeff Hartnett; Katherine Lenane; Teague Lyons; Roxanna Sun; and Colleen Taylor made key contributions to this report.

Why GAO Did This Study VA medical centers spend hundreds of millions of dollars annually on medical supplies and services. In December 2016, VA instituted a major change in how it purchases medical supplies—the MSPV-NG program—to gain effectiveness and efficiencies. GAO was asked to examine VA's transition to the MSPV-NG program and its use of emergency procurements. This report assesses the extent to which (1) VA's implementation of MSPV-NG was effective in meeting program goals, and (2) VA awards contracts on an emergency basis. GAO analyzed VA's MSPV-NG requirements development and contracting processes, and identified key supply chain practices cited by four leading hospital networks. GAO also reviewed a non-generalizable sample of 18 contracts designated in VA's database as emergency procurements with high dollar values; and met with contracting, logistics, and clinical officials at 6 medical centers, selected based on high dollar contract obligations in fiscal years 2014-2016 and geographic representation. What GAO Found The Department of Veterans Affairs (VA) established the Medical Surgical Prime Vendor-Next Generation (MSPV-NG) program to provide an efficient, cost-effective way for its facilities to order supplies, but its initial implementation was flawed, lacked an overarching strategy, stable leadership, and sufficient workforce that could have facilitated medical center buy-in. VA developed requirements for a broad range of MSPV-NG items with limited clinical input. As a result, the program has not met medical centers' needs, and usage remains far below VA's 40 percent target. VA also established cost avoidance as a goal for MSPV-NG, but currently only has a metric in place to measure broader supply chain cost avoidance, not savings specific to MSPV-NG. Also, starting in June 2015, VA planned to award competitive contracts for MSPV-NG items, but instead, 79 percent were added using non-competitive agreements. (See figure.) This was done primarily to meet VA's December 2016 deadline to establish the formulary, the list of items available for purchase through MSPV-NG. The roll-out of MSPV-NG ran counter to practices of leading hospitals that GAO spoke with, which highlighted key steps, such as prioritizing supply categories and obtaining continuing clinician input to guide decision-making. VA has taken steps to address some deficiencies identified in the first phase of implementation and is considering a new approach for this program. However, until VA addresses the existing shortcomings in the MSPV-NG program, such as the lack of medical center buy-in, it will face challenges in meeting its goals. Medical centers often rely on emergency procurements to obtain routine goods and services—some of which could be made available at lower cost via MSPV-NG. Sixteen of the 18 contracts in GAO's sample were not competed, which puts the government at risk of paying more. For instance, one medical center procured medical gas on an emergency basis through consecutive non-competitive contracts over a 3-year period. VA policy clearly defines emergency actions; however, inefficiencies in planning, funding, and communication at the medical centers contributed to emergency procurements, resulting in the contracting officers quickly awarding contracts with no competition. What GAO Recommends GAO is making 10 recommendations, including that VA expand clinician input in requirements development, calculate MSPV-NG cost avoidance, establish a plan for awarding future competitive contracts, and identify opportunities to strategically procure supplies and services frequently purchased on an emergency basis. VA agreed with GAO's recommendations.

gao_GAO-19-219

gao_GAO-19-219_0

Background According to its strategic plan for 2017 through 2022, the MEP program aims to strengthen and empower U.S. manufacturers by providing them with the information and tools to improve productivity, assure consistent quality, and accelerate the transfer of manufacturing technology into production processes and new products. MEP centers do not all offer the same services; however, across the network, their services span areas such as the following: Lean services. These services help manufacturers implement tools and practices to incorporate “lean” manufacturing principles, which involve producing more with existing resources through eliminating and reducing incidental work or non-value-added activities. Quality services. These services help manufacturers implement management systems to achieve a defined industry-specific or general quality certification or standard. Growth services. These services provide manufacturers with the tools and methods to identify and target opportunities to develop new products, markets, services, or customers. Technology/product development services. These services help manufacturers identify, develop, and diffuse technology and new products. Workforce services. These services help manufacturers recruit, retain, or develop human resources. Some centers provide services directly to manufacturers, and others, to varying extents, use external consultants to provide services. In fiscal year 2018, the 51 MEP centers served 8,425 manufacturers encompassing a variety of manufacturing subsectors (see fig. 1). To receive federal financial assistance from NIST, MEP centers must match the federal contribution with a nonfederal contribution. MEP centers provide their nonfederal contributions through various means, such as fees collected from manufacturers for services provided or in the form of cash or in-kind contributions from other sources, such as state or local governments, trade associations, or community colleges. MEP centers may receive nonfederal resources in any of those forms in excess of the amount needed to match the federal contribution. Prior to the 2017 AICA cost share adjustment, we, NIST, and others reported on issues associated with the cost share structure for the MEP program. For example: In April 2011, we reported that MEP centers identified positive and negative effects of the cost share structure in place at the time. Positive effects of the cost share structure included encouraging MEP centers to leverage resources and emphasize services relevant to manufacturers, and negative effects included MEP centers spending more time and effort seeking cost share matching funds and focusing more on larger clients that could pay higher fees and less on rural clients. In July 2013, NIST analyzed the cost share structure and found that it provided MEP centers with incentives to make strategic and operational decisions based largely on which services generated revenue rather than on which services manufacturers needed to be competitive. NIST recommended several criteria upon which to base the MEP program’s cost share, such as encouraging delivery of innovative services, providing financial stability, and enabling the program to adapt quickly to changing economic conditions and the needs of small and medium-sized manufacturers. NIST requested that the MEP Advisory Board review this analysis and provide recommendations on how best to structure the cost share requirement to provide for the long-term sustainability of the program. In October 2013, the MEP Advisory Board responded to a request in NIST’s July 2013 report with a letter to the NIST Director largely echoing the findings of the earlier reports—for example, that the cost share structure in place at the time made it more difficult to serve smaller and rural clients and drove centers to focus on larger manufacturers that could pay fees. The MEP Advisory Board recommended, among other things, adjusting the cost share ratio to 1:1. In 2014, we reported on NIST’s spending on the MEP program and found that NIST’s financial assistance to MEP centers did not take into account variations across service areas in the demand for program services and the cost of providing services. We recommended that the Secretary of Commerce revise the program’s cooperative agreements to account for such variations. Subsequently, from 2014 through 2017, NIST undertook a system-wide recompetition of MEP centers’ cooperative agreements to better align center funding levels with the national distribution of manufacturing activity and cost of providing services. As a result, NIST recompeted most MEP centers’ cooperative agreements and reduced the number of centers to 51, with a single center in each state and Puerto Rico. Additionally, the recompetition provided for a new minimum annual funding level of $500,000 per center (previously eight centers were below this mark) and nearly $20 million more in federal financial assistance for 34 of the centers. According to NIST’s 2017 congressional budget request, recompetition would increase the capacity and capability of the MEP centers to help small and medium-sized manufacturers, including very small manufacturers—those with fewer than 20 employees—and rural manufacturers. Most Centers Reported that the Cost Share Adjustment Has Helped Them Better Serve Manufacturers, but Some Officials Noted that the Impact Is Hard to Measure In response to our survey, most MEP centers reported that the AICA cost share adjustment has helped them serve manufacturers, but some center officials indicated that the impact is hard to measure. Specifically, most MEP centers we surveyed reported that the AICA’s adjustment of the cost share to 1:1 for the life of a center’s cooperative agreement has increased their financial stability and helped them serve very small and rural manufacturers. According to the survey results, centers run by nonprofit organizations reported greater impacts of the cost share adjustment than those run by states or universities. In follow-up interviews, some MEP center officials indicated that the impact is difficult to measure because of other recent changes that have also impacted their ability to serve manufacturers. Most Centers Reported that the Cost Share Adjustment Has Increased Financial Stability and Enhanced Their Ability to Serve Very Small and Rural Manufacturers In response to our survey, most of the 51 MEP centers reported that the cost share adjustment has had a positive impact on their finances, particularly by increasing their financial stability. Specifically, in their responses to an open-ended question on the effect of the cost share adjustment on the overall financial resources to support center operations, 44 centers provided examples of how the adjustment has generally helped them in areas such as improving center services (23 centers), better serving underserved manufacturers (17 centers), improving collaboration with partners (10 centers), improving planning and financial stability (10 centers), and improving ability to secure funding (10 centers). In responses to a separate question about the impact of the cost share adjustment, 41 centers indicated that the adjustment has provided a more stable financial outlook. Centers noted that in the past, meeting the 2:1 cost share often meant diverting their focus from serving manufacturers to generating and documenting revenue. Some centers provided the following examples of how the financial stability provided by the 1:1 cost share has helped them: One center stated that its staff now spend less time accounting for the hundreds of small transactions used to count toward the 2:1 cost share and can now focus their time on managing the program. One center stated that its budget is now less complicated and center staff are now less distracted by having to generate matching funds. One center stated that before the cost share adjustment, it could not plan on growing its capabilities after the third year of the cooperative agreement because of the anticipated impact of increased cost share requirements. The center noted that since the cost share adjustment, it can continue to plan for growth and has modified its strategic plan to reflect this shift. With a decreased focus on generating revenue, some MEP centers reported that they are now better able to serve manufacturers, particularly very small and rural manufacturers. Overall, 47 of the 51 MEP centers (92 percent) reported that the cost share adjustment has helped them serve manufacturers to a moderate or greater extent. In particular, in response to a question asking if MEP centers experienced certain changes as a result of the cost share adjustment, 43 (84 percent) reported conducting more work with very small manufacturers, and 39 (76 percent) reported conducting more work in rural areas. MEP centers reported that the cost share adjustment has allowed them to take a number of specific actions to serve manufacturers, such as conducting additional outreach (46 of 51), providing new services (45 of 51), offering a greater quantity of existing services (40 of 51), offering training events (39 of 51), and providing services at reduced cost (28 of 51). In follow-up interviews, officials from eight of the nine MEP centers we contacted stated that the cost share change has either already helped or should help them serve underserved manufacturers. These MEP center officials provided the following examples: One center official said that the cost share adjustment has allowed the center to donate time to help manufacturers that could not afford to pay the fees for the services provided. One center official said that the cost share adjustment could provide the financial stability to hire an additional staff person to serve rural parts of the state that were underserved before the adjustment. One center official said that the cost share adjustment has allowed the center to provide new services that it was not able to provide prior to the adjustment because the center struggled to meet its cost share requirement. For example, the center expanded its work to help manufacturers with Food and Drug Administration requirements pursuant to the FDA Food Safety Modernization Act. One center official stated that the cost share adjustment provided the center a strong financial basis upon which to begin offering Manufacturing 4.0 services throughout the state. Centers Run by Nonprofit Organizations Reported Experiencing Impacts to a Greater Extent than Centers Run by States or Universities Our analysis of survey results indicates that MEP centers run by nonprofit organizations reported impacts from the AICA cost share adjustment to a greater extent than centers run by states or universities. For instance, 22 of 26 centers (85 percent) run by nonprofits reported that the cost share adjustment has to a great or very great extent helped them serve manufacturers, compared to 14 of 25 centers (56 percent) run by states and universities. As table 1 shows, a greater percentage of nonprofit centers reported experiencing certain changes, such as an increase in center staff and the development of stronger partnerships, as a result of the cost share adjustment compared to centers run by states and universities. Officials from MEP centers run by states and universities stated that their centers are often directly funded by a state agency or educational institution and already enjoyed some degree of financial stability, which is why they generally reported fewer changes from the cost share adjustment compared to centers run by nonprofits. In a follow-up interview with the operations director of a MEP center run by a state agency, the operations director told us that one advantage of being funded by the state is that, even prior to the adjustment, the center had a steady source of income to help meet its cost share. In response to an open-ended survey question, one university-run MEP center noted that being part of a university provided access to professional services, support systems, and a network of resources that would not otherwise be available at an affordable rate. In responding to another open-ended question on the effect of the cost share adjustment on financial resources to support center operations, another university-run MEP center noted that the AICA adjustment has not resulted in significant changes to the center’s financial resources but could put some of its university funding at risk in the future. In a follow-up interview with the director of this MEP center, she told us that in a university setting her center competes against other university priorities for grant funding and being on a 1:1 cost share puts the center on a less competitive footing against other candidates because the center will no longer need additional university grant funding to meet a higher cost share ratio in the later years of its cooperative agreement. Center Officials Noted that Other Factors Have Impacted Their Ability to Serve Manufacturers, Making It Hard to Measure the Impact of the Cost Share Adjustment In survey responses and follow-up interviews, MEP center officials noted that a number of factors have impacted their ability to serve manufacturers in recent years. For example, in response to an open- ended survey question, centers provided the following as possible factors other than the cost share adjustment that could have impacted their operations: the strength of the overall economy of the nation or of the state in which they are located (19 centers), budgetary or political stability in their state (e.g., stability of state funding) (19 centers), and NIST’s recompetition of nearly all MEP centers’ cooperative agreements between 2014 and 2017 (10 centers). According to several MEP centers we surveyed or officials we interviewed, it is difficult to identify the impacts of the cost share adjustment because of the other factors that have also impacted MEP center operations. For example, in its survey response, one MEP center noted that it would not be easy to isolate the impact of the cost share adjustment from the impact of factors such as the recent recompetition that doubled the center’s federal financial assistance, new leadership at the center, and an improving economy and a tighter labor market that may have resulted in more companies needing the center’s services. In our follow-up interviews, some MEP center officials said that the recompetition, in particular, makes it difficult to isolate the effect of the cost share adjustment. Officials from several MEP centers we contacted cited effects of the recompetition, such as increased baseline funding, resetting of the cost share to 1:1, center leadership changes, and consolidation of centers within states, as reasons why it would be hard to separate the effects of the recompetition from those of the cost share adjustment. For certain centers, the impact of the cost share adjustment was clearer because they did not undergo recompetition, which meant that their cost share had not been reset to 1:1 through that process. Seven MEP centers were not included in the recompetition process that NIST began in 2014 because their cooperative agreements had recently been recompeted (i.e., within 2 years before 2014). Four of these seven “legacy” MEP centers were at or past the third year of their cooperative agreements and, as a result, were at a greater than 1:1 cost share ratio when the AICA was enacted in 2017. These four centers reported that the AICA’s cost share adjustment was helpful in the following ways: One center wrote that having to generate more matching contributions during its fourth year in operation coincided with a drop in its performance that continued until the 2017 cost share adjustment. This center said the cost share adjustment allowed it to devote additional resources to maintaining its services to manufacturers. One center wrote that it was already scaling back its plans to expand manufacturer engagement by the second and third years of its cooperative agreement in anticipation of the higher cost share ratios that would start in the fourth year of operation. This center noted that following the 2017 cost share adjustment, it revised its strategic plan to focus on growing its capabilities instead of scaling them back. One center wrote that moving to the 1:1 cost share helped it increase its focus on service delivery to clients with less concern for cost matching. One center wrote that the 2:1 cost share incentivized a focus on larger manufacturers to meet the cost share requirement. Following the cost share adjustment, the center is now able to develop new services for small and very small manufacturers. Should the cost share structure revert to what it was before the 2017 adjustment, most of the 51 MEP centers that we surveyed stated that they likely would be less able to serve manufacturers, particularly very small and rural manufacturers. In response to an open-ended survey question on the effect of changing the cost share requirement back to what it was before enactment of the AICA, 45 of the 47 MEP centers that responded to this question wrote that such a change would generally reduce their ability to serve manufacturers by causing them to do one or more of the following: shift to higher-revenue clients and services (23 centers), reduce center services and staff (21 centers), seek new revenue sources (11 centers), reduce staff (10 centers), reduce ability to collaborate with partners (7 centers), and increase fees (7 centers). NIST Data Show Some Changes in Centers’ Finances and Activities, but the Changes Generally Predate the Cost Share Adjustment Our analysis of NIST data indicates that there have been some changes in MEP centers’ finances and activities since the 2017 AICA cost share adjustment. However, these changes generally began around the time NIST recompeted the centers’ cooperative agreements, before the enactment of the AICA, and cannot necessarily be linked to the cost share adjustment. NIST data on funding for the MEP centers show that, from fiscal year 2017 to fiscal year 2018, the amount of federal assistance to the MEP centers increased and funds reported by MEP centers to meet cost share requirements decreased. However, these changes generally began around fiscal year 2013. As figure 2 shows, during the period from fiscal year 2013 through fiscal year 2018, federal assistance to MEP centers increased from about $81 million to $116 million while MEP centers’ reported nonfederal contributions decreased from approximately $195 million to $135 million. The centers’ reported nonfederal contributions generally decreased across all three of their primary sources of revenue—program income, cash contributions, and in-kind contributions. Specifically, the amount of program income centers reported to meet their cost share requirement decreased from approximately $95 million in fiscal year 2013 to $71 million in fiscal year 2018. Reported cash and in-kind contributions decreased from approximately $100 million in fiscal year 2013 to $65 million in fiscal year 2018. In particular, the MEP centers reported a substantial decrease in in-kind contributions over this time period, from approximately $25 million in fiscal year 2013 to $5 million in fiscal year 2018. Based on our analysis of NIST data, the overall changes in center financing—that is, the changes in both federal assistance and reported nonfederal contributions—were influenced by NIST’s recompetition of nearly all MEP centers’ cooperative agreements. When NIST recompeted the MEP centers’ agreements, it increased the level of federal assistance for 34 of the 51 MEP centers, constituting an overall increase in base funding amounts for federal assistance from about $90 million before recompetition began in fiscal year 2014 to about $110 million after the recompetition process was complete. Additionally, as MEP centers’ cooperative agreements were recompeted, the centers’ cost share was reset to 1:1, and the centers’ reported nonfederal contributions began to decrease. As figure 3 shows, in fiscal years 2013 and 2014, before the new cooperative agreements began taking effect, most MEP centers operated under a 2:1 cost share. After fiscal year 2014, the number of MEP centers operating under a 1:1 cost share began to increase. With enactment of the AICA, all MEP centers operated under a 1:1 cost share in fiscal year 2017. NIST data show that there also may have been some changes in MEP centers’ activities since the 2017 cost share adjustment. Our analysis of NIST data indicated that from fiscal year 2017 to fiscal year 2018, the total number of manufacturers MEP centers reported serving increased from approximately 8,000 to 8,400, very small manufacturers MEP centers reported serving increased from approximately 2,600 to 2,700, and rural manufacturers MEP centers reported serving increased from approximately 1,500 to 1,600. As with the changes in MEP centers’ finances, the changes in the numbers of total manufacturers and very small manufacturers these centers reported serving generally began before the cost share adjustment. When we analyzed NIST’s data, we found that the total number of manufacturers and the number of very small manufacturers served began increasing around fiscal year 2014, when NIST started recompeting centers’ cooperative agreements, and this increase continued through fiscal year 2018. The overall direction of the change in the number of rural manufacturers served during this period was mixed. Specifically, the number of rural manufacturers centers reported serving increased from fiscal year 2014 to fiscal year 2015, then decreased through fiscal year 2017, and then increased in fiscal year 2018. NIST officials, like MEP center officials, said that it may not be possible to separate the effects of the AICA cost share adjustment from the effects of the recompetition. NIST officials stated that a longer time span would be needed to identify trends in the manufacturers served by MEP centers; however, even then, confounding factors, such as overall economic conditions, could continue to make it difficult to analyze and isolate the effect of the AICA’s cost share adjustment. Looking forward, NIST officials said one impact of the 2017 cost share adjustment is that it will help sustain recent increases in the number of very small and rural manufacturers served by MEP centers. In addition, establishing a link between changes in MEP centers’ finances and activities and the cost share adjustment is difficult not only because the changes generally predated the cost share adjustment, but also because MEP centers likely underreport certain data to NIST. Specifically: Financial data underreporting. NIST officials stated that because MEP centers are not required to report all of their nonfederal resources in excess of the nonfederal contributions required to meet their cost share, the amount of resources available to centers is likely underreported. According to NIST officials, NIST’s Grants Management Division policy provides that centers will generally be held accountable for any amounts that they opt to pledge in excess of the 1:1 cost share. According to NIST officials, centers are thus operating rationally and legally in pledging and reporting only the amount needed to meet their nonfederal contribution for their cost share match. Activity data underreporting. NIST officials said that because MEP centers are not required to report activity data on manufacturers served if the services provided used nonfederal resources that were not directly related to meeting the MEP centers’ cost share, certain activity data, such as the number of rural manufacturers served, is likely underreported. During their discussions with some MEP centers leading up to the 2017 AICA cost share adjustment, the NIST officials learned that some centers were not reporting activity data on manufacturers served if the services provided used nonfederal resources that were not directly related to meeting the MEP centers’ cost share. An official with one such center provided us with information indicating that the number of rural manufacturers served in fiscal year 2016 was about 36 percent more than the number the center reported to NIST. According to NIST officials, centers are not obligated to report activity data on manufacturers served if those activities are not directly related to funds used to meet the MEP centers’ cost share requirements. Because of this underreporting, NIST officials stated that the amount of nonfederal resources in excess of the nonfederal contributions required to meet the cost share, as well as the total number of manufacturers served and the number of very small and rural manufacturers served, are likely higher that what centers reported to NIST. Moreover, the officials noted that, because of the recompetition and the 2017 AICA cost share adjustment, they believe that the extent of MEP centers’ underreporting may have increased in recent years as more centers began operating under a 1:1 cost share ratio. Agency Comments We provided a draft of this report for review and comment to the Secretary of Commerce. NIST provided technical comments, which we incorporated as appropriate. NIST’s comments also included some comments of a more general nature. For example, NIST highlighted the impact that the recompetition had on MEP centers. NIST also noted that while it cannot directly attribute recent increases in the number of manufacturers served to the AICA cost share adjustment, it believes the AICA cost share adjustment has fundamentally allowed MEP centers to deliver more value to clients rather than tie up resources in fundraising. We are sending copies of this report to the appropriate congressional committees, the Secretary of Commerce, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff members have any questions about this report, please contact me at (202) 512-3841 or neumannj@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix II. Appendix I: Objectives, Scope, and Methodology The objectives of our review were to describe (1) Manufacturing Extension Partnership (MEP) centers’ views regarding the extent to which the recent cost share adjustment has helped them serve manufacturers and (2) the extent to which National Institute of Standards and Technology (NIST) data show impacts of the cost share adjustment on centers’ finances and activities. MEP Centers’ Views To describe MEP centers’ views regarding the extent to which the 2017 American Innovation and Competitiveness Act (AICA) cost share adjustment has helped them serve manufacturers, we sent a survey to all 51 MEP centers and received a response from every center. We administered this survey in July and August 2018. Because this survey was not a sample survey, there are no sampling errors. As part of developing this survey, we conducted pretests over the telephone with four MEP centers to ensure that the questions were understandable, that the data collected are uniform and usable, and that the survey would place minimal burden on center officials. We pretested our survey with the MEP centers for California, Kentucky, Virginia, and Washington. Using data on MEP center characteristics provided by NIST, we selected these centers to reflect a range of characteristics in the following categories, among others: the number of manufacturers in the state, the type of MEP organization (i.e., whether the center is run by a nonprofit, state agency, or university), and NIST’s classification of the state as urban or rural. We made changes to the content and format of the survey based on the feedback we received. In the survey, we asked the MEP centers about the effects of the cost share adjustment, both experienced and anticipated, using several different types of questions. For example: We asked the MEP centers whether the cost share adjustment has resulted in or will likely result in changes such as an increase in center staff; an increased use of contractors; an increase in overall financial resources to support the centers’ operations; a more stable financial outlook; or the ability to develop stronger partnerships, conduct more work in rural areas, or conduct more work with very small manufacturers (fewer than 20 employees). If centers responded in the affirmative, we asked whether those changes have already allowed the center to take specific actions such as conducting additional outreach to manufacturers, providing new services to manufacturers, offering manufacturers a greater quantity of existing services, providing services to manufacturers at reduced cost, or offering training events to manufacturers. We asked questions that allowed the MEP centers to identify the extent to which they had experienced a change. For example, we asked, “Whether or not there were any specific types of manufacturers that may have previously underutilized the center’s services, to what extent has changing the federal/nonfederal cost share to 1:1 for future years for all centers regardless of when they began operating helped the center serve manufacturers overall?” The centers could select one of the following responses: very great extent, great extent, moderate extent, some extent, little or no extent, don’t know. We asked open-ended questions to gain additional understanding about the effect of the cost share adjustment, including the following: What, if any, other factors might contribute to the changes or lack of changes identified in ? Please consider factors such as general economic conditions in the center’s state, the recompetition of the center’s cooperative agreement with NIST, or other factors. What has been the effect of changing the federal/nonfederal cost share to 1:1 for future years for all centers regardless of when they began operating on the overall financial resources to support the center’s operations? What, if anything, would the center change about how it provides services to manufacturers in that state if the federal/nonfederal cost share were to change back to the way it was prior to enactment of the AICA in January 2017? We analyzed the survey responses using content analysis and descriptive statistics. Using content analysis, we analyzed the responses to the three open-ended questions listed above by identifying common themes in centers’ open-ended survey responses to establish categories. Two analysts independently reviewed and coded the survey responses to the categories. Then the analysts compared their coding and if there was disagreement, they discussed their assessment and reached a final determination on the categorization. We also used descriptive statistics to analyze centers’ survey responses to evaluate the impact of the cost share adjustment on different types of MEP centers. For example, we compared the number of centers responding to certain survey questions by center type (i.e., nonprofit institutions, state agencies, or universities) as well as centers whose cooperative agreements were or were not recompeted. To further understand the impacts of the AICA cost share adjustment on different types of MEP centers, we conducted follow-up interviews with officials from nine MEP centers using a standard set of questions. We selected these MEP centers based on our analysis of centers’ survey responses. Furthermore, we selected these centers to include the perspectives of a variety of MEP centers, accounting for factors such as when the center’s agreement was recompeted, number of manufacturers in the state, and whether the center is operated by a nonprofit institution, state agency, or university. During these follow-up interviews, we asked the centers questions such as the following: To what extent did the cost share change affect the center and why? Please explain. Please explain how the center meets the cost share requirement. What has changed since the AICA set the cost share at 1:1? To what extent will the cost share change help or hinder the center’s ability to reach underserved manufacturers? Is there any way that the center can isolate the changes in the cost share from the recompetition? GAO Analysis of NIST Data To describe the extent to which NIST data show impacts from the AICA cost share adjustment, we obtained NIST data on MEP centers’ finances and activities for fiscal years 2013 through 2018. We selected this period to encompass the year prior to when NIST began recompeting MEP centers’ cooperative agreements. NIST collects financial information from each MEP center, including the amount of financial assistance received from NIST, program income received from manufacturers for services provided, cash received from other sources (such as grants), and in-kind contributions. We analyzed these data to identify any changes in centers’ finances for fiscal years 2013 through 2018. We also obtained NIST data detailing the cost share under which each center was operating for fiscal years 2013 through 2017. We assessed the reliability of centers’ financial data by reviewing agency documentation, verifying some data against another data source, and interviewing NIST officials and officials from selected centers. We determined that NIST’s data on MEP centers’ federal assistance and nonfederal contributions are the best available data and are sufficiently reliable to describe general changes in these aspects of centers’ finances during this time period. However, as noted in the report, we found that some centers underreport their nonfederal resources in excess of the nonfederal contributions required to meet their cost share. As a result, we expect that centers’ total available resources—including their federal assistance, nonfederal contributions, and nonfederal resources in excess of their nonfederal contributions—are higher than what we present in the report. We determined this because the underreporting we identified with centers’ nonfederal resources in excess of their nonfederal contributions would tend to understate the amount of these resources over time and because we did not find evidence of overreporting that would contradict this pattern. In addition, we did not independently verify the nonfederal contributions reported by the MEP centers because it was outside the scope of our work. We also obtained and analyzed NIST data on MEP centers’ activities, such as data on the size, location, and number of manufacturers the centers reported serving in fiscal years 2013 through 2018. NIST guidance for MEP centers calls for centers to report various information about the manufacturers that they serve, including company name, Dun and Bradstreet number, and the North American Industry Classification System code. NIST uses the Dun and Bradstreet number to compile other information about each manufacturer, including location and number of staff. We analyzed the data to identify any changes in centers’ activities and to determine the extent to which any changes might be associated with the AICA cost share adjustment. We also reviewed NIST guidance for the MEP program and interviewed NIST and MEP center officials to gain an understanding of the MEP center activity data NIST collects. We assessed the reliability of the activity data by reviewing agency documentation and interviewing NIST officials and selected centers. As noted in the report, these efforts indicated that cost share changes caused some centers’ activity data to be underreported. While we were not able to precisely determine the extent of underreporting or precise changes in centers’ activities over time, as noted in the report, we believe the data are the best available data and are sufficiently reliable to describe general changes in centers’ activities during this time period. We determined this because the underreporting we identified would tend to understate the increases in the total number of manufacturers and the number of very small manufacturers served over time and because we did not find evidence of overreporting that would contradict this pattern. Since the number of rural manufacturers served fluctuated during this time period, however, we were unable to determine whether complete data would indicate a general increase in the number of rural manufacturers served similar to the increases in the total number of manufacturers and the number of very small manufacturers served. Other Efforts To help us understand the legal framework for the cost share adjustment, we reviewed the AICA. We reviewed other documents to provide additional context regarding MEP centers’ cost share requirements, including past GAO reports and reports from the Congressional Research Service and National Academies of Sciences, Engineering, and Medicine. We also reviewed reports on the MEP program’s cost share structure from NIST and the MEP Advisory Board. To gain additional insight on the impact of the cost share adjustment, we also interviewed NIST officials, members of the MEP Advisory Board, and the head of an association representing the MEP centers. Further, we visited a MEP Center in Bothell, Washington, and a manufacturer in Woodinville, Washington, to obtain a more in-depth perspective on the services MEP centers provide to manufacturers. We conducted this performance audit from March 2018 to March 2019 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: GAO Contact and Staff Acknowledgements GAO Contact Staff Acknowledgements In addition to the contact named above, Chris Murray (Assistant Director), Arvin Wu (Analyst in Charge), Stephen Betsock, Kevin Bray, Mark Braza, TC Corless, Ellen Fried, Jill Lacey, John Mingus, Calaera Powroznik, Sara Sullivan, David Wishard, and John Yee made key contributions to this report.

Why GAO Did This Study Small and medium-sized manufacturers are an important part of the U.S. economy. In 1988, to enhance the competitiveness, productivity, and technological performance of U.S. manufacturing, NIST established what is now called the MEP program. The program supports manufacturers through services provided by MEP centers. The centers, located in all 50 states and Puerto Rico, are operated by nonfederal organizations. The MEP centers provide assistance, either directly or through third parties, to help improve manufacturing firms' processes and productivity; expand their capacity; and help them adopt new technologies, utilize best management practices, and accelerate company growth. NIST enters into a cooperative agreement with the nonfederal organization that runs each center to provide federal financial assistance conditional upon the center contributing nonfederal matching funds—known as a cost share. The AICA included a provision for GAO to review the effect of the 2017 cost share adjustment. This report describes (1) the MEP centers' views regarding the extent to which the recent cost share adjustment has helped them serve manufacturers and (2) the extent to which NIST data show impacts of the cost share adjustment on centers' finances and activities. GAO surveyed all 51 MEP centers, analyzed NIST data on the MEP program, and interviewed NIST and MEP center officials. What GAO Found Most Manufacturing Extension Partnership (MEP) centers reported that the January 2017 American Innovation and Competitiveness Act (AICA) cost share adjustment has helped them serve manufacturers, especially very small (i.e., less than 20 employees) and rural ones. The AICA adjusted the cost share ratio to remain at 1:1, that is, $1 of nonfederal contributions for each $1 of federal assistance. Before the adjustment, MEP centers' cost share requirement increased over the course of their cooperative agreements from 1:1 to 2:1, requiring centers to obtain a greater proportion of revenue from nonfederal sources. In GAO's survey of all 51 MEP centers, 44 centers cited positive effects of the adjustment on center operations, such as helping to improve center services or better reach underserved manufacturers. Also, 41 centers indicated the adjustment increased their financial stability, which some centers stated has allowed them to focus less on revenue generation and to serve very small and rural manufacturers. However, some MEP center officials observed that the AICA cost share adjustment impact is hard to distinguish from other factors, such as the National Institute of Standards and Technology's (NIST) recompetition of nearly all centers' cooperative agreements between fiscal years 2014 and 2017. The recompetition increased the level of federal financial assistance for most centers and reset many centers' cost share ratio from 2:1 to 1:1 prior to the 2017 adjustment. Still, center officials said that if the cost share requirement reverted to what it was prior to the 2017 adjustment, centers would be less able to serve manufacturers, particularly very small and rural ones. NIST data show that there have been some changes in MEP centers' finances and activities since the AICA cost share adjustment; however, these changes generally began prior to the adjustment. For example, NIST data on centers' finances show an increase in federal assistance and a decrease in reported nonfederal contributions from fiscal year 2017 to 2018, but these changes generally began around fiscal year 2014, when NIST began the recompetition process. Similarly, NIST data on centers' activities show an overall increase in the numbers of very small and rural manufacturers served from fiscal year 2017 to 2018. While the change in the number of very small manufacturers served began around fiscal year 2014, the number of rural manufacturers served fluctuated from fiscal years 2014 through 2018. Like MEP center officials, NIST officials said the impact of the AICA cost share adjustment is intertwined with the recompetition impacts and, going forward, the AICA adjustment may help sustain recent increases in the number of very small and rural manufacturers served.

gao_GAO-19-94

gao_GAO-19-94_0

Background According to OMB, the federal government spends more than $25 billion annually for core mission support services, such as HR and financial management that are common across agencies. These services are generally supported by a wide range of activities. For example, the HR employee life cycle functions represent hiring to retirement and include activities such as payroll and other compensation and benefits management. The financial management function includes core financial activities such as making and receiving payments for goods or services. As shown in figure 1, for more than two decades, the federal government has taken actions aimed at increasing agencies’ use of shared services. Key congressional actions included new laws to create uniform standards for financial reporting, promote agency use of information technology (IT) to deliver core mission support services, and establish funding mechanisms for agencies to modernize IT systems. Presidential administrations have made it a priority to promote the use of shared services for HR and financial management activities for many years. For example, in 2014 and again in 2018, OMB established a cross-agency priority (CAP) goal of improving the use, quality, and availability of administrative shared services. Complementing the goal, the Digital Accountability and Transparency Act of 2014 is intended to standardize and increase the transparency of agencies’ spending data. At present, OMB has responsibility and authority to develop and implement government-wide shared services policy. OMB is working with GSA to develop shared services strategy, policies, and guidance, with OPM and Treasury also having important roles (see table 1). Table 1 also describes the agencies we selected and their roles in the shared services initiatives. Federal Efforts to Promote Shared Services Resulted in Some Cost Savings and Efficiency Gains, but Challenges Impeded More Widespread Adoption Efforts to Promote Shared Services for HR Activities Contributed to Cost Savings and Cost Avoidance OMB’s efforts to promote HR shared services resulted in cost savings, cost avoidance, and more consistent service delivery. OMB announced the HR Line of Business in 2004 simultaneously with the Financial Management Line of Business. These shared services initiatives shared similar goals: (1) standardize systems, business processes, and data elements to promote consistency across the federal government; and (2) reduce costs by establishing a marketplace or a system of buying and selling products and services. In this context, the marketplace would allow agencies to acquire IT systems for core mission support services through shared services solutions. OPM and FIT coordinated with their respective Chief Human Capital Officer and CFO stakeholder communities to develop data elements and business process standards for common HR and financial management activities. Setting consistent standards for data and systems can lead to benefits for shared services customers as well as providers. For example, the ability to meaningfully aggregate or compare data across the federal government increases as more agencies adopt common or standardized data elements or processes. As we have previously reported, the lack of comparable data across agencies can hinder efforts to analyze government-wide trends. Specifically, OPM reported in 2015 that the lack of standardized time and attendance data or required data components limits access to workforce data and hinders efforts to analyze government-wide trends. In addition, once providers know the standards, they can develop a solution applicable to multiple customer agencies and achieve economies of scale. OPM oversaw one of the first efforts to create a shared services marketplace in 2001, which focused on payroll. That effort resulted in cost savings, cost avoidance, and greater consistency in the interpretation and application of payroll rules. In the early 2000s, many agencies’ payroll systems were nearing the end of their estimated life cycles. As managing partner of the HR Line of Business, OPM worked with OMB to identify payroll providers. They selected 4 of the then 22 federal payroll providers to serve as FSSPs for the 116 executive branch agencies. We previously reported that, according to OPM officials who had overseen the payroll consolidation effort, OMB authorized only the chosen federal payroll providers—not other agencies—to spend money on modernizing payroll systems, thereby encouraging the shift to the selected FSSPs. OPM designated six more public- and private-sector shared services centers to provide additional HR functions to agencies in between 2005- 2008. These functions include core HR services such as personnel action processing and benefits and compensation management, as well as noncore services such as HR strategy and performance management. According to OPM, more than 99 percent of agencies migrated to a payroll provider and more than 88 percent of agencies migrated to an HR shared services center. This resulted in an estimated savings and cost avoidance of more than $1 billion between fiscal years 2002 and 2015. The consolidation of payroll providers from 22 to 4 providers also contributed to greater consistency in the way the federal government interprets and applies payroll rules. Outcome Information on Financial Management Shared Services Efforts Is Limited The federal government made progress toward establishing standards for selected financial management activities and designating providers to engage in a marketplace. However, information on outcomes is limited because data were not tracked amidst changes in the Financial Management Line of Business leadership and strategy. In 2004, OMB designated GSA’s Financial Systems Integration Office (FSIO) as managing partner of the Financial Management Line of Business. OMB also designated four FSSPs to provide financial management services to other agencies. They were: the Department of the Treasury’s Administrative Resource Center (ARC), the Department of the Interior’s Interior Business Center (IBC), the Department of Transportation’s Enterprise Service Center (ESC), and GSA’s Federal Integrated Solutions Center. Under the original Financial Management Line of Business, which was launched in 2004, federal agencies were required to either serve as a shared services provider or leverage a shared services provider when modernizing a financial system. In 2010, OMB changed this strategy. Agencies would no longer be required to adopt shared services for financial systems. In announcing this change in strategy, OMB noted concerns related to the costs and risks—such as projects that did not meet agency needs upon completion—that medium and large agencies had encountered as they pursued shared services for financial management activities. OMB also noted that agency managers were more likely to pursue shared services for less complex operations such as common website hosting, rather than more complex operations, such as financial transactions. Further, OMB announced a change in leadership. FSIO ceased operations and OMB later designated the Department of the Treasury’s Office of Financial Innovation and Transformation (FIT) as the new managing partner of the Financial Management Line of Business. FIT took steps to establish a marketplace for customers seeking shared financial management services and to develop standards for financial management activities. As part of this effort, FIT created a process to analyze the existing financial management FSSPs to identify capability gaps. FIT invited the existing FSSPs and other agencies that wanted to receive FSSP designation to apply. In 2014, FIT selected ARC, IBC, ESC, and USDA Financial Management Services (which is separate from NFC) to provide financial management services to other federal agencies. FIT also worked with the CFO community to develop more than 40 business use cases for financial management activities. Business use cases document how a common activity, such as disbursing payments, is executed, including a sequential description of each step in the process. According to a FIT official, these business use cases foster a common understanding of how to execute specific financial management functions among customers and providers, which can make it easier for customers to transition to shared services. Further, FIT identified four initiatives to expand shared services for financial transactions. FIT’s four shared services initiatives included expanding shared services for accounts payable and accounts receivable, debt collection, and payment processing. FIT officials estimated these projects could contribute to cost savings of around $620 million over 5 years, but they did not track cost savings. FIT officials also did not track the percentage of non-CFO Act agencies that migrated financial systems to a shared services provider. FSSP customer lists show that non-CFO Act agencies and commissions more frequently rely on external providers for core financial shared services than do medium and large agencies. FIT officials stated that OMB transferred many of FIT’s responsibilities, including collecting performance information, to GSA in 2016. In 2018, GSA officials published customer satisfaction data from 2017 and 2018 for administrative functions, including financial management services through the Customer Satisfaction Survey. GSA also plans to track the percentage of selected financial transactions—such as certain types of payments—completed by a shared services provider starting in 2020. However, tracking of cost data continues to be an issue, which we address later in this report. Governance and Marketplace Challenges Have Impeded Greater Progress toward Cost Savings and Other Performance Goals Wider adoption of HR and financial management shared services has been impeded by challenges in two areas. First, shared services efforts have faced persistent governance challenges, such as limited interagency collaboration, difficulty reconciling benefits and trade-offs, and limited oversight and technical support for shared services migrations. Second, the efforts have also experienced marketplace challenges, which involve difficulty obtaining funding to invest in shared services, demand uncertainty among providers, and limited choices for customers. These issues hampered efforts to establish effective and efficient shared services marketplaces. As a result, these marketplaces have not been able to consistently support sufficient competition limiting the potential cost sharing efficiencies and improved performance that could be realized with greater usage. OMB and GSA have taken steps to address these challenges, which we assess later in this report. Limited interagency collaboration. The Lines of Business governance structure limited collaboration across different mission support areas. This made it more difficult for those with expertise in acquisitions, IT, HR, and financial management policy to work together on shared services solutions. For example, although a shared payroll solution should ideally consider how to appropriately implement payroll rules, an area in which the Chief Human Capital Officers community has subject-matter expertise, it also needs the expertise of others. Specifically, the solution should also be able to integrate with an agency’s financial reporting systems, an area in which the CFOs and Chief Information Officers have expertise. Additionally, the solution should ideally leverage the government’s purchasing power, an area in which the Chief Acquisition Officers have expertise. The Lines of Business Managing Partners took steps intended to address this issue. For example, the HR Line of Business chartered the Multi-Agency Executive Strategy Committee to facilitate interagency collaboration by bringing together representatives from human capital offices across CFO Act agencies. Later in this report, we describe additional steps OMB and GSA took to promote greater collaboration across the individual Lines of Business. Difficulty reconciling benefits and trade-offs. We found that agencies have had difficulty reconciling the trade-offs associated with adopting a standardized service. OMB has issued multiple memorandums over the years directing agency officials to consider shared services solutions when researching options for replacing legacy HR or financial management systems. Despite OMB’s direction, the benefits for customers to migrate to a standardized solution were not always clear. According to ARC officials, prospective customers were invested in their legacy processes, or did not factor long-term cost savings or cost avoidance into their decision-making process, therefore limiting the full realization of standardized shared services. These difficulties are illustrated in a recent experience at Education. Education officials debated whether to migrate the department’s financial management system to a shared services provider, and spent substantial time and money determining whether it was feasible. Education has several systems which are closely integrated and dependent on one another, including financial and grants management. In considering trade- offs, officials were concerned about the costs they would incur and the impact to their grantees if they de-coupled these systems to migrate to a standardized financial system. According to Education officials, they spent about a year meeting with officials from OMB, FIT, and ARC to determine the feasibility of migrating their core accounting system to ARC. They also reported spending more than $750,000 on a feasibility study. The study noted that the cost of an internal migration would be less expensive than migrating to ARC. Ultimately, in 2016, Education officials decided that instead of migrating they would modernize their legacy system internally. GSA and OMB supported Education’s decision to modernize in house and agreed that Education did not need to move to a shared services provider at that time. However, GSA officials working with Education on their financial management modernization efforts noted that Education’s decision to pursue a customized solution that paired financial systems and grants contributed to the higher quoted cost of migrating to ARC. GSA officials also recommended that Education consider the costs and benefits of making changes to its financial management systems that would eventually facilitate the transition to a shared services solution. Education officials said they remain committed to reviewing this effort again in the future. Limited oversight and technical support. Customer and provider agencies experienced issues with project management, which contributed to delayed and costly migrations. For example, we previously reported that two recent financial management migrations—involving the Department of Housing and Urban Development (HUD) migrating to ARC and the Department of Homeland Security (DHS) migrating to IBC, the federal shared services provider within the Department of the Interior— were late, over budget, and only addressed a portion of the original project scope. In 2016, we reported that HUD migrated 4 of 14 planned financial management capabilities to shared service solutions, but ended efforts to migrate the remaining 10 planned capabilities to ARC, in part because of weaknesses in implementing key management practices. For example, HUD’s senior leaders did not recognize and fully address challenges as they arose, including those identified with scope, schedule, and program costs. As a result, HUD was unable to follow through with its plans to replace a number of its legacy financial management systems and continues to maintain those systems while seeking other new initiatives to address aspects of the remaining capabilities. HUD spent about $58 million over three years before deciding to end the migration and modernization effort in April 2016. ARC officials reported that as of November 2018, it continues to provide financial management services for the capabilities that HUD migrated. Similarly, in 2017, we reported that to address long-standing deficiencies in DHS’s financial management systems, DHS started to migrate three components to a modernized financial management system solution provided by the IBC. However, we found that significant challenges such as project management and communication problems, among others, disrupted the project, raising concerns about the extent to which objectives would be achieved as planned. In May 2016, DHS and IBC determined that the planned implementation dates were not viable. We reported that plans for DHS’s path forward on this project were delayed for 2 years. In both cases, we found that the customer agencies did not consistently follow leading project management practices, such as properly identifying potential risks and developing mitigation plans. We made four recommendations to HUD and two recommendations to DHS intended to address weaknesses in their department’s financial management systems modernization efforts. However, as of November 2018, they had not yet implemented them. OPM took steps to address this issue for the HR Line of Business. OPM officials told us that in 2007 they developed an online guide to assist customer agencies to prepare for and manage a migration of their human resources operations to a shared services center. According to OPM, the guidance contains information regarding different delivery models, the migration process, and roles and responsibilities. Further, OMB and GSA recognized that customers and providers would benefit from additional technical support and oversight. In May 2016 guidance, OMB tasked GSA with assisting agencies during implementation by publishing guidance incorporating best practices and lessons learned in project management. OMB also tasked GSA with monitoring implementations to ensure that agencies are following a disciplined process and properly assessing project risk in partnership with OMB. Later in this report, we describe steps GSA has taken to provide guidance and technical assistance to agencies. Funding challenges, demand uncertainty, and limited choices. Funding challenges, demand uncertainty among providers, and limited choices for customers are challenges that have limited the effectiveness of shared services marketplaces for HR and financial management services. We have previously reported that agencies consider obtaining the funding required for consolidation and migration efforts to be a challenge. This can affect their ability to realize cost savings and cost avoidance. GSA officials said funding challenges can be a barrier to entry into the marketplace for potential customers. In part because of funding challenges, agencies continue to rely on legacy IT systems for core mission support services. Many of these systems are increasingly at risk of failure because of aging technology and reliance on applications that are no longer supported by vendors. As a result, agencies are limited in their ability to deploy updates or make adjustments to ensure the systems support mission needs. In our 2017 High-Risk report, we found that agencies needed to establish action plans to modernize or replace obsolete IT investments across the federal government. Some FSSPs have also struggled to keep up with the capital investments necessary to modernize. We previously reported that OPM officials involved with the payroll consolidation effort said that funding had not materialized for systems modernization for the four payroll service providers, though it was expected at the outset of the initiative. The officials said this lack of funding was a major problem that put the long- term viability of the effort at risk. According to NFC officials, the HR FSSPs continue to find it difficult to keep up with the capital investments necessary to modernize. GSA officials said that federal investment in HR and financial management systems modernization lags behind the private sector. According to agency officials and subject-matter experts, federal and commercial shared services providers faced uncertainty related to customer demand, which made it difficult for them to plan and more fully participate in the shared services marketplace. For example, ARC officials said that in determining whether to invest in system improvement, they need to evaluate the impact on current customers as well as the benefits to potential customers. They also pointed out that the costs associated with systems improvements would be borne by the current customer base if potential new customers failed to materialize. On the customer side, both agency officials and subject-matter experts told us that potential customers often found it difficult to identify providers capable of meeting their needs. Some customers wanted a la carte services and others had needs which surpassed the capacity of available providers. For example, Education’s HR officials said it was difficult to find a provider to meet their needs for specific HR services. A lack of up-to- date information about providers’ services and costs complicated their search process. Education officials said they reached out to several FSSPs, but either they did not provide the specific services Education wanted, they were not taking on new customers, or the cost was not feasible for Education. We previously found that as more agencies consider transitioning to shared services providers, making pricing and performance information publically available can help agencies determine the most efficient method for obtaining services. Subject-matter experts said that large agencies also had challenges finding an FSSP capable of meeting their needs. For example, one subject-matter expert who works at a large agency with more than 350,000 employees described the challenges his agency faced identifying a provider capable of providing financial management services. He said one potential FSSP was concerned that adding a large customer would negatively impact its ability to serve other customers. In light of the difficulty in finding a provider with sufficient capacity, the agency decided to modernize its financial system internally. In light of these challenges, agency adoption of shared services has been slow and uneven. In 2015, the Association of Government Accountants (AGA) surveyed government managers and staff, and found that difficult migration experiences raised doubts among officials at other agencies contemplating shared services. AGA found that respondents considering migrating to a shared services provider were hearing enough concerns that they were not eager to undergo a substantial migration. Consequently, agencies continue to conduct common business activities in an inconsistent manner and maintain unique systems. Therefore, they may be missing opportunities to achieve cost savings offered by greater use of shared services. For example, according to OPM, there are at least 108 different systems that send time and attendance data to FSSPs. There are also an estimated 86 learning management systems across the government. We have consistently reported that duplicative and incompatible agency business systems and data prevent agencies from sharing data, or force them to depend on expensive, custom-developed systems or programs to do so. OMB and GSA Have Taken Actions to Address Governance and Marketplace Challenges, but Could Strengthen Their Implementation Approach OMB and GSA Have Taken Actions to Address Persistent Challenges Over the past several years, OMB and GSA have taken actions— including creating a new governance structure and redesigning the marketplace—to address the challenges that impeded more widespread adoption of shared services. To bolster interagency collaboration, OMB issued guidance in 2016, which designated a Shared Services Policy Officer within OMB with responsibility and authority to develop and implement government-wide shared services policy. OMB also tasked the new Unified Shared Services Management (USSM) office within GSA to bring together key stakeholders, including the managing partners of the different lines of business, and representatives from customer and provider agencies. GSA also introduced the Federal Integrated Business Framework to build on ongoing efforts by OPM and FIT to develop standards for HR and financial management data elements and business processes, among other things. As part of this effort, cross-agency working groups identified 11 end-to-end processes for mission support services. Similar to the business use cases FIT developed, these business processes document how a common administrative activity is executed, including a sequential description of each step in the process. According to GSA officials, these business processes serve as the basis for a common understanding of what services agencies need, and what shared services providers should offer. These working groups also bring together those with expertise in acquisitions, IT, HR, and financial management policies. GSA also developed guidance for selecting and migrating to a shared services provider. The new guidance identified opportunities for GSA to review agency migration materials. GSA developed the Modernization and Migration Management Playbook (M3 Playbook), a compilation of leading project management practices and lessons learned from past systems migrations, and met with agencies contemplating or undertaking migrations. The M3 Playbook divides a typical shared services migration into six phases. For each phase, the M3 Playbook identifies key steps agencies should take before they move on, such as completing a risk mitigation strategy and defining performance and success metrics. At the end of each phase, the M3 Playbook recommends a “tollgate” review to ensure both customer and provider completed the necessary steps and are ready to move to the next phase. GSA is to provide recommendations to OMB on the migrations based on observations of project status and risk from tollgate reviews. Agency officials involved with HR and financial management migrations we spoke with said they found both the Playbook and GSA’s reviews helpful. For example, Justice officials said they started to use the Playbook once it was available midway through their HR system migration to NFC. Prior to each tollgate review, Justice officials said they submitted the required deliverables so that GSA had time to review the documents prior to the meetings. Justice officials said that GSA staff reviewing their materials offered concrete suggestions such as developing and documenting success metrics, strengthening their business case, and developing a risk assessment document. According to Justice officials, these suggestions improved the migration process. Education officials also reported they appreciated the project management expertise provided by GSA staff. OMB and GSA Introduced a New Shared Services Marketplace Model, but Could Strengthen Their Implementation Approach In fiscal year 2018, OMB and GSA introduced a new marketplace model for shared services that seeks to better meet the needs of customers and providers by offering more choices for purchasing shared services. We examined their approach for the new model and found they were following some key change management practices, but there are weaknesses with the implementation. Specifically, we found OMB and GSA do not have a plan to monitor the implementation of an initiative designed to determine how well the new marketplace model works as intended. Nor have they identified and documented some key roles and responsibilities. The action plan also does not explain how OMB and GSA will provide information to customers about provider services, pricing, and performance. Lastly, OMB and GSA have not implemented a process for collecting and tracking cost-savings data. OMB and GSA described their plan for the new marketplace in an action plan, released in March 2018, along with the President’s Management Agenda. The management agenda issued a new cross-agency priority (CAP) goal to improve the effectiveness of shared services. According to the management agenda, the shared services goal will support CAP goals related to IT modernization, data accountability and transparency, and the workforce of the future. OMB and GSA are the shared services goal leaders and staff said they are coordinating with other CAP goal leaders to achieve their objectives. To oversee the marketplace and provide greater accountability for migrations, OMB and GSA are implementing a new two–tier governance structure (see figure 2). To ensure that agencies are adhering to the standards developed by the Business Standards Council and to provide greater oversight and accountability for shared services migrations, OMB and GSA are working on plans to create Task Order Review Boards (Review Boards) for different types of services, such as payroll or accounting. According to OMB and GSA’s action plan, the Review Boards will administer standards and will review all task orders for shared services purchases for compliance with the standards. The Review Boards will need to approve any requested customizations. According to GSA officials, the contracts for the various vendors providing shared technology and transaction processing services will be purchased through and managed by Service Management Offices (SMO). The SMO will be responsible for managing the integration of new commercial suppliers into the marketplace and responding to user concerns. The SMO will also be held accountable for provider performance. OMB staff noted that the details of the Review Boards depend on the shared services solutions that are identified. Figure 3 describes the different options customers will have for purchasing shared services in the new marketplace. The figure also shows how a Review Board and SMO are intended to interact with customers and providers. To determine whether the marketplace model functions as intended, OMB and GSA introduced an initiative, NewPay. In September 2018, GSA awarded a 10-year, $2.5 billion NewPay agreement to two commercial teams to provide payroll, and work schedule and leave management services using Software-as-a-Service. Software-as-a-Service—a cloud- based computing model—delivers one or more applications and all the resources—operating system, programming tools, and underlying infrastructure to run them—for use on demand. According to OMB and GSA staff, Software-as-a Service should help address some of the challenges with demand uncertainty because providers can more easily increase and decrease capacity depending on changes in demand than FSSPs have been able to do with their current technology. Our prior work on organizational transformations shows that incorporating change management practices—such as setting implementation goals and a timeline to show progress—improves the likelihood of successful reforms. Adopting key change management practices can also help managers recognize and address agency cultural factors that can inhibit reform efforts. As OMB and GSA prepared to implement the new marketplace model, they incorporated some key change management practices. For example, they defined their vision for a shared services marketplace and some of the key activities needed to achieve that future state. GSA also issued a draft statement of objectives for NewPay in December 2017. The statement includes a comprehensive list of tasks related to project management and assigns responsibility for those tasks to the prospective customers, providers, or the government agency that will fulfill the SMO role. Although OMB and GSA have incorporated some key change management practices, we found some weaknesses in OMB and GSA’s implementation of the marketplace. Monitoring. OMB and GSA do not have a finalized plan to monitor the implementation of NewPay. We have previously identified key questions for agencies that are planning and implementing transformations. In that work, we found that agencies need to monitor and evaluate their efforts to identify areas for improvement. We have also reported that effective monitoring plans should include performance goals and milestones, transparent reporting tools to help manage stakeholder expectations, and a process for capturing lessons learned to improve the management of subsequent phases. OMB and GSA staff said they are working on a plan to help them implement NewPay. However, it is not yet complete and they did not provide us with a draft to review. They said their plans continue to evolve and they anticipate having an implementation plan by spring 2019. The lack of a finalized plan with the elements listed above hinders OMB and GSA’s ability to provide sufficient oversight for this transition. Having such a plan would provide various benefits to the NewPay implementation effort. First, a monitoring plan that includes performance goals and milestones would help OMB and GSA track how many and how well customer agencies are transitioning from one provider to another. Similarly, setting performance goals related to continued delivery of services during the transition could help OMB and GSA more quickly identify gaps and make adjustments as needed. Specifically, OMB and GSA could more effectively monitor how the new approach for purchasing payroll, and work schedule and leave management systems integrates with current HR systems. Additionally, transparent reporting tools, such as web-based reporting on key milestones, could help OMB and GSA demonstrate that they are aware of challenges and are addressing them as they arise. Greater reporting transparency could also help build momentum, show progress, and help justify continuing investments in reforming shared services efforts. Finally, a process for capturing lessons learned based on NewPay could help OMB and GSA improve the process for subsequent initiatives and further minimize disruptions to agency delivery of services during these future transitions. Without a monitoring plan with performance goals and milestones, transparent reporting tools, and a process for capturing lessons learned, it will be more difficult for OMB and GSA to provide oversight of the transition and its effects on providers and customers, including whether there are interruptions to delivery of services. A monitoring plan could help OMB and GSA avoid gaps in service or costly delays as agencies transition to the new model for obtaining payroll and work management services. Roles and responsibilities. OMB and GSA have also not identified or documented some key roles and responsibilities related to the implementation of NewPay. Identifying a NewPay SMO is a crucial first step, since the SMO is supposed to play a key role driving standards and holding customers and providers accountable for performance. However, OMB and GSA have not announced which agency will serve as the SMO. Further, they have not identified which agencies or officials will serve on the NewPay Review Board. They also have not documented the authority or the resources the SMO and Review Board will have to enforce agency adoption of standards. OMB and GSA have also not yet documented which agency will be responsible for interpreting payroll rules and regulations. This has been an ongoing issue for the payroll FSSPs. According to GSA and NFC officials, the payroll FSSPs have been interpreting business rules differently, and thus have implemented new regulations inconsistently. According to NFC officials, the payroll FSSPs requested the establishment of a governing body to help standardize the process for implementing new regulations. OPM officials told us in September 2018 that they intend to start providing guidance to support payroll standardization to the extent allowed by law and regulation in the future. However, as of October 2018, OMB and GSA had not documented this decision. According to federal standards on internal control, management should establish an organizational structure, assign responsibility, and delegate authority to achieve an entity’s objectives. When the organizational structure describes overall responsibilities, and when those responsibilities are assigned to discrete units, then organizations can operate more efficiently and effectively. Moreover, in our previous body of work on enhancing interagency collaboration, we identified key practices that can help agencies mitigate challenges when they attempt to work collaboratively. For example, clarifying roles and responsibilities can enhance interagency collaboration. OMB staff and GSA officials said they are still identifying which agencies or entities will fill key roles and assume key responsibilities. They anticipate that some of this information will be finalized by spring 2019. Identifying and documenting roles and responsibilities would help ensure that key stakeholders are involved in planning and implementation activities. Until OMB and GSA clearly identify, communicate, and document key roles and responsibilities, they run the risk of not achieving their objectives. They also risk repeating past problems, such as the inconsistent implementation and interpretation of standards and migrations that encounter costly delays because agencies do not follow available guidance. Information on services, pricing, and performance. Although the action plan aims to help additional providers enter the marketplace, it does not explain how OMB and GSA will provide information to customers about provider services, pricing, and performance. According to the Association of Government Accountants, effective marketplaces require market transparency with information on services, pricing, and performance. Also, according to federal standards on internal control, managers should externally communicate the necessary quality information to achieve an entity’s objectives. As we have previously reported, reliable information on the costs of federal programs and activities is crucial for effective management of government operations. OMB staff and GSA officials said that data collection efforts are on hold as they continue to try to determine what performance metrics they will use and share with potential customers. Without up-to-date information on providers—such as the services they offer, their level of performance, and their costs—it will be time consuming and difficult for potential customers to compare providers. This lack of information could slow the rate of shared services adoption. Cost-savings data. In the CAP goal action plan for shared services, OMB and GSA established a cost savings goal of an estimated $2 billion over 10 years based on reforms to the shared services governance structure and marketplace. However, their action plan does not include steps they intend to take to collect and track cost-savings data. Such data would allow them to assess their progress toward their goal. In their action plan, OMB and GSA included performance measures for goals such as customer satisfaction. They also have output measures related to HR and financial management activities. However, they did not include a measure to gauge their progress in achieving cost savings. In our previous work on key questions for agencies that are planning and implementing transformations, we found that agencies need to have processes in place to collect the needed data and evidence to effectively measure goals of reform efforts. OMB and GSA said they are still finalizing their implementation plan. Including a process to collect and track cost savings data in the final plan would position them to assess how well their reform efforts are contributing to their cost savings goal. Cost savings data would also support oversight efforts, as OMB and GSA could better communicate to Congress and other relevant stakeholders the extent to which their reforms are contributing to cost savings goals. Earlier in this report, we described how difficult it was to determine the progress of the financial management line of business because the managing partners of that effort did not track data on cost savings. Until OMB and GSA finalize a plan for collecting the needed data and evidence to effectively measure cost-savings goals, they risk experiencing a similar challenge. OMB and GSA’s action plan to support the shared services CAP goal does not directly address funding challenges. However, new legislation intended to promote IT modernization efforts may address these challenges. In 2017, Congress enacted the Modernizing Government Technology (MGT) Act as part of the 2018 National Defense Authorization Act. The MGT Act allows agencies to create working capital funds for modernizing IT systems. Working capital funds are primarily used for business-like activities, such as purchasing consolidated or shared services within and between federal agencies. The MGT Act allows CFO Act agencies to transition legacy systems to cloud computing platforms or other innovative platforms and technologies, among other things. We have previously reported that working capital funds provide agencies with an opportunity to operate more efficiently by consolidating services and creating incentives for customers and managers to exercise cost control and economic restraint. The MGT Act also established the Technology Modernization Fund and Technology Modernization Board. Agencies can apply to the board for loans for IT modernization projects, including replacing legacy systems with shared services. In February 2018, OMB issued guidance on the initial process agencies should follow to submit proposals to the board. As of February 2019, OMB announced plans to award close to $90 million to various agencies for modernization projects. Two of these awards were for shared services: one award was to GSA for more than $20 million to help fund NewPay and one award was to USDA for $5 million to migrate 10 IT applications to a shared services cloud platform model. Conclusions When properly implemented, a shared services model for HR and financial management activities has the potential to help the federal government cut costs and modernize aging IT systems. Over the past 15 years there have been some notable shared services successes: for example, consolidating payroll services resulted in more than $1 billion in cost savings and cost avoidance over 10 years, according to OPM estimates. However, there have also been persistent governance and marketplace challenges that have impeded more widespread adoption of shared services. OMB and GSA have been involved in shared services reform efforts for decades. Their plan for a new shared services governance structure and marketplace has the potential to address some of the challenges that have previously hindered more widespread adoption of shared services. For example, their proposed marketplace model has the potential to make the marketplace more effective by reducing demand uncertainty among shared services providers and providing more choices for customers. However, several weaknesses in their implementation of NewPay could limit the initiative’s success. OMB and GSA do not have a plan to monitor NewPay’s implementation. They also have not documented key decision- making roles and responsibilities related to the implementation of NewPay. Until they develop a monitoring plan which includes performance goals and milestones, transparent reporting tools, and a process for capturing lessons learned, and documenting key roles, they risk implementation challenges that could cause gaps in service or costly delays. OMB and GSA also do not have a process to provide information to customers about provider services, pricing, and performance. Developing such a process would help minimize the challenges of transitioning to shared services on key stakeholders. Finally, OMB and GSA do not have a process for collecting and tracking cost-savings data. Until OMB and GSA finalize their plan for collecting the related data and evidence to measure their cost savings goal of an estimated $2 billion over 10 years, they will not be able to determine and report progress made. Recommendations for Executive Action We are making four recommendations to OMB to work with GSA, which is the co-goal leader for the shared services CAP goal. OMB’s Shared Services Policy Officer should work with GSA to finalize a plan for monitoring the implementation of NewPay. The plan should include: implementation goals, a timeline, and milestones for agencies to transition from one provider to another; transparent reporting mechanisms on key milestones; and a process for capturing and communicating lessons learned. (Recommendation 1) OMB’s Shared Services Policy Officer should work with GSA to document key roles and responsibilities, including which agency will be the NewPay SMO, who will be assigned to the NewPay Task Order Review Board, how the SMO, the Review Board, and other key stakeholders will work together, and which agency will be responsible for interpreting payroll rules and regulations. (Recommendation 2) OMB’s Shared Services Policy Officer should work with GSA to update provider information on services offered, pricing, and performance and share that information with prospective customers. (Recommendation 3) OMB’s Shared Services Policy Officer should work with GSA to implement a process for collecting and tracking cost-savings data that would allow them to assess progress toward the shared services cost- savings goal of an estimated $2 billion over 10 years. (Recommendation 4) Agency Comments We provided a draft of this report to the Director of OMB, the Administrator of GSA, the Acting Director of OPM, the Secretary of the Treasury, the Deputy Assistant Inspector General for Audit of the USDA, the Secretary of Education, and the Assistant Attorney General for Administration of Justice for review and comment. OMB staff did not agree or disagree with our recommendations. In comments provided by email, OMB staff stated OMB has been re-evaluating its shared services policies and may provide an updated policy in the future. OMB, GSA, Treasury, OPM, USDA, and the Department of Education provided technical comments on this report which were incorporated as appropriate. The Department of Justice did not have comments. As agreed with your office, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies to the appropriate congressional committees, the Director of the Office of Management and Budget, the Administrator of General Services Administration, the Acting Director of the U.S. Office of Personnel Management, the Secretary of the Treasury, the Deputy Assistant Inspector General for Audit of the U.S. Department of Agriculture, the Secretary of the Department of Education, and the Assistant Attorney General for Administration of the Department of Justice, and other interested parties. In addition, the report will be available at no charge on the GAO website at http://www.gao.gov. If you or your staff has any questions about this report, please contact Tranchau (Kris) T. Nguyen at (202) 512-2660 or Nguyentt@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of our report. Key contributors to this report are listed in appendix II. Appendix I: Objectives, Scope, and Methodology This report: (1) identifies the progress and challenges associated with federal shared services initiatives for selected human resources (HR) and financial management activities, and (2) assesses the Office of Management and Budget’s (OMB) and the General Service Administration’s (GSA) actions to address those challenges. To address both of our objectives, we conducted a literature review of GAO work and other relevant publications on HR and financial management shared services. In addition to GAO reports, we selected reports by think tanks and professional associations from the past 15 years, such as reports by the Partnership for Public Service and the Association of Government Accountants (AGA). We reviewed reports that described past HR and financial management federal shared services initiatives or specific shared services migrations involving HR or financial management services. These reports assessed the outcomes, challenges, or summarized lessons learned associated with those initiatives or migrations. We reviewed planning and performance documents and we interviewed officials from (1) OMB and GSA, the agencies that oversee shared services policy and guidance, and (2) the Office of Personnel Management (OPM) and the Office of Financial Innovation and Transformation (FIT) within the Department of the Treasury (Treasury), agencies that oversaw past shared services initiatives and continue to play a key role developing government-wide policy for HR and financial management shared services. Key documents we reviewed included: OMB memorandums announcing federal shared services initiatives; the Modernization and Migration Management (M3) Playbook, guidance that GSA developed and provides to agencies considering or implementing shared services migrations; strategic or operational plans for earlier shared services initiatives, such as the Human Resources and Financial Management Lines of Business; and strategic or operational plans for ongoing shared services initiatives such as the Federal Integrated Business Framework, a model GSA developed with the lines of business managing partners for moving agencies toward common, cloud-based solutions for management functions. To illustrate examples of outcomes and challenges, we selected two federal shared services providers (FSSPs), federal agencies that provide shared services to other agencies: the National Finance Center (NFC) within the U.S. Department of Agriculture (USDA) and the Administrative Resource Center (ARC) within Treasury. We also selected two customer agencies: the Departments of Justice (Justice) and Education (Education), which are experiencing different phases of shared services migrations. We made our selection based on a number of factors. To capture a range of experience and perspectives, we selected a mix of customer and provider agencies. We selected one HR and one financial management systems migration to review, as well as one migration in an earlier phase and one in a later phase. To capture an in-depth perspective of a migration, we selected one customer and provider working together on a migration. To capture perspectives on OMB and GSA’s efforts to address shared services challenges and improve outcomes, we selected provider and customer agencies that were meeting regularly with GSA in 2016 or 2017 on their shared services migration. Our selection of agencies is non- generalizable and their experiences and outcomes may not be reflective of all migrations. We reviewed guidance, planning, and performance documents at the four selected agencies. Specifically, we reviewed planning documents that describe shared services migration purpose and goals, the composition and responsibilities of the project management team, and estimated costs and savings; documented results of market research and analyses of alternatives; risk management strategies; service level agreements and performance metrics; communication plans for stakeholders; and reports that capture lessons learned. For each of the illustrative example agencies, we interviewed agency officials involved with shared services migrations. At Justice, we interviewed the project management team overseeing the HR migration to the NFC. At Education, we interviewed the officials who reviewed the Department’s HR and financial management shared services options. At the two FSSPs, we interviewed officials knowledgeable about the outcomes and challenges associated with past and ongoing federal shared services initiatives. We also interviewed subject-matter experts who were involved in public and private shared services migrations as customers, providers, or consultants. We met with members of the Shared Services Leadership Coalition, an interest group promoting shared services solutions involving commercial vendors. The members who participated in the group interview discussed shared services benefits, challenges, and lessons learned. We also met with members of the nonprofit Partnership for Public Service Shared Services Roundtable. The roundtable members who participated in the group interview are federal employees involved with shared services operations. They represented a mix of small and large agencies. To further address the second objective, we reviewed OMB and GSA’s efforts to identify and address challenges and lessons learned from past migrations, including the new shared services action plan OMB released in March 2018. We assessed the extent to which OMB and GSA’s plan and guidance are designed to facilitate better shared services outcomes using criteria such as standards for internal control in the federal government, principles identified in our previous work related to addressing major management challenges, and the Association of Government Accountants criteria for effective marketplaces. During our interviews with customer and provider agency officials and subject-matter experts, we asked for their perspectives on these efforts and the likely effect they will have on ongoing and future shared services migrations. We conducted this performance audit from June 2017 to March 2019 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments Key contributors to this report include Sonya Phillips (Assistant Director), Jessica Nierenberg (Analyst-in-Charge), Rose Almoguera, and Monique Nasrallah. Faisal Amin, Ann Czapiewski, Timothy J. DiNapoli, Jared Dmello, Robert Gebhart, Amanda Gill, Dave Hinchman, Gina Hoover, Valerie Hopkins, John Hussey, Heather Krause, Michael LaForge, Laura Pacheco, Paula M. Rascona, and Kevin Walsh also contributed to this report.

Why GAO Did This Study The federal government can reduce duplicative efforts and free up resources for mission-critical activities by consolidating mission-support services that multiple agencies need—such as payroll or travel—within a smaller number of providers so they can be shared among agencies. However, migrating to a shared services provider has not consistently increased cost savings, efficiencies, or customer satisfaction, according to OMB and others who have observed these migrations. GAO was asked to review previous shared services initiatives. This report: (1) identifies the progress and challenges associated with federal shared services initiatives for selected HR and financial management activities and (2) assesses OMB and GSA's actions to address those challenges. GAO analyzed planning and performance documents and interviewed officials from selected customer and provider agencies and from agencies involved with shared services policy and guidance. GAO also interviewed subject-matter experts familiar with shared services. GAO reviewed steps OMB and GSA are taking to identify and address challenges from past migrations to improve shared services performance. What GAO Found Efforts to promote greater use of shared services for human resources (HR) and financial management activities resulted in some cost savings and efficiency gains, but challenges impeded more widespread adoption. For example, the Office of Personnel Management estimates that shared services for HR, including payroll resulted in more than $1 billion in government-wide cost-savings and cost avoidance between fiscal years 2002 and 2015. However, challenges include limited oversight, demand uncertainty among providers, and limited choices for customers. To address these challenges, the Office of Management and Budget (OMB) and the General Services Administration (GSA), as the shared services initiative leaders, introduced a new marketplace model in 2018 meant to better meet the needs of customers and service providers by offering more choices for purchasing shared services (see figure). They are also working on plans to create Service Management Offices and Task Order Review Boards to work with agencies to adopt standards for common management activities. GAO found that OMB and GSA were following some key change management practices such as improving interagency collaboration in their design of the marketplace model. However, implementation weaknesses may limit their success. For example, OMB and GSA do not have a plan to monitor the implementation of NewPay, a 2018 payroll shared services initiative designed to determine how well the new model works. A monitoring plan which includes performance goals and milestones could help OMB and GSA avoid gaps in service or costly delays as agencies transition to the new model for obtaining shared services. What GAO Recommends GAO is making four recommendations to OMB including to work with GSA to finalize a plan for monitoring the implementation of NewPay, among other actions. OMB staff did not comment on GAO's recommendations, but noted that OMB may update its shared services policy in the future.

gao_GAO-18-278

gao_GAO-18-278_0

Background ARPA-E’s typical funding announcement and award selection process begins with the agency hiring a program director responsible for identifying a gap in energy technology research and developing a program to fill that gap. ARPA-E is required by statute to achieve its goals through energy technology projects that, among other things, accelerate transformational technological advances in areas that industry on its own is not likely to undertake because of technical and financial uncertainty, while also ensuring that its activities are coordinated with, and do not duplicate the efforts of, programs and laboratories within DOE and other relevant research agencies. ARPA-E’s efforts to identify existing energy technology research gaps and to design a program to address those gaps involve research; consultation with scientific experts, including a workshop with outside experts; and internal discussions within ARPA-E. From this process, program directors develop funding opportunity announcements that describe the technical requirements specific to each program’s technology area that applicants have to meet, as well as the four standard criteria that ARPA-E uses to guide its merit selection process. Following the issuance of a funding opportunity announcement, ARPA-E employs the following multi-stage process to merit review applications, make funding award decisions, and monitor projects: Concept paper. Applicants initially submit a 4- to 7-page abstract of their projects. Scientific experts from government, industry, and academia serve as reviewers. Full application. After reviewing concept papers, ARPA-E encourages some applicants to submit full applications. Full applications are generally quite extensive, requesting information on the technical and financial aspects of the proposed project, among other things. ARPA-E officials we interviewed noted that these applications are frequently more than 100 pages and can take 30 to 45 days for the applicant to develop. Full applications are reviewed against the selection criteria by leading scientific experts in the relevant field and assigned numerical scores. Reply to reviewer comments. After reviewing a full application, reviewers provide comments and questions to the applicants, who then have the opportunity to respond. Selection. A three- to four-person panel, chaired by the relevant ARPA-E program director, considers the reviewers’ comments and numerical scores and recommends applications for an award. The final decisions on which applicants to select for award negotiations are made by the selecting official, usually the Director of ARPA-E. Award negotiations. Once selections are made, ARPA-E program directors work closely with selectees to negotiate the terms and conditions of their award. These negotiations include, among other things, developing a project plan with technical milestones that are to be met during the 2 to 3 years that the award is being funded, a budget and management plan, and an intellectual property and data management plan. Funds are awarded once negotiations regarding the terms and conditions of the award are concluded. ARPA-E seeks to complete negotiations regarding the terms and conditions of an award within approximately 100 days of sending a letter notifying an applicant that they have been selected for award negotiations. Selectees may be allowed to begin spending money to start work on their projects up to 90 days prior to the completion of award negotiations. However, these expenditures are made with the risk that applicants may not be reimbursed if award negotiations are unsuccessful and ARPA-E does not fund the award. Monitoring. ARPA-E monitors and supports the projects it funds through quarterly reviews and site visits. At any point during the award, ARPA-E may decide whether to continue or terminate the project based on whether agreed-upon project milestones are being met. DOE Developed and Implemented a New Process to Assess the Department’s Financial Assistance against the Administration’s Priorities In 2017, DOE developed and implemented a new review process to assess DOE financial assistance for new work against the current administration’s priorities, including financial assistance for which ARPA- E had already made award selections. DOE reviewed and approved ARPA-E’s opportunities for financial assistance on a rolling basis from May to September 2017, and nearly all were approved to proceed. The formal review of DOE financial assistance officially began on May 4, 2017, when DOE’s Chief of Staff issued a memorandum stating that funding opportunity announcements and determinations of non- competitive financial assistance would be reviewed to ensure consistency with the administration’s priorities. According to the memorandum, DOE agencies that award financial assistance—referred to in this report as funding organizations—were to provide information about each competitively selected funding announcement and determination of non- competitive financial assistance by May 15, 2017. This information included, for example, a brief description of the financial assistance, the number and amount of planned awards, and the technology readiness level of the projects being funded. DOE Office of Management officials told us that the agency’s financial assistance review lasted through September 2017, as some DOE organizations continued to submit new financial assistance for review, but that the review was largely completed by August 10, 2017. However, while the formal review of DOE financial assistance began in May, award negotiations for ARPA-E-funded projects were suspended nearly 1 month earlier. Specifically, according to ARPA- E officials, DOE’s Deputy Chief of Staff verbally directed ARPA-E on April 6, 2017 to stop all ongoing award negotiations. Figure 1 shows the timeline of DOE’s review of ARPA-E financial assistance. Pursuant to the DOE Chief of Staff’s May 4th memorandum, ARPA-E and other DOE funding organizations submitted the requested information to the DOE review team, which was coordinated and facilitated by the Director of DOE’s Office of Management. Other members of the financial assistance review team included DOE’s acting Chief Financial Officer; deputy assistant secretaries, chiefs of staff, and senior advisors at several DOE funding organizations; and members of the department’s congressional affairs and public affairs staff. According to DOE Office of Management officials we interviewed, the review team assessed the department’s financial assistance against five criteria: Whether the financial assistance was statutorily mandated; Whether the financial assistance was described in congressional Whether the financial assistance was consistent with administration priorities, as identified in budget documents and other statements from the President and Secretary of Energy, among other things; What technology readiness level the financial assistance was intended to fund; and Whether the technology encompassed by the project was already being funded by the private sector or others. DOE Office of Management officials stated that the review team did not use the above criteria to assign quantitative scores to evaluate the department’s financial assistance; instead, the team collaboratively discussed each opportunity for assistance. In most cases, the review team was able to reach consensus on whether the financial assistance aligned with the administration’s priorities. DOE Office of Management officials also noted that they met with ARPA-E leadership to obtain additional information about ARPA-E financial assistance on three occasions during the course of the review. ARPA-E officials said that, in addition to those three meetings, they provided written information to address questions received from the review team and to provide additional context regarding ARPA-E financial assistance. In total, DOE’s review team assessed 6 ARPA-E fiscal year 2017 or prior- year funding opportunity announcements for which applicants had been selected for award negotiation, 7 fiscal year 2017 announcements in the earlier stages of the merit review and selection process, 2 fiscal year 2017 announcements that had not yet been released, and 17 opportunities for financial assistance where ARPA-E funded renewals or new work under a determination of noncompetitive financial assistance. According to DOE Office of Management officials, the review team worked as quickly as possible to review all of DOE’s financial assistance to minimize potential disruptions for recipients and DOE’s funding organizations. Once the review team approved an opportunity for financial assistance, DOE funding organizations were allowed to resume work, DOE Office of Management officials told us. Figure 2 shows the total cumulative funds for ARPA-E financial assistance approved by the review committee at various stages in the review. For example, as shown in Figure 2, the review team approved roughly $158.3 million (55.6 percent) of ARPA-E’s proposed financial assistance on May 18, 2017, 3 days after the deadline for DOE funding organizations to submit information to the review team. The remaining proposed financial assistance was approved in several stages from June through August 2017. As of August 25, 2017, all of ARPA-E’s competitively selected funding opportunity announcements, renewals, and determinations for noncompetitive financial assistance, where selectees had been selected for negotiation, were approved by the review team, representing roughly $265 million, or 93.1 percent, of all ARPA-E funding reviewed by the team. DOE Office of Management officials also stated that the financial assistance review team made a decision early in the course of the review to honor all existing DOE commitments to fund new work. These officials said that this extended to commitments made to entities that had been selected for award negotiations, even though the department does not officially commit to providing funds until such negotiations are completed and the award is finalized. However, according to ARPA-E selectees we interviewed, this message was generally not communicated to them, which led to uncertainty about whether their projects would be funded. In contrast, the review team recommended that the DOE Chief of Staff cancel ARPA-E’s Facsimile Appearance to Create Energy Savings funding announcement, which had accepted full applications but had not selected any applicants for award negotiation. This opportunity would have funded the development of advanced information technology that could allow for three-dimensional digital representation of a person in a room nearly indistinguishable from the person being there in real life, which might allow for increased telecommuting. DOE Office of Management officials told us that the review team reached this recommendation in part because this technology was already being funded by the private sector. As of November 2017, DOE Office of Management officials said the review team had cancelled 3 other DOE funding announcements as a result of the review. According to information we collected, DOE’s review of ARPA-E financial assistance, as part of the DOE-wide review process, did not require the President to send a special message under the Impoundment Control Act. Specifically, the delay in obligating ARPA-E funds for financial assistance examined through DOE’s review process was for programmatic reasons. DOE officials explained that the purpose of the review was to ensure that the agency’s financial assistance aligned with the priorities of the current administration. DOE’s Financial Assistance Review Created Uncertainty for ARPA-E Selectees, Which Led to Delayed Project Timelines and Staffing Difficulties, among Other Impacts According to the 10 ARPA-E selectees we interviewed, DOE’s financial assistance review process created uncertainty, which led to a variety of impacts—the most frequently cited of which were potentially delayed project timelines and difficulties staffing project teams. Selectees told us that they received little communication from ARPA-E during the review process, and they indicated that additional information about review timelines and potential effects on their awards would have helped them manage some of the uncertainty they experienced during the review process. DOE Office of Management officials said the fiscal year 2017 review process helped to better identify and coordinate future financial assistance department-wide on crosscutting issues. DOE is conducting the fiscal year 2018 review process prior to publicly issuing funding announcements. As a result, DOE Office of Management officials said, the delays and uncertainty that selectees experienced in fiscal year 2017 should be reduced. In our structured interviews with ARPA-E selectees, the most frequently cited impact of the uncertainty caused by DOE’s financial assistance review was the potential need to delay project timelines. All of the ARPA- E selectees we interviewed told us that they might need to extend their project timelines because of uncertainty caused by DOE’s review. Four of these selectees noted that the delay caused by DOE’s review could cause additional, cascading delays in their timelines. For example, 1 selectee we interviewed said that it would need to re-issue a hiring announcement it had publicized prior to the review because the review prevented it from hiring someone. In addition, the selectee would need to resubmit the hiring announcement to the university and state human resources departments for approval, which could take months to process. Another selectee said that it missed 2 months of a 3-month planting season because it could not start project work, and had the delay lasted any longer, the selectee would have missed an entire year of data collection on the project. Selectees also cited challenges to staffing project teams as a result of the uncertainty caused by the review. Selectees stated that delays caused by the review affected their ability to hire team members they had planned to hire based on their original schedule, as potential members moved on to other projects or took different jobs. For example, 9 selectees told us that they delayed hiring new project team members while DOE’s review was occurring. Four selectees we interviewed said that they had difficulty retaining staff during the review process. For example, 1 selectee had to lay off 2 of the company’s 15 staff members because of the delay in receiving funding, and several other staff members left voluntarily. Furthermore, the selectee said laying off these staff members resulted in an increase in the company’s unemployment taxes, which was expensive for a small-sized company. Four other selectees that we interviewed said they had to assign existing project team members to other funded work or general activities because they could not begin work on their ARPA-E project until they received funding. Selectees we interviewed cited additional impacts associated with the uncertainty caused by DOE’s financial assistance review. These impacts included: Delaying equipment purchases. Four selectees reported that they had to delay purchasing important equipment needed to execute their project. One selectee noted that the delay caused by the review was long enough that price quotes it had received from equipment sellers expired, and that prices could increase in later quotes. Changes to project scope. Two selectees told us that they might need to limit the planned scope of their projects to be able to complete them in the proposed timeline. For example, 1 selectee said its project involves helping to scale up three to four different technologies a year, which it might not be able to do if it has to adhere to its initial timeframes. Loss of advantage against potential competitors. Four of the selectees we interviewed said that the delay may have caused their technology to fall behind their potential competitors in some way. For example, 1 selectee noted that it was working in a competitive environment for its technology, with ongoing efforts in multiple countries, and reported that its project might have fallen behind others’ efforts as a result of delays associated with DOE’s review. However, 4 other selectees said that the review was not likely to cause any loss of competitiveness. Impacts on external project partners. Three selectees noted that DOE’s review caused uncertainty for partners on their projects, including partners that provide external funding. For example, 1 selectee told us that private investors in its technology area are most active in the fall and that its project team might not be able to seek a second round of funding if it could not demonstrate the necessary technical results of the project by then. Impacts on pre-award spending reimbursements. One selectee reported that it had to cease certain pre-award spending. The selectee said that it spent roughly $10,000 on equipment and 150 hours of labor prior to DOE’s financial assistance review, but it could not submit invoices for these expenditures to ARPA-E while the review was ongoing and would not be able to if its project was ultimately not approved. Furthermore, the selectee said that even if the award was approved, the delay might result in expenditures falling outside the 90-day window of allowable pre-award expenditures, which would require obtaining approval from ARPA-E to be reimbursed. Selectees we interviewed also stated that they received little communication from ARPA-E during the review, which contributed to the uncertainty about the status of their projects. Specifically, 6 of the selectees said that they would have liked additional information from ARPA-E on a variety of topics related to the review. For example, 4 selectees said they would have liked additional information about the review timeline and when it was planned to be completed. One of these selectees told us that a written document from ARPA-E indicating a rough time frame and next steps would have helped facilitate better planning for their project team. Two selectees we interviewed wanted additional information about whether they could renegotiate their timelines once the review was completed. Three selectees told us that they would have liked additional information about whether the review would cause them to lose their funding. ARPA-E officials we interviewed told us that they made three separate requests to DOE’s Deputy Chief of Staff to learn what they could communicate to selectees about the April 6, 2017, verbal order and the review process. ARPA-E officials told us that they were directed by the Deputy Chief of Staff not to communicate with selectees about the verbal order until receiving guidance from his office. ARPA-E developed proposed language to share with selectees but did not receive approval from the Deputy Chief of Staff to distribute it. DOE Office of Management officials we interviewed told us that they did not issue guidance to ARPA- E or other DOE funding organizations about how the organizations should communicate with selectees during the review. In contrast to its fiscal year 2017 review, DOE began its 2018 financial assistance review in August 2017, prior to publicly issuing funding announcements. On August 10, 2017, DOE’s Office of Management sent an email to DOE funding organizations directing them to submit descriptions of their proposed financial assistance by September 8, 2017. Because the review will occur prior to publicly issuing funding opportunity announcements, and thus before any recipients apply or are selected, DOE Office of Management officials said the delays and uncertainty that selectees experienced in fiscal year 2017 should be reduced. DOE Office of Management officials told us that—aside from changing the timing of its 2018 financial assistance review—the review team’s membership and evaluation criteria will be largely the same as for the fiscal year 2017 review. The officials said that they discussed the review process with senior leaders in DOE’s funding organizations to help ensure that they understood the priorities, expectations, and steps of the review process. The officials also told us that the review team developed additional guidance to clarify certain issues that arose during the fiscal year 2017 review. This additional guidance included: On August 17, 2017, funding organization managers were informed that continuation awards—those where the activity is presently being funded—would be exempt from submission to the review team and can continue to move forward. On August 29, 2017, funding organizations were informed that they should identify financial assistance that falls under one of seven crosscutting research issue areas. According to DOE Office of Management officials we interviewed, a key benefit of the fiscal year 2017 review process was that the review team noticed that DOE had several funding announcements at multiple funding organizations related to these areas. DOE’s funding organizations may be able to coordinate to issue a consolidated funding announcement in these crosscutting research issue areas, to ensure efforts are complementary and not duplicative. Furthermore, DOE Office of Management officials we interviewed said that knowing which funding organizations are funding work in these areas will support DOE meetings on crosscutting issues. Agency Comments We provided a draft of this report for review and comment to the Secretary of Energy. DOE provided technical comments, which we incorporated as appropriate. We are sending copies of this report to the appropriate congressional committees, the Secretary of Energy, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff members have any questions about this report, please contact me at (202) 512-3841 or neumannj@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix I. Appendix I: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the individual named above, Chris Murray (Assistant Director), Perry Lusk (Analyst-in-Charge), Antoinette C. Capaccio, John Delicath, Justin Fisher, Kimberly McGatlin, Dan Royer, Tind Shepper Ryen, Lauren G. Sherman, and McKenna Storey made key contributions to the report.

Why GAO Did This Study ARPA-E provides funding for research to overcome long-term and high-risk technological barriers in developing energy technologies. Since 2009, ARPA-E has awarded approximately $1.3 billion to universities, public and private companies, and national laboratories to fund energy research projects. Starting in May 2017, DOE began reviewing its financial assistance department-wide, including ARPA-E's, to determine if it met the administration's priorities. GAO was asked to examine this review process as it pertained to ARPA-E. This report describes (1) how DOE implemented the financial assistance review process; and (2) the perspectives of ARPA-E selectees on the impacts of the review process. GAO reviewed documents and interviewed officials at ARPA-E and DOE's Office of Management, which coordinated the review. GAO also interviewed a nonprobability sample of 10 of the 68 ARPA-E award selectees whose financial assistance was evaluated under the review. GAO identified selectees to interview based on representation across ARPA-E's recipient types, including universities, private companies, and national laboratories, among other criteria. While the views of selectees GAO interviewed cannot be generalized to all affected ARPA-E selectees, they provide illustrative examples of the effects of DOE's review. What GAO Found The Department of Energy (DOE) developed and implemented a new process to review its financial assistance to ensure that all new work funded by the department—including by DOE's Advanced Research Projects Agency-Energy (ARPA-E)—was consistent with the current administration's priorities. The review process covered funding opportunity announcements as well as certain other types of financial assistance. New awards were delayed until the review of the underlying financial assistance opportunity was completed. DOE reviewed and approved ARPA-E's financial assistance on a rolling basis from May through September 2017, and nearly all ARPA-E financial assistance was approved. DOE Office of Management officials met with ARPA-E officials on several occasions to discuss their review of ARPA-E financial assistance. DOE officials GAO interviewed said they wanted to complete the review as quickly as possible to minimize effects on DOE programs. GAO determined that the delay was not reportable under the Impoundment Control Act. The Impoundment Control Act requires the President to notify Congress if an agency wants to withhold the obligation of funds. GAO has separately informed Congress of an impoundment of $91 million in funds that were not allocated to any financial assistance awards, and was not related to DOE's review process. According to the 10 ARPA-E project selectees GAO interviewed, DOE's financial assistance review process created uncertainty, which led to a variety of project impacts. The impacts most commonly cited by selectees included potentially delayed project timelines, as well as difficulties in staffing their project teams, among other impacts as shown below. DOE officials GAO interviewed said that they are reviewing DOE financial assistance in fiscal year 2018. DOE officials said that a key benefit of the fiscal year 2017 review process was an opportunity to better identify and coordinate future financial assistance department-wide on crosscutting issues. However, DOE plans to review fiscal year 2018 financial assistance prior to issuing funding opportunity announcements to the public, and thus before any recipients apply or are selected. As a result, DOE officials said, the uncertainty that ARPA-E selectees experienced during the fiscal year 2017 review process should be reduced. What GAO Recommends GAO is not making any recommendations. DOE provided technical comments, which GAO incorporated as appropriate.

gao_GAO-18-198

gao_GAO-18-198_0

Background Mobile devices use wireless networks to enable voice and data communications. Mobile wireless networks comprise several components and provide coverage based on dividing a large geographic area into smaller areas of coverage known as “cells.” Each cell contains a cell site—a base station equipped with an antenna—to receive and transmit radio signals to mobile devices within its coverage area. (See fig. 1.) The cell sites are often located on a tower, rooftop, or other structure to provide coverage to a wide area. For a mobile device to transmit and receive signals, it must be within range of a cell site antenna. In many areas, a mobile device is able to transmit and receive signals from multiple cell sites. Each cell site is linked to a base station controller that manages communications between the cell site and the mobile switching center (e.g., routes and hands over calls). The mobile switching center then directs voice and data traffic to landline phones, other cell phones via a carrier’s network, or the Internet. Backhaul facilities provide transport for this voice and data traffic, and backhaul can be provided over fiber optic or copper cables or wirelessly via microwave facilities. According to FCC, there are four “nationwide” mobile wireless carriers— AT&T, Verizon, T-Mobile, and Sprint—with networks that cover most of the United States. The industry also includes dozens of other carriers, many of which provide service in a specific, sometimes rural, geographic area. According to an FCC report on the wireless industry, most consumers in the United States have the ability to choose among multiple carriers with wireless network coverage in their area, and wireless carriers typically compete on price, network quality, and the availability of mobile devices with innovative features. Federal law states that FCC must take into account whether its actions will encourage competition in mobile wireless networks. Some wireless carriers own a portion or all of the structures that host cell sites, but wireless carriers mostly lease space from independent companies that own or operate a majority of the towers and other structures that host cell sites. Mobile wireless networks face several kinds of risks that could affect the network’s physical components, resulting in disrupted service or an outage. Government reports generally identify three types of physical risks facing wireless networks: Natural disasters, such as hurricanes, tornados, wildfires, and earthquakes. Manmade events, such as terrorist attacks and damage associated with theft or another malicious act. Accidents, such as cable damage due to digging or locating errors and damage associated with a vehicle accident. The potential effects related to these physical risks include damage to wireless network components that requires wireless carriers and other providers to make repairs or replace equipment to restore service. For example, flooding, which can occur with a hurricane or heavy rain, could damage the cable or other equipment submerged in water. Wildfires can damage network components like antennas and backhaul facilities (including fiber optic and copper lines and microwave towers) as well as equipment in buildings if the buildings are damaged or destroyed. In addition to physical risks, wireless networks face risks stemming from their dependence on other sectors and providers. One key dependency identified by several government and industry reports is the reliance on commercially provided electricity, referred to in this report as commercial power. Several components—including the mobile switching center, antennas at cell sites, and consumer devices—may rely on commercial power. Therefore, loss of electric power may result in a loss of wireless communications. Another key dependency for wireless networks is backhaul used to get data from an end user to a major network. Wireless carriers can provide backhaul but typically obtain it from another communications provider, such as a local telephone company or cable company. An outage in the backhaul network can cause an outage that affects one or more cell sites or a portion of the wireless network. FCC has reported that the loss of backhaul service is a major cause of a cell site’s unavailability, which can lead to wireless outages. Also important is having clear roads and highways, as wireless carriers’ personnel or contractors need to be able to access cell sites to repair or replace equipment, or deliver fuel for generators that are sometimes located at cell sites. Resilience is the ability to prepare for and adapt to changing conditions and withstand and recover rapidly from disruptions, according to Presidential Policy Directive 21. Owners and operators of wireless networks can take a variety of actions to manage different risks, including various physical risks, according to the Communications Sector-Specific Plan. These actions can be designed to achieve different aims, including to prepare for incidents, like creating and exercising disaster recovery reduce a specific vulnerability, like elevating or moving a mobile switching center in a flood-prone area to a higher location; mitigate the consequences of an incident, like installing backup power—using batteries with a limited supply of power or generators that run on diesel or other fuel sources—to support continued wireless service during a commercial power outage; or enable efficient response and restoration following an incident, such as deploying portable cell sites on trucks and other equipment after an incident to provide wireless communications when the network experiences an outage or a significant disruption. FCC, pursuant to the Communications Act of 1934, as amended, is charged with regulating interstate and international communications throughout the United States, which means that FCC regulates wireless networks and carriers, among other responsibilities. It develops and administers policies and rules to advance the security and reliability of the nation’s communications infrastructure; this responsibility includes, among other topics, network resiliency, public safety communications, and communications infrastructure protection. FCC administers two web-based outage-reporting systems to help it oversee network reliability and resiliency: NORS: Carriers are required to report details about service disruptions or outages (e.g., cause, location, and duration) to their communications systems that meet specified thresholds set forth in regulation. FCC uses NORS data to monitor trends in communications outages and to try to identify and address any shortcomings or issues going forward. Disaster Information Reporting System (DIRS): Carriers can voluntarily report on the status of communications infrastructure during an emergency event in DIRS. For example, wireless carriers report daily on the number of cell sites, by county, that are out of service by reason (e.g., power outage, physical damage). FCC activates DIRS in response to an event and then uses these data to track network restoration during and after an emergency event. DHS also plays a role in wireless network resiliency as the lead agency for coordinating and prioritizing security and resilience activities for the communications sector. Presidential Policy Directive 21 establishes national policy to strengthen the security and resilience of critical infrastructure and states that the federal government shall work with critical infrastructure owners and operators to do so. DHS’s Office of Cybersecurity and Communications, within the National Protection and Programs Directorate, leads this coordination for the communications sector as the sector-specific agency, and this office works with the Communications Sector Coordinating Council and the Communications Government Coordinating Council to set goals, objectives, and activities for the sector. During a national emergency or disaster, DHS also coordinates response efforts for communications systems in its role as the coordinator for Emergency Support Function #2 – Communications (ESF-2). Specifically, two DHS components—the Federal Emergency Management Agency (FEMA) and Office of Cybersecurity and Communications—lead the federal government’s work to support the restoration of communications infrastructure, coordinate response efforts, and facilitate the delivery of information to emergency-management decision makers. DHS has direct access to FCC’s NORS and DIRS data to support its work. Other DHS components also have responsibilities related to wireless network resiliency. For example, the Science & Technology Directorate conducts research in the area of wireless and other communications network resiliency, although its focus is on communications for the public-safety community. Within the Department of Commerce, NIST also plays a role in promoting network resiliency by sponsoring the Community Resilience Panel. According to NIST, the Community Resilience Panel is sponsored by NIST and co-sponsored by other federal agencies to promote collaboration among stakeholders to strengthen the resiliency of infrastructure that communities rely on, including communications infrastructure. As part of this mission, the panel seeks to identify policy and standards-related impediments to community resiliency, raise awareness of sector dependencies and of the cascading effects of disasters, and identify potential resiliency metrics. Wireless Outages Caused by Physical Incidents Have Increased since 2009, and Outages due to Natural Disasters Lasted Longest Trends in Number and Reported Causes of Wireless Outages The number of wireless outages attributed to a physical incident increased from 2009 to 2016 (see fig. 2). Specifically, the number of outages with a physical incident reported as a root cause increased from 189 outages in 2009 to 1,079 in 2016. The number of outages increased substantially during the first few years of this period and then was relatively stable, which mirrored the trend for all wireless outages. According to FCC officials, the increase in reported outages was due to increases in both the number of wireless customers and wireless infrastructure over this period, as well as due to FCC’s outreach to wireless companies to clarify the thresholds for which carriers are required to report wireless outages to help ensure that carriers were consistently and fully reporting outages. From 2009 to 2016, about one- third of all wireless outages reported to FCC (6,002 of 18,325) were attributed to physical incidents. Of wireless outages reported to FCC that were attributed to physical incidents, most were due to accidents, described below: Accidents—which include cable damage due to a backhoe cut, among other causes—were the root cause for 74 percent of wireless outages attributed to a physical event. Natural disasters—including tornados and wildfires—were the root cause for 25 percent of wireless outages attributed to a physical incident. Manmade events—which include damage associated with theft or other intentional damage to facilities—were the root cause for the remaining 1 percent of these outages. FCC typically suspends NORS reporting requirements in areas where FCC activates DIRS reporting for an emergency event, generally a natural disaster. For example, when FCC activated DIRS reporting for all counties in Puerto Rico and the U.S. Virgin Islands in response to Hurricane Maria in September 2017, FCC suspended NORS reporting requirements for those counties. As a result, FCC officials said that NORS data can undercount the number of wireless outages due to natural disasters. For a large natural disaster, however, FCC still can receive NORS reports for wireless outages outside the DIRS reporting area that are due to the natural disaster. Hurricane Irma In September 2017, Hurricane Irma made landfall as a Category 3 hurricane in Florida, having previously tracked near Puerto Rico and the U.S. Virgin Islands. The hurricane produced sustained winds of nearly 115 miles per hour as it made landfall in Florida. In the days that followed, the hurricane’s impact was felt over the southeastern United States, with nearly 16 inches of rain falling over portions of Florida and high winds observed in five states. The President issued disaster declarations covering portions of Puerto Rico, the U.S. Virgin Islands, Florida, and Georgia. The damage from Hurricane Irma—both damage to wireless network infrastructure and damage resulting in power outages—created wireless service disruptions and outages in certain impacted areas. In particular, over half of cell sites were out of service for 3 or more consecutive days in five counties in Puerto Rico and in two counties in the U.S. Virgin Islands, according to data from wireless carriers reported to FCC. Within a week, only 6 percent of cell sites were out of service in reporting counties in Puerto Rico, but a majority of cell sites remained non-operational in the U.S. Virgin Islands; in one county, St. John, 90 percent of cell sites remained out of service a week and a half after landfall. In southern Florida, three counties had more than half of cell sites out of service for 4 straight days. The number of out-of-service cell sites decreased over time, so that less than 20 percent of cell sites were out of service in these counties within a week. Looking more broadly across all counties for which FCC collected data in Florida, Georgia, and Alabama, about 13 percent, 2 percent, and 1 percent of cell sites in the reporting area were out of service 4 days after Hurricane Irma’s landfall, respectively. cause and contributing factor fields. Looking across cause fields, wireless outages citing an accident were most common, particularly from 2010 to 2016, as shown in figure 3. Wireless outages citing a natural disaster were less common, although there were several spikes in the number of outages citing a natural disaster. Some of these spikes correspond with major natural disasters like the derecho affecting Midwest and Mid-Atlantic states in 2012 or Hurricane Matthew in 2016. Manmade events were rarely reported as the cause or contributing factor. Duration of Wireless Outages Attributed to Physical Incidents While less common than accidents, wireless outages attributed to natural disasters lasted much longer than outages attributed to other physical incidents. Specifically, figure 4 shows that outages where a natural disaster was cited as the root cause were often twice as long as outages attributed to an accident or manmade event. From 2009 to 2016, the annual median duration of wireless outages attributed to accidents ranged from 8 hours to 16 hours, compared to natural disasters, which ranged from 19 to 36 hours. Due to this longer duration, wireless outages attributed to natural disasters have a greater impact on the public as it is left without key means of communication for longer periods of time. In addition, an industry association told us that even though public safety officials primarily use dedicated communication networks, like land mobile radio networks, to carry out their work, they also rely on their mobile devices that use commercial wireless networks for maps and other applications. Ten of 24 stakeholders we interviewed said that natural disasters pose the greatest risk to wireless networks as they have the most intense consequences. Natural disasters can result in physical damage to or flooding of critical network components, and fallen trees and debris can temporarily block transportation routes, keeping repair crews from inoperable cell sites and other network components, as described in the Community Resilience Planning Guide for Buildings and Infrastructure Systems. Further, the failure of other systems like commercial power upon which wireless networks depend can lead to cascading failures in communications networks. One industry association we spoke with said that natural disasters are the primary risk to wireless network resiliency as these events usually create the largest outages with the longest durations. Location of Wireless Outages Attributed to Physical Incidents By location, the number of wireless outages attributed to physical incidents increased from 2009 to 2016 in some states, including several of the most populous states such as California and Texas (see fig. 5). Most of the recent expansion of wireless networks has tended to be in the most populous states, as those states contain the most customers and the highest densities of customers, according to FCC officials. In addition, the thresholds for which carriers are required to report wireless outages in NORS are such that many outages that affect primarily rural areas will not accumulate enough user minutes to be reportable. However, the number of wireless outages with a physical incident as the root cause was relatively steady in many states or had spikes that generally corresponded with major natural disasters like the 2012 derecho. For more detailed information on the location of all reported wireless outages that occurred from 2009 through 2016, including the cause and number of users associated with these outages, see an interactive graphic which can be viewed at http://www.gao.gov/products/gao-18-198. Wireless Network Dependencies Power failures and failures in other providers’ networks (e.g., backhaul) played a role in the majority of wireless outages attributed to physical incidents. When an accident, natural disaster, or manmade event was the root cause for an outage, we found that wireless carriers often also reported a failure in one of these two key dependencies for wireless networks: Regarding power, 8 percent of wireless outages with a physical incident as the root cause (465 of 6,002 outages) cited power failure as the direct cause of the outage. Nearly all these outages were attributed to a natural disaster. Regarding failures in other providers’ networks, 87 percent of outages attributed to a physical incident (5,206 of 6,002 outages) were due to a failure in another provider’s network, which includes backhaul connecting cell sites to mobile switching centers and onto the broader network. Most of these outages—4,111—cited an accident (i.e., a digging error resulting in cable damage) as the root cause. In 2014, a working group from an FCC-chartered federal advisory committee concluded that there is little to no shared, last-mile transport infrastructure for backhaul that wireless carriers (or other providers) could share dynamically to mitigate the effect of a backhaul failure. Thus, a backhaul outage will often result in a wireless outage. However, the working group identified existing best practices that providers can employ to help reduce or lessen the impact of failure in last-mile backhaul. Federal Agencies and Industry Have Taken Some Actions to Improve Wireless Network Resiliency, but FCC’s Oversight of Industry Actions Is Limited Federal Agencies Largely Continue to Use Existing Mechanisms to Improve Resiliency Since 2013, FCC and DHS have both continued to take action using a range of existing mechanisms to improve wireless network resiliency. These mechanisms include leading communications-specific planning activities and topic-specific research to develop and to share best practices. While these mechanisms are not new, FCC and DHS report updating and adapting these activities based on emerging needs and lessons learned from exercises and emergency events. FCC and DHS actions include the following: Chartering advisory committees that examine resilience: DHS and FCC charter federal advisory committees that have studied how agencies and industry could improve resiliency. For example, one such committee is FCC’s Communications Security, Reliability, and Interoperability Council (CSRIC), whose members include representatives from wireless carriers and other communications companies, industry associations, and federal, state, and local agencies. CSRIC working groups often develop best practices for industry and make recommendations to wireless carriers, FCC, and others to improve network resiliency. One example is a working group that studied how industry could share backup power resources in 2014. FCC maintains a database of best practices and publicizes these through presentations at conferences and in public reports, as it did in a report on communications outages caused by the 2012 derecho. Six stakeholders we interviewed said best practices represent a valuable means to improve resiliency, as for example, best practices are flexible and enable providers to adapt practices as communications networks evolve. One stakeholder attributed CSRIC’s effectiveness in issuing and promoting best practices and information in part to its affiliation with the industry’s regulator, FCC. Other advisory committees that examined resiliency include DHS’s National Security Telecommunications Advisory Committee and FCC’s Technological Advisory Council. Developing and implementing procedures to respond to physical incidents: DHS leads emergency communications response and recovery efforts, as coordinator for ESF-2. For example, within DHS, the National Coordinating Center for Communications (NCC) holds weekly calls with government and industry partners to exchange information as part of NCC’s work to continuously monitor events that may affect communications. These weekly calls sustain relationships and promote readiness that can be leveraged to coordinate a response to an emergency incident, according to DHS and FCC officials and members of the Communications Sector Coordinating Council. During an incident, NCC reports that it holds these calls on a daily basis to understand the status of wireless and other networks—along with FCC outage data and other information collected from carriers—and to support industry response efforts. For instance, NCC officials said that during an incident they can help carriers find available generators or work with local governments to enable carriers to enter disaster areas to make repairs if carriers are denied access. Two carriers we interviewed said the NCC works well to support industry response to and recovery from incidents, as NCC has established response processes to help the communications sector to coordinate with the power industry. According to DHS, NCC participates in the Energy Priority Restoration Group that is dedicated to determining power restoration priority following an incident. While this group includes many sectors, it enables communications providers to help prioritize power restoration for critical communications components, like mobile switching centers. Analyzing wireless outage data to identify trends and areas for further study: As noted above, FCC collects and regularly analyzes data on wireless outages during the regular course of business and during emergency events. FCC meets with each nationwide wireless carrier annually to discuss trends in the carrier’s outages and any issues related to how the carrier completes NORS reports, according to FCC officials and an industry association we interviewed. FCC also analyzes and shares its analysis of NORS data with industry at quarterly meetings of the Alliance for Telecommunications Industry Solutions’ Network Reliability Steering Committee. Specifically, FCC presents trends in NORS outage data for the last 3 years for different types of outages. Such data include trends in the total number and duration of wireless outages. The Network Reliability Steering Committee, at FCC’s request or its own initiative, establishes teams to examine NORS trends and to make recommendations that may increase network reliability and reduce network outages. Based on this work, the team may identify relevant best practices that carriers could use to reduce or eliminate outages or suggest refining or creating a new best practice. Representatives from two industry associations said that FCC meets with industry to discuss outage data and is receptive to feedback on how to improve data-reporting processes and data quality. In addition to these existing mechanisms, federal agencies have initiated some new activities to enhance wireless network resiliency since 2013. Community Resilience Panel: NIST issued the Community Resilience Planning Guide for Buildings and Infrastructure Systems in October 2015 and sponsors the Community Resilience Panel, which aims to reduce barriers to achieving community resilience by promoting collaboration among stakeholders to strengthen the resilience of buildings, infrastructure, and social systems upon which communities rely. The panel held its first meeting in November 2015. The planning guide provides a process that communities can use to improve their resilience by setting priorities and allocating resources to manage risks based on their prevailing hazards. The guide also devotes sections to key infrastructures; the communications section describes components of communications networks, the regulatory environment, and industry standards that can help inform community planning. The Community Resilience Panel also has a Communication Standing Committee comprised of industry and government members. This committee is currently creating additional resources to support communities, including a methodology that communities could use to involve wireless carriers and other communications providers in resilience planning activities. While communities have started to use the guide, NIST officials said it is too soon to measure or point to specific outcomes attributable to the Community Resilience Panel’s work. Post Hurricane Sandy hearings and proposed rule: In 2013, FCC held field hearings on network reliability and continuity. The goals of the hearings were to improve network resiliency, improve restoration, empower the public, and unleash technological solutions. The two hearings included a wide range of panelists including representatives from FCC and FEMA, state and local agencies, consumer groups, wireless carriers, and other communications providers. Following the hearings, FCC issued a notice of proposed rulemaking to promote transparency to the public on how wireless carriers compare in keeping their networks operational during emergency events. Specifically, FCC’s proposed rule would publicly report the number and percentage of each carrier’s cell sites that remained operational during an emergency event to enable consumers to compare wireless carriers when purchasing service. Based on our review of comments, public safety and consumer groups tended to support the proposed rule while industry expressed concerns, in particular, that the public reporting in the proposed rule would not accurately portray the service available during an emergency or be a useful measure to help consumers choose among wireless carriers. FCC decided not to issue a final rule, stating in December 2016 that a voluntary industry approach, described below, provided a more appropriate path to improve network resiliency. We asked stakeholders about the results of the actions taken by FCC and DHS, and across the 24 stakeholders we interviewed, there was no consensus regarding needed improvements in DHS and FCC guidance, coordination, or research on wireless resiliency. Seven stakeholders said they did not think there were any gaps or needed improvements from FCC, and six stakeholders said they did not think there were any gaps or needed improvements from DHS. Although most stakeholders identified a need for further federal agency action, they tended to identify dissimilar actions. However, the three state and two local agencies we interviewed noted that more real-time data on wireless outages would help aid their efforts to respond to an incident, which we discuss further below. Voluntary Industry Framework Aims to Improve Wireless Network Resiliency, but FCC Has Limited Plans to Monitor This Framework In April 2016, an industry coalition consisting of CTIA, a wireless industry association, and five wireless carriers announced the Wireless Network Resiliency Cooperative Framework (framework) in response to FCC’s 2013 notice of proposed rulemaking on wireless network resiliency. The framework is a voluntary initiative designed to advance wireless service continuity and information sharing during and after emergencies by enhancing coordination and communication, both among carriers and between carriers and government. The framework has five elements; some elements are specific to disaster response while other elements focus on preparedness and education. Industry has taken steps related to all five elements of the framework, as described in table 1. Furthermore, CTIA representatives told us they have ongoing meetings with representatives from the public-safety community, the outcome of which they expected to be a series of best practices concerning planning before disasters, addressing coordination during and after emergencies, and developing education and awareness strategies in fall 2017. The threshold to trigger the response elements is when DHS activates ESF-2 and FCC activates DIRS. At the time of our review, four events—Hurricane Matthew in October 2016 and hurricanes Harvey, Irma, and Maria in late 2017—had met the threshold to trigger the response elements. Prior to the three events in 2017, FCC officials and three stakeholders we interviewed told us it was too soon to know the effectiveness or results of the framework. FCC is responsible for administering policies to improve resiliency, which include monitoring actions taken by industry, and federal standards for internal control state that management should establish and operate monitoring activities, evaluate and document the results of ongoing monitoring, and then identify changes that either have occurred or are needed. Federal standards for internal control also state that agencies should define objectives clearly and that objectives should be in specific and measurable terms that allow for the assessment of performance. In December 2016, FCC said it would continue to engage with industry on the implementation and use of the framework, and FCC has taken some steps to monitor the framework’s implementation. Specifically, FCC developed a plan to track certain tasks related to the framework in August 2017. This plan tracks the completion of initial tasks related to the framework, such as tracking industry’s publication of best practices to enhance municipal preparedness and resiliency, and confirming the five signatory wireless carriers’ commitment to the framework, and notes that FCC will update its emergency response documents to ensure that the documents reflect the framework and include checklists to validate that carriers take these actions during emergency events (e.g. instituting roaming, providing mutual assistance). In August 2017, FCC also issued a public notice inviting carriers beyond the five signatory wireless carriers to sign on to the framework. However, we found FCC’s plan does not include any steps to document and assess the effect of the framework on the resiliency of wireless networks. In particular, FCC’s plan does not track any outputs or outcomes over time that speak to the results of the framework, such as the number of roaming requests made and fulfilled during an emergency event. FCC’s plan to monitor the framework is still new. According to FCC officials, FCC did not decide what division would lead its monitoring of the framework until August 2017 because it needed to determine which division should have responsibility for the framework. Since the plan was created, FCC has met with industry groups and individual carriers to gather additional information and has updated its plan with this information to track implementation tasks. Overall, by monitoring the outputs and outcomes of the framework, FCC could determine where further changes are needed to help ensure that wireless networks are resilient. In 2016, FCC reported that the framework could produce benefits such as bolstering FCC’s situational awareness and providing consumers with a means to hold carriers accountable for service continuity during emergency events. Yet, seven stakeholders we interviewed, including wireless carriers, said the framework largely codified actions that carriers already generally take to prepare for and respond to an emergency event. In addition, comments submitted to FCC in 2016 were split on whether the framework represented a sufficient path forward, and some stakeholders noted specific issues that they believed could limit the effectiveness of the framework, for example: Four stakeholders we interviewed—an industry association, local agency, state agency, and consumer group—cited the lack of federal agency enforcement or monitoring. Two industry associations stated in joint comments that there was no assurance that all carriers would conduct adequate testing to enable roaming under disasters. A local agency said in comments that the threshold to trigger the response elements was too high; as such, carriers would not be- obligated to implement the elements for more local events. Therefore, monitoring the outputs and outcomes of the framework would help FCC understand the effect of industry formalizing these actions in the framework. Furthermore, although FCC and industry documents that describe and endorse the voluntary framework include broad goal statements, there are no specific measures for what the framework hopes to achieve. As a result, FCC lacks specific and measurable terms to monitor the effect of the framework. The CTIA- and carrier-released public summary of the framework said it aims to advance wireless service continuity and information sharing during and after emergencies and disasters, as well as help consumers be better prepared for future disasters. FCC, when endorsing the framework, said it was a reasonable approach to achieve FCC’s stated goals for the 2013 proposed rule, including promoting availability of wireless mobile services in the event of natural disasters and increasing provider transparency around wireless resiliency. FCC officials told us they have not discussed possible measures to monitor the effect of the framework with industry participants. As the creators of the framework, industry participants could provide insight into such measures. However, FCC officials acknowledged that it will be important to determine what the results of the framework have been in light of the 2017 hurricanes, and that developing measures to assess industry’s efforts under the framework would be beneficial. In addition, FCC has not communicated the framework to all state and local public-safety officials and wireless carriers, potentially limiting its effectiveness. At the time of our review, CTIA and the signatory wireless carriers had released a high-level summary of the framework but no additional documentation on the scope of wireless carriers’ obligations under the framework. Based on our interviews, we found that knowledge of the framework was not widespread. Six stakeholders we interviewed, including representatives of state agencies we interviewed and a non-signatory wireless carrier, were either unaware of the framework or unaware of whether industry had taken actions on any elements of the framework since its announcement. For example, a state emergency manager in one state affected by Hurricane Matthew was unaware of the framework and that FCC, based on one element of the framework, had posted daily status updates on wireless service following the hurricane. This manager noted that those updates would be useful for response efforts. Federal standards for internal control state that federal agencies should externally communicate necessary, quality information to achieve the agency’s objectives and that open communication can also help enable a federal agency to obtain information from external parties. Among the stated objectives of FCC is to advise and assist public safety entities on wireless communications issues and to develop and administer policy goals and plans to promote reliable communications for public safety and disaster management. Moreover, one of FCC’s current strategic objectives is to promote access to effective public-safety communications services used by government as well as all consumers in need. To address this and other objectives, FCC stated that it will facilitate discussions and share information among key constituencies. FCC uses several mechanisms and standing forums to share information and educate constituencies. For example, FCC gives presentations about FCC activities at conferences on public safety communications that include state and local officials. In addition, FCC participates in the regular conference calls hosted by DHS’s NCC through which government and industry exchange information and the Network Reliability Steering Committee’s public quarterly meetings, as noted above. Without greater awareness of the framework, state and local public safety officials may continue to be unaware of tools or other improvements available through the framework that could help them prepare for or respond to an emergency, such as the posting of daily updates on the number of out-of-service cell sites or best practices that could aid resilience. Also, smaller and rural (non-signatory) wireless carriers might be unaware of commitments made by the signatory carriers, such as committing to roaming under disasters that could benefit them and the citizens they serve during an emergency event but may require entering into and testing a roaming arrangement. By actively communicating information about the framework, FCC could also increase the likelihood of receiving information from industry or state and local public-safety officials about any implementation issues or positive results from the framework. In August 2017, FCC created a website that summarized the framework and, as noted above, issued a public notice inviting additional carriers to sign on to the framework. Only two carriers, as of October 2017, beyond the carriers involved in creating the framework publicly informed FCC of their intent to participate in the framework. As of October 2017, FCC officials told us they did not have additional plans to promote awareness of the framework, but noted that it would be important to inform relevant stakeholder groups about the framework, especially those who might remain unaware of it. Stakeholders Cited Advantages and Disadvantages for Options Aimed at Improving Wireless Network Resiliency We identified options that federal agencies could take to further improve wireless network resiliency based on agency reports, federal advisory committee recommendations, peer-reviewed literature, and other reports. The options we identified could be implemented either alone or in combination and are not meant to be exhaustive. We categorized them by their aim—preparedness, response, and awareness. FCC, as the regulator for wireless communications, would be the likely agency to implement many of the options, although DHS or other federal agencies could play a role in implementing some of the options. We asked stakeholders to comment on the advantages and disadvantages, including the feasibility—technical, legal, or other—of each option. The tables below describe identified options by category, along with the most frequently cited advantages and disadvantages. FCC has previously suggested and discussed some of these options, most recently during its notice of proposed rulemaking in 2013. FCC noted that its proposed rule sought to comply with guidance from the Office of Management and Budget to use disclosure requirements or transparency measures where possible in place of prescriptive regulations. However, as noted above, FCC declined to issue a final rule, stating that the proposed rule was problematic in light of substantial concerns raised about proposed metrics and disclosure requirements. Preparedness Two options identified in agency reports and literature intend to improve resiliency by focusing on actions to be taken ahead of an emergency or disaster, as described in table 2. Twelve stakeholders we interviewed raised concerns about the feasibility of the option to require a minimum level of backup power at cell sites due to technical or legal issues. In 2007, FCC adopted a requirement for wireless carriers to provide 8 hours of backup power at cell sites. That requirement was vacated after the Office of Management and Budget disapproved the rule’s information collection requirements. In contrast, nine stakeholders we interviewed were more positive about the feasibility of guidance. Further, FCC created the Broadband Deployment Advisory Committee in January 2017 to provide recommendations on how to accelerate broadband deployment. Two of the committee’s five working groups focus on state and local regulatory barriers and model language for state and municipal code, both of which could provide a model for wireless network infrastructure. Response As shown in table 3, agency reports and literature also included options related to response activities during and after an emergency event. For the first option, FCC officials and six stakeholders we interviewed noted that wireless carriers have on occasion opened up their networks in prior emergency situations, which indicates that the option is technically feasible. For the second option, every state and local agency we spoke with noted the value of having real-time information on wireless outages during an emergency event. FCC collects DIRS data, and these data are confidential when provided to FCC. According to FCC, if outage data were shared with a state or local agency, it may be subject to open records laws that provide a means for the public to gain access to government documents. Awareness Other options are intended to improve wireless network resiliency by fostering transparency, as described in table 4. For some options below, transparency would involve making information publicly available so consumers could use this information when choosing a wireless carrier. Such transparency could give industry an incentive to improve the resiliency of their networks. For example, by setting performance standards or requiring wireless carriers to disclose their efforts to improve resiliency, consumers could compare the performance or practices of wireless carriers. However, some of these options would require defining specific parameters, whether a metric or the specific information to disclose, and seven stakeholders we interviewed noted this could be difficult given factors such as the variation in carriers’ wireless networks and the pace of technological change. For other options below, transparency would involve more selectively sharing information with other public safety agencies to improve coordination and aid planning for possible disruptions to wireless networks during emergencies. Conclusions During natural disasters and other emergencies, wireless network outages can make emergency communications, such as making 911 calls, nearly impossible for the vast number of people who rely solely on wireless communications. The wireless industry sought to enhance resiliency by improving the continuity of wireless service and information sharing during and after emergency events by introducing a voluntary framework. Although FCC stated that this voluntary framework would have many benefits, neither industry nor FCC has identified any specific, measurable objectives that could be used to determine whether the framework meets its broad goals, and FCC has limited plans to monitor the framework’s implementation and use. Absent sufficient monitoring, including identifying specific, objective measures for the framework, FCC lacks information on the framework’s outcomes and overall effectiveness; such information could help FCC identify whether it needs to take steps to address challenges or take other action to further promote wireless network resiliency. Furthermore, FCC does not have any plans to actively communicate information about the framework to public safety officials and industry representatives. A concerted effort by FCC to promote awareness of the framework could help more public safety officials and other industry participants use the framework to prepare for or help mitigate the risks to wireless networks posed by natural disasters and other emergencies. Recommendations for Executive Action We are making the following three recommendations to FCC: The Chairman of FCC should work with industry, to the extent practical, to develop specific and measurable objectives for the Wireless Network Resiliency Cooperative Framework, such as outputs to measure the extent of the framework’s use. (Recommendation 1) The Chairman of FCC should develop a plan to monitor the outputs and outcomes of the Wireless Network Resiliency Cooperative Framework and document the results of its monitoring to evaluate its effectiveness and identify whether changes are needed. (Recommendation 2) The Chairman of FCC should promote awareness about the elements of and any outcomes from the Wireless Network Resiliency Cooperative Framework among state and local public safety officials and other industry stakeholders, such as through existing outreach mechanisms and government-industry forums. (Recommendation 3) Agency Comments We provided a draft of this report to FCC, DHS, and the Department of Commerce for comment. In its comments, reproduced in appendix III, FCC agreed with the recommendations. FCC also provided technical comments, which we incorporated as appropriate. DHS and the Department of Commerce had no comments. As agreed with your office, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies to the appropriate congressional committees, the Chairman of FCC, the Secretary of Homeland Security, and the Secretary of Commerce. In addition, the report will be available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-2834 or goldsteinm@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix IV. Appendix I: Objectives, Scope, and Methodology This report examines federal agency and industry efforts to improve the resiliency of mobile wireless networks in response to natural disasters and other physical incidents since Hurricane Sandy, a natural disaster that caused significant communications outages across several states in late 2012. Specifically, this report examines: (1) trends in mobile wireless outages attributed to physical incidents since 2009 as reported to the Federal Communications Commission (FCC), (2) the actions federal agencies and industry have taken since 2013 to improve wireless network resiliency, and (3) options that federal agencies could take to improve network resiliency and their advantages and disadvantages. This report focuses on the physical risks facing wireless networks; that is, the potential for an unwanted effect from an incident on a network’s infrastructure like towers, antennas, and switches. Therefore, we did not examine cyber risks facing wireless networks. To determine the trends in wireless outages, we analyzed data from FCC’s Network Outage Reporting System (NORS) on wireless outages that occurred from 2009 through 2016. We chose this timeframe to cover 4 years of data before and after Hurricane Sandy. Communications providers, including wireless carriers, are required in regulation to submit outage reports in NORS for network service disruptions that reach certain thresholds. Given our focus on wireless outages, we examined reports in NORS for outages (1) that were reported by wireless companies that identified themselves as either a wireless carrier or Voice over Internet Protocol (VoIP) provider and (2) that were cited, as the reason the outage was reportable, a reporting requirement applicable to a wireless carrier or VoIP provider in 47 C.F.R. Part 4. We analyzed NORS data for wireless outages to determine the total number and the causes of wireless outages that occurred from 2009 through 2016. FCC provides carriers a list of 19 main categories from which a carrier selects a root cause, direct cause, and contributing factor(s) for an outage. We examined the root cause and direct cause for each wireless outage reported to FCC. We collapsed several of the FCC categories to create 9 categories for ease of presentation. Table 5 shows the crosswalk between the 19 FCC categories and our 9 collapsed categories. We also analyzed the data to identify the share of all wireless outages attributed to a physical incident—that is, a natural disaster (e.g., flooding, earthquake, wildfire); accident (e.g., backhoe cut); or manmade event (e.g., theft, malicious act). FCC provides carriers a list of categories from which they select a root cause, direct cause, and contributing factor(s) for an outage. Using these FCC categories, we created three new categories for natural disasters, accidents, and manmade events (see table 6). We based these three new categories on the description and categorization of physical risks in FCC and DHS reports, including the Communications Sector-Specific Plan. To understand the distribution of causes across all wireless outages in our time frame, we focused primarily on the root cause for each wireless outage. However, we also examined the root and direct cause reported for each outage to better understand the multiple factors that may have led to an outage and to understand wireless network dependencies (e.g., power, backhaul). Finally, we examined the number of outages, by month and by year, for which a wireless carrier reported a physical event as the root cause, direct cause, or contributing factor to better understand the total number of outages related to a physical incident during our time period. We also analyzed other characteristics of wireless outages such as location, duration, and whether the failure occurred in another company’s network. To examine location, we focused on three NORS fields—city, state, and description of location—to identify a city(ies) and state for each outage, and then we determined the latitude and longitude for each outage. We also examined NORS data in conjunction with two other data sets. First, data on events for which FCC activated its Disaster Information Reporting System (DIRS) or “DIRS-lite” to understand any correlation between wireless outages and major physical incidents. Second, data from CTIA’s annual wireless survey on the number of wireless subscribers and other measures of wireless networks’ size to understand any correlation between wireless outages and the size of the wireless industry. To assess the reliability of NORS data, we reviewed FCC’s data glossary and other FCC documentation on the NORS data system and data elements. We interviewed agency officials responsible for collecting and analyzing NORS data to understand the manual and automated controls used to review carrier-reported outage information and any potential limitations in the data. We also reviewed relevant data elements for missing data, outliers, and errors. We found the data were sufficiently reliable for the purpose of describing the number and type of wireless outages reported to FCC that were attributed to a physical incident. To determine the actions federal agencies have taken since 2013 to improve the resiliency of mobile wireless networks, we reviewed reports and documents from FCC, the Department of Homeland Security (DHS), and the National Institute of Standards and Technology (NIST) within the Department of Commerce. Specifically, we reviewed transcripts and papers from hearings and a workshop FCC held in 2013 on communications reliability and continuity. We also analyzed agency orders and comments submitted in FCC’s 2013 proceeding on wireless resiliency. In addition, we reviewed communications sector planning reports, such as the 2015 Communications Sector-Specific Plan and 2013 National Infrastructure Protection Plan, and other DHS communications sector-specific documents, as well as the NIST Community Resilience Planning Guide for Buildings and Infrastructure Systems and related documents. We also examined reports from federal advisory committees and partnership councils that cover wireless network resiliency, including reports from the Technological Advisory Council; Communications, Security, Reliability, and Interoperability Council; National Security Telecommunications Advisory Committee; and the Communications Sector Coordinating Council. To ensure we covered relevant agency actions and to seek any information on the results of these actions, we interviewed officials from DHS’s Office of Cybersecurity and Communications within the National Protection and Programs Directorate, including officials from the Stakeholder Engagement and Cyber Infrastructure Resilience division— the sector-specific agency that leads federal efforts to protect and secure the communications critical infrastructure—and National Cybersecurity and Communications Integration Center—the center that continuously monitors incidents that may impact communications. We also interviewed officials from DHS’s Federal Emergency Management Agency and Science and Technology Directorate, FCC, and NIST. Beyond federal agency officials, we interviewed 24 stakeholders to further understand federal agency and related industry actions to improve wireless network resiliency since 2013 and any results from these actions. Stakeholders included wireless carriers and other owners of wireless network infrastructure, industry associations, consumer groups, and state and local government officials. We selected stakeholders to ensure we covered different perspectives (e.g., industry and consumer groups, associations that represent state and local public safety officials). In particular, we selected industry associations and individual companies to cover both wireless carriers—which operate networks and own some network infrastructure—and communications tower companies—which own and operate towers and sites and then lease space to wireless carriers. We selected wireless carriers to include both nationwide and regional carriers. We selected state agencies to include states directly affected by two events—flooding in Louisiana and Hurricane Matthew— for which industry had implemented elements of the framework at the time we began our review. The views presented in our report are not generalizable to those of all stakeholders. See table 7 for a list of interviewed stakeholders. We reviewed documents describing the Wireless Network Resiliency Cooperative Framework (framework)—a voluntary, industry initiative announced in April 2016. We interviewed CTIA and three of the five wireless carriers that collectively proposed the framework to learn about the impetus for, status of, and any outcomes or lessons learned from use of the framework to date. We also interviewed FCC and DHS about each agency’s awareness of and role monitoring industry use of the framework, and we reviewed FCC plans to monitor and share information about the framework. Finally, we asked stakeholders we interviewed, as described above, about their knowledge of and experience with the framework, including any observed outcomes from its use to date. We assessed FCC’s efforts to monitor implementation of the framework against Standards for Internal Control and FCC’s current strategic plan. To determine what options exist for federal agencies to improve wireless network resiliency, we examined federal agency reports, literature, and other sources. First, we reviewed filings in FCC’s 2013 proceeding examining wireless resiliency, including FCC’s orders and comments filed by various parties, for proposed options that federal agencies could take. Second, we conducted a literature review of peer-reviewed articles, government reports, industry publications, and think tank publications from the last 5 years to identify additional options. Third, we examined reports from the aforementioned federal advisory committees on wireless network resiliency, the NIST Community Resilience Planning Guide for Buildings and Infrastructure Systems, and the Hurricane Sandy Rebuilding Task Force for recommendations made to federal agencies to enhance wireless network resiliency. From these sources, we identified 11 proposed options that federal agencies could take to improve wireless network resiliency. We eliminated one option—requiring wireless carriers to disclose outage information to the public—as this was the only option that FCC specifically proposed as a new rule in its 2013 proceeding, but ultimately FCC decided not to move forward on this proposal when it decided to not issue a final rule. The identified options were primarily those that FCC could implement, as FCC is the regulatory agency for wireless communications, although DHS or NIST could implement several of the options. We interviewed a variety of stakeholders, described above, to obtain their views on the advantages, disadvantages, and feasibility of each of the identified options. We used open-ended questions to solicit input on each option rather than provide a list of advantages and disadvantages to stakeholders. We also asked stakeholders if there were additional options for federal agencies to ensure we had a thorough list of options for federal agencies. Based on interviews with stakeholders and federal agencies, we decided not to present two options in our report—establish more formal, ongoing collaboration between wireless carriers and power companies and create a program to facilitate collaborative restoration between wireless carriers and power companies—as federal agencies told us they were already taking actions on these fronts. Therefore, we included federal agencies’ actions related to these two options while describing actions taken by federal agencies since 2013. We analyzed information collected through the interviews with stakeholders to identify the most commonly cited advantages and disadvantages, and to determine the number of stakeholders that supported or did not support each option. The information collected from stakeholder interviews is not generalizable to all industry stakeholders. Appendix II: Analysis of FCC Data on Wireless Outages The figures below provide results from our analysis of Federal Communications Commission (FCC) data on wireless outages from 2009 through 2016. The data is from the Network Outage Reporting System (NORS), the system that wireless carriers and other communications providers use to report information on outages meeting certain threshold as required by regulation. The figures below present information on the number, cause, and duration of all wireless outages reported to FCC for this period. To describe wireless outages by cause, we use nine categories that collapse several of the FCC categories from which wireless carriers select the root cause, direct cause, and contributing factor(s) when reporting an outage. The following provides a brief description of these nine categories. Appendix I contains information on the scope and methodology for this analysis, including these nine collapsed categories. Cable damage/failure includes outages caused by an error locating or digging that resulted in cable damage, by an aerial cable that was damaged or ceased to function, and by loss of transmission in a cable due to aging, among other causes. FCC categories: cable damage, cable damage/malfunction. Equipment failure contains outages caused by the failure of a hardware component (e.g., circuit pack or card in a processor) or by a problem with the design of firmware, hardware, or software (e.g., failure for firmware to reset or restore after initialization, logical errors in software). FCC categories: design-firmware, design-hardware, design-software, hardware failure. Network robustness includes outages caused by, for example, a failure to provide or maintain diversity, thus preventing single points of failure. FCC categories: diversity failure, simplex condition. Maintenance includes outages caused by a needed spare part not being on hand or available, a vendor or contractor lacking updated procedures for its work, a service provider not providing adequate or up-to-date training, and scheduled maintenance to upgrade a network component or fix a known problem, among other causes. FCC categories: spare, procedural-other vendor/contractor, procedural- service provider, procedural-system vendor, planned maintenance. External environmental contains outages caused by earthquakes, wildfires, flooding, and other natural disasters as well as vandalism, theft, vehicle accidents that impair or destroy a component, and animal damage. FCC category: environment (external). Internal environmental contains outages caused by contamination due to dirt or dust that leads to overheating, by water entering manholes or vaults that destroys or impairs a component, and by other damage related to the condition of buildings and structures housing network equipment. FCC category: environment (internal). Other/Insufficient data includes outages for which there is not enough information for a failure report or investigation to determine the cause of the failure, service was restored before the cause could be determined, and the cause cannot be determined or proven. FCC categories: insufficient data, other/unknown. Power failure includes outages due to a commercial power failure (including power failures that extend beyond any backup power capabilities), a generator running out of fuel, a power system that was insufficiently sized for its purpose, and batteries not functioning as designed. FCC category: power failure (commercial and/or backup). Traffic/System overload contains outages where a network is overloaded or congested because of an unplanned, external event, or because of under-engineering the network due to changing demand or technologies. FCC category: traffic/system overload. Appendix III: Comments from the Federal Communications Commission . Appendix IV: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contacts named above, Sally Moino (Assistant Director); Joanie Lofgren (Analyst in Charge); Enyinnaya David Aja; Stephen Brown; David Hooper; Richard Hung; Joshua Ormond; Amy Rosewarne; Andrew Stavisky; and Timothy Young made key contributions to this report. Jon Ludwigson, John Mortin, Mark Pross, Pam Snedden, James R Sweetman, Jr., and Joe Thompson also made contributions to this report.

Why GAO Did This Study Americans increasingly rely on mobile wireless communications for safety-related communications like calling 911 and receiving weather alerts. Mobile wireless networks face risks from physical incidents including extreme weather events and intentional and accidental damage. For example, in 2017 several major hurricanes damaged wireless network infrastructure, leaving many U.S. citizens without reliable access to wireless communications. GAO was asked to review federal efforts to improve the resiliency of wireless networks following natural disasters and other physical incidents. This report examines: (1) trends in mobile wireless outages reported to FCC since 2009 and (2) actions federal agencies and industry have taken since 2013 (after Hurricane Sandy) to improve wireless network resiliency, among other objectives. GAO analyzed wireless outage data from 2009 to 2016 (4 years before and after Hurricane Sandy); reviewed FCC, DHS, and industry documents; and interviewed stakeholders who represented a variety of perspectives, such as industry, public safety, and consumer groups. GAO assessed FCC's efforts to monitor an industry initiative to improve wireless network resiliency against federal internal control standards. What GAO Found The number of wireless outages attributed to a physical incident—a natural disaster, accident, or other manmade event, such as vandalism—increased from 2009 to 2016, as reported to the Federal Communications Commission (FCC). During this time, the number of outages substantially increased from 189 to 1,079 outages, with most of the increase occurring from 2009 to 2011. FCC officials said this increase was due in part to growth in wireless customers and wireless infrastructure. Almost all outages attributed to a physical incident were due to an accident, such as damage to a cable due to a digging error (74 percent) or a natural disaster (25 percent). However, outages due to a natural disaster had a longer median duration (ranging from 19 to 36 hours), which was more than twice as long as outages caused by an accident. Power failures and failures in other providers' networks also play a role in wireless outages attributed to physical incidents. For instance, carriers reported that 87 percent of wireless outages attributed to a physical incident were due to a failure in another provider's network on which they rely. Since 2013, federal agencies and industry have taken actions to improve the resiliency of wireless networks. For example, the Department of Homeland Security (DHS) and FCC charter federal advisory committees that have examined resiliency issues and potential solutions, such as sharing infrastructure during emergencies. FCC also proposed a rule that would disclose how individual wireless carriers' networks performed during emergency events. In response, an industry coalition announced an initiative—the Wireless Network Resiliency Cooperative Framework—whereby carriers agreed to allow roaming on each other's networks and aggregated statistics to be published on how networks performed during emergency events. This initiative prompted FCC to not adopt its proposed rule. FCC said it would engage with industry about the framework's implementation and use, but FCC has limited formal plans to oversee or spread knowledge of the framework: FCC developed a plan to track the completion of initial implementation tasks outlined in the framework, but this plan does not include steps to track or evaluate any outputs or outcomes from the framework. FCC and industry documents describe broad goals for the framework, such as advancing information sharing during and after emergency events, but neither FCC nor industry has set any specific measures to help determine whether the framework achieves these broad goals. Although some public safety officials and other stakeholders GAO contacted were not aware of the framework, FCC did not have plans to actively communicate information about the framework to these audiences. More robust measures and a better plan to monitor the framework would help FCC collect information on the framework and evaluate its effectiveness. Such steps could help FCC address any challenges or decide whether further action is needed. Also, by promoting awareness about the framework, FCC would help public safety officials and other industry participants to be well positioned to use the framework to help them prepare for or respond to emergency events. What GAO Recommends FCC should work with industry to develop specific performance measures for the Wireless Network Resiliency Cooperative Framework, monitor the framework's outcomes, and promote awareness of it. FCC agreed with the recommendations.

gao_GAO-18-136

gao_GAO-18-136_0

Background Democracy Assistance Program Areas The U.S. government supports various types of democracy assistance activities, which USAID and State categorize under the DRG portfolio. USAID and State use their Updated Foreign Assistance Standardized Program Structure and Definitions to categorize and define DRG program areas. As updated in April 2016, this document defines the aims of DRG as “to advance freedom and dignity by assisting governments and citizens to establish, consolidate, and protect democratic institutions, processes, and values, including participatory and accountable governance, rule of law, authentic political competition, civil society, human rights, and the free flow of information.” Prior to the 2016 update, DRG program areas were (1) rule of law and human rights, (2) good governance, (3) political competition and consensus-building, and (4) civil society. Each program area features different program elements, as shown in table 1. USAID, NED, and State Roles and Responsibilities Related to Democracy Assistance Overseas Multiple bureaus and offices in USAID and State, as well as NED, provide funding for democracy assistance programs, as shown in table 2. USAID provides democracy assistance through contracts, grants, and cooperative agreements, while NED provides democracy assistance only through grants. INL was the only State bureau that reported providing a significant amount of democracy assistance through contracts in addition to grants and cooperative agreements, while other bureaus primarily use grants and cooperative agreements. USAID, NED, and State Democracy Assistance Funding during Fiscal Years 2012 through 2016 Combined allocations for democracy assistance administered by USAID and State ranged from about $2 billion to about $3 billion per year, and NED funding ranged from about $100 million to about $170 million annually during fiscal years 2012 through 2016, as shown in figure 1. USAID’s and State’s combined allocations for democracy assistance varied by account in fiscal years 2012 through 2016. Economic Support Fund was the largest account ranging from 50 to 63 percent of the total in fiscal years 2012 through 2016, as shown in figure 2. Laws, Regulations, and Polices Relevant to Award- Type Decisions The following laws, regulations, and policies are related to agencies’ decisions to use a contract, grant, or cooperative agreement to implement democracy assistance programming: According to the Federal Grant and Cooperative Agreement Act of 1977, one of the purposes of the act is to promote a better understanding of government expenditures and help eliminate unnecessary administrative requirements on recipients of government awards by characterizing the relationship between executive agencies and contractors, states, local governments, and other recipients in acquiring property and services and in providing government assistance. The act provides agencies with criteria to be considered when making award-type decisions, including the intended nature of the relationship between the agency and recipient, as well as whether the principal purpose of the award is to benefit the federal government or to transfer a thing of value to a recipient to carry out a public purpose of support or stimulation authorized by law. The Competition in Contracting Act of 1984 requires agencies to obtain full and open competition for contracts through the use of competitive procedures in procurements unless otherwise authorized by law. The Federal Acquisition Regulation (FAR) establishes uniform policies and procedures for all executive agencies for acquisition through contracts. For example, the FAR includes policies and procedures to promote the requirement to obtain full and open competition for contracts. It defines the circumstances under which it is permissible for agencies to limit competition for contracts, including when there is an unusual or compelling urgency or when doing so is necessary for reasons of public interest or national security. The Office of Management and Budget’s “Uniform Administrative Requirements, Cost Principles, and Audit Requirements for Federal Awards,” as codified in the Code of Federal Regulations (C.F.R.), establishes government-wide requirements for federal agencies administering grants and cooperative agreements with nonfederal entities. This regulation includes policies and procedures for award elements, including monitoring and reporting as well as cost sharing. USAID established agencywide guidance for making award-type decisions in its Automated Directives System (ADS), Chapter 304 (ADS 304). In addition to referencing the criteria established in the Federal Grant and Cooperative Agreement Act, this ADS guidance lists indications for when a specific award type should be used and also identifies factors that should not be primary considerations in making award-type decisions. According to USAID guidance, agreement officers and contract officers are individuals representing the U.S. government who are responsible for documenting the final determination of award-type decisions. USAID further outlines policies and procedures for administration of grants and cooperative agreements in ADS Chapter 303 (ADS 303) and for contracts in ADS Chapter 302 (ADS 302). State also established agencywide guidance for making award-type decisions in its Federal Assistance Directive. It references relevant legislation and instructs contracting and agreement officers to consult with State’s Office of the Procurement Executive if disagreements regarding award-type decisions arise. The Consolidated Appropriations Act, 2016, required USAID and State to each establish guidelines for clarifying program design and objectives for democracy programs, including the use of contracts versus grants and cooperative agreements, for programs carried out with funds appropriated by the act. For more information on USAID and State guidance related to award-type decisions, see appendix III. USAID’s Award Process and Award Life Cycle USAID officials are to make award-type decisions based on applicable laws, regulations, and policies, some of which are described above. Figure 3 provides an overview of the considerations in making this determination based on USAID guidance. ADS 304 provides the following definitions and guidance to USAID personnel as to what award type to select: A contract is a mutually binding legal instrument in which the principal purpose is the acquisition, by purchase, lease, or barter, of property or services for the direct benefit or use of the federal government, or in the case of a host country contract, the host government agency that is a principal, signatory party to the instrument. According to ADS 304, USAID personnel shall use a contract when the principal purpose of this legal relationship is the acquisition of property or services for the direct benefit of a federal government agency. A grant is a legal instrument used when the principal purpose is the transfer of money, property, services or anything of value to a recipient in order to accomplish a public purpose of support or stimulation authorized by Federal statute and when substantial involvement by USAID is not anticipated. USAID personnel are instructed to use a grant when the principal purpose of the relationship with an awardee is to transfer money, property, services, or anything of value to that awardee to carry out a public purpose of support or stimulation authorized by federal statute; and the agency does not anticipate substantial involvement between itself and the awardee during the performance of the activity. A cooperative agreement is a legal instrument used when the principal purpose is the transfer of money, property, services, or anything of value to a recipient in order to accomplish a public purpose of support or stimulation authorized by federal statute and when substantial involvement by USAID is anticipated. According to ADS 304, USAID personnel must use a cooperative agreement when the principal purpose of the relationship with an awardee is to transfer a thing of value to that awardee in order to carry out a public purpose; and the agency anticipates substantial involvement between itself and the awardee during the performance of the activity. The active engagement of USAID officials with awardees in certain programmatic elements of a project constitutes substantial involvement. Such activities include approval of the awardee’s implementation plan and of specified key personnel. In addition to awarding contracts, grants and cooperative agreements to private organizations (such as a for-profit business or a nongovernmental organization), USAID makes awards to federal agencies and public international organizations. Under USAID guidance, a public international organization is an international organization composed principally of countries or other related organizations designated by USAID. USAID maintains a list of public international organizations and international agricultural research centers that are considered public international organizations. These organizations include the United Nations and related organizations, such as the Food and Agriculture Organization, and international financial institutions, such as the World Bank Group. USAID officials noted that public international organizations normally receive grants. Under USAID guidance, awards to public international organizations and interagency agreements do not require the same award-type decisions as those required by ADS 304 for contracts, grants, and cooperative agreements. Awards made to public international organizations are governed by USAID guidance separate from the guidance that applies to awards to other types of organizations, and interagency agreements are governed by guidance separate from contracts, grants, and cooperative agreements. According to USAID’s guidance, the award-type decision should occur early in the preaward stage within the life cycle of an award. Award type- decisions impact other elements of awards because different regulations and guidance are applicable based on award type. For example, competition and oversight requirements differ for contracts compared with grants and cooperative agreements. Similarly, award-type decisions affect whether the recipient of an award is eligible to make a profit. The award life cycle contains preaward and award implementation stages, as shown in figure 4. USAID Obligated $5.5 Billion and NED Obligated $610.2 Million in Democracy Assistance Funding; Total Funding State Obligated Cannot Be Reliably Determined During fiscal years 2012 through 2016, USAID obligated $5.5 billion and NED obligated $610.2 million in democracy assistance funding, and the total such funding that State obligated cannot be reliably determined. In providing democracy assistance, USAID obligated more through grants and cooperative agreements combined than contracts, but its obligations through different award types varied by fiscal year and DRG program area. NED provided democracy assistance only through grants, and its obligations remained generally constant by fiscal year but varied by DRG program area. State bureaus that were able to provide reliable data provided democracy assistance primarily through grants and cooperative agreements. INL was the only State bureau that reported providing a significant amount of democracy assistance through contracts in addition to grants and cooperative agreements, but INL was one of the three State bureaus unable to provide reliable data. USAID Obligated $5.5 Billion in Democracy Assistance through Contracts, Grants, and Cooperative Agreements USAID Obligated About Two- Thirds of Its Democracy Assistance Funding through Contracts and Cooperative Agreements, and One-Third through Grants to Public International Organizations USAID obligated $5.5 billion in democracy assistance funding during fiscal years 2012 through 2016, about 31 percent through contracts; about 33 percent through cooperative agreements; about 4 percent through grants, excluding grants to public international organizations (PIO); and about 32 percent through grants to PIOs. Of the $5.5 billion in democracy assistance, USAID obligated over $1.7 billion of all its democracy assistance through grants to PIOs. The three countries for which USAID obligated the most funds for democracy assistance projects were Afghanistan, Iraq, and South Sudan. Democracy assistance projects in Afghanistan received over $2 billion or 37 percent of USAID’s total democracy assistance obligations during fiscal years 2012 through 2016. Moreover, two grants to the World Bank for the Afghanistan Reconstruction Trust Fund totaling $1.5 billion during fiscal years 2012 through 2016 accounted for 85 percent of the total democracy assistance funds USAID obligated through grants to PIOs during that period. For both total obligations and number of awards, USAID awarded more of its democracy assistance through grants and cooperative agreements combined than through contracts, as shown in figure 5. Contracts and cooperative agreements each accounted for roughly one- third of total obligations, while grants, excluding those to PIOs, accounted for 4 percent of total obligations during fiscal years 2012 through 2016. Excluding grants to PIOs, the number of grants and obligations for grants on average were significantly less than cooperative agreements, as shown in table 3. USAID’s democracy assistance obligations through contracts, grants, and cooperative agreements have varied during fiscal years 2012 to 2016, with significant increases in USAID’s obligations through grants to the World Bank in fiscal years 2012 and 2015, as shown in figure 6. These increases were driven by two large grants to the World Bank for the Afghanistan Reconstruction Trust Fund. Specifically, the World Bank received more than $820 million in fiscal year 2012 and more than $360 million in fiscal year 2015. During fiscal years 2012 to 2016, the World Bank accounted for 93 percent of grants to PIOs. For more details on USAID obligations through different award types by fiscal year and DRG program area, see appendix IV. USAID’s democracy assistance obligations for good governance varied the most compared with the other three DRG program areas, rule of law and human rights, political competition and consensus-building, and civil society, as shown in figure 7. This variation was again due to two large grants to the World Bank for the Afghanistan Reconstruction Trust Fund, which were categorized under good governance. As shown in figure 8, USAID provided more democracy assistance in the area of good governance, over $2 billion more than the next largest program area. Excluding USAID obligations through grants to PIOs, USAID obligated more democracy assistance through contracts than through grants and cooperative agreements combined for the two program areas of good governance and rule of law and human rights. For the two other program areas—civil society and political competition and consensus-building—USAID obligated less through contracts. NED Obligated Over $610.2 Million in Democracy Assistance Funding through Grants NED obligated over $610.2 million in democracy assistance funding through a single award type—grants—during fiscal years 2012 through 2016. The three countries for which NED obligated the most funds for democracy assistance are in Eurasia and Asia. NED’s obligations remained generally constant in the past few fiscal years, as shown in figure 9. NED’s approved funding varied across the four DRG program areas. According to NED officials, NED does not maintain obligations data for awards by DRG program areas, as defined by USAID and State. Therefore, NED categorized its grants into DRG program areas for projects when funds were approved rather than when funds were obligated to provide a general sense of funding by DRG program area. NED approved the most funding in the area of good governance followed closely by political competition and consensus-building and then by civil society. NED’s approved funding in all program areas, except for civil society, increased over the years, as shown in figure 10. State Reported Obligating About $3 Billion in Democracy Assistance through Grants, Cooperative Agreements, and Contracts, but Data for Roughly Half of State’s Obligations Were Unreliable State reported obligating approximately $3 billion in democracy assistance funding during fiscal years 2012 through 2016 primarily through grants and cooperative agreements, but also through contracts. Seven of 10 State bureaus that were able to provide reliable data obligated $1.7 billion primarily through grants and cooperative agreements; the remaining three bureaus that were unable to provide reliable data reported obligating about $1.4 billion through all three award types. Seven State Bureaus Providing Reliable Data Obligated $1.7 Billion in Democracy Assistance Funding Primarily through Grants and Cooperative Agreements The seven State bureaus that were able to provide reliable data collectively obligated $1.7 billion for fiscal years 2012 through 2016 primarily through grants and cooperative agreements, as shown in table 4. Of these State bureaus, the Bureau of Democracy, Human Rights, and Labor obligated the most with about $1.2 billion in democracy assistance through 547 grants and 56 cooperative agreements for that period. The three regions for which the Bureau of Democracy, Human Rights, and Labor obligated the most funds for democracy assistance were the Near East, East Asia and Pacific, and the Western Hemisphere. Three State bureaus—INL, EUR, and SCA—were unable to provide reliable data on democracy assistance obligations for fiscal years 2012 through 2016. Collectively, these three bureaus reported obligating about $1.4 billion in democracy assistance during this period: INL, about $1.1 billion; EUR, about $150 million; and SCA, about $160 million. INL was the only State bureau that reported providing a significant amount of democracy assistance through contracts in addition to grants and cooperative agreements. We deemed data from these three bureaus unreliable because the data were incomplete, nonstandard, or inaccurate. For example, INL did not provide democracy assistance data for Colombia, Egypt, and Kenya until we identified these countries as potentially missing based on our comparison of INL data with USAID data. According to data INL subsequently provided, the democracy assistance projects in these three countries received about $49 million of the approximately $1.1 billion in democracy assistance obligated by INL in fiscal years 2012 through 2016. According to INL officials, the initial data INL provided did not include records of awards for these countries because awards were miscoded when the data were entered; for example, some awards were coded under the broad category of law enforcement rather than under specific DRG program areas. According to INL officials, this erroneous law enforcement code was used for all of Colombia’s programs and for some programs in other countries such as Egypt and Kenya. According to INL officials, for two additional countries, Tunisia and Morocco, the regional post did not always use codes associated with DRG program areas or personnel entered incorrect codes. INL also provided incomplete data for multiple data fields, including the dates for periods of performance. INL was missing the start date for 74 percent of records and the end date for almost 75 percent of records for fiscal years 2012 through 2015. A September 2014 State Office of Inspector General report on INL found, among other things, that because State’s budgeting and accounting systems are not designed to manage foreign assistance, INL staff were required to engage in time-consuming, inefficient, and parallel processes to track the bureau’s finances. According to INL officials, INL has made improvements in its data since the Inspector General report was published. However, INL was missing the start date for 69 percent of records and the end date for almost 71 percent of records for fiscal year 2016. According to INL, data fields such as these were incomplete because contract officers and agreement officers were not required to enter values for these data fields into State systems until October 2016. EUR and SCA also initially provided incomplete, inaccurate, or nonstandard data for multiple data fields. According to State officials, this was due to manual data entry and transfer errors. For example, dates were in various formats and recipient names were sometimes listed in the field intended for recipient categories, which did not allow for the systematic analysis of records. While EUR generally provided more complete and standard data for fiscal year 2016 compared with fiscal years 2012 through 2015, EUR still provided nonstandard codes to identify award subtype for 5.3 percent of its fiscal year 2016 records. For example, “ESF,” an abbreviation for the Economic Support Fund, was listed as the award subtype for multiple contracts. Furthermore, we identified 145 duplicate EUR records. EUR officials in Washington, D.C., noted that some of the duplicates resulted from their efforts to validate the data they had collected from staff in each country. Subsequently, these officials—who manually merged, analyzed, and validated data to correct it—identified additional duplicates beyond the 145 that we had identified. According to EUR officials, the bureau’s obligation data for democracy assistance awards were maintained in separate databases at posts, rather than in a centralized database. In validating the data they had collected, EUR officials identified duplicate records amounting to at least 5 percent of the records during fiscal years 2012 through 2016. On the basis of our independent analysis of the same dataset, we were able to confirm that about 4 percent of the EUR records were duplicate records. Data on democracy assistance awards are maintained in the countries where the awards are made. To obtain award level data, EUR headquarters personnel had to ask staff in each country to manually compile and report award data. In addition, SCA did not initially provide data for Afghanistan and Pakistan, including award-type data. Records associated with these two countries accounted for about 92 percent of SCA’s total democracy assistance funding. We identified these countries as potentially missing based on our comparison of SCA data with USAID data. SCA subsequently provided the missing data on democracy assistance awards made in Afghanistan and Pakistan; the data resided within a separate database. SCA democracy assistance awards are allocated across three offices within SCA and EUR, and information regarding democracy assistance programs is not currently managed through a centralized database. According to SCA officials, due to the lack of a centralized database, they would need to carefully coordinate across the three offices. However, despite the coordination efforts of these offices, SCA did not include Afghanistan and Pakistan in their initial submission of data to us, and the additional data SCA subsequently submitted through EUR for Central Asia still contained nonstandard and missing values. A June 2017 State Office of Inspector General report determined that State cannot obtain timely and accurate data necessary to provide central oversight of foreign assistance activities and meet statutory and regulatory reporting requirements. For example, the report said that State cannot readily analyze its foreign assistance by country or programmatic sector. Similarly, we found that State cannot readily analyze its foreign assistance agencywide by country or for its DRG portfolio since INL, EUR, and SCA did not provide reliable DRG award data, including incomplete or duplicative data associated with certain countries. According to the report, this lack of data hinders State’s leadership from strategically managing foreign assistance resources, identifying whether programs are achieving their objectives, and determining how well bureaus and offices implement foreign assistance programs. In September 2014, State began the Foreign Assistance Data Review to better understand and document issues with its agencywide data and multiple budget, financial, and program management systems, but State does not plan to complete its Foreign Assistance Data Review until fiscal year 2021. The Consolidated Appropriations Act, 2016, requires State to report on its use of the various award types, and the Office of Management and Budget’s Bulletin No. 12-01 requires State to report quarterly on its foreign assistance activities. Given these reporting requirements, State would not be able to provide accurate and complete data on democracy assistance unless INL, EUR, and SCA took immediate steps to address their data deficiencies. Federal internal control standards call for agencies to use quality information from reliable sources to achieve intended objectives and to effectively monitor activities. Without reliable democracy assistance data from all relevant bureaus, State cannot effectively monitor its democracy assistance programming and report reliable data externally. USAID Generally Lacked Complete and Timely Documentation of Award-Type Decisions USAID generally did not document award-type decisions in a complete and timely manner for the awards in our sample. Specifically, USAID provided complete and timely documentation of the award-type decision for 5 of the 41 awards we reviewed. For the remaining 36 awards, the documentation was either incomplete, not timely, or both. According to ADS 304, contract and agreement officers must determine whether to use a contract, grant, or cooperative agreement, including a rationale based on criteria outlined in the Federal Grant and Cooperative Agreement Act. USAID Did Not Have Complete Documentation of Award-Type Decisions for 14 of the 41 Awards in Our Sample Consistent with the requirements of ADS 304, USAID personnel documented the rationale for using a contract, grant, or cooperative agreement for 27 of the 41 awards we reviewed. As table 5 shows, the number of awards in our sample with complete and incomplete documentation of the award-type decision varies by award type. ADS 304 requires contract and agreement officers to document the selection of an award type, including the rationale for the award-type decisions based on the requirements of the Federal Grant and Cooperative Agreement Act. USAID provided documentation of the award-type decision for 31 of the awards in our sample but lacked such documentation for 10 awards. However, for 4 of the 31 awards with documentation of the award-type decision, the documentation was not complete because it did not include a rationale for choosing between grants, cooperative agreements, and contracts on the basis of criteria in the Federal Grant and Cooperative Agreement Act, as required by USAID guidance. The documentation of the award-type decision for these 4 awards, which were all contracts, outlined the rationale for selecting a particular type of contract, information that is required by the FAR. However, the documentation for these 4 awards did not address the decision to use a contract rather than a grant or cooperative agreement, including a rationale based on the requirements outlined in the Federal Grant and Cooperative Agreement Act, as required by ADS 304. For example, documentation for one contract provided a rationale for selecting a firm- fixed-price contract based on the level of risk, which is in accordance with requirements of the FAR. However, the documentation did not indicate the rationale for deciding to use a contract rather than a grant or cooperative agreement as required by ADS 304. Without documentation of the rationale for award-type decisions as required under USAID guidance, USAID cannot demonstrate that award-type decisions are made based on the requirements outlined in the Federal Grant and Cooperative Agreement Act. USAID Lacked Timely Documentation of Award- Type Decisions for 25 of the 31 Awards for Which It Provided Documentation For the 31 awards in our sample for which USAID provided documentation of the award-type decision, 6 met the timeliness standard set by USAID guidance, and 25 did not, as shown in table 6. While 5 award-type decisions were both timely and complete, one award that met the timeliness standard lacked a required component. According to ADS 304, contract and agreement officers must document the final award-type decision before a solicitation is issued or before USAID initiates communications with a potential sole source recipient. We found that 25 awards lacked timely documentation of the award-type decision because the decision was documented after the solicitation was issued or timeliness was indeterminate because the documentation lacked a date or other indication of when in the process this determination was documented. Without this, we could not determine whether the award-type decisions were documented prior to solicitation or before USAID initiated communications with a potential sole source recipient. Instances in which final award-type decisions were documented after the issuance of a solicitation or communication with a potential sole source recipient include the following: Solicitation for one of the contracts in our sample occurred in 2011, but the award-type decision was not documented until 2013. The award-type decision for one of the grants in our sample was documented after the grant was awarded, which occurs after the solicitation is issued. Solicitation for one cooperative agreement in our sample occurred in 2010, but the award-type decision was not documented until 2012. According to USAID officials, the agency’s practice prior to October 2016 was to include award-type decisions in a comprehensive document that was intended to record all key decisions made throughout the award process. This document was finalized at the end of the award process. However, USAID officials also stated that they have introduced new processes and procedures, including making updates to relevant guidance, templates, and instructions that they believe will result in more timely and complete documentation of award-type decisions. Specifically, in 2016 USAID issued an update to ADS 304 that includes examples of when to use contracts, grants, and cooperative agreements and provides additional information about the legal framework for making award-type decisions. In 2017, USAID also issued revised templates to guide the documentation of award-type decisions. According to USAID officials, in addition to clarifying the ADS 304 guidance and developing new templates, USAID is also developing specific guidance for DRG programs that it expects to release at a future date. For additional information about this DRG-specific guidance, see app. III. USAID has taken steps to improve documentation for award-type decisions by updating its guidance and templates but has not assessed whether these updates have resulted in timely and complete documentation of award-type decisions. USAID officials stated that assessments are conducted at the sub-bureau or mission level, rather than by specific sectors, such as for DRG programs. As a result, USAID officials do not have plans to assess whether the newly updated processes and procedures have resulted in more timely documentation of DRG award-type decisions. It is important that USAID document the award-type decision before it publishes a solicitation for the award because award-type decisions impact other award elements, such as the requirements for competition and oversight and whether profit is permissible under the award. Until USAID assesses its updated processes and procedures, it cannot know if the steps it has taken have resulted in complete and timely documentation of award-type decisions as required by USAID guidance. USAID Contracts Differed from Grants and Cooperative Agreements for Selected Award Elements For the awards in our sample, contracts generally differed from grants and cooperative agreements in terms of competition, scope of work, cost sharing and profit, and oversight requirements, among other characteristics. We identified differences in three award elements— competition, cost sharing and profit, and oversight requirements—that were generally consistent with the unique requirements provided for in procurement regulations and agency guidance. We also identified differences between the award types with regard to scope of work, and found certain activities were conducted under all three award types. USAID Used Competition Procedures for a Greater Proportion of Contracts than for Grants and Cooperative Agreements in Our Sample, due to Differences in Award Types’ Applicable Legal Frameworks USAID awarded most, but not all, of the contracts in our sample using full and open competition, according to USAID data. Different federal and USAID requirements are in place regarding the use of competition procedures to award contracts than apply to grants and cooperative agreements. In accordance with the FAR, executive agencies such as USAID are required to promote and provide for full and open competition in awarding contracts, with only limited exemptions. USAID did not require full competition for any of the grants in our sample and required it for only about one-third of the cooperative agreements, according to USAID data. For the 41 awards in our sample, table 7 shows how many of each award type used full competition, limited competition, or no competition, based on USAID data. Below are examples of the rationale USAID provided for limiting competition for selected contracts, grants, and cooperative agreements: USAID limited competition for one of the contracts in our sample because of potential impairment to a foreign aid program, and another contract was limited to local competitors. This exemption to full and open competition is based on a unique statutory authority available to USAID and other agencies operating foreign assistance programs, which has been implemented in the USAID Supplement to the FAR. USAID also exempted one of the contracts in our sample from full and open competition using a provision in the FAR that allows for solicitation from a single source when the purchase falls below a threshold of $150,000. However, USAID officials indicated that they erroneously cited FAR 13.106-1(b), which permits sole source awards for acquisitions not exceeding the simplified acquisition threshold if only one source is reasonably available, when they should have cited FAR 13.501(a)(2)(i), which permits sole source acquisitions of commercial items (including brand-name items) for acquisitions greater than $150,000. For two of the grants in our sample, USAID limited the awards to local competition, according to USAID officials. For one cooperative agreement in our sample, competition was limited, according to USAID data, but USAID did not provide additional information on how the award competition was limited. The recipient of this award had submitted an unsolicited application, which under ADS 303, may be included in a relevant competition for an award, if USAID finds that the unsolicited application reasonably fits an existing program. USAID found that this unsolicited application was responsive to an existing solicitation and thus provided no additional justification. For more information on the rationales USAID used to exempt contracts, grants, and cooperative agreements in our sample from full and open competition, see appendix V. Contracts More Often Included a Greater Number of Activities Working with Host-Country Governments or Other National Institutions, While Grants and Cooperative Agreements More Often Included a Greater Number of Activities Working with Civil Society Organizations We found that the scope of work for contracts, grants, and cooperative agreements included similar types of activities. We also found that contracts more often included a greater number of activities working with the host-country government or other major national institutions, and grants and cooperative agreements more often included a greater number of activities working with civil society organizations. Seven of the 13 contracts in our sample included more activities focused on engaging with host-country governments and national institutions, while only 2 of the 13 contracts included more activities focused on engaging civil society organizations. Grants and cooperative agreements, by contrast, more often included a greater number of activities to support civil society organizations and media organizations than government or major national institutions of the country of performance. Three of the 5 grants in our sample included more objectives or activities focused on engaging civil society organizations, rather than engaging with host government or other major national institutions, while none of the grants included more objectives or activities related to the host government or other major national institutions. Cooperative agreements slightly more often included a greater number of objectives or activities to engage civil society organizations than they did to work with host government and national institutions, with 9 cooperative agreements including more objectives or activities focused on engaging civil society organizations and 7 with more objectives or activities focused on engaging host governments or other national institutions. Below are some examples of the activities and types of parties engaged with as stated in the awards in our sample: One contract in our sample provided various advisors to assist the government of a foreign country in implementing transparent policies, laws, and systems to strengthen public financial management and provide for a well-regulated financial sector, among other things. For more information on program objectives for selected democracy assistance awards by contract type, see appendix VI. A grant in our sample sought to increase the capacity of civil society organizations and the media to promote transparent democratic elections and political processes, among other things. Activities under the scope of work for this award included building alliances with stakeholders, conducting election-day observations, and analyzing electoral results. A cooperative agreement in our sample was intended to support a political transition through, among other things, organizational capacity development and grant-making opportunities for civil society organizations working to raise awareness about electoral events. In addition, for our award sample, we found that activities such as technical assistance, training, and local capacity building were conducted under grants, cooperative agreements, and contracts. Estimated Profits Ranged from 1 to 6 Percent of Estimated Contract Cost; Cost Sharing Ranged from Less Than 1 to 74 Percent of Total Value of Grants and Cooperative Agreements Eight of the 13 contracts in our sample were cost-plus-fixed-fee contracts, under which the contractor is reimbursed its costs in implementing the program in addition to a fee (profit) that is fixed at the outset. For these 8 contracts, the estimated percentage of profit ranged from about 1 to 6 percent of the estimated contract cost. According to the FAR, under cost-plus-fixed-fee contracts, the fee cannot exceed 10 percent of the contract’s estimated cost excluding fee. The average estimated fixed fee percentage for these contracts was about 5 percent of the estimated contract cost. While USAID contracts may be structured to provide for contractor profit in accordance with the FAR, USAID guidance does not allow profit under grants and cooperative agreements. For the grants and cooperative agreements in our sample, the awards did not specifically provide any fee (profit), and the awardees often agreed to contribute to the cost of the program through cost sharing. In addition, USAID guidance identifies cost sharing—whereby an awardee contributes to the total cost of an agreement—as an important element of the USAID-awardee relationship for grants and cooperative agreements. According to this guidance, although there is no general requirement for the awardees of grants and cooperative agreements to share in providing the costs of programs, cost sharing can be a mechanism to help awardees build their organizational capacity. For the awards in our sample, USAID included provisions for cost sharing in 3 of the 5 grants, and the awardees agreed to contribute about 11 percent, 13 percent, and 74 percent of the respective total award funding, including the cost share amount. USAID also included cost sharing provisions in 10 of the 23 cooperative agreements, with the awardee contribution ranging from less than 1 percent to 36 percent of the total award funding, including the cost share amount. All of the grants and cooperative agreements that included cost sharing provisions were awarded to nonprofit organizations, according to USAID data. Some of these awardees agreed to contribute to cost sharing by covering in-kind costs, such as donated time from volunteer legal specialists, and others agreed to contribute cash to cover some of the direct costs of implementing programs, such as personnel and benefits. According to USAID officials, cost sharing is rarely used under USAID contracts because under a cost sharing contract the contractor agrees to absorb a portion of its costs in expectation of substantial compensating benefits, such as certain research and development efforts, and these circumstances rarely occur under USAID’s programming. USAID did not include cost sharing provisions in any of the 13 contracts in our sample. For additional information about profit in our sample, see table 8. Table 9 provides additional information about cost sharing under the awards in our sample. Below are some examples of profit and cost sharing arrangements included in contracts, grants, and cooperative agreements in our sample: A contract in our sample sought to, among other things, improve the access of vulnerable and disadvantaged populations to the country’s legal system by engaging in activities such as working to build the capacity of government and civil society organizations to be more responsive to the needs of these populations. Under this award, the contractor was to receive approximately $1.7 million in profit, which was 4 percent of the estimated value of the award. The awardee for a grant in our sample agreed to provide $2.1 million of the program costs, about 74 percent of the total cost of the program, which sought to develop public opinion survey research capacity in the host country, among other things. USAID’s grant to this awardee funded additional support for the program, which the awardee was already executing prior to USAID assistance. A cooperative agreement in our sample included a requirement for the awardee to contribute about 9 percent of program expenditures, or about $3 million, for a program that sought to improve access to health services, as well as strengthen health delivery systems and health governance. Contracts Had Different Oversight and Evaluation Requirements Than Grants and Cooperative Agreements, due to Differences in Award Types’ Applicable Legal Frameworks We found that USAID oversight requirements differed for contracts compared with grants and cooperative agreements for the awards in our sample. This is because contracts (1) at times required more frequent reporting and (2) more often required evaluations of the contractor’s performance. Reporting requirements: We found that while most awards in our sample required quarterly financial and performance reporting, some contracts required these reports to be submitted monthly. USAID required quarterly financial and performance reporting for the majority of grants and cooperative agreements in our sample. None of the grants or cooperative agreements in our sample included requirements for financial reporting more frequently than quarterly, and no grants and only one cooperative agreement included a more frequent performance reporting requirement. According to Title 2 of the Code of Federal Regulations (C.F.R.), Section 200.327, under grants and cooperative agreements, financial reports must be collected by agencies with the frequency required by the award, but no less frequently than annually and no more frequently than quarterly, except in unusual circumstances, such as where more frequent reporting is necessary for effective monitoring of the award. USAID officials confirmed that there would have to be a reason to justify quarterly or more frequent reporting requirements for grants or cooperative agreements. For example, considerations related to risk could result in the need for more frequent reporting for grants and cooperative agreements. Table 10 shows the financial and performance reporting requirements for the contracts, grants, and cooperative agreements in our sample. Evaluations of performance: For the majority of contracts in our sample, USAID included provisions for evaluation of the contractor’s performance at the conclusion of performance. According to the FAR, evaluations of a contractor’s performance shall be prepared at the time the work under the contract is completed, and, for contracts longer than 1 year, interim evaluations should be prepared at least annually. USAID officials indicated that there is no similar government-wide or USAID requirement for grants and cooperative agreements. None of the grants and only a few of the cooperative agreements in our sample included such evaluation provisions. However, USAID officials noted that, in accordance with USAID policy, the past performance of a potential awardee is considered in conducting risk assessments for grants and cooperative agreements. Table 11 shows the number of USAID contracts, grants, and cooperative agreements in our sample that included provisions for evaluation of the contractor or awardee’s performance. For most of the contracts in our sample, award documentation indicated that the contractor’s performance would be assessed on a variety of factors such as quality of service, cost control, timeliness of performance, and effectiveness of key personnel. These evaluations form the basis of the contractor’s performance record for the contract. Only one contract in our sample had no requirement for a performance evaluation of the contractor, and that award was for the rental of a hotel ballroom and related services for an event. For three of the cooperative agreements in our sample that included provisions for the evaluation of the awardee’s performance, the award documentation indicated that USAID officials were to ensure prudent management of the award and to make the achievement of program objectives easier by, among other things, evaluating the awardee and its performance. One cooperative agreement included a provision that USAID will fund or conduct an external midterm evaluation during the second year of the project. For the one remaining cooperative agreement with an evaluation provision, documentation indicated that the evaluation would be used to inform a decision about a potential follow-on award. Conclusions Democracy assistance has been a key component of U.S. foreign assistance, supporting activities related to rule of law and human rights, good governance, political competition and consensus-building, and civil society. USAID and State together have allocated about $2 billion annually for democracy assistance in fiscal years 2012 through 2016. USAID’s information systems enable it to track and report the amount of democracy assistance funding through contracts, grants, and cooperative agreements. However, State lacks the ability to provide comparable agencywide data. The quality of democracy assistance award data provided by 10 State bureaus and offices varied, and three of these bureaus were unable to provide reliable data. Of the State bureaus, INL is the only bureau that regularly makes use of contracts, and it provided unreliable data. Without reliable data from INL, State cannot accurately report on its use of the various award types. In addition, since EUR’s and SCA’s award data are maintained across embassies, offices, and the two bureaus, opportunities for data errors may increase when regional data needs to be compiled. Without reliable data from all relevant bureaus, State cannot be sure that it is fully and accurately reporting on democracy assistance awards, which limits, among other things, congressional oversight of democracy assistance funding. While USAID requirements for complete and timely documentation of award-type decisions have existed since at least 2011, for our sample of 41 USAID awards for which an award-type decision was required, only 5, or about 12 percent, had both complete and timely documentation of the award-type decision. USAID recently introduced processes and procedures to improve the documentation of these decisions. However, until USAID assesses its updated processes and procedures, it cannot know if the changes resulted in award-type decisions being documented in a complete and timely manner, as required by its guidance, or if additional steps are needed. Recommendations for Executive Action We are making three recommendations, two to State and one to USAID. The Secretary of State should direct the Bureau of International Narcotics and Law Enforcement Affairs to identify and address factors that affect the reliability of its democracy assistance data, such as miscoded or missing data. (Recommendation 1) The Secretary of State should direct the Director of the Office of U.S. Foreign Assistance Resources to implement a process to improve the reliability, accessibility, and standardization of democracy assistance data across the geographic regions of the Bureaus of European and Eurasian Affairs and South and Central Asian Affairs, such as utilizing a centralized database for award data. (Recommendation 2) The USAID Administrator should direct the Office of Acquisition and Assistance to assess whether current processes and procedures as outlined in revised guidance result in complete and timely documentation of award-type decisions for democracy assistance. (Recommendation 3) Agency Comments and Our Evaluation We provided a draft of this report to State, USAID, and NED for review and comment. State, USAID, and NED provided technical comments on the draft, which we incorporated as appropriate. State and USAID also provided written comments in letters that are reproduced in appendices VII and VIII, respectively. In their written comments, both State and USAID concurred with our recommendations. State also requested that the report provide more information about its commitment and efforts to improve accountability of foreign assistance under its Foreign Assistance Data Review process. We have added more details about these efforts, including a discussion of State’s recent report to Congress on the outcomes of Phases One and Two of its four-phase review, which is expected to be completed in fiscal year 2021. State’s letter also described other efforts to improve the quality and accessibility of data at the bureau- level and at posts. In its written comments, USAID stated that it will take steps to assess documentation of award-type decisions and planned to complete this assessment by September 30, 2018. USAID also underscored certain details regarding required documentation of award-type decisions for some awards in our sample of 41 USAID democracy assistance awards. USAID noted that three contracts in our sample consisted of task orders, which do not require award-type decision documentation separate from their base awards under USAID guidance, according to agency officials. The draft report included these details, and we added more information to the report to further clarify them. We are sending copies of this report to the appropriate congressional committees, the Secretary of State, the Administrator of the U.S. Agency for International Development, and the President of the National Endowment for Democracy. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-3149 or gootnickd@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix IX. Appendix I: Objectives, Scope, and Methodology This report (1) examines funding that the U.S. Agency for International Development (USAID), National Endowment for Democracy (NED), and U.S. Department of State (State) obligated for democracy assistance through contracts, grants, and cooperative agreements; (2) evaluates USAID documentation of award-type decisions; and (3) compares USAID contracts with grants and cooperative agreements across selected award elements. To examine funds obligated by USAID, NED, and State for democracy assistance by award types, we obtained data on awards that USAID, NED, and State administered during fiscal years 2012 through 2016 under the Democracy, Human Rights, and Governance (DRG) portfolio. The data we obtained included awards to public international organizations (PIO). However, awards to PIOs are governed by USAID guidance separate from the guidance that applies to awards to other types of organizations. The data we obtained also included interagency agreements. However, interagency agreements are governed by separate USAID guidance that does not require the same award-type decision as when agencies obligate funds to entities through contracts, grants, and cooperative agreements. We analyzed the award data for fiscal years 2012 through 2016 but did not include fiscal year 2011 data in our analysis because State did not consistently track obligations data at the award level prior to fiscal year 2012, according to State officials. We assessed the reliability of these data by reviewing related documentation; interviewing knowledgeable officials; and conducting electronic or manual data testing for missing, nonstandard, or duplicative data; among other things. We determined that data provided by USAID, NED, and State, except for data from State’s Bureau of International Narcotics and Law Enforcement Affairs (INL), Bureau of European and Eurasian Affairs (EUR), and Bureau of South and Central Asian Affairs (SCA), were sufficiently reliable for the purposes of our report. For the USAID, NED, and State data that were sufficiently reliable, we analyzed the amount of funding by award type, among other variables. We assessed State’s data reliability challenges against federal internal control standards. To evaluate USAID’s award-type decisions, we reviewed relevant regulations and agency policies, and we interviewed knowledgeable agency officials about these polices. State and NED were not included in our sample because most State bureaus did not regularly use all three types of awards and NED only provides assistance through grants. In addition, three State bureaus were unable to provide reliable data from which to select a sample. We also selected a roughly proportional, random, nongeneralizable sample of 41 awards—13 contracts, 5 grants, and 23 cooperative agreements. These awards were selected based on characteristics, such as award type, DRG program area, and place of performance. The sample focused on the 14 countries for which USAID obligated the most democracy funding. Democracy assistance projects in these 14 countries received over 70 percent of USAID’s democracy assistance funding. The sample was also limited to contracts, grants, and cooperative agreements that were awarded by USAID in fiscal years 2012 through 2015 because fiscal year 2015 was the most recent fiscal year for which data were available at the time of our sample selection. We excluded (1) grants made to PIOs because these awards are governed by USAID guidance separate from the guidance that applies to awards to other types of organizations; (2) interagency agreements because engaging other federal agencies through interagency agreements does not require the same award-type decision under USAID guidance as when agencies obligate funds to entities through contracts, grants, and cooperative agreements; and (3) awards that fell below the simplified acquisition threshold, which is $150,000, because there are different acquisition procedures allowable for awards that fall below the threshold. For the selected awards, we obtained and analyzed preaward documentation relevant to the award-type decision and evaluated this documentation against the relevant regulations and agency guidance. To ensure accuracy, we cross-checked information from the documentation for the selected awards with USAID’s award data. In collaboration with subject-matter experts, we selected four award elements—competition, cost sharing and profit, scope of work, and oversight requirements—for a comparison of contracts with grants and cooperative agreements. To compare USAID contracts with grants and cooperative agreements across selected award elements, we obtained and conducted a review of documentation associated with the same sample of 41 USAID awards. Additionally, we obtained information about award recipients from a public database maintained at SAM.gov. Using information collected from the documentation, we analyzed the selected awards’ competition, cost sharing and profit, scope of work, and oversight activities. Subsequently, we reviewed the documentation and applicable legal frameworks, including federal regulations and guidance pertaining to the award elements we selected, to compare differences between award types. We also interviewed relevant agency officials as well as the leading industry organizations that represent implementers of foreign assistance programs to better understand the use of various award types. We conducted this performance audit from July 2016 to December 2017 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: USAID’s Use of Various Award Types for Democracy Assistance by Place of Performance during Fiscal Years 2012 through 2015 Our nongeneralizable sample of U.S. Agency for International Development (USAID) awards was limited to fiscal years 2012 through 2015 and to the 14 countries for which USAID obligated the most democracy funding, which accounted for over 70 percent of USAID’s democracy assistance funding. Total USAID democracy assistance funding for projects in Afghanistan was greater than for any other country, amounting to almost 39 percent of USAID’s total democracy assistance obligations during fiscal years 2012 through 2015. The total USAID democracy assistance funding for projects in Afghanistan included obligations to public international organizations (PIOs) of more than $827 million in fiscal year 2012, more than $55 million in fiscal year 2013, more than $48 million in fiscal year 2014, and more than $369 million in fiscal year 2015. USAID’s use of award types for democracy assistance varied across these 14 countries during fiscal years 2012 through 2015, as shown in figure 11. Appendix III: USAID and State Guidance in Response to the Consolidated Appropriations Act, 2016 The Consolidated Appropriations Act, 2016, states that not later than 90 days after enactment of the act, the U.S. Department of State (State) and U.S. Agency for International Development (USAID), following consultation with democracy program implementing partners, shall each establish guidelines for clarifying program design and objectives for democracy programs, including the use of contracts versus grants and cooperative agreements in the conduct of democracy programs carried out with funds appropriated by the act. The joint explanatory statement accompanying the act further elaborated that the act requires the development of guidelines for the use of contracts versus grants and cooperative agreements for the unique objectives of democracy programs, and that the guidelines should assist contracting and agreement officers in selecting the most appropriate mechanism for democracy programs, among other things. In 2016, USAID released its revised agencywide guidance, Automated Directives System (ADS) Chapter 304 (ADS 304), on how to make award- type decisions between contracts, grants, and cooperative agreements. According to USAID officials, USAID expects to release guidance further clarifying ADS 304 at a future date. USAID intends to issue the guidance after it completes final consultations with implementing partners, the Congress, and other stakeholders. It includes scenarios and examples to further clarify existing government-wide and agencywide guidance. According to USAID officials, in drafting its guidance to further clarify ADS 304, USAID pursued multiple rounds of review within USAID, and with implementing partners, the Congress, and other stakeholders. According to State, it met the requirement to establish additional guidelines for democracy assistance through State’s release of a Program Design and Performance Management Toolkit in fall 2016 and State’s updating of its Federal Assistance Directive in May 2017. The aim of the Program Design and Performance Management Toolkit was to clarify program design and objectives for foreign assistance programs broadly. The Federal Assistance Directive combined both policies and procedures from the Federal Assistance Policy Directive and the Procedural Desk Guide into one document and clarified appropriate mechanisms for all programs. Although applicable to democracy programs, neither of these actions was specific to democracy programs. According to State, the Bureau of Democracy, Human Rights, and Labor; the Bureau of International Narcotics and Law Enforcement; and other relevant State bureaus that work closely with democracy assistance implementing partners consult regularly with and provide guidance to implementing partners on the use of the guidelines. Appendix IV: USAID Obligations through Different Award Types by Fiscal Year and Democracy, Human Rights, and Governance (DRG) Program Area U.S. Agency for International Development (USAID) democracy assistance obligations through different award types varied by fiscal year and DRG program area, as shown in tables 12, 13, and 14. Appendix V: Regulations and Policy Allowing USAID to Limit Competition for Contracts, Grants, and Cooperative Agreements Regulations, law, and policy enable the U.S. Agency for International Development (USAID) to limit competition in awarding contracts, grants, and cooperative agreements under certain circumstances. One source of USAID’s authority to limit competition for contracts is the Competition in Contracting Act of 1984, as implemented in the Federal Acquisition Regulation (FAR), which outlines policies and procedures for acquisition by all federal agencies, including policies and procedures pertaining to exemptions from competition. In addition, for contracts awarded under USAID programs, the FAR, among other regulations and legislation, contains specific provisions on exemptions from competition. For grants and cooperative agreements, USAID’s Automated Directives System Chapter 303 outlines circumstances under which competition can be limited. In accordance with applicable policies, procedures, and guidance, USAID can use some exemptions from competition only for contracts and others only for grants and cooperative agreements. For example, USAID can limit competition for contracts for the sake of public interest or when circumstances are such that competition would compromise U.S. national security; however, according to USAID officials, they rarely have cause to use these grounds for limiting competition. USAID guidance outlines some unique exemptions to competition for grants and cooperative agreements. For example, USAID can exempt follow-on awards, which are the same or substantively similar to recently completed awards, if the awardee will be the same, or can exempt awards from competition in certain instances when USAID has received an unsolicited application. For the awards in our sample, USAID limited competition for only 3 of the 13 contracts, based on USAID data. However, for one of these contracts, USAID officials indicated that they erroneously cited FAR 13.106-1(b), which permits sole source awards for acquisitions not exceeding the simplified acquisition threshold if only one source is reasonably available, when they should have cited FAR 13.501(a)(2)(i), which permits sole source acquisitions of commercial items (including brand-name items) for acquisitions greater than $150,000. For the exemptions from competition that USAID used for these awards, see table 15. USAID exempted from full competition all five of the grants and 15 of the 23 cooperative agreements in our sample. Table 16 outlines exemptions from competition that USAID may use for grants and cooperative agreements in our sample. Appendix VI: Program Objectives of Selected USAID Democracy Assistance Awards by Contract Type For the contracts in our sample, we found that program objectives varied by type of contract. For example, the firm fixed price award and three of the four indefinite quantity awards in our sample procured goods and services with specific deliverables that were directly for the U.S. Agency for International Development’s (USAID) benefit. Nearly all of the cost- plus-fixed-fee contracts sought to achieve improvements in the public sector of the country of performance through activities such as supporting developments in public policy or strengthening national institutions. For examples of differences in program objectives by contract type in our sample, see table 17. Appendix VII: Comments from the U.S. Department of State Appendix VIII: Comments from the U.S. Agency for International Development Appendix IX: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Mona Sehgal (Assistant Director), Justine Lazaro (Analyst-in-Charge), Lindsey Cross, Christopher Hayes, Carl Barden, Karen Cassidy, David Dayton, Timothy DiNapoli, Justin Fisher, Alexandra Jeszeck, Heather Latta, Madeline Messick, Natarajan Subramanian, Alex Welsh, and Bill Woods made key contributions to this report.

Why GAO Did This Study Supporting efforts to promote democracy has been a foreign policy priority for the U.S. government. In recent years, USAID and State have allocated about $2 billion per year toward democracy assistance overseas. Congress required USAID and State to each establish guidelines for and report on the use of contracts, grants, and cooperative agreements for certain democracy programs. GAO was asked to review U.S. democracy assistance. This report (1) examines funding USAID, NED, and State obligated for democracy assistance primarily through contracts, grants, and cooperative agreements and (2) evaluates documentation of USAID award-type decisions, among other objectives. GAO analyzed USAID, NED, and State democracy assistance award data for fiscal years 2012–2016. GAO also reviewed relevant regulation and agency policies and analyzed documentation for a nongeneralizable sample of USAID awards selected based on factors such as award type, program area, and country. What GAO Found In fiscal years 2012–2016, the U.S. Agency for International Development (USAID) obligated $5.5 billion and the National Endowment for Democracy (NED) obligated $610.2 million in democracy assistance funding. The total funding the Department of State (State) obligated for democracy assistance could not be reliably determined. One-third of all USAID obligations were provided through public international organizations (PIOs), which under USAID guidance are composed principally of countries or other organizations designated by USAID; 94 percent of PIO obligations were provided to the World Bank for democracy assistance projects in Afghanistan. The remaining two-thirds of USAID obligations were provided through contracts, grants (excluding PIOs), and cooperative agreements. Of the 10 State bureaus providing democracy assistance, 3 were unable to provide reliable funding data for fiscal years 2012–2016. Data from these bureaus were incomplete, nonstandard, or inaccurate. Federal internal control standards call for agencies to use quality information from reliable sources to achieve intended objectives and to monitor activities. Without such data, State cannot effectively monitor its democracy assistance programming and report reliable data externally. For the awards GAO sampled, USAID generally did not document decisions about whether to award a contract, grant, or cooperative agreement (known as award-type decisions) in a complete and timely manner. According to applicable USAID guidance, agency officials were required to (1) document the final award-type decision with their written determination, including a rationale based on the requirements of the Federal Grant and Cooperative Agreement Act, and (2) complete this documentation before award solicitation occurs or, for noncompetitive awards, before USAID initiated communications with a potential sole-source awardee. However, USAID provided both complete and timely documentation of the award-type decision for 5 of the 41 awards GAO sampled. For the remaining 36 awards, the documentation was either incomplete, not timely or timeliness was indeterminate, or both (see table). While USAID has taken steps to improve documentation for award-type decisions by updating its guidance and templates, it has not assessed whether these updates have resulted in complete and timely documentation. It is important that USAID document these decisions in advance of solicitation because the selection of an award type may affect requirements for administering the award, including competition and oversight requirements and whether or not profit is permissible. What GAO Recommends State should improve the reliability and completeness of its democracy assistance funding data, and USAID should assess whether steps taken are resulting in complete and timely documentation of democracy assistance award-type decisions. State and USAID concurred with GAO's recommendations and described actions planned or under way to address them.

gao_GAO-18-486

gao_GAO-18-486_0

Background UI Program Administration and Funding The federal-state UI program provides temporary cash benefits to eligible workers who lose their jobs through no fault of their own. Under this arrangement, states administer their own programs according to certain federal requirements and under the oversight of DOL’s Office of Unemployment Insurance. States have considerable flexibility to set benefit amounts and their duration, or the maximum period of time that the state pays benefits, and establish eligibility requirements. UI benefits are funded primarily through state payroll taxes on employers, and administrative costs are primarily funded through a federal payroll tax on employers. The states collect taxes that will be used to pay UI benefits, and the U.S. Department of the Treasury holds these funds in trust on behalf of the states in the Unemployment Trust Fund. DOL certifies for payment to the states administrative grants to operate their UI programs, which amounted to about $2.7 billion in fiscal year 2017. DOL is responsible for ensuring that state UI laws include certain provisions, which is a condition of the state receiving its UI administrative grant. Individuals typically claim their UI benefits by filing claims with their state UI agency online or by phone on a weekly or bi-weekly basis. In fiscal year 2017, the average weekly UI benefit was about $350, and claimants remained on the program for an average of 15 weeks, according to DOL data. Work Search Requirements for UI Claimants Federal law establishes a work search requirement for UI eligibility, but the specific work search activities UI claimants are expected to conduct vary by state, according to a DOL report. To be eligible for unemployment benefits, individuals are generally required to actively search for work under federal law. The Middle Class Tax Relief and Job Creation Act of 2012 amended the Social Security Act to, among other things, require states to have work search requirements for UI claimants specified in their laws as a condition of eligibility for the states’ UI administrative grants. Specifically, states must have laws that require UI claimants to be “actively seeking work” as a condition of eligibility for unemployment compensation for any week. Because federal law does not specifically define actively seeking work, states have some discretion to establish a reasonable definition, according to DOL’s 2013 guidance to states. For example, a state can specify a minimum number of weekly contacts a claimant must have with potential employers. Acceptable work search activities might also include searching for jobs online, submitting job applications, visiting a job center, attending a networking event, or establishing a LinkedIn account, according to a DOL report. Depending on the state, UI claimants may be directed to register for work with their state’s Employment Service, which provides job search assistance, job placement assistance, and referrals to employers. In addition, in some cases UI claimants may be directed to participate in reemployment services at an American Job Center. In 2017, DOL provided $115 million in grants to states to provide Reemployment Services and Eligibility Assessments (RESEA). RESEA services include in-person reemployment services and eligibility assessments in American Job Centers for ex-service members and UI claimants determined to have a high likelihood of exhausting their UI benefits. RESEA-funded activities include developing an individual reemployment plan, providing labor market information, identifying job skills and prospects, and reviewing the claimant’s continued eligibility for UI benefits. Process for Identifying Work Search Overpayments in State Benefit Accuracy Measurement Audits DOL uses its Benefit Accuracy Measurement (BAM) system to determine the accuracy of UI benefit payments and estimate the amount and rate of improper payments. Under the BAM system, each state reviews a number of randomly selected cases on a weekly basis and reconstructs the UI claims process to assess the accuracy of the payments that were made. The state determines what the benefit payment should have been according to its laws and policies. States report the results of their BAM case reviews to DOL—including overpayments and underpayments—through an online data system. DOL uses the data to estimate improper payment rates by state, as well as to calculate a nationwide rate. State BAM audits involve reviews of existing records in the state’s UI claims information system as well as original fact-finding by the state BAM investigator. DOL requires states to use a standard claimant questionnaire when conducting BAM audits. The questionnaire includes numerous questions about the claimant’s circumstances—including their work search efforts—during the week under review. The questionnaire includes questions that could indicate that a claimant qualifies for an exemption from work search requirements, or made specific job contacts and the results of the job contacts, such as whether the claimant submitted an application and received a job offer. State BAM investigators are also expected to take steps to verify the information reported by the claimant by collecting documentation from claimants and contacting employers or other third parties. According to DOL’s 2016 BAM annual report and BAM procedures, state BAM investigators are to review a sufficient number of work search activities to determine whether the claimant has complied with the state’s minimum requirement for the number of weekly work search activities. The BAM program assigns one of three classifications to each of the work search activities reviewed: Acceptable – Documentation exists that the work search activity reported by the claimant, such as an employer contact, employment application, or other state approved work search activity, was made by the claimant and was acceptable according to the state’s law or policy. Unverifiable – The investigator was unable to establish sufficient information to make a judgment of whether the work search activity was either acceptable or unacceptable according to the state’s law or policy. Unacceptable – Written documentation exists that the work search activity reported by the claimant was not made at all by claimant, or was made but was unacceptable according to the state’s law or policy. According to DOL’s BAM annual report, work search activities classified as acceptable or unverifiable may be considered in calculating whether the claimant has satisfied the state’s required number of work search activities for purposes of BAM. If the state investigator finds that the claimant’s work search is unacceptable and does not meet the state’s requirements, he or she may determine the claimant was ineligible for benefits and establish an overpayment, depending on state law (see fig. 1). Currently, several states have formal warning policies and provide claimants warnings for the first instance of noncompliance with work search requirements, whereas states without these policies count these cases as overpayments, according to DOL. Reported Causes of UI Improper Payments Since 2002, federal agencies have been required to identify and report improper payments. The leading reported cause of UI improper payments in fiscal year 2017 was overpayments to claimants who failed to meet work search requirements. DOL data show that states made an estimated $1.36 billion in overpayments to such claimants in fiscal year 2017. Other major reported causes of UI improper payments in fiscal year 2017 included payments made to individuals who continue to make claims even after returning to work (benefit year earnings) and payments made to claimants who were determined ineligible due to disqualifying job separations, such as quitting a job without good cause or being discharged for misconduct (separation issues). (See fig. 2. Table 8 in app. II provides greater detail.) According to DOL officials, many UI improper payments cannot be prevented given certain legal requirements that states pay claims in a timely manner and provide claimants with due process when the state finds an eligibility issue. Specifically, according to DOL, federal law requires that when an eligibility issue is detected, the claimant has a right to receive notice and provide the state information before being denied benefits. In addition, if an eligibility issue associated with work search, or any other matter, is detected but not resolved, the state is still required to pay for a claimed week no later than the week after an eligibility issue is detected, according to DOL. The time it takes to work through the necessary due process steps can prevent states from stopping the payment before it must be paid. Trend in Reported Work Search Overpayments Nationally, the estimated work search overpayment rate and the estimated amount of work search overpayments have risen in recent years. Specifically, in fiscal year 2013, approximately $1 billion in estimated work search overpayments were made to claimants who were not actively searching for work and, in fiscal year 2017, the amount increased to close to an estimated $1.36 billion (see fig. 3). The national work search overpayment rate for such claimants also increased during this time. (See table 9 in app. II for additional details.) According to DOL officials, some states implemented more stringent work search requirements, which may account for the recent trend. As work search requirements become more stringent, the opportunities for non- compliance and errors increase and thus higher improper payment rates, according to DOL officials. While the national work search overpayment rate was 4.5 percent in fiscal year 2017, state work search overpayment rates varied widely from an estimated 0 to 41 percent of the UI benefits that states paid in fiscal year 2017. (See table 10 in app. II for state-by- state estimates of work search overpayment rates and amounts.) States use various methods to recover overpayments to UI claimants, including setting up payment plans, off-setting UI benefits, or deducting refunds from federal or state income tax returns. Like all recovered UI overpayments, recoveries of work search overpayments must be deposited in the unemployment trust fund of the state that recovered the money and can be used only for the payment of UI benefits, according to DOL officials. National data are not available on the amount of work search overpayments that states have recovered because, although DOL collects recovery data from states, it does not require states to separate out work search overpayment recoveries from other types of recoveries in their reporting. Conducting More Work Search Investigations Is Associated With Lower Estimated Work Search Overpayment Rates, As Is the Use of Formal Warnings Based on our analysis of DOL data, we found that certain state administrative practices, including investigating a higher percentage of claimant-reported work search activities and frequent use of formal warnings, were associated with lower reported state work search overpayment rates. However, DOL recently determined that federal law does not permit states to warn claimants the first time they failed to meet work search requirements (i.e., issue formal warnings) instead of establishing that an overpayment was made. Additionally, a higher percentage of claimants required to search for work is associated with higher reported state work search overpayment rates. Investigating Claimant- Reported Job Contacts Is Associated with Lower Work Search Overpayment Estimates, but the Extent to Which States Verified Contacts Varied One of the administrative practices significantly associated with lower work search overpayment estimates was investigations of claimants’ reported job contacts. Specifically, a higher percentage of cases with claimants whose contacts were investigated by the state UI agency as part of the state’s BAM audit was associated with a lower work search overpayment rate estimate. According to our analysis, for every 1 percentage point increase in the percentage of cases with claimants’ whose job contacts were investigated, there was a 0.072 percentage point decrease in the work search overpayment rate estimate. The extent to which states attempted to verify claimants’ reported job contacts through these investigations varied, according to our analysis of DOL data. Nationally, in fiscal year 2017, states investigated job contacts in about 80 percent of BAM cases where claimants were required to search for work. However, among the states, the proportion of cases in which job contacts were investigated was less than 50 percent in 5 states. (Table 11 in app. II shows the percentage of contacts that were investigated for each state.) Furthermore, states often were not able to verify the information claimants reported. Of the job contacts that were investigated, states reported that about 48 percent of the job contacts were acceptable, about 8 percent were unacceptable, and about 45 percent could not be verified (see fig. 5). Our analysis of BAM data for fiscal year 2017 also shows that for the overpayments that states were able to detect, that a large portion were found through investigating and verifying claimants’ work search contacts. Specifically, 47 percent of reported work search overpayments were found through this practice. Interviewing claimants about their work search was the next most common way states detected work search overpayments. States reported identifying 32 percent of work search overpayments in fiscal year 2017 using this practice. Although overpayments can be the result of actions taken by the claimant or the agency administering the program, states reported that most work search overpayments are associated with claimants. For example, claimants may provide inadequate or incorrect information needed by the UI agency to determine if the claimant met work search requirements. In fiscal year 2017, states attributed about 99 percent of overpayments at least partially to claimant action, while they attributed about 2 percent at least partially to administrative errors at the state agency. Frequent Use of Formal Warnings Is Associated with Reporting Lower Work Search Overpayments, but DOL Recently Determined Their Use Is Legally Impermissible According to DOL data, 22 states issued formal warnings to one or more claimants at some point between fiscal year 2013 and fiscal year 2017 for failure to meet work search requirements instead of finding that the claimants were overpaid. Although the states that made use of these warnings varied over this period, the number of states issuing formal warnings has generally increased over time from 13 states issuing formal warnings in fiscal year 2013 to 19 states in fiscal year 2017. Overall, states that most frequently issued formal warnings had lower reported work search overpayment rates than states that did not issue formal warnings. However, their work search overpayment rates are lower because, under their state policies, they did not count an overpayment when they issued a formal warning. Our analysis indicates that states which issued formal warnings frequently—in 75 percent or more of cases involving work search errors— had estimated work search overpayment rates that were 3.5 percentage points lower, on average, than states that did not issue formal warnings. However, states that use formal warnings less frequently—in fewer than 75 percent of cases involving work search errors—reported work search overpayment rates that were between 3 and 4 percentage points higher than states that did not issue formal warnings. See appendix I for a detailed discussion of our econometric analysis. Table 1 shows the average work search overpayment rate estimates for each of these groups of states when formal warnings are not counted as overpayments, and the potential average work search overpayment rate when formal warnings are counted as overpayments. Excluding formal warnings, two of the groups—frequent and low users of formal warnings—had average work search overpayment rate estimates lower than the average for states that did not use formal warnings. However, when formal warnings are included in the overpayment rate, only low users of formal warnings have a work search overpayment rate estimate lower than the average for states that did not use formal warnings. GAO’s analysis of DOL data shows that in fiscal year 2017, estimated work search overpayments were nearly $1.4 billion, but potentially would have been an estimated $1.8 billion greater if states had not issued formal warnings and established overpayments. Our analysis further shows that if formal warning cases had been included in DOL’s calculation of the UI overpayment rates for fiscal year 2017, the nationwide UI overpayment rate would have increased by about 6 percentage points, from an estimated 12 percent to an estimated 18 percent. Moreover, these figures represent an increase from the fiscal year 2016 figures we presented in our previous report on states’ use of formal warnings related to work search requirements. At that time, we found the nationwide UI overpayment rate would have increased by about 5 percentage points from an estimated 11 percent to an estimated 16 percent with the inclusion of formal warning cases. The amount of UI payments made to claimants for weeks in which they received formal warnings in fiscal year 2016 was about $1.6 billion. Table 12 in appendix II shows how the estimated UI overpayment rate in each state issuing formal warnings in fiscal year 2017 may have increased if formal warnings were not used. State use of formal warnings has resulted in inconsistent reporting of work search overpayments, which affects DOL’s reported improper payment rate for the UI program. Specifically, states that issue formal warnings have not counted as overpayments cases in which claimants did not actively search for work and received a UI benefit payment. On the other hand, states that did not have formal warning policies counted such cases as overpayments, which are factored into DOL’s reported improper payments rate. The variation among states related to formal warning policies makes it difficult for DOL and others to understand the reasons behind states’ reported work search overpayments. DOL has determined and documented in its FY 2017 Agency Financial Report that the use of formal warnings is no longer allowed under the 2012 federal law, which generally requires UI claimants to actively seek work. DOL officials told us in July 2017 they would soon issue a letter to states to inform them that they are no longer permitted to use formal warnings when they determine that claimants failed to meet work search requirements. To date, DOL has not issued such a letter. In May 2018, DOL officials told us that they expect to issue the letter by the end of calendar year 2018. Federal internal control standards direct agency management to remediate identified internal control deficiencies on a timely basis. Federal internal controls standards also state that management should externally communicate the necessary quality information to achieve the entity’s objectives. Additionally, these standards state that agency management should establish and operate monitoring activities to monitor the internal control system and evaluate the results. According to DOL officials, they began discussing the need for a potential discontinuation of formal warnings at conferences with states in the first half of 2017. However, we found that states continued to implement their formal warning policies, potentially resulting in an increase in the estimated amounts of overpayment dollars associated with formal warnings between fiscal year 2016 and 2017. Until DOL informs states of the need to discontinue the use of formal warnings through a letter or another mechanism, states will continue to be inconsistent in whether they count as overpayments cases in which claimants who failed to search for work in any week were provided benefits. Additionally, once DOL provides additional information to states on formal warnings, it should monitor states’ responses to help ensure that DOL achieves its desired results. Furthermore, having more consistent information on overpayments related to work search issues could help DOL assess how the program is working nationwide and whether further federal and state actions would be needed to address this leading source of reported improper payments in the UI program. DOL officials stated that the national work search overpayment rate is likely to increase in the future as states begin to eliminate their formal warning practices. Officials also stated that this may take some time as some states may need to amend laws or regulations in order to do so. Higher Percentages of Claimants Required to Search for Work Are Associated with Higher Work Search Overpayment Rate Estimates State work search overpayment rate estimates were higher for states where a higher proportion of claimants were required to search for work. Based on our analysis, for every one percentage point increase in the fraction of cases with claimants required to search for work, there was a 0.084 percentage point increase in the work search overpayment rate estimate on average, all else being equal. Nationwide in fiscal year 2017, states reported that work searches have been required in 80 percent of cases, with requirements in individual states ranging from 38 to 100 percent of cases. Three states reported requiring fewer than 50 percent of claimants to perform work searches, while five states reported requiring more than 95 percent of their claimants to perform work searches. The most common reasons states exempted claimants from work searches were because claimants were “job-attached” (e.g. temporarily laid-off, recalled), or they had union deferrals because they were seeking employment through their union, according to DOL data. The map in figure 6 shows the range among all the states in the percentage of UI claimants required to search for work, according to our analysis of DOL data for fiscal year 2017. Selected States Used Multiple Approaches to Address Work Search Overpayments, but Cited Challenges Verifying Claimants’ Work Search Activities Selected states used multiple approaches to address work search overpayments, including online systems to facilitate the work search reporting and verification process, work search audits beyond the BAM audits, and messaging to inform claimants of their work search responsibilities. For example, three of six states in our review had online systems where claimants could report specific work search activities as part of filing their weekly claims, according to state officials (see table 2). Some of the approaches states used were specifically designed for UI claimants participating in state reemployment programs. State officials cited several benefits of the approaches they use to address work search overpayments. The online systems, work search audits, and messaging helped prompt work search activities and prevent work search overpayments in some cases, according to state officials. According to officials from the selected states that used them and a study on work search improper payments, online systems can facilitate the work search reporting and verification in several ways: Automatically documenting the claimants’ work search activities. Online reporting of work search activities can help prevent overpayments because their work search is documented in the online claims system, which means the claimant does not need to keep a work search log. In addition, online job search/training systems can be used to track the work search activities completed by claimants, making it easier for the state to verify that the work search was completed. For example, officials in Mississippi and Indiana told us they piloted an online system called NextJob and required RESEA program participants to conduct work search activities through the system. Mississippi officials reported that NextJob motivated claimants to conduct their job search and increased the speed of reemployment among these individuals. Similarly, New Jersey officials told us that UI claimants selected to participate in New Jersey’s RESEA program are required to use an online job search and training system called OnRamp, which, for example, allows job seekers to create or upload their resume on the website, search for jobs, access online training, and receive email alerts on potential job matches. In addition, officials in Nevada said that the online reporting system is beneficial because the work search activities are documented in the system and are more reliably retrieved if the claimant is selected for a BAM audit because few claimants maintain and retain their work search effort logs. Performing automated checks on data the claimants submit. The online claims systems can identify potentially duplicate job contacts and check whether the claimant reported the required number of job contacts. For example, Utah officials reported that if a claimant enters job contacts from another week, officials would follow up with the claimant by phone after it is flagged by their online system. If claimants report self-disqualifying information, such as an insufficient number of job contacts, the system can automatically put a hold on a claim until the issue is resolved. Facilitating communication with the claimant. Some states added messages to their online claims system that pop up if the claimant enters incorrect or insufficient information. For example, Mississippi officials stated that they used messaging to better inform claimants of their responsibilities and to encourage them to report accurate information. According to the officials, if claimants do not report the required number of work search activities in the online system, a questionnaire will pop up requesting that the claimants explain why they did not do so. The system requires the claimants to enter more information in order to submit their claims and to receive their benefit payments. Mississippi officials also stated that adding targeted messaging resulted in fewer denials of benefits due to claimant failure to meet work search requirements and also reduced the number of appeals related to this type of denial. Some state officials said that messaging encouraged accurate reporting. For example, Indiana officials told us that at the end of the online claim filing process, claimants receive a message notifying them of the state’s work search requirements and informing them that they are required to search for work to continue receiving benefits. In Utah, officials developed a video that covers claimants’ responsibilities, including work search requirements, which claimants must view before receiving their initial benefit payment, according to state officials. Officials in three of the states we reviewed reported using additional work search audits beyond BAM to help reinforce their state policies. Pennsylvania officials reported conducting 7,182 work search audits for RESEA program participants in 2016. As a result, officials reported that the state issued 1,300 warnings to claimants. The two other states— Mississippi and Utah—report that their random work search audits, coupled with their online systems, helped prevent work search overpayments as they are able to disqualify claims before the payment goes out. Mississippi and Utah officials also reported that they were also able to identify and recover some work search overpayments (see table 3). Despite implementing these approaches, state officials in five of the six states we contacted told us they face challenges with verifying work search activities. These officials stated that they have difficulty verifying work search activities as some claimants do not understand the work search requirements or do not keep accurate records of their work search activities, which makes it difficult for the state to confirm compliance with the state requirements. In addition, state officials also said that many employers do not keep records of job seekers’ inquiries or do not respond to state requests for information when they are trying to verify claimants’ work search activities. For example, Nevada officials said that work search contacts are often virtually unverifiable as many companies outsource their hiring processes to contractors who refer the job candidate for a job posting and keep the job application. Officials said these contractors also rarely respond to state inquiries about claimants’ job applications. As discussed later, DOL has provided states tools to help address this issue, such as a messaging toolkit to help states improve communications with claimants and employers. DOL Monitors States’ Work Search Overpayment Rate Estimates and Provides Assistance, but Lacks Clear Procedures on Work Search Verification DOL Monitors States’ Work Search Overpayment Rate Estimates and Has Identified Strategies and Provided Tools to Help States Reduce Their Rates DOL uses UI performance data to monitor state progress in reducing the estimated improper payment rate, including data on overpayments to claimants who failed to meet work search requirements. To do so, DOL requires states to submit State Quality Service Plans, which serve as the performance reporting and grant application documents through which states receive administrative funding. The plans include a summary of state performance on various measures related to operating the UI program, including the improper payment rate, according to DOL documentation we reviewed. States with estimated improper payment rates of 10 percent or more are required to submit corrective action plans to DOL. For example, data from two of the six states we reviewed– Nevada and New Jersey–had estimated improper payment rates above 10 percent during DOL’s most recent planning cycle and developed corrective action plans. Nevada’s corrective action plan noted that the state expects their rate to decline due to their June 2017 implementation of online work search reporting as part of their UI claims system. New Jersey’s corrective action plan noted that the state plans to implement new online tools that will help them verify wage and employment information. DOL separately monitors each state’s estimated work search overpayment rate. In addition, all states, including those who estimated improper payment rates of less than 10 percent, are required to prepare a state-specific action plan that describes the root causes of improper payments and the state’s strategies to address them. According to agency officials, DOL reviews plans to monitor state performance and help states identify strategies to improve performance. Although DOL requires states with estimated improper payment rates of 10 percent or more to develop corrective action plans, according to DOL, the agency has limited options to require state UI agencies to take actions to respond to high improper payment rates. DOL officials told us that, beyond routine monitoring and providing states with technical assistance to help reduce their improper payments rate, their enforcement options are limited to withholding the state’s administrative funding or removing federal tax credit reductions, which is, in effect, a tax increase for the state’s employers. According to DOL officials, both are considered extraordinary sanctions that require significant due process. The agency has not withheld state administrative funding to address improper payments, according to DOL officials. DOL officials also told us that they are concerned about the effects on UI claimants if they were to withhold administrative funding. The administration’s fiscal year 2019 congressional budget justification includes a legislative proposal that would authorize the Secretary of Labor to require states to implement corrective action measures for poor state performance in the UI program, helping to reduce improper payments in states with the highest estimated rates. DOL has identified strategies to address the leading causes of UI improper payments—including work search issues—and provided states tools and funding to help implement them. For example: Pathway to Reemployment Framework. In 2016, DOL and the National Association of State Workforce Agencies published a framework that contained a broad menu of work search options that states could adopt to better reflect how individuals search for work, such as allowing use of online job search tools to count as an approved work search activity. The framework also includes suggestions for how states could document or verify claimants’ work search activities for eligibility purposes. Messaging toolkit. In 2012, DOL published a messaging toolkit designed in part to improve claimants’ understanding of work search requirements as a condition of eligibility for benefits. According to DOL, claimants who fully understand their responsibilities and the consequences of not fulfilling them may be more likely to complete the required work search activities, thereby reducing instances of claimants’ failing to search for work. DOL provided states supplemental funding to support improved messaging and tracked state implementation of the strategies. Online tool to record work search activities. In 2011, DOL provided supplemental funding to New York to develop an online work search record that could be replicated by other states. This tool is designed to reduce improper payments that result from inadequate documentation of work search activities. Claimants can use the tool to record their work search online when they file their UI claims. The work search record is automatically shared with state job centers, which in turn act to enhance claimants’ work search and connect claimants with jobs. New York used open source technologies when designing the tool in an effort to help more states replicate the product at a lower cost. UI Integrity Center. The DOL-funded UI Integrity Center of Excellence (UI Integrity Center) was established as a way to help states develop and share innovative strategies to prevent and detect UI improper payments, reduce fraud, and improve program integrity. For example, the UI Integrity Center funded a pilot project in 2016 to support a state in implementing an online tool that trains claimants on how to effectively search for jobs and allows claimants to use the tool to complete their work search activities. The UI Integrity Center also hosted national conferences in 2016 and 2018 that included presentations on state practices to address work search improper payments. DOL Directs States to Investigate a Sample of Claimants’ Work Searches, but Has Not Clarified Procedures on Verifying Work Search Although DOL monitors states’ work search overpayment rate estimates and has provided assistance to help states address such overpayments, it has provided limited direction to states on the level of effort states must make to verify whether claimants are actively searching for work. Specifically, DOL’s BAM procedures direct states to investigate a sufficient number of work search contacts in its BAM sample of UI cases to determine if the claimants met the state’s work search requirement. However, DOL has not provided states any additional direction on what is considered “sufficient.” As previously mentioned, the extent to which states attempted to verify job contacts from claimants’ work searches varied across states. Some states reported to DOL that they did not investigate work search activities for a majority of their BAM cases even though work search was required and the claimant reported job contacts. Federal internal control standards state that agency management should externally communicate the necessary quality information to achieve its objectives, including addressing related risks. According to DOL officials, the agency plans to clarify its BAM procedures, to include providing more definitive instructions to states on work search verification requirements, after the agency issues its planned letter to states about discontinuing the use of formal warning policies. As of May 2018, the agency said it plans to issue the letter on formal warning policies by the end of calendar year 2018. However, as previously stated, it has been nearly a year since DOL initially told us they were planning to issue the letter to states. Effective monitoring of state compliance with the clarified work search verification requirements will also be important once DOL revises its BAM procedures. Federal internal control standards state that agency management should establish and operate monitoring activities to monitor the internal control system and evaluate the results. Monitoring states’ responses could help ensure that DOL achieves its desired results. By providing clear direction to states about work search verification requirements and monitoring states’ implementation of these requirements, DOL would have greater assurance that states are complying with its requirements. Conclusions The health of the UI program depends, in part, on the ability of states to control its benefit payments by accurately determining individuals’ eligibility. Improper payments, including overpayments, in the UI program have led to billions of federal and state funds being used inappropriately. Actively seeking work has generally been an eligibility requirement for individuals receiving unemployment benefits under federal law since 2012, but states have not consistently implemented the requirement for claimants in similar circumstances. States with formal warning policies reported lower work search overpayments not necessarily because they are better at ensuring claimants’ compliance with requirements, but because they are not counting cases where claimants receive formal warnings as overpayments. Furthermore, although DOL determined in 2017 that states are not permitted to issue formal warnings rather than reporting an overpayment, it has not officially told states to stop using warnings. Without providing states with this information and monitoring their response, states will continue to report inconsistent information on the extent of work search overpayments. State efforts to check whether claimants are meeting work search requirements also vary. Our evidence suggests that states making a greater effort to investigate work search activities tend to have lower overpayment rate estimates associated with this issue. However, some states are not investigating claimant-reported work search activities as part of their BAM audits despite DOL’s procedures directing them to do so. Until DOL provides clear direction to states about verifying work search and monitors state compliance, DOL has little assurance that states are complying with its requirements. Recommendations for Executive Action We are making the following four recommendations to the Department of Labor: The Assistant Secretary of DOL’s Employment and Training Administration should provide states with information about its determination that the use of state formal warning policies is no longer permissible under federal law. (Recommendation 1) The Assistant Secretary of DOL’s Employment and Training Administration should monitor states’ efforts to discontinue the use of formal warning policies. (Recommendation 2) The Assistant Secretary of DOL’s Employment and Training Administration should clarify information on work search verification requirements in its revised Benefit Accuracy Measurement procedures. The revised procedures should include an explanation of what DOL considers to be sufficient verification of claimants’ work search activities. (Recommendation 3) The Assistant Secretary of DOL’s Employment and Training Administration should monitor states’ compliance with the clarified work search verification requirements. (Recommendation 4) Agency Comments We provided a draft of this report to the Department of Labor (DOL) for review and comment. In its written comments, reproduced in appendix IV, DOL agreed with all four of our recommendations and stated that it would take action to address them. DOL also provided technical comments, which we incorporated as appropriate. Additionally, we provided excerpts of the draft report to state UI officials in the selected six states we included in our review. We incorporated their technical comments as appropriate. As agreed with your office, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies to the appropriate congressional committees, the Secretary of the Department of Labor, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at 202-512-7215 or brownbarnesc@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix V. Appendix I: Econometric Analysis of Estimated State Work Search Overpayment Rates Data We used the Department of Labor’s (DOL) Unemployment Insurance Benefit Accuracy Measurement (BAM) data. The BAM program is designed to determine the accuracy of paid and denied claims for unemployment insurance (UI) in three major UI programs—state UI, Unemployment Compensation for Federal Employees (UCFE), and Unemployment Compensation for Ex-Servicemembers (UCX). The BAM data covers claimants in the 50 U.S. states, the District of Columbia, and Puerto Rico. Each week, state workforce agencies select random samples of paid and denied unemployment insurance claims (i.e., cases). State BAM investigators then audit these cases to determine whether the claimant was properly paid or was properly denied benefits in the week for which the claim was made (i.e., key week). The bases for determining whether paid and denied claims were accurate are federal and state law, regulations, and policy. We used BAM cases for paid claims for all UI programs and all states for fiscal years 2013 through 2017. The number of cases for each state is determined by DOL for each year. Cases are chosen randomly each week from the population of claims for that week. The normal weekly number of paid claims sampled in most states is 9, with a minimum of 6, for an annual sample of around 480 cases. For the 10 states with the smaller UI workloads, the normal weekly number of paid claims sampled is 7, with a minimum of 5, for an annual sample of around 360 cases. For each paid claim case, the BAM data include variables describing the claimant, as well as information on their UI benefit year, separation from their last job, monetary eligibility, benefit payment history, employment services registration and work search, and the outcome of the BAM investigation. For cases for which BAM auditors identify errors, the BAM data also include information on the errors. In 2016, work search issues, benefit year earnings, and separation issues were the most common causes of reported overpayments: Work search issues. These occur when the state finds that claimants did not actively search for work during the key week. Federal law generally requires people receiving unemployment compensation to be actively searching for work. States have discretion to establish requirements for what constitutes active work search, and these requirements vary by state. Benefit year earnings issues. These occur when claimants have earnings that exceed the threshold for UI eligibility in their state or when these earnings are not properly reported. Separation issues. These occur when claimants are ultimately determined to be ineligible for UI due to disqualifying job separations, such as quitting a job without good cause or being discharged for misconduct under the state UI law. Other causes of overpayments include incorrect reporting of wages used to calculate benefits, able and available to work issues, employment service registration issues, and other issues. For some cases, BAM investigators identify multiple errors with different causes. When this occurs, BAM investigators determine the overpayment amount associated with each cause. BAM sample data is weighted to make inferences about the population. In accordance with DOL’s method, we calculated the weight on each BAM sample case as (1) the number of unemployment compensation payments to claimants for the state and the week from which the BAM sample case was selected divided by, and (2) the number of completed BAM sample cases for that week, as long as there were two or more completed cases for the week. Because each state followed a probability procedure based on random selections to pick cases, the BAM sample is only one of a large number of samples that they might have drawn. Since each sample could have provided different estimates, we express our confidence in the precision of the BAM sample’s results using confidence intervals. For example, a 95 percent confidence interval is the interval that would contain the actual population value for 95 percent of the samples that could have been drawn. In some tables we provide the margin of error instead of the confidence interval, where the margin of error is the half-width of the confidence interval. Our analysis sample consists of cases with no missing or invalid values of variables used in our analysis that also have positive sample weights. Our analysis sample includes 98.1percent of all cases. Estimated Overpayment Rates Due to Failure to Meet Work Search Requirement In accordance with DOL’s method, the estimated overpayment rate is equal to the amount of UI benefits overpaid as a percentage of the total amount of UI benefits paid. The amount of UI benefits overpaid includes fraud, nonfraud recoverable, and nonfraud nonrecoverable overpayments; and overpayments from all causes and responsible parties. The amount overpaid excludes overpayments that DOL considers as technically proper. An overpayment may be considered technically proper by DOL under a finality rule, which generally means that too much time has passed between the decision to pay the claimant and the detection of the eligibility issue, or for some other reason. Our estimated work search overpayment rate is calculated using the same method as the official overpayment rate and is generally comparable to work search overpayment rates reported by DOL. More specifically, we used the BAM data to identify cases with overpayments and with overpayments due to work search, excluding formal warnings and other payments DOL considers to be technically proper. Next, we applied DOL’s proration algorithm to allocate overpayment amounts for each error associated with a case so that the total amount of overpayments from all errors does not exceed the key week amount paid. Then, we calculated the total overpayment amount and work search overpayment amount for cases with overpayments and work search overpayments, respectively. Finally, we tracked the key week amount paid for each case. To estimate the work search overpayment rate for each state and fiscal year, we calculated the weighted sum of work search overpayment amounts as a percentage of the weighted sum of amounts paid. Table 4 shows the estimated national work search overpayment rate, the average estimated state work search overpayment rate, and the median estimated state work search overpayment rate for each fiscal year. The national work search overpayment rate estimate has increased from 2.4 percent in 2013 to 4.5 percent in 2017. Over the same time period, the average state work search overpayment rate estimate has increased from 2.6 to 4.6 percent, and the median state work search overpayment rate estimate has increased from 1.4 percent to 2.7 percent. Most states’ work search overpayment rate estimates are less than 5 percent, but some states’ rates are more than 10 percent. To identify state practices and other factors that may be associated with state work search overpayment rates, we reviewed DOL documents describing the BAM program and summarizing key features of states’ UI programs, as well as research on factors associated with time claimants spend searching for employment. Based on our review, we identified factors that may be associated with work search overpayments that can be measured using variables in the BAM data: Exemptions from work search requirements. According to DOL, while federal law generally requires claimants to be actively seeking work, it does allow states to exempt claimants in some circumstances. For example, according to DOL, because states may not deny benefits to an individual in approved training, all states provide an exemption from the requirements to be able and available for work and conducting an active work search for any week the individual is in approved training. In addition, according to DOL, some states allow work search exemptions if the worker is union-attached and finds work through the union hall, or if a separation is classified as a temporary lay-off and there is a reasonable expectation that the worker will return to work soon. According to DOL, other state work search exemptions include that the worker has a specified start date for new employment, has jury duty, has a compelling personal reason, is in a labor dispute with the employer, is the victim of domestic violence, or labor market or other information indicates no suitable employment. It is possible for a case with a claimant who indicated he or she was not required to search for work to have a work search error if the claimant provided an invalid reason for being exempt from his or her state’s work search requirement. For example, the claimant may have said that he or she was a member of a union with a hiring hall and obtained employment through union referrals or that he or she had a definite recall date, and therefore, the work search requirement was waived. However, the BAM investigator’s verification with the union found that the claimant was not in good standing, or the investigator’s verification with the employer found that the claimant had no definite recall date. In such a situation, the claimant might be held ineligible for a failure to conduct an active work search because the exemption was invalid. Claimant response. According to DOL, the claimant interview anchors the BAM audit and is a major error detection point, and the claimant questionnaire is a required standard form. Claimants may provide information for the BAM audit either in person, over the phone, or via mail or some other method. Some claimants may not respond to the audit at all. For the period from 2013 to 2017, of the work search overpayments that BAM auditors detected in their BAM sample of cases, about 37 percent of cases were uncovered during the claimant interview. State investigation of job contacts. Claimants are asked to provide information about job contacts as evidence that they were actively searching for work as part of the required claimant questionnaire. In addition, even if the claimant does not respond to the BAM audit, job contacts may be available for BAM auditors to investigate if the state’s continued claim process captured the claimant’s work search information. BAM staff must investigate a sufficient number of contacts to establish whether the claimant has met the state’s work search requirement. For the period from 2013 to 2017, of the work search overpayments that BAM auditors detected in the BAM sample of cases, about 45 percent were uncovered by investigating claimant- provided job contacts. State use of formal warnings. If the BAM audit of a case determines that the claimant’s work search during the key week was not acceptable, the state might issue a formal warning to the claimant instead of finding that the claimant received a work search overpayment, and these cases are not included in the calculation of the work search overpayment rate for that state. Over the period from 2013 to 2017, of the BAM sample of cases for which work search errors were identified, state workforce agencies issued formal warnings for about 46 percent. Claimant demographic and other characteristics. Characteristics of claimants that are associated with the amount of time claimants spend searching for work include age, education, gender, length of time between initial claim and key week, and weekly benefit amount as a percentage of the normal weekly wage in the claimant’s industry. To the extent that they affect time spent searching for work, these characteristics may also be associated with the likelihood of claimants receiving a work search overpayment. Some factors that may be associated with work search overpayments cannot be measured using variables in the BAM data and, thus, are excluded from our analysis. Examples include: Minimum number of work search activities. According to DOL, the minimum number of work search activities required per week varies across states and can vary based on labor market conditions, which can, for example, produce different requirements in a rural area versus an urban area. However, some states do not specify a required number of work search activities and instead require that the number of work search activities be “reasonable,” according to a DOL report. Type of required employer contacts. According to DOL, some states allow claimants to search for part-time work as well as full-time work, while others do not. In addition, according to DOL, some states specify that participation in reemployment services counts as a contact, and some states require claimants to make at least one contact through the state online system. Frequency and method of reporting. According to DOL, some states require weekly claimant reporting of work search activities, while others require bi-weekly reporting. In addition, according to DOL, some states require claimants to report online as part of their continued claim, while others require claimants to keep a log of their work search activities. Econometric Mode Our econometric model is the following: work search overpayment rates, = α + β∗work search requireds, + γ*n-person responses, + γ*telephone responses, + γ*mail, email, fax, or other responses, + δ*contacts investigateds, + φ + X’Θ + year indicators + ε, where s and y denote state and year, respectively, and the explanatory variables in the model are the following: Work search required is the estimated fraction of cases with claimants for whom work search is required, according to the BAM data. In-person response, telephone response, and mail, email, fax, or other response are the estimated fractions of cases with claimants who responded to the BAM audit in person, by telephone, or by mail, email, fax, or other method, respectively. Contacts investigated is the estimated fraction of cases for which BAM auditors investigated one or more job contacts. Low, moderate, and frequent formal warning use are binary indicator variables equal to 1 if the state issued a low, moderate, or high number of formal warnings in the fiscal year as a percentage of BAM cases with work search errors, respectively, and 0 otherwise. We defined low formal warning states as those that issued formal warnings for some but less than 25 percent of the cases with work search errors, moderate formal warning states as those that issued formal warnings for at least 25 percent but less than 75 percent of cases with work search errors, and high formal warning states as those that issued formal warnings for anywhere from 75 to 100 percent of cases with work search errors. The omitted group is states that issued no formal warnings during the fiscal year. X is a list of other characteristics of claimants that may be associated with work search, including the estimated distributions of cases across claimants’ age groups, education levels, gender, length of time between initial claim and key week, and weekly benefit amount as a percentage of the normal weekly wage for their occupation. ε is an error term. The parameters of interest in our econometric model are β, γ, γ, γ, δ, φ is an estimate of the change in the work search overpayment rate associated with a 1 percentage point increase in the fraction of cases with claimants who responded to the BAM audit in person, all else being equal. The parameters γ and γ are estimates of the change in the work search overpayment rate associated with a 1 percentage point increase in the fraction of cases with claimants who responded to the BAM audit by telephone and by mail, email, fax, or other method, respectively, all else being equal. The parameter δ is an estimate of the change in the work search overpayment rate associated with a 1 percentage point increase in the fraction of cases for which BAM auditors investigated one or more job contacts, all else being equal. The parameters φTable 5 shows descriptive statistics for the explanatory variables, other than formal warnings. Table 6 shows descriptive statistics for states’ use of formal warnings. Our analysis is subject to several limitations, and the results we discuss below should be interpreted with caution. No causality. Our econometric approach can establish correlations between state work search overpayment rate estimates and the factors we analyzed, but it cannot establish causal relationships. This limitation is especially important when it comes to interpreting the relationship between work search overpayment rate estimates and the fraction of cases for which BAM auditors investigated job contacts. BAM investigations can only detect work search overpayments, not prevent them, because BAM investigators are reviewing cases after the payment has already been made. Preventing work search overpayments is the only way to reduce the work search overpayment rates. Thus, the relationship between investigation of claimant- provided contacts and work search overpayment rates likely reflects not a causal relationship but an equilibrium relationship in which claimants who know that their job contacts will be investigated are more likely to search for work, all else being equal. Nongeneralizability. Our results do not generalize to other time periods with different labor market conditions, different rules and regulations about unemployment insurance, or other differences. Omitted variables. As discussed above, we have excluded from our models some state practices that may be associated with work search overpayments because those variables were not included in the BAM data. For example, we do not account for the fact that, according to DOL, some states require claimants to engage in more work search activities than other states, which could affect the likelihood that a claimant receives a work search overpayment. In addition, according to DOL, some states require claimants to report their work search activities in order to continue receiving benefits, which might prevent some work search overpayments from occurring. We have also likely excluded some relevant claimant characteristics as well. For example, total income for a claimant’s household and a claimant’s number of dependents may affect their work search efforts. However, variables describing a claimant’s household income and composition are not included in the BAM data and thus are omitted from our analysis. To the extent that these factors are correlated with the factors we included, our estimates of the relationships between work search overpayment rates and the included factors could be biased. Measurement error. Our variables may have been measured with error, which could bias our coefficient estimates. If the measurement error is random, our coefficient estimates would be biased down, but if the measurement error is systematic, then we cannot say whether our coefficient estimates would be biased down or up. Results Our baseline specification suggests that some state practices are associated with state work search overpayment rates estimates (see column 1 of table 7). We used the 10 percent level of significance as our threshold for statistical significance because we have a relatively small number of observations (259). State work search overpayment rate estimates were higher in states with more cases with claimants who were required to search for work. A 1 percentage point increase in the fraction of cases with claimants required to search for work was associated with a 0.084 percentage point increase in the work search overpayment rate estimates on average, all else being equal. Work search overpayment rate estimates were lower in states with more cases for which the BAM auditors investigated one or more job contacts. A one percentage point increase in the fraction of cases for which the BAM auditors investigated contacts was associated with a 0.072 percentage point reduction in the work search overpayment rate on average, all else being equal. Compared to states that did not issue any formal warnings, work search overpayment rate estimates were higher in states that made low or moderate use of formal warnings but lower in states that issued formal warnings frequently. Work search overpayment rate estimates in states that were low and moderate users of formal warnings were 3.9 and 3.1 percentage points higher on average than those in states that issued no formal warnings, all else being equal. Work search overpayment rate estimates in states that were frequent users of formal warnings were 3.5 percentage points lower on average than those in states that issued no formal warnings. Our state-level analysis also suggests that some characteristics of cases are not associated with state work search overpayment rate estimates, but others are. State work search overpayment rate estimates were not significantly associated with the distribution of cases by claimant age, gender, duration of unemployment benefits receipt, or key week amount as a percentage of the normal weekly wage in their occupation, or with the number of cases. However, state work search overpayment rate estimates were higher in states with more cases with claimants with some college or an associate’s degree relative to cases with claimants without a high school degree or GED. A one percentage point increase in the estimated fraction of cases with claimants with some college or an associates degree was associated with a 0.192 percentage point increase in estimated work search overpayment rates on average, all else being equal. To assess the sensitivity of our results, we estimated fractional response models with robust standard errors instead of clustered standard errors (see column (2) of table 7). We also estimated fractional response models that replaced the fractions of cases by claimant response method with the fraction of cases with claimants who responded to the BAM audit by all methods combined, with both types of standard errors. Finally, we estimated linear models, also with both types of standard errors. While the direction and magnitude of the relationship between estimated work search overpayment rates and the fraction of cases with claimants required to search for work was generally similar across specifications, it was not always statistically significant. The direction, magnitude, and statistical significance of the relationship between estimated work search overpayment rates and the fraction of cases for which BAM auditors investigated job contacts was generally similar across specifications, with the exception of the linear model with clustered standard errors, where it was not statistically significant. The directions, magnitudes, and statistical significance of the relationship between estimated work search overpayment rates and estimated state formal warning use were generally similar across specifications. Appendix II: Additional Benefit Accuracy Measurement Data Tables This appendix contains several tables that show the underlying data used throughout this report, using the Department of Labor’s Benefit Accuracy Measurement (BAM) data for fiscal years 2013 through 2017. According to DOL, because the BAM data are based on a statistical survey, estimates produced from our analysis of the BAM data are subject to sampling and non-sampling errors. Sampling errors are errors that arise in a data collection process as a result of taking a sample from a population rather than using the whole population. Non-sampling errors are errors or biases that arise in a data collection process as a result of factors other than taking a sample, such as the timeliness of data collection, data entry errors, biased questions in fact-finding, biased decision-making, and inappropriate analysis and conclusions completed by state investigators or false or inaccurate information provided by survey respondents. We express our confidence in the precision of our results by reporting the margins of error associated with 95 percent confidence intervals. This is the interval that would contain the actual population value for 95 percent of the samples the respective agency could have drawn. In addition, it may be misleading to compare one state's work search overpayment rates with another state's rates. According to DOL, no two states' written laws, regulations, and policies specifying eligibility conditions are identical, and differences in these conditions influence the potential for error. The following tables and information are included in this appendix: Table 8: The estimated overpayment amounts for the Unemployment Insurance (UI) program by cause in fiscal year 2017 (also represented in fig. 2 in the letter portion). Table 9: The national work search overpayment rate estimates and dollar amounts for fiscal years 2013 through 2017 (also represented in fig. 3 in the letter portion). Table 10: The work search overpayment rates estimates and dollar amount of work search overpayments in each state in fiscal year 2017, excluding cases where formal warnings were given. Table 11: The percentage of cases in which claimants were required to search for work and the percentage of those cases in which job contacts were investigated in each state in fiscal year 2017. Table 12: For states that issued formal warnings in 2017, the UI overpayment rate estimates excluding and including cases in which formal warnings were issued, and the difference in the dollar amount of overpayments if formal warnings were not used. Appendix III: Summary of Six States’ Work Search Requirements for Unemployment Insurance Claimants Work search requirements for Unemployment Insurance claimants vary by state. According to our review of program documents in our six selected states, the minimum number of work search activities required per week ranged from not specified to four. Three of the states limited the work search activities to applying for jobs or contacting employers in other ways and the three other states had broader definitions of what would qualify as a work search activity. See table 13 below for a summary of the work search requirements, confirmed by state officials, for the six states included in our review. Appendix IV: Comments from the Department of Labor Appendix V: GAO Contact and Staff Acknowledgments GAO Contact Cindy Brown Barnes 202-512-7215 or brownbarnesc@gao.gov. Staff Acknowledgments In addition to the contact names above, Danielle Giese (Assistant Director), Cathy Roark (Analyst in Charge), Carl Barden, Rachel Beers, Deborah Bland, Beryl Davis, Holly Dye, Alex Galuten, Dana Hopings, Joel Marus, Phillip McIntyre, Sheila R. McCoy, Jean McSween, Courtney LaFountain, Stacy Spence, Almeta Spencer, Matt Valenta made significant contributions to this report.

Why GAO Did This Study The UI program, which is overseen by DOL and administered by states, paid $30 billion to about 5.7 million individuals in 2017. Under federal law, to be eligible for benefits, individuals are generally required to actively search for work, but the specific work search requirements vary by state. Yet, states found that some benefits were overpaid to UI claimants who were ineligible because they were not meeting work search requirements. GAO was asked to review improper payments due to UI claimants' failure to actively search for work. Building on GAO's prior work ( GAO-18-133R ), this report examines (1) state administrative practices associated with work search overpayments; (2) selected states' approaches to address work search overpayments; and (3) DOL's oversight and support of states' efforts. GAO analyzed DOL data, including the results of state reviews of a representative random sample of UI payments made from fiscal years 2013 through 2017. GAO also reviewed UI information from six states selected for variation in work search requirements and overpayment rates, interviewed DOL and state officials, and reviewed relevant federal laws, regulations, and guidance. What GAO Found GAO's analysis of Department of Labor (DOL) data found that certain state administrative practices, such as reviewing a higher percentage of claimant-reported work search activities and frequent use of formal warnings, were associated with lower estimated work search overpayment rates for the Unemployment Insurance (UI) program. According to DOL data, 22 states were warning claimants after the first discovered occurrence of their failure to meet work search requirements (i.e., issuing formal warnings) rather than reporting that an overpayment was made, while the other states were reporting such cases as overpayments. In 2017, DOL determined that federal law does not permit states to use such policies. GAO's analysis of DOL data shows that in fiscal year 2017, estimated work search overpayments were nearly $1.4 billion (see fig.), but would have been an estimated $1.8 billion (+/-$0.2 billion) greater if states had not issued formal warnings and established overpayments. DOL officials told GAO in July 2017 that the agency would issue a letter to states informing them that federal law does not permit them to warn claimants instead of establishing an overpayment. To date, DOL has not issued the letter. Until DOL provides states with such notification, states may continue to report inconsistent information on overpayments. State officials GAO interviewed reported using multiple approaches to address work search overpayments, including using their online systems that automate collecting information on claimants' work search activities; conducting audits of claimants work search activities beyond those required; and sending automated messages to claimants regarding their work search requirements. Officials said that their approaches encouraged claimants to conduct a more active work search and prevented work search overpayments in some cases. DOL monitors states' work search overpayment rate estimates and has helped states address such overpayments, but lacks clear procedures for how states should verify claimants' work search activities. DOL directs states to verify a “sufficient” number of work search activities during their audits but has not provided information on what is considered sufficient. DOL data show that some states did not review claimants' work search activities for a majority of audited cases. DOL officials said that the agency plans to clarify its procedures after issuing a letter about formal warnings. By clarifying these procedures, DOL will have greater assurance that states are complying with verification requirements. What GAO Recommends GAO is making four recommendations to DOL, including that it provide states information about its determination that the use of state formal warning policies is no longer permissible and clarify its work search verification requirements. DOL agreed with GAO's recommendations and stated that it would take action to address them.

gao_GAO-18-97

gao_GAO-18-97_0

Background The United States railroad system consists of a vast network of operations that includes more than 780 railroads operating across 220,000 miles of track—including about 212,000 grade crossings. Both freight and passenger railroads operate across the system. The freight railroad industry is dominated by the seven largest railroads, referred to as class I railroads, whereas passenger rail service includes Amtrak and 29 commuter railroads. FRA is responsible for providing regulatory oversight of the safety of both freight and passenger railroads. To accomplish this oversight, FRA issues and enforces numerous safety regulations, including requirements governing track, signal and train control systems, grade crossing warning systems, and railroad-operating practices. FRA monitors railroads’ compliance with federal safety regulations through routine and special emphasis inspections on railroads’ systems. FRA’s inspectors generally specialize in one of five areas. These inspection areas are called disciplines and include: (1) operating practices, (2) track, (3) hazardous materials, (4) signal and train control, and (5) motive power and equipment. FRA also has specific responsibilities related to the safety of grade crossings, including issuing regulations regarding the use of train horns at grade crossings. FRA issued regulations in August 2006, after FRA’s analysis illustrated the dangers of whistle bans. Federal regulations require that train horns be sounded in advance of all public grade crossings. However, the regulations also provide an opportunity for public authorities to reduce the effects of noise associated with the train horn by establishing quiet zones. While railroads are directed to cease the routine sounding of the train horn at-grade crossings within quiet zones, the final rule states that train horns may still be sounded in emergency situations and to comply with other federal regulations and railroad operating rules. As of June 2017, there were 570 new quiet zones located across 42 states (see fig. 1). Train engineers are generally required to sound the horns at least 15 seconds, and no more than 20 seconds, in advance of all grade crossings. Train horns must be sounded in a standardized pattern of 2 long, 1 short and 1 long blast, with the volume ranging from 96 decibels to 110 decibels. Each pedestrian crossing or private crossing to an active commercial or industrial site must be reviewed by a diagnostic team and equipped or treated in accordance with its recommendations. The public authority must invite the state agency responsible for grade crossings’ safety and all affected railroads to participate in the diagnostic review. FRA is not required to participate in diagnostic reviews. The Notice of Intent provides railroads and state agencies with an opportunity to provide comments and recommendations on the quiet zone. A complete and accurate U.S. Department of Transportation Grade Crossing Inventory Form must be on file with FRA for all crossings within the quiet zone to reflect the current conditions at each crossing. A Notice of Quiet Zone Establishment must be issued to FRA, applicable railroads, and relevant state agencies indicating a quiet zone is being established at least 21 days prior to the establishment date. Throughout the process public authorities may work with a number of stakeholders who have roles and responsibilities related to grade crossings. These include: FRA: In addition to issuing rules and regulations governing train horns and quiet zones, FRA has staff—in headquarters and in FRA’s eight regional offices—that review public authority applications for use of ASMs, issue guidance on implementing federal regulations, answer questions from the public, and provide technical assistance related to the establishment of quiet zones. For example, FRA’s 19 regional GCMs serve as subject matter experts on the train horn regulations and respond to questions from public authorities, while FRA program officials approve ASMs and conduct required annual reviews of quiet zones established relative to the Nationwide Significant Risk Threshold to ensure they equal or fall below this risk index. Railroads: Railroads work with public authorities to: (1) identify appropriate safety measures at grade crossings; (2) participate in diagnostic review meetings when the quiet zone includes public, private, or pedestrian grade crossings; (3) receive and comment on public authority’s quiet zone notifications (e.g., the Notice of Intent and Notice of Quiet Zone Establishment); (4) install safety measures on railroad property; and (5) direct train crews not to sound horns in established quiet zones. State departments of transportation and rail regulatory agencies: These agencies receive and comment on Notices of Intent, public authority applications, and Notices of Quiet Zone Establishment; review, and in some cases approve grade crossing modifications; and participate in diagnostic reviews. Private industry consultants: In some cases, public authorities hire consultants to provide subject matter expertise on establishing quiet zones. Consultants may perform such tasks as determining the feasibility of a quiet zone; arranging diagnostic reviews; assessing quiet zone risks; and identifying appropriate safety measures. According to FRA officials, federal funding is available to reduce the risks of accidents at grade crossings, but funding specific to quiet zones is limited and no dedicated source exists. The primary source of federal funding to improve grade crossings’ safety is the Federal Highway Administration’s (FHWA) Railway-Highway Crossings (Section 130) Program, which received a set-aside of $230 million for fiscal year 2017 from amounts authorized for the Highway Safety Improvement Program. While the funds are not specific to quiet zones, Section 130 funds may be used to upgrade crossing infrastructure, an upgrade that may result in a public authority’s being more easily able to establish a quiet zone. However, according to FRA program officials, the program is competitive and funding must be used for safety projects. They said projects are selected on a safety priority basis, and quiet zones are generally considered a quality of life issue, not a safety improvement. Hence, it is unlikely that many public authorities will obtain these funds to establish quiet zones. Further, the officials said that while other federal funding is available for which grade crossing improvements may be an eligible expense, none is dedicated to quiet zones. According to FRA officials, limited federal funding is available because quiet zones are not a national issue. They produce highly localized quality-of-life benefits and little or no improvement in the level of safety at grade crossings, but rather the safety measures are installed to compensate for silencing the sound of a train horn at grade crossings. As a result, public authorities seeking to establish quiet zones generally fund the installation of SSMs and ASMs. Given limited funding, public authorities determine whether the benefits of establishing a quiet zone outweigh the costs to establish them. Benefits of Quiet Zones Have Not Been Quantified, and Costs Depend on Many Factors While Benefits of Quiet Zones Have Not Been Quantified, Selected Stakeholders Highlighted Benefits for Communities Benefits derived from establishing quiet zones and reducing noise from the train horn have not been quantified in research we reviewed or by the public authorities (i.e., communities) that we interviewed. Specifically, our review of literature did not identify any studies that had quantified the benefits resulting from public authorities establishing quiet zones at grade crossings where the horn was previously sounded. Further, FRA has not quantified benefits associated with quiet zones, but did note in its RFIA that quiet zones would likely result in localized quality-of-life benefits from silencing of the horn at locations where it had previously been sounded. Finally, none of the public authorities we interviewed have conducted any analysis that has quantified benefits associated with quiet zones or were aware of any studies that quantified these benefits. While the benefits of quiet zones have not been quantified, the majority of stakeholders whom we interviewed stated that quiet zones do provide benefits for communities. The most commonly cited benefit (35 of 40 stakeholders) was the reduction in noise due to the absence of routine sounding of the train horn. Stakeholders told us this noise reduction led to improvements in quality of life from, for example, the ability to sleep better at night, as well as a reduction in residents’ noise complaints. To a lesser extent, stakeholders also cited economic development and safety as benefits for communities. Almost half of the stakeholders (19 of 40) we interviewed told us that areas with new quiet zones saw an increase in economic development from such things as new businesses or residential developments. Similarly, almost half of the stakeholders (17 of 40) said that quiet zones increased safety along rail lines, given the addition of new safety measures at the grade crossings. While the benefits associated with quiet zones have not been measured, more generally, researchers have analyzed the effect of transportation noise on property values and health to understand the effects. Property values: Our review identified two studies that analyzed the effect of freight train noise on property values in selected communities and found mixed results. In one study, the authors looked at the effect of a freight rail line on home prices and concluded that, while for smaller homes results suggest a negative and statistically significant effect on sale prices, results for medium and larger units were mixed. In the second study, the author examined the effect of a railroad’s decision to ignore whistle bans and found that proximity to rail lines and crossings had a negative and statistically significant effect on residential property values in some communities, with the effect varying depending on distance to the rail line. The author concluded that the crossing effects were largely temporary, because over time, buyers less sensitive to noise would likely move into the area, reducing or eliminating any long-term effect of the railroad’s decision. However, both of these studies have limitations, are based on data almost two decades old, and the results might not be representative of the economic effects associated with quiet zones. Health effects: In 2002, FRA summarized available academic literature on the undesirable effects of noise—primarily focusing on transportation noise associated with aircraft, highways, and railroads. According to the research, transportation noise can cause undesirable psychological health effects, such as annoyance, and physiological health effects, such as hearing impairments and sleep disturbance on individuals. Costs to Establish Quiet Zones Depend on Many Factors Total costs to establish quiet zones depend on many factors and vary widely. Prior to issuing regulations, in the RFIA, FRA identified the types of costs associated with establishing quiet zones that can be incurred by public authorities, states, railroads, and FRA. These factors included such things as upgrading signals at grade crossings; purchasing, installing, and maintaining safety measures like flashing lights and gates; developing, reviewing, and evaluating quiet zones; and designing public education and awareness efforts. The actual cost that public authorities incur to establish quite zones will vary and depend on these and other factors. Both FRA program officials and FRA guidance has stated that, in general, the factors that affect the costs include such things as the number of grade crossings in a quiet zone, the geography of the area in which the quiet zone is established, and the types of safety measures a public authority decides to install. For example, some grade crossings may require upgrades to constant-warning-time devices or installation of complex and costly SSMs (e.g., four-quadrant gates), whereas other grade crossings may require fewer upgrades or less complex safety measures (e.g., traffic channelization devices). In 2013, FRA published guidance for public authorities in which it estimated that the capital costs public authorities may incur to establish quiet zones may range from about $30,000 to more than $1 million per grade crossing, depending on the types of safety improvements and existing infrastructure at grade crossings. The RFIA stated that, because grade crossings may differ significantly, public authorities must analyze the characteristics of each and the safety measures needed to accurately estimate costs to establish quiet zones. Public authorities we interviewed confirmed that the costs to establish quiet zones do vary and depend on many factors. All 13 public authorities we interviewed often said that in establishing quiet zones they incurred costs for identifying safety measures for grade crossings, purchasing and installing these safety measures, and maintaining quiet zones, among other things. According to the public authorities we spoke with in our review, the cost to establish quiet zones ranged from about $14,000 to several million dollars. However, this range also reflects different levels of quiet zone activity; for example, one public authority established a quiet zone at a single grade crossing, while another established a quiet zone that encompassed 60 grade crossings. In addition, railroads, states, and FRA may incur costs as part of establishing quiet zones. For example, officials from seven of the eight railroads we interviewed stated that they incur costs for such things as (1) participating in diagnostic reviews, (2) commenting on Notice of Intents and Notice of Quiet Zone Establishments; and (3) notifying and training crews not to sound horns in quiet zones. States may also incur costs. Two states included in our review—California and Colorado—have public utility commissions that told us they are required to review and approve any modifications to grade crossings in their states, including those associated with quiet zones. Finally, FRA incurs costs related to quiet zones. This cost includes reviewing quiet zone applications, participating in diagnostic reviews when invited, and the time GCMs or other FRA staff spends providing technical assistance to public authorities and others on establishing quiet zones. While public authorities are generally responsible for paying the costs to establish quiet zones, about half of the public authorities we interviewed (10 of 13) said they obtained funding from outside sources to help pay for the zones, for example: Federal funds: Six of the public authorities we interviewed reported receiving federal funds to help establish their quiet zones. In particular, one public authority that we interviewed reported receiving a $3.3 million Transportation Investment Generating Economic Recovery grant to establish a quiet zone. Alternatively, public authorities in the remaining five communities were eligible for grade crossing safety improvement efforts that were designated by the state through FHWA or other programs. State or railroad funds: For three of the public authorities we interviewed, quiet zones were established in conjunction with larger state department of transportation highway or railroad projects and these entities paid a portion of the costs. Grade crossing incentive funds: Four of the public authorities we interviewed received grade-crossing incentive funds from railroads or state departments of transportation to close grade crossings that were part of a quiet zone. Private funds: In two communities, private investors provided financial assistance to public authorities for a quiet zone. For example, a private developer paid for a quiet zone in order to facilitate the building of residential developments. Selected Public Authorities and Other Stakeholders Reported Several Challenges in Establishing Quiet Zones and Suggested Potential Improvements Public authorities and other stakeholders that we spoke with reported several types of challenges with establishing quiet zones. These stakeholders noted three primary challenges, which included the cost to establish quiet zones, obtaining stakeholder cooperation, and the process to establish quiet zones. As aforementioned, public authorities generally incur costs to establish quiet zones, so cost plays a major role in a public authority’s decision of whether to pursue a quiet zone or not. The most commonly cited challenge was cost (29 of 40 stakeholders). In some cases, officials whom we interviewed reported that costs were the main reason that public authorities delayed or discontinued the process to establish a quiet zone. In addition to cost, stakeholders cited two other primary challenges to establishing quiet zones—obtaining cooperation among quiet zone participants and the process for establishing quiet zones—and suggested a variety of improvements related to bolstering the process. Cooperation among quiet zone participants (18 of 40): Although stakeholders we spoke with cited a number of cooperation issues, including difficulties in getting private grade crossing owners to participate and lack of state cooperation, over half (10 of 18) cited cooperation between public authorities and railroads as a challenge. Such cooperation is important since both must typically work together to establish quiet zones. However, there are natural tensions between public authorities and railroads with respect to establishing quiet zones. As discussed earlier, stakeholders we spoke with supported quiet zones believing they not only maintain safety, but improve quality of life. On the other hand, all eight railroads told us that the train horn is the most effective safety measure. The process for establishing quiet zones (16 of 40): In general, the stakeholders we spoke with cited a variety of process related challenges, including that the train horn regulations are difficult to understand, FRA waivers are difficult to obtain, and that the quiet zone process could be better explained by FRA. In particular, over half of the stakeholders whom said that process was a challenge (10 of 16) explained that the quiet zone process was either difficult to understand or navigate or that the requirements to establish a quiet zone were confusing. For example, one public authority told us that rules for establishing a quiet zone can be difficult to interpret and that this difficulty could impact public authorities’ establishment of quiet zones. Four of the 16 stakeholders also told us the process was time consuming and, in some instances, can take years to complete. FRA program officials said the turnaround time for FRA reviews depends on the quality of materials provided. They said it generally takes 90 to 120 days for FRA to complete its review, but it can take longer if there is missing information or other problems with a public authority’s application, as is often the case. Stakeholders we spoke to suggested three types of process-related improvements: administrative changes to improve the efficiency of the process, improvements to FRA’s role in the quiet zone process, and improvements to FRA guidance that public authorities use to establish quiet zones. Administrative improvements: Twenty-five of the 40 stakeholders that we interviewed identified one or more types of administrative process improvements to improve the efficiency of the process for establishing quiet zones or better facilitate their establishment. These suggested improvements included: Making the quiet zone process more user-friendly (11 of 40 stakeholders that offered suggestions related to the quiet zone process): Stakeholders we interviewed identified various improvements that could streamline some administrative requirements of the quiet zone process. These steps include standardizing or automating the quiet zone process, developing sample Notices of Intent or Notices of Quiet Zone Establishment that public authorities could use to input information, and making quiet zone materials available electronically. For example, GCMs in one FRA region told us that by standardizing the paperwork all regions would receive the same documents, a step that would make review easier. In addition, these officials said public authorities often forget to include key information in the Notice of Intent and with a standard form this may not occur. Requiring diagnostic reviews for all quiet zones (7 of 40): As discussed earlier, when there are private grade crossings that allow public access to active commercial or industrial sites or pedestrian grade crossings in a quiet zone, a diagnostic review is required. The regulations require public authorities to provide state agencies and affected railroads, among others, the opportunity to participate in diagnostic reviews. According to FRA program officials, FRA is not required to participate in diagnostic reviews. Diagnostic reviews evaluate conditions at proposed quiet-zone crossings and a diagnostic review team makes recommendations about measures that are needed to protect safety at these crossings. Seven stakeholders we interviewed suggested that diagnostic reviews should be required for all quiet zones, not just instances when there are private or pedestrian crossings. For example, one GCM told us conducting a review for all grade crossings provides a better idea of what safety measures are needed and is a prudent action to protect public safety. FRA’s Role in the Process: About half of the stakeholders we spoke with (21 of 40) suggested improvements related to FRA and its role in the quiet zone process: Increase FRA oversight and inspection of quiet zones (11 of 40): In general, these stakeholders believe FRA should be more involved with inspections and oversight of quiet zones, particularly between when a Notice of Quiet Zone Establishment is issued and when a quiet zone is established. Most of the railroad stakeholders we spoke with (6 of 8) believe there is a need for increased FRA involvement with quiet zones’ oversight. Among the railroad concerns were that without additional FRA oversight, quiet zones may not achieve compliance with the train horn regulations, and that public authorities may not actually install the safety measures identified in the Notice of Quiet Zone Establishment. A GCM in one FRA region told us that officials discovered noncompliant safety measures and missing signs after quiet zones had been established in this region, and that safety measures that were supposed to be installed were not. We discuss quiet zone oversight issues later in this report. Expedite FRA’s review of quiet zone applications (10 of 40): As discussed earlier, FRA plays a role in the quiet zone process, in part, by reviewing quiet zone applications when ASMs are used. The 10 stakeholders felt that FRA should expedite its review process. For example, a GCM in one FRA region suggested FRA shorten the review time by developing a list of frequently used ASMs and their safety effectiveness ratings and posting them online, a process that would save FRA time when reviewing ASMs. Guidance about the process: Finally, stakeholders we spoke with suggested guidance on the quiet zone process could be improved (17 of 40). In particular, 13 of the 17 stakeholders whom offered suggestions about guidance said that FRA’s quiet zone guidance should be clearer or that training about the quiet zone process is needed. As previously discussed, some stakeholders told us the quiet zone process is difficult to understand or navigate, or that FRA could better explain the process. In particular, two public authorities suggested some form of step-by-step guide is needed to better describe the process, and GCMs in three FRA regions also suggested classes or other types of education were needed to better help public authorities understand the quiet zone process. According to FRA program officials, FRA’s quiet zone guidance consists of its user guide and a document on how to create a quiet zone. The train horn regulations also specifies how public authorities are to establish quiet zones and includes steps to follow under the public authority designation or public authority application processes. Moving forward, FRA is in the process of conducting a retrospective regulatory review and deciding what, if any, changes may be needed. In March 2016, FRA issued a Notice of Safety Inquiry, which, according to FRA, is a retrospective review of the train horn regulations. The Notice of Safety Inquiry solicited comments about many aspects of the train horn regulations, including whether FRA can decrease the barriers public authorities encounter when establishing a quiet zone. Among other things, the inquiry seeks comments about whether there should be an online process for submitting notices and other required quiet zone paperwork, whether diagnostic reviews should be required for all quiet zones, and if the regulations should be amended to include common ASMs in the list of approved SSMs. The Inquiry is also looking at other aspects of the quiet zone process and guidance. As of July 2017, FRA was still in the process of reviewing comments received in response to the notice. FRA program officials did not indicate what, if any, changes may result from this inquiry, but said any changes that are made would be handled through a rulemaking. However, FRA program officials noted that a rulemaking would not be necessary for the agency to provide public authorities with additional tools to aid in the development of a quiet zone, such as guidance. FRA Has Conducted Analyses of Safety in Quiet Zones and Is Formalizing Quiet Zone Inspections, but Limitations Exist FRA’s Analyses Generally Indicate That Grade Crossings in Quiet Zones Are As Safe As The Same Grade Crossings When the Train Horn Was Sounded, but Methodology Has Limitations One way FRA evaluates the effectiveness of its train horn regulations is through conducting analyses of data on the safety of grade crossings in quiet zones. Those analyses show that grade crossings in quiet zones are generally as safe as the same grade crossings when the train horn was sounded. Specifically, FRA conducted analyses in 2011 and 2013 to assess whether there was a statistically significant difference in the number of accidents before and after implementation of quiet zones. The results showed that there was generally no statistically significant difference in the number of accidents that occurred before and after quiet zones were established. To conduct the analyses, FRA grouped quiet zones by the number of years of available data since establishment of the quiet zone, using an equal number of months before and after establishment. FRA’s analyses in 2011 and 2013 included 359 and 203 quiet zones, respectively. While FRA’s analyses of quiet zones generally showed that grade crossings in quiet zones were as safe as the same grade crossings when the train horn was sounded, in 2013 FRA identified one exception that FRA program officials reported resolving in a subsequent analysis. Specifically, while FRA’s 2011 analysis did not show any differences in safety after establishment of the quiet zones, in 2013 FRA concluded that for quiet zones established from May 2010 through April 2011, there was a statistically significant increase in the number of accidents that occurred after the establishment of the quiet zones. Specifically, FRA found that accidents doubled from 11 accidents before establishment of the quiet zones to 22 accidents following the establishment of the quiet zone. After that finding, FRA program officials conducted a preliminary analysis for 2017 and reported that the results did not show a statistically significant increase in accidents for any period of quiet zones, including those established from 2010 through 2011. In addition to looking at quiet zones by establishment year, FRA’s 2013 analysis also grouped quiet zones by how they were established, such as with safety measures at all crossings or against FRA’s risk indexes. Results from this analysis did not show an increase in accidents by any establishment method analyzed. As a result, FRA program officials told us that they believe the result in 2013 for quiet zones established from 2010 through 2011 was likely an anomaly and that those quiet zones are as safe as other crossings. Before-and-after analysis is a methodologically acceptable practice, but the reliability of the results decrease over time because unlike other types of analyses, they do not control for factors that may change over time. In particular, FRA’s analyses assume that the number of accidents experienced before the quiet zone is established is a good estimate of the number of accidents that would be expected in the future had the quiet zone not been established. However, FRA’s before-and-after analyses have limitations because, unlike other methodologies, they do not take into account changes to characteristics of grade crossings over time. For example, a multivariate method can control for changes to characteristics at grade crossings that may impact safety. These characteristics can include changes to train or vehicle traffic, train or vehicle speeds, time of day when train activity occurs, number of highway lanes, the number of tracks in use, or other changes to surrounding roads or infrastructure at a crossing. For example, if train or vehicle traffic increased over time, it is possible that the number of incidents would increase, while the risk of an accident would stay the same. Specifically, closing a grade crossing near a quiet zone or increases in traffic from new businesses around a quiet zone could increase traffic after the establishment of a quiet zone; however, these changes would not be factored into FRA’s current methodology for conducting safety analyses. This inherent limitation is exacerbated over time, because the assumption that there would be no changes to relevant characteristics of the grade crossings is less likely to be the case as more time passes. FRA also conducts annual reviews of selected quiet zones to ensure their safety, and FRA program officials told us that this review further validates its before-and-after analyses. As mentioned previously, FRA conducts annual reviews of quiet zones established against the Nationwide Significant Risk Threshold because the measure is variable and subject to change over time. According to FRA program officials, about 11 percent of all quiet zones are established against the Nationwide Significant Risk Threshold and are thus included in this annual review. To ensure that established quiet zones fall at or below the Nationwide Significant Risk Threshold, FRA is required to recalculate this measure on an annual basis and notify a public authority if the Quiet Zone Risk Index no longer falls at or below the threshold. By doing so, FRA program officials told us that they are further validating that the grade crossings in quiet zones are as safe as other grade crossings. While this annual review may provide FRA with additional support that grade crossings in quiet zones are as safe as others, it does not address the underlying limitations of a before-and-after analysis. While the reliability of a before-and-after analysis may decrease over time, FRA has no plans to revise its methodology. In fact, as mentioned previously, FRA program officials told us that preliminary results for their 2017 safety study mirror results from 2011, showing that there was no statistically significant difference in accidents before and after the establishment of quiet zones. According to FRA program officials, the agency is not required to conduct this analysis, but moving forward, program officials plan to conduct the same analysis on a biennial basis to internally validate that grade crossings in quiet zones are as safe as others. By continuing to rely on the current methodology, FRA’s future analyses may continue to provide the agency with information that does not account for changes in characteristics of grade crossings over time. The Standards for Internal Control in the Federal Government states that management should use quality information to make informed decisions. This requirement can be satisfied by, for example, obtaining relevant data from reliable sources, obtaining that information on a timely basis, and processing that data into quality information that accurately represents what it purports to represent. Furthermore, a previous FRA study that the agency relied on in developing the final rule has reported that changes in grade crossings’ characteristics can affect the results of analyses used to predict accidents at grade crossings. As a result, FRA’s Rail-Highway Crossing Resource Allocation Procedures recommended that analyses used to predict accidents at grade crossings only include accident data for the most recent 5 years because older accident history information may be misleading due to changes that occur in grade crossings’ characteristics over time. While FRA’s recommendation was not developed to analyze the safety of grade crossings in quiet zones, the agency’s recommendation that accident data older than 5 years may be misleading because of changes that occur to grade crossings’ characteristics over time is relevant to those analyses. Nevertheless, FRA program officials told us that they have no plans to revise the methodology because it effectively compares the safety of grade crossings in quiet zones to other grade crossings. By continuing to use the same methodology, the agency may be missing an opportunity to fully understand the safety of grade crossings in quiet zones. FRA Has Taken Steps to Formalize Quiet Zone Inspections, but Lacks Formal Guidance In addition to conducting studies, FRA also oversees quiet zones by inspecting them to ensure their safety and compliance with train horn regulations. According to FRA program officials, FRA is not required to inspect quiet zones; rather, public authorities, in conjunction with the railroads, are responsible for maintaining quiet zones and ensuring compliance with train horn regulations. Until recently, FRA has utilized its GCMs to, among other things, informally inspect quiet zones and work with public authorities to resolve issues affecting the safety of quiet zones—issues such as foliage covering signage, maintenance issues with safety devices, or outdated pavement markings. In fact, GCMs in all eight regions told us that they informally inspect quiet zones. According to FRA program officials, the agency has recently identified the need for “more eyes on the ground” to more systematically address maintenance issues within quiet zones and to ensure compliance with train horn regulations. As a result, FRA is transitioning its informal inspection program for quiet zones to a more formal inspection process. As of August 2017, FRA had not terminated any quiet zones because of violations or fined any entities for quiet zone violations. August 2017 that they planned to hire 24 new Inspectors. As of August 2017, FRA had also developed the Inspector training curriculum, and began training three Inspectors. FRA program officials expressed uncertainty over when the remaining 21 Inspectors will be hired because of uncertainty regarding FRA’s hiring and training priorities, among other things. In September 2017, FRA program officials told us that one of the newly hired Inspectors had completed the training and had begun inspecting quiet zones. While FRA has started conducting formal quiet-zone inspections, we found that FRA has not developed guidance on how the inspections should be conducted, including guidance on how frequently these inspections should be conducted and what should be examined. As a result, such guidance is not included as part of the training curriculum developed for Inspectors. According to FRA program officials, this guidance has not been developed because program officials are still finalizing the inspection program. Although no guidance has been developed, FRA program officials told us that they are considering inspecting all new quiet zones between when the public authority submits a Notice of Quiet Zone Establishment and when the quiet zone is established. Additionally, FRA program officials told us that existing quiet zones would be inspected based on mission requirements, risk, and the availability of resources, but ideally every 3 years. With respect to how the quiet zones are to be inspected, FRA program officials said they plan to develop guidance for Inspectors that is akin to the other FRA safety disciplines. FRA program officials told us that they are working toward establishing an Audit Division, which would be responsible for developing this guidance. However, as of August 2017, FRA program officials had not provided a timeline for when this division or guidance would be completed. The absence of guidance on inspections is inconsistent with internal control standards. Specifically, the Standards for Internal Control in the Federal Government states that management should implement control activities through its policies that document each unit’s responsibility, or further delineates day-to-day procedures. These procedures may also include the timing of when a control activity occurs and state that management should communicate these policies to its staff. Without this type of guidance, FRA cannot have reasonable assurance that inspections are being conducted consistently across FRA’s eight regions and as FRA intends. Conclusions Grade crossing collisions are one of the leading causes of fatalities in the railroad industry, and ensuring safety in these areas, including those within quiet zones, is a vital part of FRA’s mission. While public authorities are primarily responsible for safety in quiet zones, FRA can help ensure that grade crossings in quiet zones are as safe as others. However, the methodology FRA uses to assess the safety of quiet zones has limitations because it does not account for changes to grade crossings’ characteristics over time. By continuing to rely on this methodology, FRA may be missing an opportunity to ensure that established quiet zones are providing the same level of safety as when train horns were sounded. In addition to its safety studies, FRA is also taking steps to formalize its process for conducting physical inspections of quiet zones. While FRA has started hiring and training a few Inspectors, it lacks guidance on how and when quiet zone inspections are to be performed. Without this guidance, FRA cannot ensure that quiet zones will be inspected consistently across FRA’s eight regions. Recommendations for Executive Action We are making the following two recommendations to FRA: The Administrator of FRA should revise the methodology for the analysis of safety in quiet zones to take into account relevant changes over time— including changes in train and automotive traffic, or in the physical characteristics of the grade crossing. (Recommendation 1) The Administrator of FRA should develop guidance for Inspectors on the nature and frequency of quiet zone inspections. (Recommendation 2) Agency Comments and Our Evaluation We provided a draft of this report to the Department of Transportation for review and comment. The department provided a written response (see app. II), as well as technical comments that we incorporated as appropriate. The department concurred with the second recommendation regarding developing guidance for quiet zone inspectors and partially concurred with the first recommendation regarding revising the methodology for analyzing the safety of quiet zones. The department said it would consider our recommendation to revise its methodology as it explores options for updating its methodology. We are encouraged that FRA is willing to consider revising its methodology for analyzing the safety of grade crossings in quiet zones. However, we continue to believe that our recommendation is valid and that to fully understand quiet zone safety FRA needs to revise its methodology to account for relevant characteristics of quiet—zone grade crossings. As we state in the report, the reliability of FRA’s current methodology will likely decrease over time because it does not control for relevant changes to grade crossings in quiet zones including changes to vehicle or train traffic or speeds. These and other factors are critical determinants of grade-crossing safety. Further, developing a methodology that incorporates characteristics that affect safety at grade crossings in quiet zones may also provide FRA insight into the safety of grade crossings more generally. Since grade-crossing accidents are a major source of fatalities and, according to the department, are expected to increase as train- and highway-traffic increases, it will become increasingly important to have reliable information about grade-crossing safety, both in quiet zones and across grade crossings more generally. We will send copies of this report to appropriate congressional committees, the Secretary of Transportation, and the Administrator of the Federal Railroad Administration. In addition, we will make copies available to others upon request, and the report will be available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-2834 or flemings@gao.gov. Contact points for our Office of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix IV. Appendix I: Objectives, Scope, and Methodology The Fixing America’s Surface Transportation Act included provisions for GAO to review the effectiveness of the Federal Railroad Administration’s (FRA) final rule governing the use of train horns at highway-rail grade crossings. The objectives of this report were to determine: (1) what is known about the benefits and costs of quiet zones, (2) what challenges, if any, public authorities and others encounter in establishing quiet zones, and (3) how, if at all, FRA is evaluating the effectiveness of federal train horn regulations. The scope of this report was limited to new quiet zones—that is, quiet zones that were established since FRA published the final rule in August 2006. Federal regulations govern the use of train horns at public-highway-rail-grade crossings (grade crossings) and provide public authorities—typically a city, town, or county—with the opportunity to create quiet zones where train horns are not sounded. We focused on new quiet zones to better understand the benefits, costs, challenges, and safety impacts associated with the regulations. To obtain information about quiet zones, we reviewed FRA’s data on quiet zones established from 2005 through 2017. To assess the reliability of these data, we examined FRA’s reports, analyzed the data to identify any outliers, and interviewed FRA officials about how the data were collected and used. We determined that the data were sufficiently reliable for our purposes. For each of our objectives, we reviewed pertinent law and FRA regulations and documents; interviewed FRA program officials in headquarters; and conducted in-depth interviews with a nongeneralizable sample of 40 stakeholders. This sample included stakeholders from 8 freight railroads, 5 private industry consulting firms with experience helping public authorities establish quiet zones, 6 state agencies, 13 public authorities within these six states, and FRA Grade Crossing Managers (GCMs) in each of FRA’s 8 regions. The railroads selected included all seven class I railroads, plus the Florida East Coast Railway. The latter was selected due to its previous experience with whistle bans, and it was located in a state where we conducted interviews. The private industry consultants were selected based on several factors, including (1) experience with assisting public authorities in establishing quiet zones, (2) recommendations from FRA and other stakeholders we interviewed, and (3) geographic dispersion. We selected six states as part of a nongeneralizable sample for interviews. These states included California, Colorado, Florida, Illinois, Maryland, and Texas. The states were selected based on a variety of factors, including the number of new quiet zones and the number of grade crossings in new quiet zones. Five of the six states accounted for about 48 percent of new quiet zones (California, Colorado, Florida, Illinois, and Texas). We also conducted interviews in Maryland before we conducted other interviews to test our interview protocol. Maryland was selected for this purpose to, among other things, minimize resources. Within these states, we conducted interviews with 13 judgmentally selected public authorities (see table 1). The public authorities were also selected based on factors such as the number of new quiet zones and recommendations from FRA and other stakeholders we interviewed. For all our objectives we also conducted a literature review of pertinent studies in scholarly/peer-reviewed journals, conference papers, non-profit or think tank publications, and trade publications or industry articles to identify research on quiet zones. We restricted our review to results published between January 1, 1996, and October 17, 2016, and our search yielded 99 results. Of these 99 results, we reviewed each abstract or full article if available, to determine whether it was relevant to any of our objectives. Our analysis identified 10 results pertaining to safety, 11 results related to benefits and costs, and 1 result related to challenges. With respect to the articles related to costs and benefits, we also looked at citations within the studies we reviewed to identity whether any of these were relevant to our objective on costs and benefits of quiet zones. Using this approach we identified one additional study. Each abstract was reviewed by two analysts to determine whether it seemed relevant. Where disagreement existed with respect to whether the abstract was relevant, we included the abstract in our request for the complete study. We then developed criteria/requirements for each objective and reviewed each study against our criteria/requirements. Namely, we were only interested in studies that quantified the benefits or costs of quiet zones or that used data or analytics to measure safety at grade crossings in quiet zones or compared safety at-grade crossings in quiet zones to grade crossings where train horns sound. Further, each study was reviewed by an analyst and a statistician or economist to determine its relevance. With respect to our objective on the effectiveness of the train horn regulations, we determined that none of the studies met our underlying criteria. Specifically, none of the studies measured the safety at grade crossings in quiet zones or compared the results to grade crossings where the train horn sounded. Conversely, with respect to our objective on the costs and benefits of quiet zones, we determined that six studies were relevant. To assess the reliability and methodological soundness of the studies we determined were relevant, we compared the studies with general guidelines based on standards for assessing research and analysis from the literature, past GAO reports on evaluating research programs, and our internal expertise in research design. These guidelines include, for example, examining a study based on: (1) the extent to which it was well designed and the methodology supports the objectives; (2) whether the assumptions were reasonable and explicitly stated; (3) whether the study used the best available data; and (4) whether the conclusions and recommendations were balanced and supported by data analysis. To determine what is known about the benefits and costs of quiet zones, we reviewed the literature search discussed above and analyzed any studies obtained using the methodology described above. We also reviewed FRA’s Regulatory Evaluation and Regulatory Flexibility Assessment for Use of Locomotive Horns at Highway-Rail Grade Crossings Final Rule (RFIA). The RFIA was issued before the final rule and analyzed the potential economic effects of requiring the train horn to be sounded at all public grade crossings, of eliminating whistle bans, and of providing conditions under which the train horn can be silenced at- grade crossings. To review the RFIA, we compared it to selected principles from Office of Management and Budget’s (OMB) guidance for developing regulatory analyses. These principles included whether the analysis considered alternatives; whether the analysis estimated the incremental effect of the rule compared to a business-as-usual baseline; and whether the analysis analyzed uncertainty. In evaluating the RFIA, an analyst and economist independently reviewed the analyses and subsequently came to consensus about each element’s adherence to OBM guidance. We also reviewed FRA’s September 2013 user guide for quiet zones. This guide provides a high-level overview of the quiet zone process, including an estimated cost range to establish quiet zones. We discussed the cost range with FRA, including the source of the information and its reliability. Since FRA program officials told us it was an “order of magnitude” estimate and not meant to represent actual costs to establish quiet zones, we did not determine the reliability of the information. As a result, the cost range information is used for illustrative purposes only, and we included a disclaimer about its reliability. Finally, we interviewed FRA GCMs in all eight of FRA’s regional offices, states, public authorities, railroads, and private industry consultants about the benefits and costs of establishing quiet zones. Some of these stakeholders provided information about costs to establish quiet zones, but this was anecdotal, and we did not attempt to verify its completeness or accuracy. To determine the challenges encountered by public authorities and other stakeholders in establishing quiet zones and improvements stakeholders suggested to the quiet zone process, we interviewed FRA GCMs, states, public authorities, railroads, and private industry consultants. We asked these stakeholders to identify the primary challenges in establishing quiet zones and for suggested improvements to the quiet zone process. We then analyzed the information obtained to identify common themes of challenges or suggested improvements. Based on this analysis, an initial list of categories for each challenge and improvement was then developed along with their definitions. The definitions identified specific types of comments to be included in each challenge or improvement category. After reviewing the initial list for overlaps and duplication, as well as to keep the list manageable, a final consolidated list was developed that consisted of eight types of challenges and seven types of improvements (see table 2). Using this list, an analyst then reviewed each interview and judgmentally assigned the information into one of the categories. A second analyst then independently reviewed these assignments using the consolidated list of categories and definitions. Any differences were then reconciled by the two analysts. To further enhance our understanding of quiet zone challenges and improvements, we reviewed guidance issued by FRA about quiet zones and the train horn rule. This included FRA’s How to Create a Quiet Zone document (posted to the FRA website in September 2012) and FRA’s user guide about quiet zones published in September 2013. Additionally, we reviewed FRA’s regulations governing train horns and quiet zones. We also interviewed FRA program officials about the quiet zone process, application processing, various aspects of the train horn rule, and obtained information from FRA about quiet zone guidance. To determine how FRA is evaluating the effectiveness of the federal train horn regulations, we reviewed FRA’s analysis of the safety of quiet zones at highway-rail-grade crossings completed in 2011 and 2013, which compared the safety of grade crossings in quiet zones to the safety of grade crossings where the train horn is sounded. We also discussed with FRA program officials the methodologies used to prepare these studies, and concerns with the data, conclusions, and plans to conduct future analyses. To assess the reliability and methodological soundness of the studies, we used the same approach as above. Both analyses were reviewed by a statistician and economist to corroborate the review. In addition to developing criteria for reviewing the analyses, we also reviewed guidance by FRA and others regarding analyzing incident data at grade crossings and about the limitations of a paired t-test—FRA’s methodology for comparing the grade crossings. To assess the extent to which FRA’s methodology generally reflects internal control principles, we reviewed it against practices for presenting accurate information and communicating with internal and external stakeholders outlined in the Standards for Internal Control in the Federal Government. We also conducted data reliability assessments with respect to the underlying data FRA used in its analyses. FRA’s analyses used data that originated from two distinct FRA databases: ccmMercury (CCM) and the Safety Data Analysis website. CCM is a correspondence management system which includes all data on quiet zones—such as the establishment date and grade crossings included, among others. This information is contained in the Notice of Quiet Zone Establishment that the public authority establishing the quiet zone is required to provide to FRA. Alternatively, the Safety Data Analysis website contains two datasets: the Grade Crossing Inventory System (GCIS) and the Railroad Accident/Incident Reporting System (RAIRS). The GCIS contains information on every crossing in the nation and was used to identify the characteristics of the individual crossings within the quiet zone, whereas the RAIRS contains details about each crossing collision incident that has occurred. To assess the reliability of the data used in our review, we examined FRA reports, reviewed prior GAO data reliability material, and interviewed FRA stakeholders about how the data were collected, stored, and used. We determined that the data were sufficiently reliable for the purposes of our objectives. Finally, to understand how FRA conducts oversight of quiet zones, we interviewed FRA program officials about oversight of quiet zones, guidance to staff and public authorities, and any planned changes for how the agency conducts oversight of quiet zones. We also interviewed GCMs in each of FRA’s eight regions to understand how they carry out oversight of quiet zones and to learn about the extent to which differences exist across regions. We also reviewed prior GAO reports that summarized FRA’s oversight approach to the rail industry, including its more traditional inspection disciplines. We also asked stakeholders included in our sample of FRA GCMs, states, public authorities, railroads, and private industry consultants about the challenges of establishing quiet zones and potential improvements to the quiet zone process. We then assessed FRA’s oversight approach using the Standards for Internal Control in the Federal Government. We conducted this performance audit from July 2016 to October 2017 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Comments from the Department of Transportation Appendix III: List of Organizations Contacted by GAO Appendix IV: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Susan Zimmerman (Assistant Director), Krister Friday, Sarah Gilliland, Timothy Guinane, Richard Jorgenson, SaraAnn Moessbauer, Malika Rice, Amy Rosewarne, Melissa Swearingen, Larry Thomas, and Crystal Wesco made significant contributions to this report.

Why GAO Did This Study Accidents at grade crossings are a major source of fatalities in the railroad industry. FRA—the federal agency responsible for providing regulatory oversight of grade-crossing safety—–issued regulations on the use of train horns in 2005. Railroads generally support sounding the horn, whereas, communities often support quiet zones to reduce noise. Congress included a provision in statute for GAO to examine FRA's train horn regulations, including those on quiet zones. Among other things, this report: (1) describes benefits and costs of quiet zones, and (2) examines how FRA evaluates the effectiveness of its train horn regulations. GAO analyzed FRA's documentation on quiet zones, including FRA's train horn regulations and 2011 and 2013 studies on quiet zone safety; reviewed literature; and interviewed FRA program officials in headquarters, Grade Crossing Managers in FRA's 8 regions, and a nongeneralizable sample of another 32 stakeholders from 6 states, railroads, public authorities, and private industry consulting firms. State and public authorities were selected based on the number of quiet zones, geographic diversity, and FRA's recommendations. What GAO Found GAO found that the benefits of quiet zones—–i.e., highway-rail at-grade crossings (grade crossings) where train horns are not sounded—have not been quantified and that the costs to establish quiet zones vary. The Federal Railroad Administration's (FRA) train horn regulations allow public authorities (e.g., cities or towns) the opportunity to establish quiet zones if they install safety measures that reduce risks associated with the absence of the train horn (see fig.). While GAO did not identify any research that has quantified the benefits of quiet zones, most stakeholders GAO interviewed said that these quiet zones provide benefits to communities, such as reducing noise or increasing economic development. According to FRA guidance, the factors that affect the costs to establish quiet zones can vary based on the number of grade crossings and types of safety measures used. Public authorities, which typically incur the costs and receive the benefits of quiet zones, must therefore decide whether the benefits of quiet zones exceed the costs. To evaluate the effectiveness of its train horn regulations, FRA has analyzed data on grade crossings in quiet zones and is transitioning to a formal process for inspecting quiet zones. Analyses: FRA's analyses showed grade crossings in quiet zones were generally as safe as they were when train horns were sounded. However, these analyses did not control for changes to grade crossings' characteristics over time—–e.g., train speeds or frequency. Such changes may decrease the analyses' reliability. A revised methodology that accounts for these changes could provide FRA with better information on the long-term effects of the train horn regulations, including the safety of quiet zones. Inspections: Recognizing the need for additional oversight, FRA has taken steps to formalize its process for inspecting quiet zones. FRA has primarily relied on public authorities to oversee quiet zones and ensure compliance with the train horn regulations, in addition to informal inspections by FRA's Grade Crossing Managers. In September 2017, FRA began conducting formal inspections of quiet zones using Grade Crossing Inspectors. However, FRA has not developed guidance for how inspections are to be conducted, including how frequently inspections are to be performed or what should be examined. Without guidance, FRA cannot ensure that inspections are being conducted consistently across FRA's eight regions. What GAO Recommends GAO recommends that FRA: (1) revise its methodology for analyzing the safety of quiet zones, and (2) develop guidance on conducting quiet zone inspections. The Department of Transportation partially concurred with the first recommendation, saying it would consider it, and fully concurred with the second. GAO continues to believe changes to the methodology are needed, as discussed in the report.

gao_GAO-18-477

gao_GAO-18-477_0

Background Changing Petroleum Markets Oil and petroleum products markets have changed substantially in the years since the establishment of the SPR. Specifically, U.S. domestic crude oil production has generally been increasing, consumption has been relatively stable, and crude oil and petroleum products markets have become increasingly global. Additionally, U.S. crude oil production is projected to rise further in the future, according to EIA and IEA projections, further reversing a decades-long decline. Recent technological improvements have made onshore production from shale formations economically viable, and domestic crude oil production began to rise in about 2008. The combination of increasing production and relatively stable consumption has resulted in declining net crude oil and petroleum products imports, from a high of about 12 million barrels per day in 2005 to fewer than 4 million barrels per day in 2017. Since these trends are expected to continue, the IEA and EIA both project net U.S. crude oil and petroleum products imports will decline to zero sometime in the late 2020s and the United States will become a net exporter shortly thereafter. Since the IEA 90-day reserve obligation is based on a country’s net imports, there is no such obligation for net exporters; therefore, the United States would have no 90-day reserve obligation as long as it is a net exporter, though it would still be obligated to release reserves in response to supply disruptions. Over the longer term, EIA’s projections show U.S. net exports peaking in 2037 and the United States again becoming a net importer between 2040 and 2050. At the time of the Arab oil embargo, price controls in the United States prevented the prices of oil and petroleum products from increasing as much as they otherwise might have, contributing to a physical oil shortage that caused long lines at gasoline stations throughout the United States. In addition, in the 1970s, oil prices were often set in long-term contracts, which meant that prices would not automatically rise in the face of greater scarcity. This generally reduced incentives for producers to expand production and sales as well as for consumers to reduce consumption in the face of greater scarcity caused by a supply disruption. Now that crude oil and petroleum product markets are global, the prices of these commodities are determined in the world market, primarily on the basis of supply and demand. In the absence of long-term contracted prices or price controls, scarcity from a supply disruption is generally expressed in the form of higher prices, as purchasers are free to bid as high as they are willing to pay to secure oil supply. In a global market, a large enough supply disruption anywhere in the world raises prices everywhere. This creates incentives for producers unaffected by the disruption to increase their production and release existing inventories and for consumers everywhere to reduce consumption in the ways they find most efficient and least disruptive. While it can take time for some of these actions to affect crude oil and petroleum product markets—according to DOE officials, it can take approximately 6 months from when a producer drills an oil well until oil production comes on line—all these actions tend to mitigate the effects of supply disruptions. Strategic Petroleum Reserve The Energy Policy and Conservation Act of 1975 authorized the creation of the SPR, partly in response to the Arab oil embargo of 1973-1974 that caused a shortfall in the international oil market. The purposes of the SPR are, among other things, to reduce the impact of disruptions in supplies of petroleum products and to carry out obligations of the United States under the international energy program. Specifically, the 1974 International Energy Program Agreement, a joint strategy and treaty, established the IEA to address oil security issues on an international scale. The SPR is owned by the federal government, managed by DOE’s Office of Petroleum Reserves, and maintained by Fluor Federal Petroleum Operations LLC. The SPR stores crude oil in underground salt caverns along the Gulf Coast in Louisiana and Texas. The SPR currently maintains four storage sites—Bayou Choctaw, Big Hill, Bryan Mound, and West Hackberry—with a design capacity of 713.5 million barrels. Under conditions prescribed by the Energy Policy and Conservation Act, as amended, the President has discretion to authorize the release of petroleum products from the SPR to minimize significant supply disruptions. When oil is released from the SPR, it is distributed through commercial pipelines or on waterborne vessels to refineries, where it is converted into gasoline and other petroleum products, and then transported to distribution centers for sale to the public. According to DOE documents, well-functioning infrastructure is fundamental to the SPR’s ability to maintain operational readiness and meet mission requirements. However, most of the critical infrastructure for moving SPR oil has exceeded its serviceable life, which has led to increasing maintenance costs and decreasing system reliability. Specifically, the reserve relies on a complex system of salt caverns, pipelines, wells, and pumps, with other infrastructure and equipment. Any failures, such as ruptured pipelines, could affect the readiness of a site for an oil release. According to DOE officials, a growing backlog of major maintenance needs raises concerns about the ability of the system to operate as designed. In addition, there have been equipment failures that have rendered parts of the system temporarily inoperable. For example, the SPR has experienced at least five major equipment failures since fiscal year 2013, including the Big Hill site pipe failure shown in figure 1. Regional Petroleum Product Reserves The United States has two regional petroleum product reserves—the Northeast Home Heating Oil Reserve and the Northeast Gasoline Supply Reserve. The Northeast Home Heating Oil Reserve, which is not part of the SPR, holds 1 million barrels of ultra low sulfur distillate, a petroleum product essentially equivalent to diesel fuel but that is also used for heating oil. The Northeast United States is heavily dependent on the use of heating oil in winter months. The distillate is stored in leased commercial tank storage in terminals in Connecticut, Massachusetts, and New Jersey. In 2000, the President directed the creation of the reserve to hold approximately 10 days of inventory, the time required for ships to carry additional heating oil from the Gulf of Mexico to New York Harbor. The Northeast Gasoline Supply Reserve, a part of the SPR, holds 1 million barrels of gasoline for consumers in the northeastern United States. According to DOE’s website, this region is particularly vulnerable to gasoline disruptions as a result of hurricanes and other natural events. For example, Hurricane Sandy caused widespread gasoline shortages in the region in 2012. DOE conducted a test sale of the SPR in 2014 and used a portion of the proceeds from the sale to create the reserve. The gasoline is stored in leased commercial tank storage in terminals in Maine, Massachusetts, and New Jersey. IEA Obligations The SPR helps the United States meet its IEA obligation to hold the equivalent of 90 days of net imports of crude oil and petroleum products. In order to meet the IEA 90-day reserve obligation, countries, including the United States, can count existing private reserves of crude oil and petroleum products in addition to public reserves (in the United States, the SPR). In most years, the United States has met its 90-day reserve obligation with a combination of SPR and private reserves. The days of import protection may vary based on actual net U.S. crude oil and petroleum products imports as well as the inventory levels of the SPR and private reserves. As discussed previously, because the IEA 90-day reserve obligation is based on a country’s net imports, there is no such reserve obligation for countries that are net exporters of crude oil and petroleum products. The United States also relies on the SPR to meet its IEA obligation to release reserves in the event of a collective action to respond to a supply disruption. Countries contribute to an IEA collective action based on their share of IEA oil consumption, and they can meet their obligation by whatever measure they choose, including release of public or private reserves, or demand restraint. IEA collective actions are designed to mitigate the negative effects of sudden supply shortages by making additional crude oil and petroleum products available to the global market through a combination of emergency response measures, which include increasing supply and reducing demand. In the event of a global market disruption, IEA member countries can call for a collective action after reaching consensus on whether a response is needed. DOE stated that the collective action IEA obligation is more relevant to the SPR’s mission of protecting the U.S. economy from severe petroleum supply interruptions than the 90-day reserve obligation. The United States has participated in each of the three IEA collective actions. In 1991, with the commencement of Operation Desert Storm, DOE released 17.3 million barrels of SPR crude oil. After Hurricane Katrina in 2005, DOE released 11 million barrels of SPR crude oil. Most recently, in June 2011, in response to crude oil supply disruptions driven by hostilities in Libya, DOE released 30.6 million barrels of crude oil from the SPR. The Libya collective action is an example of how, in practice, member countries participate according to national circumstances. After consultations with IEA member countries, all IEA member countries agreed to the Libya collective action, under which 12 of the 28 members at that time contributed to the action. In addition to the three IEA collective actions, the SPR has been used 10 times in response to U.S. domestic supply disturbances that were not IEA collective actions, most notably in response to severe weather events. In Contrast with the United States, Most IEA Members Rely on Private Reserves to Meet Reserve Obligations and Hold Significant Proportions of Their Reserves as Petroleum Products In terms of how they meet their IEA obligations, most other IEA members differ from the United States in two basic ways. Specifically, as of December 2017, most IEA members rely at least in part on private rather than public reserves to meet their obligations, and most hold significant proportions of these reserves as petroleum products rather than as crude oil. In December 2017, before Mexico joined the IEA in early 2018, there were 29 member countries. Of these 29 countries, 25 IEA members had two common attributes: (1) as net importers, they had a 90-day reserve obligation and met that obligation, and (2) they had formal processes for holding and releasing these reserves. As of December 2017, 18 of these 25 members relied entirely or in part on private reserves to meet their reserve obligations. Specifically, based on IEA data as of December 2017, these 18 countries met their 90-day reserve obligation through private reserves and either had no public reserves or had public reserves of less than 90 days. According to a 2014 IEA report, some of these countries require industry to hold reserves and, when needed, release them. For example, according to a 2014 IEA report and documentation provided by government officials, the United Kingdom meets its entire obligation by requiring private industry to hold reserves. In contrast, New Zealand had publicly held reserves amounting to 26 days of net imports, according to IEA data as of December 2017. According to a 2014 IEA report, New Zealand relied on industry reserves held for commercial purposes to meet the rest of its 90-day reserve obligation, although New Zealand does not formally require industry to hold reserves specifically for this purpose. Unlike the 18 countries that rely at least in part on private reserves, as of December 2017, the United States and 6 other IEA members met the 90- day reserve obligation exclusively through public reserves. Specifically, according to IEA data on member reserves, Estonia, Finland, Germany, Hungary, Ireland, Japan, and the United States held public reserves equal to 90 days or more of net imports. Although the United States currently meets its IEA 90-day reserve obligation solely with public reserves, for most of the SPR’s existence, public reserves were insufficient to meet this obligation, so the United States also had to rely on private reserves. Specifically, according to EIA data, the United States has relied, at least in part, on private reserves together with the SPR to meet the 90-day reserve obligation with the exception of two time periods (1984-1987 and 2012-present), when the United States has relied solely on the SPR. The United States does not require industry to hold reserves for the purposes of meeting IEA obligations. Figure 2 compares the United States’ reserves in days of net imports to the IEA’s 90-day reserve obligation. According to a 2014 IEA report, most IEA members hold at least a third of their reserves as petroleum products, such as gasoline and diesel fuel, rather than as crude oil. Holding petroleum products can be advantageous during certain disruptions because such reserves can be directly distributed to consumers, whereas crude oil must first be refined and turned into products, adding response time. According to the IEA’s 2014 report, Germany’s stockholding agency holds 55 percent of its reserve as petroleum products. Similarly, France holds only petroleum products that are distributed geographically across the country so that the reserves can be used quickly in the event of a supply disruption. In contrast, more than 99 percent of the SPR (665.5 million barrels as of March 2018) is held as crude oil, all of which is stored at the four storage sites in Louisiana and Texas. The exception is the Northeast Gasoline Supply Reserve, which, as mentioned previously, is a 1 million barrel gasoline reserve in terminals in Maine, Massachusetts, and New Jersey that was established in 2014 after Hurricane Sandy and that is considered part of the SPR. According to DOE officials, there are several reasons the SPR holds predominantly crude oil, including that it is more costly to store petroleum products than crude oil and that the United States has the largest refining capacity of any IEA member country. Because of the large U.S. refining sector, crude oil from the SPR can be domestically refined into petroleum products to meet demand. Some IEA member countries store some of their reserves abroad, though the United States does not. According to a 2014 IEA report, some IEA member countries allow part of their reserves to be stored abroad to leverage spare storage capacity or more cost-effective storage by utilizing available storage space or excess private reserves in other countries. For example, approximately 30 percent of Ireland’s reserves are held in other European Union countries. In some of these cases, countries use short-term contracts, also known as tickets, instead of directly acquiring and storing oil and petroleum products. For example, according to documents provided by government officials, since 1995 the United Kingdom has increased its reserves held under ticket agreements outside of the country from around 10 percent of its total reserves to more than 25 percent. In addition, unlike the United States, some IEA countries specify the size of their public or private reserves in terms of net imports or consumption, rather than a specific volume. In the United States, the total volume of crude oil and petroleum products held in the SPR is the result of amounts historically purchased to fill the reserve and subsequent sales as mandated by Congress or released in response to a supply disruption. According to DOE, it cannot otherwise reduce or increase volumes held in reserve without congressional action—either through requirements to purchase additional oil or laws authorizing or mandating sales. On the other hand, some IEA countries have tied their reserves’ volumes of crude oil and petroleum products to a metric such as days of net imports or a percent of consumption. For example, according to documentation provided by government officials, in 2015 Japan changed how it specifies its target reserves from a specified amount to days of net imports. In specifying the size of reserves in this way, the amount held is adjusted as market conditions change—for example, if net imports change and require more or fewer reserves to meet the IEA 90-day reserve obligation, or when other underlying factors affecting a nation’s energy security needs change. DOE Has Not Identified the Optimal Size for the SPR or the Potential Need for Regional Product Reserves While DOE has examined a range of sizes for the SPR, it has not identified the optimal size for the SPR to meet U.S. energy security needs and IEA obligations, and DOE’s analysis of SPR sizes was limited in three ways. DOE also has not identified whether additional regional petroleum product reserves should be part of the SPR in U.S. regions identified as vulnerable to fuel supply disruptions. DOE Examined a Range of Sizes for the SPR but Has Not Identified the Optimal Size for SPR and the Agency’s Analysis Was Limited in Three Ways DOE has not identified the optimal size for the SPR and though the agency examined a range of SPR sizes, its analysis was limited in at least three ways. In response to direction from Congress and recommendations from GAO and the DOE Inspector General, DOE developed and published a long-term strategic review of the SPR in August 2016. In DOE’s 2016 review, the agency examined the expected economic benefits of SPR sizes ranging from 430 million to 695 million barrels of oil over a 25-year time horizon (2016 through 2040), but it did not recommend an optimal size for the reserve. DOE’s review did not identify the optimal size for the SPR because of three limitations: DOE did not fully evaluate implications of market fluctuations and estimate needs. DOE did not fully evaluate the implications of falling net imports of crude oil and petroleum products with respect to meeting IEA obligations to hold the equivalent of 90 days of net imports and to respond to collective actions. As mentioned previously, the United States is expected to become a net exporter of crude oil and petroleum products by the late 2020s. Since the IEA 90-day reserve obligation is based on a country’s net imports, this means that at that point the United States would not have a 90-day reserve obligation. However, even as a net exporter, the United States would still have to meet the IEA obligation to respond to a collective action. Yet, DOE’s analysis did not evaluate the SPR’s configuration as it relates to projected fluctuations in net imports or estimate the minimal amount of reserves needed to meet potential future collective actions. Without considering projected fluctuations in net imports or providing an analysis of how much oil is estimated to be needed to meet IEA collective actions, DOE cannot fully advise Congress on the optimal size of the SPR. DOE did not consider private-sector response. DOE’s analyses in its 2016 review focused on the publicly held reserves in the SPR as the only means to respond to oil supply disruptions and did not consider a response from the private sector or through consumers reducing demand. According to DOE’s 2016 review, the underlying analysis for the benefits of the SPR did not consider a response from the private sector for three reasons: (1) while U.S. commercial stocks could conceivably address part of a supply disruption, private industry could also hold oil inventories in a crisis instead of releasing them; (2) unlike most other IEA member countries, the United States does not require private-sector response; and (3) research on the exact nature of private-sector response during a disruption is needed. DOE officials told us the agency has not studied the extent to which SPR releases of crude oil displace what would otherwise have been private releases of inventories. As we reported in September 2014, changing market conditions— most importantly the significant increase in domestic production of oil—have implications for the SPR’s size because increased production has led to increasing private reserves. According to IEA data as of December 2017, U.S. private reserves held the equivalent of 194 days of net import protection coverage, up from about 59 days in 2006. Further, private reserves in the United States consist of both crude oil and petroleum products with more than half in the latter category. For example, as of January 2018, total private reserves of crude oil and petroleum products were about 1.215 billion barrels, of which about 420 million barrels were in the form of crude oil and 795 million barrels were petroleum products, according to the EIA. As of 2013, these private reserves were distributed across the entire country in more than 1,400 terminals, according to the EIA. As we reported in December 2007, international trade in oil and petroleum products has expanded significantly over the past 2 decades, making markets for gasoline and other petroleum products increasingly global in nature. In such a global oil market, higher levels of private reserves can benefit the United States and the rest of the world by helping mitigate a supply disruption. Most experts and stakeholders we interviewed generally agreed that the private sector is in a better position to respond to supply disruptions than they were when the SPR was created. With regard to demand response, DOE officials told us they do not consider this because there is no mechanism to require industry to respond to supply disruptions or consumers to reduce demand in response to a supply disruption. However, DOE has not studied how voluntary response to changes in petroleum product prices affects the need for or efficacy of strategic releases. Without conducting an analysis of how private parties respond to supply disruptions, DOE cannot advise Congress on the optimal size of the SPR because it cannot know how effective such private responses could be in mitigating supply disruptions. DOE did not fully examine costs of differently sized reserves. DOE’s review of the expected economic benefits of differently sized reserves did not fully examine the corresponding costs of those sizes. According to DOE officials, there was no requirement or need to conduct a formal cost benefit analysis of the SPR because the SPR’s oil acquisition and initial capital costs to create the reserve are sunk costs and the ongoing operational costs to maintain the reserve are minimal in comparison. However, this does not take into account the opportunity cost to the government that holding reserves represents; as Congress has mandated several times recently, crude oil from the reserve can be sold to fund other federal priorities. Without additional analysis, such as of the costs and benefits of SPR’s size, DOE cannot fully advise Congress on the optimal size of the SPR. When we reviewed the SPR in 2006 and 2014, we found that DOE had not periodically re-examined the strategic reserves. In 2006, we recommended that the Secretary of Energy reexamine the appropriate size of the SPR. In its response to our recommendation, DOE stated that its reexamination had taken the form of more “actionable items,” including not requesting expansion funding in its 2011 budget and canceling and redirecting the prior year’s expansion funding to general operations of the SPR, based on the Administration’s decision that the SPR’s current size at the time was adequate. Similarly, as previously mentioned, in 2014 we found that changing market conditions have implications for the size, location, and composition of the SPR, but DOE had not reexamined the SPR’s size since 2005. Accordingly, we recommended that the Secretary of Energy undertake a comprehensive reexamination of the appropriate size of the SPR. In response to our recommendation, the 2014 DOE Inspector General recommendation mentioned previously, and the Bipartisan Budget Act of 2015, DOE published its 2016 review. As previously mentioned and reported, crude oil and petroleum markets are constantly changing, but DOE conducted its full evaluations of the SPR more than a decade apart. According to DOE officials, there is no formal policy to periodically reevaluate the SPR. We previously found that federal programs should be reexamined if there have been significant changes in the country or the world that relate to the reason for initiating the program. In that report, we found that many federal programs and policies were designed decades ago to respond to trends and challenges that existed at the time of their creation. Moreover, the Office of Management and Budget Circular A-94 for benefit-cost analysis of federal programs includes guidelines that apply to any analysis used to support government decisions to initiate, renew, or expand programs or projects that would result in a series of measurable benefits or costs extending for 3 or more years into the future. Given changing market conditions and future projections, without conducting additional analysis to supplement its 2016 review and thereafter periodically reexamining the SPR to take into account changes in market conditions and include a thorough consideration of the costs and benefits of a wide range of SPR sizes, DOE cannot provide information to Congress to inform decisions about the appropriate size of the SPR and risks holding too much or too little in the SPR to meet the United States’ evolving energy security needs and IEA obligations. DOE Has Not Identified Whether Additional Regional Petroleum Product Reserves Should Be Part of the SPR DOE has also not fully identified whether additional regional petroleum product reserves should be part of the SPR. Because the SPR stores oil nearly exclusively along the Gulf Coast, the SPR is configured primarily to respond to global oil supply disruptions. However, as we reported in November 2017, the SPR has primarily been used in response to domestic disruptions. The SPR is limited in its ability to respond to domestic disruptions because reserves are almost entirely composed of crude oil and not refined petroleum products, which may not be effective in responding to disruptions that affect the refining sector. For example, as we reported in November 2017, Hurricanes Harvey, Irma, and Maria damaged infrastructure and property, caused the loss of life, and disrupted the operations of refineries representing at least 15 percent of the nation’s refining capacity. DOE has identified regions subject to product supply vulnerabilities as shown in Figure 3. The Quadrennial Energy Review of 2015 recommended that the agency analyze the need for additional or expanded regional product reserves by undertaking updated cost-benefit analyses for all of the regions of the United States that have been identified as vulnerable to fuel supply disruptions. In response to this recommendation, DOE studied the costs and benefits of regional petroleum product reserves in the West Coast and Southeast Coast. According to DOE officials, weather events in the Southeast Coast are of higher probability but lower consequence, and events in the West Coast are of lower probability but higher consequence. DOE did not finalize its 2015 studies on regional petroleum product reserves and make them publicly available. However, the draft 2015 studies concluded that a product reserve in the Southeast would provide significant net economic benefits to the region and the United States, particularly in the event of a major hurricane, while further analyses are needed to determine the potential benefits of a reserve on the West Coast. A prior DOE study also suggests that petroleum product reserves merit consideration—in 2011, DOE carried out a cost-benefit study of the establishment of a refined product reserve in the Southeast and estimated that such a reserve would reduce the average gasoline price rise by 50 percent to 70 percent in the weeks immediately after a hurricane landfall, resulting in consumer cost savings, according to the Quadrennial Energy Review of 2015. According to DOE officials, the agency has no plans to conduct additional studies. DOE’s 2016 review of the SPR did not fully assess whether there is a need for additional regional product reserves in other U.S. regions identified as vulnerable to fuel supply disruptions, as recommended by DOE’s studies and the 2015 Quadrennial Energy Review. Without completing studies on the costs and benefits of regional petroleum product reserves for all the vulnerable U.S. regions and publicly releasing the results, DOE cannot ensure that it and Congress have the information they need to make decisions about whether additional regional product reserves are needed. DOE Has Taken Steps to Update Its Modernization Plans, But Is Hindered by Uncertainty Regarding the SPR’s Long-term Size DOE Has Taken Steps to Update Its Modernization Plans for Currently Mandated Sales DOE has taken steps to take into account the effects of congressionally mandated oil sales in its plans for modernizing the SPR, though DOE’s current plans are based on information largely developed prior to the most recent congressionally mandated oil sales. According to DOE, the SPR modernization program is focused on a life extension project to modernize aging infrastructure to ensure the SPR will be able to meet its mission requirements for the next several decades. The project’s scope of work has undergone several revisions since its inception in response to changing conditions and requirements, according to the agency. DOE has estimated the total cost for the SPR’s modernization at up to $1.4 billion. DOE raised about $323 million for modernization through the sale of SPR oil in fiscal year 2017, and the Consolidated Appropriations Act of 2018 provided that DOE is to draw down and sell an amount of crude oil not to exceed $350 million for modernization in fiscal year 2018. As of the end of February 2018, DOE has spent $22 million on modernization efforts and the additional funds will allow DOE to continue moving forward with the project, according to agency officials. According to DOE’s modernization plans, the first major construction is scheduled for fiscal year 2019. However, these plans are largely based on information DOE analyzed before recent congressionally mandated sales of an additional 117 million barrels of oil. Since the most recent mandated sales, DOE has taken steps to update its modernization plans and has changed its assumptions for SPR’s modernization. For example, DOE now assumes that the reserve will hold about 405 million barrels of oil and that one of the four SPR sites may close after congressionally mandated sales are completed at the end of fiscal year 2027, according to agency officials. However, DOE has not fully updated the SPR’s modernization plans based on these assumptions. According to DOE officials, in March 2018, DOE commenced a study—the SPR post-sale configuration study—to examine potential future reserve configurations. This study is to take into account the effects of congressionally mandated sales on the reserve and its modernization, and is targeted for completion in October 2018, according to agency officials. Information from the study will inform DOE’s updates to the SPR’s modernization plans, according to DOE officials. As part of its post-sale configuration study, DOE plans to examine how the agency may handle the potentially excess SPR facilities created by the mandated sales. In January 2017, the SPR had a design capacity to hold 713.5 million barrels of oil and actually held 695 million barrels. As shown in figure 4, without action by DOE to reduce the SPR’s design capacity or otherwise use SPR facilities, congressionally mandated sales will cause excess storage capacity to grow to 308 million barrels or more by the end of fiscal year 2027—meaning that about 43 percent of the SPR’s total design capacity to store oil would be unused. DOE plans to explore some options to use these potentially excess SPR assets in its ongoing post-sale configuration study. In withdrawing oil to meet congressionally mandated oil sales currently in place (290 million barrels through fiscal year 2027), DOE could close at least one SPR site based on our analysis of projected excess storage capacity. For example, if DOE were to close the smallest SPR site, Bayou Choctaw, the agency could also explore selling the connected pipeline and marine terminal, which are currently being leased to a private company. DOE could also consider leasing excess storage capacity to other countries so that they could store oil at the SPR. DOE has not entered into any such leases with other countries and has not considered such leases because, according to DOE, the SPR has historically lacked capacity to store additional oil. DOE has not proposed any of these options or explored the revenue the agency could generate by selling or leasing these assets. According to DOE officials, the agency will examine the feasibility of such options in the ongoing SPR post-sale configuration study. Uncertainty Has Hampered DOE’s Efforts to Account for Potential Future Mandated Sales As DOE takes steps to plan for the SPR’s modernization, ongoing uncertainty regarding the SPR’s long-term size and configuration have complicated DOE’s efforts. According to DOE officials, this uncertainty makes it extremely difficult to effectively perform any mid-to long-range planning efforts for the SPR’s modernization project, including the execution of major maintenance projects. Congress has generally set the SPR’s size by mandating purchases or sales of oil, and has established and amended the minimum size of the SPR as it pertains to the release of oil for emergency protection. Since 2015, Congress has, across six pieces of legislation, mandated 290 million barrels in additional oil sales. However, DOE developed its modernization plans in 2016. DOE officials told us they do not know whether additional sales will be mandated over the next 10 years or whether other changes may be required to the configuration of the reserve. Any additional congressionally mandated sales or direction to pursue additional petroleum product reserves would require DOE to again revisit its modernization plans and assessments of the potential uses of any excess SPR assets. Oil market projections also have implications for the future of the SPR. Under current projections, the United States may fluctuate between being a net importer and net exporter over the next several decades. Specifically, the United States is projected to become a net exporter by the late 2020s and would then no longer have a 90-day reserve obligation, but it is projected to return to being a net importer between 2040 and 2050. These projected fluctuations could affect the desired size of the SPR in the future. This uncertainty creates risks for DOE’s modernization plans, as DOE may end up spending funds on facilities that later turn out to be unnecessary should Congress ultimately decide on a larger- or smaller-sized SPR than DOE anticipates. Having a long-term target for the size and configuration of reserves helps other IEA member countries manage their reserves. For example, as previously discussed, unlike the United States, some other IEA members have specified in dynamic terms the amount of reserves to be held, such as days of net import protection or days of consumption, rather than specifying a specific static volume amount. Under such approaches, the amount held varies over time as entities managing the reserve acquire or sell reserves in order to meet the target. Setting a long-term target for the size and configuration of the SPR—taking into account projections for oil production, consumption, and IEA obligations—could better position DOE to ensure that funds spent on the SPR’s modernization do not modernize a system that is no longer needed and that DOE is able to adequately plan for potentially excess SPR assets. In the course of our work, we also identified other options for handling potentially excess SPR assets that DOE is not planning on examining, largely because DOE does not currently have the authority to pursue them, according to agency officials. First, DOE could explore leasing storage capacity to private industry. U.S. oil production has generally increased over the last decade. As a result, the private sector may want to lease excess SPR capacity, which may be cheaper than above-ground storage, according to a representative of a private company we spoke with. Fees for doing so could help defray public reserve storage costs. However, officials told us that the Energy Policy and Conservation Act gives DOE authority to lease underutilized storage to other countries, but not to the private sector. Second, if Congress determines that the SPR holds oil in excess of that needed domestically, DOE could explore selling contracts or tickets for the excess oil rather than selling the oil outright. Australian and New Zealand officials told us that if DOE were to sell tickets for SPR oil, tickets would help these countries meet their IEA 90- day reserve obligations. Australian officials told us they have discussed this option with DOE. Currently the United States and Australia have agreed, through an arrangement, to allow Australia to contract for petroleum stocks located in the United States and controlled by commercial entities. According to DOE officials, the arrangement would permit Australia to receive credit from the IEA for tickets it purchases from the U.S. private sector. While the arrangement does not cover government-owned oil in the SPR, if it did, based on our analysis, DOE could generate up to approximately $15 million annually if Australia purchased the maximum allowable amount of oil specified in an arrangement through tickets for excess SPR oil. However, although the Energy Policy and Conservation Act allows DOE to lease underutilized storage to other countries, DOE lacks the authority to sell tickets and does not plan to seek this authority, according to DOE officials. DOE officials told us that they do not plan to examine these options. According to DOE’s real property asset management order, the agency is to identify real property assets that are no longer needed to meet the program’s mission needs and that may be candidates for reuse or disposal. Once identified, the agency is to undertake certain actions, including determining whether to dispose of these assets by sale or lease. As part of its SPR post-sale configuration study, DOE plans to determine whether it is appropriate to close SPR facilities, and the relative benefit of any closures would be informed by potential lease revenues from maintaining sites so they could be leased, according to officials. However, without examining a full range of options in the post-sale configuration study, DOE risks missing beneficial ways to modernize the SPR while saving taxpayer resources. Conclusions Given changing crude oil and petroleum product market conditions and the constrained budget environment, it is important that DOE ensures the SPR is effective at meeting U.S. energy security needs and IEA obligations while being managed and maintained in an efficient manner. In response to congressional direction and recommendations from GAO and DOE Inspector General, DOE conducted a long-term strategic review of the SPR in 2016 after its last comprehensive examination in 2005. In its review, DOE did not determine an optimal size for the SPR, and its analysis was limited in several ways. In particular, DOE did not fully consider recent and expected future changes in crude oil and petroleum market conditions such as the implications of projected fluctuations in U.S. net imports or the role that increased levels of private reserves could play in responding to supply disruptions. DOE also did not perform a full cost-benefit analysis of holding different volumes of reserves. Without supplementing its 2016 strategic review by conducting additional analysis, and periodically conducting such analyses going forward, DOE cannot provide information to Congress to inform decisions about the appropriate amounts of crude oil and petroleum products to hold in the SPR and risks holding too much or too little in the SPR to meet the United States’ energy security needs and international obligations. Such information is needed on a timely basis, to reflect the pace of change in oil and petroleum markets and other relevant factors that affect the optimal size of the SPR. Though the SPR has primarily been used in response to domestic supply disruptions, such as hurricanes, the reserve is limited in this role because it is almost entirely composed of crude oil, and not petroleum products. In this regard, the Quadrennial Energy Review of 2015 recommended that DOE analyze the need for additional regional product reserves for U.S. regions that have been identified as vulnerable to fuel supply disruptions. DOE has not identified whether additional regional product reserves should be part of the SPR or completed studies of all vulnerable U.S. regions, and it has no plans to do so, according to DOE officials. Without conducting or completing studies for all the vulnerable U.S. regions and releasing the results, DOE cannot ensure it and Congress have the information they need to make decisions about potential additional regional product reserves. In the face of declining net U.S. imports, Congress has taken repeated steps to reduce the size of the reserve. Given that net imports are projected to continue to decline through the late 2020s and fluctuate in the future, there may be additional congressionally mandated SPR oil sales. This has created long-term uncertainty regarding the future size and configuration of the SPR. Congress could address this uncertainty by identifying a long-term target for the size of the SPR—either by volume or in terms tied to factors, such as consumption or net import protection, that affect the country’s energy security needs and IEA obligations. Setting such a long-term target could better position DOE to ensure the efficiency and efficacy of federal funds spent on the reserve. DOE has recently begun to study the potential effects of congressionally mandated sales on its modernization plans. As part of its SPR post-sale configuration study, DOE plans to determine whether it is appropriate to close SPR facilities, and the relative benefit of any closures would be informed by potential lease revenues from maintaining sites so they could be leased, according to officials. However, we identified other options for handling potentially excess SPR assets that DOE is not planning to examine in its study, inconsistent with the agency’s order on real property asset management. Although DOE does not currently have the authority to implement these options, according to officials, examining their potential use, including possible revenue enhancement, could inform Congress as it examines whether it should grant such authority. Without examining a full range of options in the post-sale configuration study for handling potentially excess SPR assets, DOE risks missing beneficial ways to modernize the SPR while saving taxpayer resources. Matter for Congressional Consideration We are making the following matter for congressional consideration: Congress may wish to consider setting a long-range target for the size and configuration of the SPR that takes into account projections for future oil production, oil consumption, the efficacy of the existing SPR to respond to domestic supply disruptions, and U.S. IEA obligations. (Matter 1) Recommendations for Executive Action We are making four recommendations to DOE: The Secretary of Energy should supplement the agency’s 2016 long-term strategic review by conducting an additional analysis that takes into account private-sector response, oil market projections, and costs and benefits of a wide range of different SPR sizes. (Recommendation 1) The Secretary of Energy should take actions to ensure that the agency periodically conducts and provides to Congress a strategic review of the SPR that, among other things, takes into account changes in crude oil and petroleum product market conditions and contains additional analysis, such as the costs and benefits of a wide range of different SPR sizes. (Recommendation 2) The Secretary of Energy should conduct or complete studies on the costs and benefits of regional petroleum product reserves for all U.S. regions that have been identified as vulnerable to fuel supply disruptions, and the Secretary should report the results to Congress. (Recommendation 3) The Secretary of Energy, in completing DOE’s ongoing study on the effects of congressionally mandated sales, should consider a full range of options for handling potentially excess assets and, if needed, request congressional authority for the disposition of these assets. (Recommendation 4) Agency Comments and Our Evaluation We provided a draft of this report to DOE for review and comment. DOE provided written comments, which are reproduced in appendix I. Of the four recommendations, DOE agreed with two, partially agreed with one, and disagreed with one. Regarding our recommendation that DOE supplement its 2016 long- term strategic review with an additional analysis that takes into account private sector response, oil market projections, and costs and benefits of a wide range of different SPR sizes, the agency partially agreed with the recommendation. DOE agreed to conduct an additional analysis to assess the purpose, goals, and objectives of the SPR, taking into account private sector response, oil market projections, and any other relevant factors, that will lead to an evaluation of possible optimal sizes of the SPR in the future. In response to taking into account the costs and benefits of a wide range of different SPR sizes, DOE stated that the agency determined the projected benefits of a wide range of different SPR sizes ranging from 430 million barrels of oil to 695 million barrels of oil in its 2016 review. However, the minimum SPR size considered by DOE is greater than the projected SPR size after congressionally mandated sales have occurred. Further, the SPR size after congressionally mandated sales is projected to be far in excess of the IEA obligation to hold a minimum of 90 days of net imports. DOE must also consider the minimum size needed to meet its IEA obligations in the event of a collective action. In conducting additional analysis, DOE should consider a smaller lower bound, in line with congressionally mandated sales, for the size of the SPR, and more fully consider the size needed to meet the IEA 90-day net import and collective action obligations. Regarding our recommendation that DOE conduct periodic reviews of the SPR, the agency agreed with the recommendation. DOE stated that a 5-year time interval between reviews would strike an appropriate balance between the need to periodically conduct a strategic assessment and evaluation of the SPR and the limitations on resources to plan and conduct such a review. Regarding our recommendation that DOE conduct or complete studies on the costs and benefits of regional petroleum product reserves, the agency disagreed. DOE stated that it is the agency's position that government owned and operated regional petroleum product reserves are an inefficient and expensive solution to respond to regional fuel supply disruptions. DOE further stated, based on studies done in 2015 that DOE officials told us were pre-decisional and therefore could not be reported, that there are additional concerns associated with government-owned and operated regional refined petroleum product reserves, including little to no storage capacity for lease in commercial terminals and high costs for government owned and operated regional product reserves. However, these same studies took these concerns into account, and concluded that a product reserve in the Southeast would provide significant net economic benefits (benefits minus costs) to the region and the United States in the event of a major hurricane. These studies also concluded that additional analyses are required to inform decisions regarding the potential benefits of a similar reserve on the West Coast. Further, the Quadrennial Energy Review of 2015 recommended that similar analyses be completed for other areas deemed by DOE to be vulnerable to fuel supply disruptions. Therefore, we continue to believe that conducting these analyses, as recommended in the Quadrennial Energy Review of 2015, will provide Congress with information needed to make decisions about regional product reserves. Regarding our recommendation that DOE consider a full range of options for handling potentially excess assets, DOE agreed with the recommendation. DOE stated that in its ongoing study, the agency will include an assessment of disposition options for any potential excess or underutilized SPR assets, to include the need for new legislative authority, as necessary, for the disposition of assets. DOE expects this study to be completed in October 2018. DOE also provided technical comments, which we incorporated, as appropriate. As agreed with your offices, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies to the appropriate congressional committees, the Secretary of Energy, and other interested parties. In addition, the report will be available at no charge on the GAO website at http://www.gao.gov. If you or your staff members have any questions about this report, please contact me at (202) 512-3841 or ruscof@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix II. Appendix I: Comments from the Department of Energy Appendix II: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the individual named above, Quindi Franco (Assistant Director), Nkenge Gibson (Analyst-in-Charge), Philip Farah, Ellen Fried, Cindy Gilbert, Greg Marchand, Celia Mendive, Patricia Moye, Camille Pease, Oliver Richard, Dan Royer, Rachel Stoiko, and Marie Suding made key contributions to this report.

Why GAO Did This Study More than 4 decades ago, Congress authorized the creation of the SPR to reduce the impact of disruptions in supplies of petroleum products. DOE manages the SPR. As a member of the International Energy Agency, the United States is obligated to maintain reserves equivalent to at least 90 days of the previous year's net imports (imports minus exports). The SPR's storage and related infrastructure is aging, and DOE has plans to modernize these facilities. Since 2015, Congress has mandated crude oil sales. As of March 2018, the SPR held about 665 million barrels of crude oil. GAO was asked to examine the SPR's ability to meet U.S. energy security needs. This report examines, among other things, the extent to which (1) DOE has identified the optimal size of the SPR, and (2) DOE's plans for modernizing the SPR take into account the effects of congressionally mandated crude oil sales. GAO reviewed DOE's plans and studies, and interviewed agency officials and nine experts selected based on prior work, referrals, and a literature review. What GAO Found The Department of Energy (DOE) has not identified the optimal size of the Strategic Petroleum Reserve (SPR). In 2016, DOE completed a long-term strategic review of the SPR after its last comprehensive examination conducted in 2005. The 2016 review examined the benefits of several SPR sizes, but it did not identify an optimal size and its review was limited in several ways. In particular, DOE did not fully consider recent and expected future changes in market conditions, such as the implications of falling net imports, or the role that increased levels of private reserves (reserves held by private companies for their own purposes) may play in responding to supply disruptions. These changes have contributed to SPR and private reserves reaching historically high levels on a net imports basis (see figure). These changes are expected to continue to evolve—according to government projections, the United States will become a net exporter in the late 2020s before again becoming a net importer between 2040 and 2050. GAO has found that agencies should reexamine their programs if conditions change. Without addressing the limitations of its 2016 review and periodically performing reexaminations in the future, DOE cannot be assured that the SPR will be sized appropriately into the future. DOE has taken steps to take into account congressionally mandated sales of SPR crude oil in its $1.4 billion modernization plans for SPR's infrastructure and facilities. The SPR is projected to hold 405 million barrels of oil by the end of fiscal year 2027. However, DOE's current plans are based on information analyzed prior to recently mandated sales. According to DOE officials, the agency began a study in March 2018 to assess the effects of these sales on the SPR's modernization. However, this study is not examining all options for handling any excess SPR assets that may be created by currently mandated sales or any additional sales that may be mandated in the future, inconsistent with an agency order on real property asset management that calls for identifying excess assets. For example, DOE does not plan to examine the potential to lease unused SPR storage capacity to the private sector because DOE is not currently authorized to enter into such leases, according to agency officials. If authorized, leasing capacity could generate revenues that could help offset the costs of modernization. By not examining a full range of options, DOE risks missing beneficial ways to modernize the SPR while saving taxpayer resources. What GAO Recommends GAO is making four recommendations, including that DOE (1) supplement the 2016 review by conducting an additional analysis, (2) ensure that the agency periodically reexamines the size of the SPR, and (3) consider a full range of options for handling potentially excess assets as it conducts its study, among other things. DOE agreed with two, partially agreed with one, and disagreed with another recommendation on refined product reserve studies. GAO maintains that the recommendations are valid.

gao_GAO-18-53

gao_GAO-18-53_0

Background The HCV program, administered by HUD, subsidizes housing costs for low-income households in the private rental market. Because HUD provides HCV assistance directly to the household, participants are able to find their own housing, including single-family homes, townhouses, and apartments. If the household moves out of the unit, it can move with continued assistance to another private rental unit. PHAs administer the HCV program at the local level, while HUD administers funding and furnishes technical and professional assistance to PHAs in planning, developing, and managing the program. Approximately 2,200 PHAs across the country administer the voucher program on HUD’s behalf, managing day-to-day operations in the HCV program, including the application and voucher distribution processes, as well as housing inspection and approval. PHAs are responsible for ensuring that rents are reasonable, determining households’ eligibility, calculating and periodically redetermining households’ incomes and rental payments, and making subsidy payments to landlords. In addition, PHAs perform basic program functions, such as establishing and maintaining a waiting list, processing tenant moves, conducting landlord and tenant outreach, and reporting to HUD. Local PHAs determine the eligibility of households, approve applications, and distribute vouchers. In general, to be eligible to participate in the HCV program, households must have very low incomes—that is, incomes not exceeding 50 percent of the area median income. Moreover, at least 75 percent of new voucher program participants must have extremely low incomes, not exceeding 30 percent of the area median income. Once a household is approved by a PHA to participate in the program and finds a rental unit, that household pays 30 percent of its monthly income, after certain adjustments, toward rent. The remaining portion of the rent is paid through the HUD-subsidized voucher. PHAs can pay subsidies to cover between 90 percent and 110 percent of the fair market rent for their areas. The HCV application and rental process is displayed in figure 1. The following applies in the HCV application process: Households seeking to enter the HCV program may wait years for their local PHA to announce an open application period. PHAs may establish waiting lists if the number of applicants to the program exceeds available vouchers, and may close the waiting list if it contains more households than the PHA can assist in the near future. Therefore, prospective applicants in some locations can wait years for a local PHA to determine the eligibility of those already on the waiting list—and provide vouchers to eligible individuals—before reopening waiting lists to new applicants. During open application periods, applicant households may encounter processes and requirements that vary amongst PHAs. Applying for a voucher from a local PHA may take place in person or online, while PHAs may determine an applicant’s priority to receive a voucher by varying methods, such as a random lottery amongst all applicants, or on a first-come-first-serve ordering of when applicants applied. Moreover, PHAs can establish local preferences for selecting applicants from their waiting lists. For example, PHAs may give preference to a household that (1) is homeless or living in substandard housing, (2) is paying more than 50 percent of its income for rent, or (3) has an older-adult household member. Regardless of methods to determine eligibility, apply for, and obtain a voucher within the HCV program, it is free to participants. When an open application period ends and before determining household eligibility, PHAs may initially put applicant households on a waiting list. Because the demand for vouchers may exceed the supply available to the local PHA, households that have already waited to apply to the program may also wait years to receive a determination of eligibility and receive a voucher. After receiving a voucher, households must find eligible private- market rental housing within a limited time frame. A PHA will make contact with and issue a voucher to a household that is determined to be eligible and is subsequently selected from the waiting list. Households receiving vouchers use them to subsidize their rents in private apartments or houses available in the rental market. Households must find housing quickly—generally within 60 days— unless the PHA grants an extension. In some cases, PHAs direct voucher holders to websites dedicated to rentals in the HCV program, where private landlords list available units. When a voucher-holding household finds a unit that it wishes to occupy—and reaches an agreement with the landlord over the lease terms—the PHA inspects the dwelling and determines whether the rent requested is reasonable. To be eligible, a rental unit must meet HUD minimum housing-quality standards, and must provide an acceptable level of health and safety. After the unit is inspected and deemed eligible, the household signs a contract with HUD, and both HUD and the household sign contracts with the landlord. The contract stipulates that the PHA will make the housing-assistance payment to the landlord and the household will pay the difference between the housing-assistance payment and the rent. Landlord participation in the HCV program is free, and landlords do not pay to maintain compliance with the program. Moreover, the HCV program provides for the use of vouchers across locations. Once a household receives a voucher, it may use the voucher in any location in which a PHA administers the voucher program, as long as it remains eligible. Various Reported Fraud Schemes against Housing Choice Voucher Participants Can Occur, but PHAs Reported Limited Incidents, and GAO Covert Testing Found Limited Online Indicators of Fraud Waiting-List, Rental, and Side-Payment Fraud Are among the Reported Types of Schemes That Can Affect Voucher Participants Reported fraud schemes against program participants—including prospective applicants, individuals on waiting lists, current voucher holders, and landlords providing rental units—can occur at each point in the HCV application and rental process, according to program officials and our analysis of FTC complaint data. On the basis of our reviews of fraud alerts issued by PHAs and complaints submitted to the FTC, the type of fraud that participants, including older adults, may encounter depends on where they are in the process and whether they are landlords or renters. Fraudsters perpetrate reported schemes in a variety of ways, such as through in-person impersonation of PHA staff or by manipulating telephone numbers to convince landlords to make unnecessary payments. Reported types of fraud schemes and when they could potentially occur in the HCV application and rental process are displayed in figure 2. As shown in figure 2, various reported fraud types can be carried out against HCV participants. While some fraud types are specific to the HCV program, participants may also be victims of general rental-housing fraud. The reported fraud types, which we identified through interviews with PHAs and others and through an analysis of PHA fraud alerts and Consumer Sentinel complaints, include the following: Waiting-List Placement Fraud. In online or in-person settings, fraudsters may claim they can provide a voucher, place applicants onto a waiting list, or move individuals to a higher position on the waiting list. In exchange, fraudsters may request a payment, or may request information (such as name, credit-card number, and e-mail address) that may put participants at risk of credit-card fraud or identity theft. Waiting lists maintained by PHAs may open infrequently, and program application processes and requirements vary from location to location. Reported fraud schemes may take advantage of applicant unfamiliarity with program rules, and target those seeking to enter the program or awaiting a voucher. Rental-Advertisement Fraud. Because they rent in the private market, voucher holders are susceptible to online rental-fraud schemes. Those who place online rental advertisements may request wire transfers from prospective renters to secure fake rentals, or steer potential renters to suspect credit-reporting services that offer commissions to the scammers or realtor services that charge users a onetime or recurring monthly fee. Side-Payment Fraud. Officials from PHAs and other organizations characterized side payments as two distinct activities—alternatively, as landlord fraud against tenants on one hand, and mutually beneficial agreements between tenants and landlords on the other. In exchange for property rental or successful inspection, landlords or building inspectors may fraudulently request additional payments or pressure participants for other favors from voucher holders. Landlords may ask tenants for a monthly payment above the agreed rent, or may require HCV participants to pay for utilities when not required to in their rental agreement. For example, a Midwest PHA we interviewed reported being aware of coercive demands by landlords for side payments or sexual favors in exchange for a rental unit. Side payments may also be a mutual arrangement between landlord and voucher holders. For example, voucher holders may make a payment above their monthly rent—in violation of program rules—and in exchange the landlord agrees not to report that there are unauthorized occupants living in the unit, again in violation of program rules. A West Coast PHA we interviewed characterized most side payments it is aware of as mutual agreements of this type. Security-Deposit Fraud. Because they rent in the private market, voucher holders may encounter fraudsters advertising a rental and requiring a security deposit from one or several prospective renters even if there is no rental unit available or only one of the prospective renters will ultimately obtain the rental. Program-Compliance Fraud against Landlords. Fraudsters may take advantage of landlord unfamiliarity with HCV program rules. In calls to HCV program landlords, fraudsters mask their phone number with that of the local PHA, and direct the landlords to make a credit-card payment over the phone to purchase materials or to make a payment in order to remain in compliance with program rules. PHAs Are Aware of Limited Instances of Fraud Affecting Voucher Participants in Their Jurisdictions, with Side- Payment Fraud Reported Most Often An overwhelming majority of surveyed PHAs did not report awareness of any occurrences of most fraud types that could affect HCV program participants, while those that were aware of fraud against participants reported few instances, according to our survey results and interviews. We surveyed a nationally representative sample of PHAs representing approximately 1.9 million households. We inquired about incidents occurring within their area of jurisdiction from spring 2016 through spring 2017. We asked about fraudsters promising placement onto or a higher place on a waiting list, selling vouchers, stealing security deposits, or offering suspect credit-report services; voucher holders and landlords engaging in side payments; and landlords and building inspectors illegally soliciting favors. Apart from incidents of side payments (discussed in detail below), on the basis of PHA responses to our survey we estimate that between 3 and 10 percent of all PHAs with 1,000 or more vouchers were aware of any occurrences of the types of fraud schemes included in our survey (see fig. 3). Further, when PHAs were aware of such fraud schemes, most reported between 2 to 5 cases in their local area of jurisdiction from spring 2016 through spring 2017. Our other sources of evidence were consistent with our survey results. For example, we interviewed officials with two PHA associations—representing approximately 1,900 total PHAs—about fraud against HCV participants (other than side payments). The associations reported that they were unaware of widespread instances of these types of fraud against participants. PHAs were much more likely to report awareness of incidents of side payments than the other types of fraud included in our survey, according to our analysis of survey responses. As noted above, side-payment fraud involves agreements—mutual or compelled—in which the voucher holder pays additional rent or other payments to the landlord in return for benefits, to secure a rental, or to avoid eviction. On the basis of survey responses, we estimate that 41 percent of all PHAs with 1,000 or more vouchers were aware of incidents of side payments in the prior year, as shown in figure 3 above. Of PHAs reporting side payments, we estimate that the vast majority (93 percent) were aware of 1 to 10 instances of side payments in the prior year (spring 2016–spring 2017), with most reporting between 2 and 5 incidents in the past year. Further, we estimate that 7 percent were aware of 11 or more instances in the prior year. Officials from all eight PHAs we interviewed similarly told us that they were aware of side payments, but some said that participants rarely report cases of side payments to them. Because violation of HVC rules could result in termination of a lease or loss of voucher for the recipient, it is possible that side payments are not always reported to PHAs. Two experts providing legal services to low-income individuals said that, in regard to fraud affecting HCV participants, landlord requests for side payments is relatively more common than other types of fraud. In addition, a current voucher holder told us about personal experiences involving requests for such payments by landlords, but also said that he or she had not experienced any other types of fraud. In response to our open-ended survey question on other fraud not specifically mentioned in the survey, three PHAs in two regions reported variations of a type of fraud that intends to convince landlords that they are not in compliance with HCV rules, and that they must make a payment over the phone. Although this type of fraud was not included explicitly in our survey and is therefore not included in figure 3 above, in survey comments one West Coast PHA and two PHAs in the Southeast reported instances of this type of fraud. In open-ended survey comments, a West Coast PHA reported being aware of two attempts of similar fraud, in which fraudsters called landlords and asked them to make a credit-card payment over the phone to maintain program compliance. Similarly, two southeastern PHAs also reported being aware of instances of similar fraud schemes in the last year, although neither provided the number of cases reported to them. Furthermore, another West Coast PHA that was not included in our survey issued an online alert about scammers calling landlords, masking their actual phone number with the PHA’s phone number, and stating that in order for the landlord to maintain program compliance, the landlord must make a credit-card payment over the phone to purchase a program manual. In an interview, officials from this PHA reported being aware of 36 attempts of this fraud type against landlords from September 2015 to April 2017. According to PHA officials, fraud schemes generally have not targeted older-adult HCV participants. In survey responses, a limited number of PHAs reported fraud against older adults. For example, we estimate that of PHAs reporting awareness of side-payment schemes, very few PHAs (about 8 percent) were aware of instances in which landlords targeted older adults in side-payment requests. Moreover, an official from one PHA we interviewed stated that because older adults are likely to have advocates helping them to find housing, they are less likely than other HCV participants to be victims of fraud. Similarly, one expert providing legal services to older adults in the HCV program indicated that project- based participants—who rent units only in specific buildings—are more likely than HCV participants to be targets of in-person fraud because they are located in identifiable properties. Several PHAs issued fraud alerts about schemes against participants, but officials at PHAs we interviewed about some of the alerts told us that incidents of these fraud types were limited. Specifically: A Midwestern PHA issued an alert about an individual promising a voucher for a fee and meeting victims in person to receive payment. In a follow-up interview, PHA officials stated that, in total, they received five to seven reports about this fraud. Each of these cases occurred while the PHA’s waiting list was open in 2015. A nearby PHA also reported that when its waiting list was open in 2011, an individual with fake PHA credentials fraudulently took payment from individuals and promised to move them to the top of the waiting list. The PHA was aware of 10 individuals who were victims of the fraud, and estimated that they each paid about $200 to the perpetrator. A West Coast PHA issued an alert about a website charging applicants to submit a program application. However, in an interview the PHA reported awareness of only one case over the last 3 years of fraud committed by outside parties against voucher participants. Although instances of fraud against HCV participants reported to PHAs appear relatively rare, participants who provide personal information to unknown individuals are still at risk for identity theft, according to some experts we interviewed. Two identity-theft experts stated that credit-card fraud is likely if individuals enter payment information in unverified sites. Furthermore, one expert stated that low-income individuals can be targets for identity theft because fraudsters can use stolen identities with low credit scores to obtain high-interest loans that they do not intend to pay off. Few Indicators of Potential Fraud Found in Online Covert Testing In covert testing using undercover tools and techniques, we found no indicators of fraud in rental advertisements posted online, and few indicators of fraud in commercial websites offering information to participants about the HCV program. Through our online covert testing, we found no indicators of potential fraud in 350 advertisements—selected using a random-selection methodology—posted in online marketplaces across six cities. On the basis of an academic study of online fraud schemes on rental marketplaces and information provided by PHAs, we developed a list of indicators of potential fraud. These indicators include requests for a wire transfer of security deposit or first month rent without offering to provide an in-person viewing of the property, or requesting an up-front or monthly side-payment agreement as a condition of rental. In searches of one rental website, we came across a small number of advertisements that initially appeared to contain an indicator of potential fraud—specifically, links to a website for specialized realtor services; covert testing of that link did not find further indicators of potential fraud. None of the 26 commercial websites we covertly tested contained text explicitly stating that they would place or move someone up a waiting list for a fee. Further, in e-mail correspondence with every website, we asked whether they could help place us on a waiting list. Some website operators never replied and some stated they could not do so, with none agreeing to place us on a list. However, some websites used HUD’s Equal Housing Opportunity logo, which might make them appear to be associated with official government programs, while others requested payment for suspect services and products. For example: One PHA fraud alert specifically named one of the websites we tested covertly, and indicated that the website fraudulently offered to submit an application for the HCV program for a onetime registration fee. Our covert testing found that this website displayed HUD’s Equal Housing Opportunity logo, and charges a fee for a “guide” about the voucher program, but at the time of our testing the website did not offer to submit an application for a fee. Payment to the website resulted in access to a guide and online forum containing a list of open PHA waiting lists for the HCV program and links to publicly available PHA websites. One website we covertly tested requested payment for an e-book guide to assist with the HCV application process. We made payment but never received the e-book. One website subject to covert testing stated that landlords in the HCV program may deny or refuse to rent to a potential tenant based on his or her credit-report information, and referred us to a suspect website offering credit reports. The website claimed to offer a “free credit report” and requested personal information including a Social Security number and credit-card number. As part of our covert testing, we provided a credit-card number. We also entered all zeroes as a Social Security number on the site, prompting the site to state that it could not provide us with a credit report, as we had not provided a valid Social Security number. Despite that fact, the site charged our credit card a recurring payment. In a phone call, a representative of the website stated that, in the terms-of-use for the website, users are informed that they must explicitly request that recurring payments be terminated or that those payments would continue. HUD and PHA Antifraud Efforts Focus Mainly on Fraud against the HCV Program; Some PHAs Voluntarily Provide Information about Fraud against Participants HUD regulations and guidance and PHA informational materials pertinent to fraud primarily focus on protecting the HCV program rather than protecting participants from fraud committed by external parties. For example, where the regulations mention fraud explicitly, it is generally in relation to mitigating program violations by owners and voucher program participants, recovering program losses from fraud, and assessing participants, applicants, and owners for participation or continued participation in the program. Apart from requiring that PHAs inform participants about a prohibition against side payments to landlords (a program rule violation), HUD’s antifraud guidance, as outlined in the Housing Choice Voucher Program Guidebook, generally focuses on preventing fraud against the program, as opposed to fraud against participants. For example, an applicant misrepresenting income and assets to obtain an HCV voucher and a landlord bribing a PHA employee to approve substandard rental housing are types of program fraud. See figure 4 for examples of HCV program- related fraud listed in the guidebook. Education and outreach requirements for PHAs specified in the HCV guidebook largely focus on providing adequate public notice of waiting-list openings; an oral briefing when the PHA selects a family to participate in the program; and a written briefing packet for participants, which must include a variety of subjects related to program administration, leasing a unit, and family obligations. Consistent with the guidebook, written or online briefing materials from the eight PHAs we interviewed mention various types of program violations. All but one specifically state that side- payment agreements between landlords and tenants are prohibited, which, as discussed above, can be viewed as both fraud affecting the participant and against the program. HUD directs PHAs to inform participants that landlord–participant side payments are prohibited. HUD provides guidance on how a PHA should handle a situation in which the landlord is collecting side payments. If the PHA finds that the landlord is collecting side payments, the PHA must notify the landlord to immediately cease collecting these payments and require repayment to the tenant of the full amount collected. The PHA must determine whether the landlord also collected side payments from other participants and follow up to require repayment. The amount can be repaid by offsetting the amount due against future housing-assistance payments. At its discretion, the PHA may terminate the housing-assistance payments contract with the landlord immediately, even if the landlord has repaid amounts due the tenant, but the PHA must cancel the contract if the landlord fails to repay. Although not required, several PHAs we visited or contacted voluntarily provided informational materials to program participants that included targeted messages and alerts notifying them of certain housing- assistance fraud by outside parties, such as voucher-sale fraud, or fraud involving being placed on or moved up a waiting list. For example, in program briefing documents given to participants, a northeastern PHA warns participants of housing-assistance scams, and specifically advises participants not to pay to, among other things, (1) be placed on or be moved up a waiting list or (2) receive an HCV voucher or voucher extension; a different northeastern PHA advises HCV participants not to give their voucher to anyone, including the apartment owner, agent, or property manager; and not to give any money to the apartment broker, owner, or agent until the PHA approves the selected apartment; and a West Coast PHA advises HCV applicants and participants to be aware of fraud, particularly schemes that require a payment to file an application or to move up a waiting list. In addition, PHAs share best practices that could include these and other issues. For example, two industry associations representing approximately 1,900 PHAs provide mechanisms for PHAs to share information and best practices about HCV administration and issues affecting HCV program participants and their communities. Both regularly hold conferences, meetings, and other events that provide a venue for members to discuss relevant issues. In addition, one of these associations has published reports on issues affecting older adults and connecting housing and community services, among other issues, while the other published a report on issues related to rental reform proposals. Agency Comments We provided a draft of this report for review and comment to HUD, FTC, the Consumer Financial Protection Bureau, the Department of Health and Human Services, and the U.S. Postal Service. We received e-mails from HUD, the Consumer Financial Protection Bureau, and the Department of Health and Human Services in which liaisons to GAO for those agencies stated they had no comments on the report. We received technical comments from FTC and the U.S. Postal Service, which we incorporated in the report as appropriate. We are sending copies of this report to the appropriate congressional committees, the Secretary of Housing and Urban Development, the Federal Trade Commission, the Director of the Consumer Financial Protection Bureau, the Secretary of Health and Human Services, and the Postmaster General of the U.S. Postal Service. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff members have any questions about this report, please contact me at (202) 512-6722 or shear@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made major contributions to this report are listed in appendix III. Appendix I: Objectives, Scope, and Methodology In this report, we describe (1) the types of reported fraud schemes committed against Housing Choice Voucher (HCV) participants (including older adults), awareness by Public Housing Agencies (PHA) and other relevant organizations of fraud incidents and how often they occur, and indicators of such schemes online; and (2) antifraud regulations, guidance, and informational materials, if any, that the Department of Housing and Urban Development (HUD) and PHAs have in place to identify and mitigate fraud against program participants. To address our first objective we used a variety of methods (see fig. 5). Details on our use of these methodologies are described below. First Objective Types of Reported Fraud Fraud Alerts Some PHAs issue online alerts and post these on their websites to inform HCV applicants and participants about potential fraud schemes. We initially identified PHA fraud alerts by performing online searches about fraud in the HCV program. We then developed a structured search method for identifying additional fraud alerts. To do this, in 2016 and 2017, we searched fraud alerts issued by a nongeneralizable sample of 60 PHAs; in total, 20 PHAs published 22 alerts about fraud affecting HCV applicants and participants on their websites. We identified our nongeneralizable sample of 60 PHAs using 2015 fourth- quarter Picture of Subsidized Households data from HUD, which contain information on subsidized housing units by several types of programs including the HCV program. The initial population contained PHAs ordered by total number of HCV program vouchers available. We selected our sample of PHAs based on those with the most vouchers. Consumer Complaint Data We reviewed consumer complaint data from the Federal Trade Commission’s (FTC) Consumer Sentinel database from calendar years 2011 through 2016. The date range of the data represents the most- recent years available at the time of our request. Our review of the data focused on complaints related to the HCV program and companies that offer rental housing services. We assessed the reliability of the data by interviewing officials and reviewing related documentation and found the data sufficiently reliable for the purposes of our reporting objectives. We developed several categories for reviewing complaints based on criteria on fraud schemes affecting HCV program applicants and participants. To develop an initial list of categories, we selected a subsample of the first entries in the data and independently created categories that could be used to categorize the complaints in the subsample. On the basis of this methodology, we identified a set of defined coding categories, which were as follows: 1. HCV-specific fraud, 2. housing-related fraud (HCV not mentioned), 3. housing-related fraud involving the purchase of foreclosed property, 4. housing-related credit-report fraud, 5. housing-related fraud requesting electronic wire transfer of funds, 6. HCV-specific complaints where fraud is not mentioned or the nature of 7. housing-related complaints where fraud is not mentioned or the nature of fraud is unclear, and 8. complaints not related to the scope of the engagement. We also separately coded whether the subject matter of each complaint specifically affected an older adult. We applied a two-person data-coding process to ensure intercategorization reliability. FTC delivered the data to us in batches organized along search terms we provided. For several of the initial batches we received, as a first step in the coding process, a coder categorized each complaint into one of the categories above, and simultaneously identified any complaints that contained relevant housing- related or HCV-related fraud types that we had not already discovered. As a second step, a reviewer assessed a nonrandomized sample of the data to determine whether coding was correct, and whether the coder had identified any previously unknown fraud types in the batch. We repeated this process for three of the five batches that we received, and reviewed over 600 total complaints. Upon finding no new fraud types in the coded data, we ceased analysis and did not code the remaining two batches we received, which we deemed to contain complaint categories unlikely to reveal new types of fraud. We also interviewed PHA officials from eight PHAs (selected using a methodology discussed below). Additionally, we reviewed an academic study describing fraud against prospective renters in online marketplaces, which allowed us to identify several fraud types that could be used against HCV participants searching for rental units. The study used crawling and automated interaction to identify fraud types. We interviewed an author of the study to clarify research techniques. We assessed that the individual was sufficiently independent. Our methodological specialist assessed the study, and found its conclusions to be sufficiently valid and reliable for our purposes. Awareness of Fraud Incidents Survey We conducted a web-based survey with a nationally representative stratified random sample of executive directors overseeing PHAs. In the survey, we asked PHA executive directors and their staffs to provide information on known fraudulent activities by fraudsters or impersonators, fraudulent activities by landlords and building inspectors, and any other information on fraudulent activities adversely affecting HCV program applicants and participants from spring 2016 through spring 2017. We administered our survey from April to May 2017. Estimated percentages of the responses for all closed-ended questions from the survey are included in appendix II. We identified the population of PHA executive directors using 2015 fourth-quarter Picture of Subsidized Households data from HUD, which contain information on subsidized housing units by several types of programs including the HCV program. The Picture of Subsidized Households data also contain the percentage of households using these programs by factors such as age, income, and disability. We assessed the reliability of the data for use as our sampling frame by reviewing technical documentation, conducting electronic testing, and interviewing officials who oversee the data system; we found the data sufficiently reliable for our purposes. Our initial population list contained a total of 2,243 PHA executive directors, and our sample contained 278 PHA executive directors. We stratified the population by size of PHAs as follows: We drew (1) a certainty sample of 83 executive directors who oversaw at least one PHA with 5,000+ vouchers (“large” PHAs) and (2) a probability sample of 195 executive directors who oversaw at least one PHA with 1,000–4,999 vouchers (“medium” PHAs). For purposes of discussion, we refer to the experiences of PHAs in our analysis, although our sampling unit was the executive directors of the PHAs. To formulate our survey questionnaire on the types of fraud potentially adversely affecting HCV program applicants and participants, we conducted research on the topic of fraud by interviewing PHA officials, reviewing fraud alerts publicly posted on the Internet by large PHAs, and reviewing consumer complaint data. On the basis of the results of our research, we developed our survey questionnaire to include questions on external fraud such as (1) fraudsters or impersonators promising placement on a voucher waiting list, (2) fraud offering higher placement on voucher waiting lists, (3) fraud offering fake vouchers, (4) fraud offering suspect credit-report services, (5) landlords requiring prohibited side payments, (6) illegal solicitation of favors by landlords and building inspectors, and (7) illegal solicitation of rental-unit security deposits by landlords. We pretested our survey instrument with four PHAs located in Maryland, Michigan, Ohio, and Virginia. We revised our questionnaire language and format based on input received by officials in these four PHAs in order to improve the clarity of the questions. An independent survey specialist within GAO also reviewed a draft of the questionnaire prior to its administration; it is available in appendix II. We administered a web-based questionnaire accessible through a secure server. When we completed the final survey questions and format, we sent an e-mail announcement of the survey to 278 PHAs in April 2017. The PHA points of contact were notified that the questionnaire was available online and were given unique passwords and usernames. We sent follow-up e-mail messages twice in May 2017 to those who had not yet responded. We contacted remaining nonrespondents by telephone, beginning in May 2017. The questionnaire was available online until mid- May 2017. We obtained a weighted overall response rate of 84 percent, and the response rate by stratum was 86 percent for our first stratum (“large” PHAs) and 83 percent for our second stratum (“medium” PHAs). Because we followed a probability procedure on random selections, our sample is only one of a large number of samples that we might have drawn. Since each sample could have provided different estimates, we express our confidence in the precision of our particular sample’s results as a 95 percent confidence interval (e.g., plus or minus 10 percentage points). This interval would contain the actual population value for 95 percent of the samples we could have drawn. Confidence intervals are provided along with each sample estimate in the report. All survey estimates presented in this report are generalizable to the population of large and medium PHAs, or to either the population of large PHAs or medium PHAs analyzed separately. Unless otherwise noted, estimates for the full population of large and medium PHAs have a margin of error for a 95 percent confidence interval within +/-4.5 percentage points or less. Unless otherwise noted, estimates for the medium PHAs analyzed separately have a maximum margin of error for a 95 percent confidence interval of +/-5.4 percentage points or less. Unless otherwise noted, estimates for the large PHAs analyzed separately have a maximum margin of error for a 95 percent confidence interval of +/-4.5 percentage points or less. Some questions had too few respondents to generate reliable estimates. In these cases, we report the raw frequencies of respondents to our survey. To minimize nonsampling errors, and to enhance data quality, we employed recognized survey design practices in the development of our survey questionnaire and in the collection, processing, and analysis of the survey data. To minimize errors arising from differences in how survey questions might be interpreted and to reduce variability in responses that should be qualitatively the same, we conducted pretesting of our survey questionnaire; see discussion on pretesting above. To reduce nonresponse, a source of nonsampling error, as mentioned above we followed up by e-mail and by telephone with PHAs who had not responded to the survey to encourage them to complete it. To analyze open-ended comments provided by those responding to the survey, we conducted a content analysis for the purpose of identifying fraudulent activities against HCV program participants not addressed in our survey questionnaire. We analyzed open-ended responses to identify fraud types not directly addressed in our survey. We identified two additional types of fraud. One type of fraud involved fraudsters posing as PHA officials, calling landlords to convince them to make unnecessary payments. This type of fraud against landlords is discussed in the report. The other type of fraud identified was not related to participation in the HCV program, and so is not discussed in the report. Interviews of Government Officials, Nongovernment Officials, and Others As part of our site visits, we interviewed officials from eight PHAs located in three U.S. regions. On the East Coast, we interviewed officials with the New York City Department of Housing Preservation and Development, the New York City Housing Authority, and New York State Homes and Community Renewal. In the Midwest, we interviewed officials with the Cuyahoga Metropolitan Housing Authority, the Detroit Housing Commission, and the Flint Housing Commission. On the West Coast, we interviewed officials with the Housing Authority of the County of San Bernardino and the Housing Authority of the City of Los Angeles. We identified our interviewee selection on the basis of ensuring geographical representation, budgetary considerations, metropolitan cities with a large population, PHAs’ issuance of fraud alerts, PHA size—large, medium, and small, PHA in states with a large number of older adults, and consideration for overlap of other GAO ongoing work in the area of the HCV program. Our sample of PHA interviewees is nongeneralizable. Moreover, we interviewed an HCV voucher holder about the voucher holder’s knowledge of fraud against participants. We also interviewed government and nongovernment officials and others based on their knowledge and expertise on the topic of fraud in general; fraud education campaigns; fraud adversely affecting HCV program applicants and participants; fraud affecting older adults; or identity theft. Specifically in reference to government agencies, we interviewed officials from the U.S. Federal Trade Commission (FTC), the Consumer Financial Protection Bureau, the U.S. Postal Inspection Service, and the Department of Health and Human Services about fraud types and about practices used by federal agencies to inform the public about fraud- related issues. We also interviewed officials from the HUD Office of Inspector General (OIG) about any past or ongoing work related to the scope of our reporting objectives. For our interviewee selection, we considered recommendations from other organizations such as PHAs and legal-assistance organizations and reviewed prior GAO work on the issue of older-adult financial exploitation. We also considered organizations’ characteristics in terms of fraud prevention or work performed in assisting potential fraud victims. These characteristics include whether the organization has an investigative unit that may have data on fraud schemes, posts fraud alerts on its Internet websites, has data on fraud cases, collaborates with other groups on fraud awareness, has a fraud or complaint hotline, works on fraud prevention and provides support to victims of fraud, or works with vulnerable populations including low-income individuals or the older-adult population on social or legal issues. Indications of Potential Fraud Online Covert Testing On the basis of an academic study about fraud in online rental marketplaces (discussed above) and PHA-provided information, we developed a list of indicators of potential fraud that might appear in online advertisements. We used covert tools and techniques to test a nongeneralizable sample of advertisements posted on commercial websites commonly used by HCV voucher holders and landlords for rental property listings, counting advertisements as potentially fraudulent if the originator of the advertisement did any of the following: Requested a wire transfer of security deposit or first month’s rent, or both, without offering to provide an in-person viewing of the property. For example, the person who posted the advertisement might state that the rental is available, but that the person is currently not in the country. Provided a link to a suspect credit-report site within the advertisement or subsequent correspondence. While requiring a credit report is a normal part of the rental process, fraudsters may post a fake rental advertisement and redirect victims to a credit-score company. If the victim pays for the credit score, the credit-score company will pay the fraudster a commission. Stated in correspondence that the rental unit is no longer available, but recommended a suspect site providing rental search, broker services, or monthly payments toward purchase of a foreclosed property. Requested an up-front or monthly side-payment agreement as a condition of rental. Included an e-mail address directly in the text of the ad. According to one academic expert on fraud, it was rare to see a legitimate advertisement poster embed an e-mail address in a post, because most people do not want to expose that information on the Internet. Advertisements or correspondence containing an indicator of fraud do not necessarily reveal the presence of a fraud scheme. For example, the presence of an e-mail address directly in the text of an advertisement may also indicate that the advertisement was posted by a realtor service. We selected 6 geographically diverse cities nationwide for covert testing of online rental advertisements. To generate a list of cities for possible selection, we identified cities containing the 20 PHAs with the largest number of HCV vouchers. We identified one commonly used online housing-rental marketplace for the general public and another online rental marketplace specifically dedicated to HCV rentals. We then determined the total number of advertisements available across the two rental websites in those cities. We used the following criteria to select cities for covert testing: From the 20 cities described above, we selected 3 cities with the most available advertisements across the two online rental marketplaces. From the 20 cities described above, we selected 2 cities where a nearby PHA had closed its HCV waiting lists in the last half of 2016. Outside of the 20 cities described above, we also selected 1 city with a large number of available advertisements where we had previously completed a site visit. In each metropolitan area selected for covert testing, we identified relevant online rental marketplaces operating in the area. We then used a random-selection methodology to identify advertisements for testing from amongst all current advertisements on each marketplace. In total, we responded to 350 advertisements. On sites specifically dedicated to HCV rentals, we generally sampled from among all current ads. On sites featuring a mix of private rental ads, we developed a list of search keywords and sampled only among ads that explicitly stated that they would accept an HCV voucher. By e-mail, we contacted the originator of each of the 350 advertisements we covertly tested and engaged in correspondence. To determine the extent of detected potential fraud in online sites, we covertly tested websites identified as offering information or assistance with the HCV program; note that this methodology is distinct from that described above to covertly test advertisements. To discover sites for investigation, we performed web searches with a variety of relevant search terms using two popular search engines. We clicked through the first few pages of each set of search results, and collected the names of commercial sites that appeared either within search results or in ads accompanying the results. Finally, we entered the address of each of the websites we found into a separate search portal. This search portal suggests possible competitor and similar websites for the target website; we added websites discovered through this method to our list of websites for investigation. In total, we tested 26 websites. If the website requested a payment of any kind, we made the payment. We also corresponded with each website, asking explicitly whether it could place us on an HCV waiting list. Second Objective To address our objective regarding antifraud regulations, guidance, and informational material, if any, that HUD and PHAs have in place to identify and mitigate potential fraud against program participants, we reviewed HUD regulations on the HCV Program and the Section 8 Management Assessment Program. These regulations outline the requirements for PHAs to perform education and outreach on the HCV program. We also reviewed HUD’s HCV Program Guidebook. This guidebook provides direction to PHAs administering the HCV Program on informing applicants about program-related fraud. We also reviewed written or online briefing materials for participants developed by the eight PHAs we interviewed. These briefing materials must include a variety of subjects, related to program administration, leasing a unit, and family obligations. We conducted this performance audit from May 2016 to December 2017 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. We conducted our related investigative work from January 2017 to July 2017 in accordance with investigative standards prescribed by the Council of the Inspectors General on Integrity and Efficiency. Appendix II: Results of GAO’s Survey of Public Housing Agencies To determine awareness of incidents of fraud schemes among PHAs, we conducted a web-based survey to a nationally representative sample of executive directors overseeing Public Housing Agencies (PHA) from April 2017 to May 2017. We solicited input on executive directors’ familiarity with and awareness of fraud schemes against U.S. Housing and Urban Development (HUD) Housing Choice Voucher (HCV) Program applicants and participants. We distributed the survey to 278 PHAs, of which 233 (84 percent) responded. We stratified the population by size of PHA as follows: (1) a certainty sample of 83 executive directors who oversaw at least one PHA with 5,000+ vouchers (“large” PHAs); and (2) a probability sample of 195 executive directors who oversaw at least one PHA with 1,000–4,999 vouchers (“medium” PHAs). Results of our survey are generalizable to the entire population of large and medium PHAs. For a more-detailed discussion of our survey methodology, see appendix I. The results of our survey provide the input of PHA executive directors and their staffs at the time they completed the survey in April and May 2017. The questions we asked in our survey are presented below. Our survey comprises seven top-level, fixed-choice questions; three subquestions for each “yes” response to top-level questions; and one open-ended question. In this appendix, we include all survey questions, and the estimated percentages for the responses to the top-level questions. Because of the limited number of respondents answering “yes” to the top- level survey questions, we could not generate reliable estimates for the survey subquestions; therefore we present only raw frequency counts for all subquestions except those corresponding to survey question 5 where we present both estimated percentages and raw frequency counts for those responses. In our survey open-ended question, we asked PHA executive directors to provide information on fraudulent activities affecting HCV program participants other than those covered by the seven fixed-choice questions. This element was our attempt at identifying fraud types not directly addressed in our survey. While we are not providing the responses to the open-ended question, our analysis of those responses identified two additional types of fraud. One type of fraud involved fraudsters posing as PHA officials, calling landlords to convince them to make unnecessary payments. The other type of fraud identified was not related to participation in the HCV program, and so is not discussed in the report. Fraudulent Activities by Fraudsters or Impersonators Fraudulent Activities by Landlords and Building Inspectors Survey Question 5 Responses Tables 17–23 below present results for PHAs that responded “yes” to question 5. Because of the number of respondents answering “yes” to question 5, we were able to generate reliable estimates for question 5 subquestions and present both estimates and raw frequency counts. Appendix III: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Kathy Larin (Director), Tonita Gillich (Assistant Director), Scott Hiromoto (Analyst-in-Charge), Maurice Belding, Yue Pui Chin, Colin Fallon, Dennis Fauber, Maksim Glikman, Ronald La Due Lake, Jill Lacey, Won Lee, Robert Letzler, Barbara Lewis, Olivia Lopez, Maria McMullen, Anna Maria Ortiz, Sabrina Streagle, Adam Windram, and Helina Wong made key contributions to this report. Also contributing were Marcus Corbin, Cory Marzullo, Wayne McElrath, Josephine Perez, Samuel Portnow, and Paul Schmidt.

Why GAO Did This Study With the goal of providing safe, decent, affordable housing, HUD provides rental assistance to low-income households through its HCV program, administered locally by approximately 2,200 PHAs around the country. In fiscal year 2016, the HCV program received approximately $20 billion in funding and provided rental assistance to approximately 2.4 million households. Local demand in the program may exceed voucher supply, and individuals may wait years before receiving a voucher. After receiving a voucher, participants have a limited amount of time to secure a rental. Accordingly, PHAs have issued alerts about criminals targeting program participants with fraud schemes, such as by claiming to offer admission to the program for a fee. This report describes (1) the types of reported fraud schemes against HCV participants, including older adults, PHAs' awareness of such schemes and their frequency, and indicators of such schemes online; and (2) HUD's and PHAs' antifraud regulations, guidance, and information related to fraud risks affecting program participants. GAO reviewed online fraud alerts and consumer complaint data from calendar years 2011 to 2016; conducted a generalizable survey of PHA officials about their awareness of fraud against participants; interviewed agency officials and experts; and conducted online covert tests of 350 rental ads and 26 commercial websites. GAO visited eight PHAs, selected based on size and location, among other factors. GAO is not making recommendations in this report. HUD had no comments on a draft of this report. What GAO Found Public Housing Agencies (PHA) have reported various types of fraud schemes against Housing Choice Voucher (HCV) participants, including older adults, but were aware of limited instances of such schemes. Similarly, GAO identified few potential indicators of these schemes in online covert testing of rental ads and websites. According to GAO's analysis of fraud alerts and complaint data, the type of fraud participants may encounter—such as waiting-list, rental, and side-payment fraud—depends on where they are in the HCV process and whether they are applicants, voucher holders, or landlords, as shown in the figure below. For example, side-payment fraud involves agreements—mutual or compelled—in which the voucher holder pays additional rent or other payments to the landlord for benefits, for example to secure a rental or avoid eviction. According to GAO's survey of PHAs representing approximately 1.9 million households, PHAs reported few incidents of the various fraud types, although side-payment fraud, a program violation, was noted most frequently. Specifically, GAO estimates that while 41 percent of PHAs were aware of instances of side-payment fraud in the prior year, most reported 2 to 5 incidents in the prior year. In addition, 3 to 10 percent of PHAs were aware of instances of the other types of fraud GAO identified. GAO's online covert testing also found few indicators of potential fraud. For example, some websites requested payment for information about the HCV application process, but none explicitly offered to do something prohibited by program rules, such as placing someone on a waiting list for a fee. The Department of Housing and Urban Development's (HUD) and PHAs' antifraud regulations, guidance, and information largely focus on efforts to protect the HCV program. For example, PHAs are required by HUD to inform families of program-related fraud and abuse, including the prohibition against side payments. In addition, GAO found that several PHAs voluntarily provide targeted messages to participants about fraud schemes by outside parties. Through industry associations, PHAs have mechanisms through which they share best practices that could include these and other issues.

gao_GAO-18-483

gao_GAO-18-483_0

Background Puerto Rico’s Territorial Status and Residents’ Status Puerto Rico is the most populous U.S. territory with approximately 3.3 million residents. Puerto Rico and its residents are generally subject to the same federal laws as the states and their residents, except in cases where specific exemptions have been made, such as with certain federal programs. Individuals born in Puerto Rico are U.S. citizens and can migrate freely to the states. Hurricane Maria On September 20, 2017, Hurricane Maria, a category 4 storm, devastated Puerto Rico and left nearly all its residents without potable running water and electricity. In addition, the existing infrastructure for cellular and wireless service was rendered virtually useless, hampering communication. Four months after Hurricane Maria, more than a third of Puerto Rico’s energy customers remained without power. The lack of power and communication impeded residents’ ability to return to work. According to the Federal Emergency Management Agency (FEMA), rebuilding will take years. Puerto Rico Oversight, Management, and Economic Stability Act (PROMESA) PROMESA established a Financial Oversight and Management Board for Puerto Rico (Oversight Board), and granted it broad powers of fiscal and budgetary control over Puerto Rico. The Oversight Board is comprised of seven members appointed by the President of the United States from a list of recommendations from House and Senate Leadership and one ex- officio member designated by the Governor of Puerto Rico. PROMESA also established a mechanism through which the Oversight Board could petition U.S. courts on Puerto Rico’s behalf to restructure debt. Government Assessment of Puerto Rico’s Economic Condition Under Puerto Rico law, the Puerto Rico Planning Board (Planning Board) has the legal responsibility of developing an economic outlook and a detailed analysis of the economy, including gross domestic product (GDP), and producing an annual Economic Report to the governor and to the legislature. The Planning Board Chairperson releases GDP measures only after approval from the governor’s office, according to Planning Board officials. The Department of Commerce’s Bureau of Economic Analysis (BEA) produces economic accounts statistics that enable government and business decision-makers, researchers, and the American public to follow and understand the performance of the nation’s economy. To do this, BEA collects source data, conducts research and analysis, develops and implements estimation methodologies, and disseminates statistics to the public. BEA calculates GDP for the United States, including for the territories of American Samoa, Guam, the U.S. Virgin Islands and the Commonwealth of the Northern Mariana Islands. Since 2009 the Department of the Interior’s Office of Insular Affairs has reimbursed BEA for estimating and publishing GDP for these territories. This office carries out the administrative responsibilities of the Secretary of the Interior and the Assistant Secretary for Insular Areas by coordinating federal policy for these territories, but does not for Puerto Rico. Census in cooperation with the Department of Labor’s Bureau of Labor Statistics (BLS) produces the Current Population Survey (CPS), which provides statistics on work, earnings, and education. CPS is one of the oldest, largest, and most well-recognized surveys in the United States, according to Census. In addition to being the primary source of monthly labor force statistics, the CPS is used to collect data for a variety of other studies that provide information on economic and social well-being factors. The CPS does not collect or report data for Puerto Rico or any of the other U.S. territories. Census also produces the American Community Survey (ACS). It is an ongoing survey that provides national information on a yearly basis that includes information for the States, as well as for Puerto Rico. The ACS includes data on jobs and occupations, educational attainment, veterans, whether people own or rent their homes, and other topics. Information from the survey generates data that help determine how more than $675 billion in federal and state funds are distributed each year. Federal Labor Laws DOL’s Wage and Hour Division (WHD) administers the wage, hour, and child labor provisions of the Fair Labor Standards Act of 1938 (as amended) that sets the minimum wage and overtime pay standards applicable to most U.S. workers. The Fair Labor Standards Act (FLSA) requires employers to compensate employees who are covered by the act and not specifically exempt from its provisions, at least federal minimum wage (currently $7.25 per hour) and with premium pay (at one- and-one-half the regular rate) for overtime hours worked in excess of 40 hours in a workweek. There are a number of exemptions from the requirements of the FLSA. For example, employees working in a “bona fide executive, administrative, or professional capacity” (EAP) are not entitled to premium pay for overtime. The FLSA was enacted to address problems associated with substandard working conditions by, in part, establishing a floor on wages and a ceiling on hours, beyond which the employer is required to pay extra wages. With a requirement for overtime pay, employers would either have to hire more workers or assume extra wage costs in order to achieve the same amount of work. Employees would be assured additional pay to compensate them for the burden of a workweek in excess of 40 hours. The Minimum Wage Study Commission of 1981 justified the EAP exemption in part because these employees are associated with higher base pay, higher promotion potential, and greater job security than most of the U.S. labor force. For employers and employees, the practical effects of the exempt employee classification can be important. An exempt employee may be required to work as many hours as it takes to complete a task. Although this may be more than 40 hours per week, the employee will not be entitled to overtime pay. Thus, an exempt financial manager may be required to work 60 hours a week and be paid a set weekly salary. On the other hand, a nonexempt bookkeeper may be required to work 60 hours per week, but must be paid for 20 hours of overtime, in addition to a set weekly salary. The FLSA authorizes DOL to define EAP exemptions. Balancing the competing interests of expanding exemptions and restricting them, DOL regulations establish specific tests that must be met before an employee may be classified as an EAP and exempt from overtime. In general, there are three tests: Salary Basis Test. The employee must be paid on a salary basis, rather than an hourly basis. This means that the employee must be paid at least the guaranteed amount, regardless of the number of hours actually worked and the quality or quantity of worked performed. Salary Level Test. The employee must meet a minimum salary level that indicates managerial or professional status. Duties Test. The employee must have duties and responsibilities associated with an exempt EAP position. In 2003, DOL reviewed the regulations for EAP exemptions in response to a GAO recommendation. Based on its review, in 2004 DOL increased the minimum “salary level” threshold for an employee to be exempt from receiving overtime pay to $23,660. In May 2016, DOL again updated minimum the salary level threshold for EAP employees to be exempt from receiving overtime pay to $47,476 in the 2016 Overtime Rule (see fig. 1). Status of the 2016 Overtime Rule In July 2015, DOL proposed updating the overtime regulations relating to the EAP exemption, and published a notice of proposed rulemaking. After receiving approximately 294,000 comments, the Secretary of Labor published the final rule on May 23, 2016 (2016 Overtime Rule). The major changes included increasing the salary level threshold from $455 per week ($23,660 annually) to $913 per week ($47,476 annually) and providing an automatic update to the salary level every 3 years. DOL estimated that about 4.2 million EAP employees in the states would become newly entitled to overtime pay under the revised salary level threshold. At the time of publication, the 2016 Overtime Rule would have applied to Puerto Rico; however, on June 30, 2016, prior to the rule’s effective date of December 1, 2016, PROMESA was enacted which, in part, delayed the applicability of this rule to Puerto Rico. Prior to the 2016 Overtime Rule going into effect, several states and various business groups challenged the rule in the Federal District Court of the Eastern District of Texas. On November 22, 2016, this court issued a nationwide preliminary injunction preventing DOL from implementing and enforcing the 2016 Overtime Rule for the duration of the case. In the interim, the 2004 Overtime Rule salary level threshold for EAP employees of $23,660 remained in effect. In July 2017, DOL published a Request for Information to gather additional information to begin the rulemaking process to replace and update the overtime regulations. In August 2017, the district court determined that the 2016 Overtime Rule was unlawful and ordered it invalidated. In October 2017, DOL filed a motion to appeal that ruling with the Fifth Circuit Court of Appeals. In November 2017, DOL filed a motion to stay the appeal pending the outcome of its rulemaking, and the Fifth Circuit granted this motion. DOL’s comment period for the Request for Information ended on September 25, 2017, and the agency currently is reviewing submissions. DOL plans to publish a Notice of Proposed Rulemaking on the salary level threshold for EAP employees in October 2018. Meanwhile, the 2004 Overtime Rule continues to remain in effect as of today, while the appeal and rulemaking are pending. Puerto Rico’s Labor Laws In addition to FLSA, workers and employers in Puerto Rico may be subject to various other federal, Puerto Rican, and local labor laws or regulations depending on eligibility, exemptions, and other limitations. In some cases, including sick leave, vacation leave, mandatory meal period, weekly day of rest, and maternity leave, these laws may be more generous to workers than federal law, according to Puerto Rico Department of Labor officials. Sick leave. Non-exempt employees in Puerto Rico are entitled to accrue at least 1 day of paid sick leave after working at least 130 hours per month. Vacation leave. Non-exempt employees in Puerto Rico are entitled to accrue paid vacation after working at least 130 hours per month. Non- exempt employees hired before January 26, 2017, are entitled to a minimum monthly vacation leave accrual rate of one-and-a-quarter days. Non-exempt employees hired on or after January 26, 2017, are entitled to a minimum monthly vacation leave accrual rate of a half- day during the first year of service; three-quarters of a day after the first year of service up to the fifth year of service; 1 day after the fifth year of service up to the fifteenth year of service; and one-and-a- quarter days after the fifteenth year of service. However, in the case of Puerto Rico resident employers who have less than 12 employees, the minimum monthly vacation leave accrual rate is a half-day. Mandatory meal period. Non-exempt employees in Puerto Rico are entitled to a mandatory meal period between the third and sixth consecutive hour of work. In general, any employer that employs or allows an employee to work during the meal period is required to pay said period or fraction thereof at a pay rate equal to twice or one and one-half times the regular pay rate, as applicable. Weekly day of rest. Non-exempt employees in Puerto Rico are entitled to a mandatory weekly day of rest for every six consecutive days of work. Work performed during the day of rest is considered overtime and requires extraordinary compensation, regardless of the total number of hours that the non-exempt employee worked in the preceding 6 days. Maternity leave. Pregnant women in Puerto Rico are entitled to paid maternity leave 4 weeks before and 4 weeks after childbirth. Working mothers may opt to take only 1 week of pre-natal leave and extend post-natal leave up to 7 weeks. Women who adopt a child 5 years old or younger are entitled to 8 weeks of maternity leave. Possessions Tax Credit The Tax Reform Act of 1976 created the possessions tax credit to assist Puerto Rico and other insular areas in obtaining employment-producing investments. The credit effectively reduced federal taxes on income earned by qualifying U.S. corporations from operations in U.S. insular areas. However, the credit was repealed in 1996, but existing claimants were allowed to continue to use the credit during a 10-year phaseout period ending in 2006. In 2006, we reported that U.S. corporations claiming the credit dominated Puerto Rico’s manufacturing sector in the late 1990s and that after the tax credit began to phase out in 1996, the activities of these corporations decreased significantly. Unreliable Economic Data Make Conditions in Puerto Rico Difficult to Evaluate Prior to Hurricane Maria, Puerto Rico Had Already Experienced Prolonged Economic Decline Overall Economic Decline Puerto Rico Planning Board (Planning Board) data show that Puerto Rico has been in an economic decline for more than a decade. From 2005 to 2016, Puerto Rico’s GDP decreased by over 9 percent, after adjusting for inflation. Beginning in 2006, Puerto Rico’s economy experienced declines in real output in 9 of the next 11 years, as measured by real GDP (see fig. 2). While we have concerns about the precision of the Planning Board’s real GDP measure from year to year, as discussed later, we are confident in the downward direction of growth. Puerto Rico officials described the economic contraction as a downward spiral, where negative economic growth spurred outmigration by skilled workers, leading to decreased tax revenue and thereby increasing public debt per capita. This, in turn, they said decreases new investment and the cycle repeats. Five Main Factors that Contributed to Puerto Rico’s Economic Condition In May 2018, GAO reported on five main factors it identified through discussions with officials and experts and a review of literature. The factors were: Outmigration and diminished labor force. Some experts tied Puerto Rico’s negative economic growth to a steady decline in its population and labor force since 2005. According to Census data, Puerto Rico’s aging population means there are proportionally fewer individuals of working age. Regulatory challenges of doing business in Puerto Rico. Some experts cited the high cost to businesses of complying with Puerto Rico’s regulations, such as the permitting process for new businesses, and federal laws, such as the minimum wage law. High cost of importing goods and energy. Many of the goods used by businesses in Puerto Rico must be imported, significantly increasing their costs and in turn the cost of doing business. Petroleum, the main source of electronical energy generation, is a good whose high cost was particularly consequential to Puerto Rico’s economic struggles, according to Puerto Rico government officials, experts, and a literature reviews. Phaseout of the possessions tax credit. The loss of the tax credit was been cited by some as a potential cause of Puerto Rico’s economic decline since 2006; however, there was no consensus as to the magnitude. Banking and housing struggles. Puerto Rico’s banks have struggled and several have closed. Puerto Rico’s housing prices peaked in 2009, but fell 25 percent by January 2017, according to Federal Housing Finance Agency data. In our May 2018 report examining the Puerto Rico debt crises, we spoke with officials and experts, and conducted a literature review, and identified five main factors contributing to Puerto Rico’s current economic condition: outmigration and a diminished labor force; regulatory challenges of doing business in Puerto Rico; the high cost of importing goods and energy; the phaseout of the possessions tax credit; and banking and housing struggles (see sidebar). Hurricane Maria’s Economic Impact Puerto Rico was already experiencing a long economic contraction when Hurricane Maria made landfall in September 2017. Previous U.S. natural disasters, such as Hurricane Katrina in the Gulf Coast, have had significant adverse impacts on the economies of the affected regions, including significant outmigration. Immediately following Hurricane Katrina, the Gulf Coast experienced a number of challenges to its economy including a rise in unemployment; an increase in outmigration and decrease in housing units; a decline in state tax revenue; and a decline in imports and exports. Puerto Rico may experience similar challenges. For example, a February 2018 Federal Reserve Bank of New York press briefing on the impact of Hurricanes Maria and Irma characterized the 4 percent local job losses in Puerto Rico as substantial. Further, the briefing indicated that the true economic cost may be understated because some workers who are still employed likely suffered a drop in income, there may be unmeasured effects on the informal economy, and the value people place on quality of life issues are not measured. The substantial damage to the territory also accelerated outmigration and will likely worsen its economic condition. A January 2018 report from the Puerto Rico government identified the 2017 hurricanes as having a significant impact on the economy and projected that the population will decline by 10 percent over the next 2 years and could decline by nearly 20 percent over the next 5 years as people leave the island due to poor economic conditions. Initial data from the U.S. Bureau of Transportation Statistics show that 92,284 more people flew out of Puerto Rico with one- way tickets than flew into Puerto Rico in October 2017, the first full month after Hurricane Maria. That number represents a 255 percent increase over similar statistics in August 2017 and a 1,195 percent increase over October 2016 (see fig. 3). By December 2017, 17,281 more people flew out of Puerto Rico with one-way tickets than flew into Puerto Rico. This is 149 percent increase over similar statistics for December 2016. While the extent to which citizens of Puerto Rico may return to the territory is unclear, the initial outmigration could prolong negative economic growth. Methods Currently Used To Measure Puerto Rico’s Economy Are Outdated and Lead to Unreliable Measures Measuring GDP in Puerto Rico Outdated methods for measuring GDP make it difficult for the Puerto Rico government to fully analyze specific economic needs and develop long- range plans. There is no federal statistical measure of Puerto Rico’s GDP. The U.S. Census’ Economic Census of Island Areas provides some limited insights into Puerto Rico’s economic performance by industry, including revenue, payroll, employee count, and inventories. The Economic Census of Island Areas is updated every 5 years, but does not include total GDP. Instead, each year, BEA calculates GDP for four other territories and is reimbursed by the Department of the Interior’s Office of Insular Affairs for the estimation and publication of this information. In contrast, Puerto Rico’s Planning Board calculates GDP, but its methods are outdated and therefore unreliable, as they do not provide a precise measure of economic activity. Specifically, a 2011 White House Task Force Report examining Puerto Rico’s economic challenges found the Planning Board’s methods were outdated because they did not follow the same standards used for the rest of the United States. The Task Force also found that the methodology was not in line with modern statistical techniques, resulting in a less precise measure of Puerto Rico’s economic activity. Accurately calculating GDP is necessary to adequately measure total output of goods and services in Puerto Rico. GDP is also useful in measuring productivity and conducting monetary policy, and may be used to develop and apply appropriate policies for promoting economic growth. For example, a reliable and timely measure of GDP helps government officials calculate more accurate projections of tax revenue. The Planning Board’s method for calculating GDP does not effectively adjust for inflation because, the methodology uses a fixed-weighted index method that assumes the structure of the economy—what is being produced and prices of what is being produced relative to each other—is roughly constant over time. Further, the Planning Board is using this method to report inflation adjusted GDP based on the prices in a 1978 “market basket”—a fixed set of goods and services that people buy for day-to-day living. The Planning Board then uses 1954 as the reference year in its inflation adjustment to report GDP based on the price of goods and services. Consequently, the Planning Board’s real GDP measure may not be accurately adjusted to reflect current purchasing patterns and inflation in the prices of purchased products. BEA provides Puerto Rico’s Planning Board with some support in its calculation of GDP, but does not verify the accuracy of the calculation. In response to the 2011 White House Task Force findings, BEA began providing technical assistance and support to the Planning Board in updating its methods to adjust GDP for inflation and developed a report with recommendations for updating economic accounts. BEA found that the Planning Board’s methods did not comply with the internationally agreed upon standards for compiling measures of economic activity. Officials said that BEA was helping the Planning Board update its methods; however, a change in the level of communication slowed the update from 2013 through 2014. As a result, the Planning Board continued to use the same outdated methods. In January 2017, the Planning Board and BEA signed an agreement to modernize Puerto Rico’s economic accounts and align them with international guidelines. The agreement also tasked the Planning Board with providing deliverables in regular intervals beginning in spring 2017, including publication of alternative estimates of GDP that implement steps towards modernization. BEA officials told us that they are providing support to the Planning Board, and Planning Board officials told us they are working on updating the methodology. However, as of March 2018, the Planning Board had not yet produced all of the agreement deliverables, including publication of alternative GDP estimates. Given the impact of Hurricane Maria, it may be challenging for Puerto Rico to modernize its GDP measures. Planning Board officials told us in August 2017 that they were working to update their GDP methodology, so that it is similar to the one used by BEA, and that they would be updating to a 2007 “market basket.” Board officials said the new GDP figures were expected to be completed in December 2017. Their release was delayed in the aftermath of Hurricane Maria, but officials said they now expect to release GDP measures using the new methodology in summer 2018. Officials added that they plan to continue publishing GDP measures using the old methodology along with the new one for trend comparisons. The Planning Board and BEA estimated the cost to the Puerto Rico government to modernize its GDP measure is $2 million — including staff time and computing infrastructure. A 2016 bi-partisan Congressional Task Force on Economic Growth in Puerto Rico (2016 Congressional Task Force) recommended BEA calculate GDP for Puerto Rico as it does for the states and other territories, and BEA’s long-term goals include this objective. Further, in February 2018, the Financial Oversight and Management Board for Puerto Rico recommended that the Governor of Puerto Rico support efforts to implement the Congressional Task Force recommendation. BEA officials told us one of the agency’s long-term goals is calculating GDP for Puerto Rico and they have discussed including Puerto Rico in its reporting of GDP. Officials noted that including Puerto Rico in GDP reporting would require additional funds similar to reimbursements it received for the other four territories’ calculations. BEA’s mission is to promote a better understanding of the entire U.S. economy by providing the most timely, relevant, and accurate economic accounts data in an objective and cost-effective manner. BEA has provided technical assistance and support for 6 years; however the Planning Board has not yet modernized its methods to report a reliable GDP measure, and BEA has not included Puerto Rico in its reporting efforts. Federal standards for internal control state that management should use quality information to achieve the entity’s objectives. The lack of a federal GDP measure for Puerto Rico makes it difficult to make reasoned policy recommendations, adds uncertainty around issues affecting Puerto Rico’s economy, and makes it more difficult to identify fiscal and economic recovery plan priorities. Without modernized GDP methods, it remains difficult to compare Puerto Rico’s GDP with the rest of the United States and the other four territories for which BEA calculates GDP. Finally, without such a measure of GDP, federal policy makers and private investors must rely on various and sometimes unreliable data sources to try to establish common facts about Puerto Rico’s economic condition—an impediment in reaching consensus, engaging in meaningful policy discourse, and investment. U.S. Government Labor Statistics for Puerto Rico Federal labor statistics for Puerto Rico are incomplete because the Current Population Survey (CPS) does not include Puerto Rico and four other U.S. territories, and the American Community Survey (ACS) primarily provides data on population and housing, rather than labor. PROMESA recognized this and recommended that Census consider the feasibility of including Puerto Rico, and the other territories in the CPS. Specifically, PROMESA suggested that Census conduct a study to determine the feasibility of expanding data collection to include Puerto Rico and the other four U.S. territories in the CPS and if necessary, request the funding required to conduct this feasibility study as part of its budget submission to Congress for fiscal year 2018. Census officials told us they estimate a feasibility study including all of the U.S. territories will cost $1.1 million in fiscal year 2018, but did not request funding. The 2016 Congressional Task Force also recommended that BLS and Census take reasonable steps to include the territories. Federal standards for internal control state that management should use quality information to achieve the entity’s objectives. CPS data are intended to provide a comprehensive body of labor data that can be used to keep the nation informed about the economic and social well-being of its people, but Census and BLS are unable to report on the economic and social well-being of a segment of the nation and its people. Census officials told us they are concerned about unduly burdening Puerto Rico citizens with data collection efforts that would provide state level estimates. However, Census has not studied the feasibility of including Puerto Rico in the CPS, which would inform officials’ decision on whether to include Puerto Rico and the other territories in the CPS. By conducting such a study, Census would better understand the tradeoffs of including or continuing to omit Puerto Rico from CPS, including the extent to which it can be considered in public policy decisions, such as the 2016 Overtime Rule. Increasing the Overtime Threshold Would Affect a Small Percentage of Workers in Puerto Rico and Could Affect Employment Conditions The 2016 Overtime Threshold Increase Would Likely Affect Less Than 5 Percent of the Workforce in Puerto Rico Our estimates suggest that a larger percentage (about 4.5 percent) of Puerto Rico’s total workforce would have been affected by the Overtime Rule than the states (about 2.6 percent), based on our analysis of ACS data and DOL’s analysis of CPS data. Specifically, DOL’s analysis estimated that of 159.9 million wage and salary workers in the states, about 4.2 million (or about 2.6 percent) might be directly affected by the 2016 Overtime Rule. In our analysis of the Overtime Rule for Puerto Rico, we estimated that about 47,250 (about 4.5 percent) of 1.06 million wage and salary workers in Puerto Rico would have been directly affected (see fig. 4). The lack of data from CPS on Puerto Rico and the effects of Hurricane Maria hinder our ability to fully assess the potential effect of the 2016 Overtime Rule on Puerto Rico. Instead, we used data from the 2015 5- year ACS to estimate the impact of the Overtime Rule on Puerto Rico. The ACS employment data lack multiple variables available in the CPS; hence, we were limited in what we could estimate. For example, DOL’s estimate for the states included the wealth transfer from employers to employees, which is important for understanding the economic effects of the 2016 Overtime Rule. We do not provide similar insights because of the difference in variables in the ACS and CPS. (See table 2 in appendix I for the differences between our analysis and that of DOL.) Additionally, DOL estimated the effect the Overtime Rule would have on the probability that a worker had multiple jobs, but the limitations of the data we used kept us from performing this analysis. Our analysis estimates that the impact of the 2016 Overtime Rule in Puerto Rico would have been largely concentrated in four industries: education and health services, wholesale and retail trade, public administration, and financial activities. We estimated that in these four industries about 76 percent (about 36,000) of our approximate 47,250 total workers would have been directly affected (see table 1). The largest directly affected industry, education and health services, makes up about 43 percent (about 20,000) of this total. Raising the Overtime Threshold Could Increase Wages for Some, but Reduce Overall Hours and Employment in Certain Industries Depending on how employers respond to an increase in the overtime threshold, the effect of amending the overtime regulations could vary across employees. Similar to employers in the states, employers in Puerto Rico could respond to changes in the overtime regulations based on the current employee’s salary and work schedule by: 1) making no changes, 2) paying overtime, 3) raising salaries, or 4) adjusting hours worked (see fig. 5). In its analysis of the impact of the 2016 Overtime Rule in the states, DOL estimated the largest impact would be an aggregate transfer of income from employers to employees, which would be seen as a positive for some employees (e.g., increased pay, fewer hours for same pay, or new hires) and a negative for others (e.g., employers in our facilitated discussion groups said they would have layoffs, move employees from salaried to hourly, or lower benefit amounts). While we were unable to conduct a full impact analysis identical to DOL’s because of data limitations, we held facilitated discussion groups, to gain insight into how employers would have responded if the 2016 Overtime Rule was implemented in Puerto Rico. Views reported by participants in these groups may not be representative of all Puerto Rico employers, but they provide illustrative examples of the types of steps employers might have taken if the 2016 Overtime Rule were implemented. Employers in 2 of our 10 discussion groups said they would make staff adjustments by increasing the salary of some employees while, in some cases, minimizing the role of others. However, employers in 9 of our 10 of our discussion groups said they might also need to convert the remaining employees to part-time or hourly work, reduce their hours, or lay them off. “We have 125 exempt employees. The rule change would impact our labor costs a lot. We would need to minimize employees/hours to reduce the labor costs. We may adjust our contribution to medical plans to make up for the increased labor costs.” “With the new rule, half would need to be paid overtime or increase salary. This would leave us not enough flexibility to cover the hours or operation.” “We have a total of 850 employees. About 8 would be affected by the change. Of those 8, 3 we would boost their salary; the others will be switched to hourly. This may affect their benefits.” Through these facilitated discussion groups, we also learned that employer responses to the 2016 Overtime Rule may differ by industry. Employers in 3 of 10 industry discussion groups said they would be able to absorb some of the higher costs associated with an increase in the Overtime Rule threshold. For example, some manufacturers told us it would not be difficult for their businesses to absorb these additional costs, particularly if the salary threshold was at a somewhat lower level. Similarly, some hotel employers told us they would be able to absorb costs associated with the change across their many hotel locations, but others said they may not be able to assimilate the threshold increase. Hospital employers who participated in our facilitated discussion groups told us they have lower margins and face threats of closure even without the threshold increase. Some restaurant and hotel employers who participated in our facilitated discussion groups said they may be unable to pass associated increased labor costs to consumers; some would have to require exempt workers to work longer hours and reduce the number of full time employees or hours (see sidebar). Employers could respond by adjusting staff if the 2016 Overtime Rule goes into effect, but the impacts to employers may be limited and the workforce could benefit from the 2016 Overtime Rule change according to our interviews with 1 economist and 1 labor group official. One economist suggested that instead of having two employees who work 60 hours each, an employer might hire a third employee so that each works 40 hours. Additionally, this economist said an increase to the threshold would not be as burdensome to business, because employee wages have risen above the current overtime salary threshold. One labor group representative suggested that the 2016 Overtime Rule would have a minimal impact because very few workers in Puerto Rico earn enough to meet the 2004 salary threshold. Further, one economist said that under the current threshold, workers work excessive hours and do not have the same bargaining power. An increased overtime threshold would improve these dire working conditions. Specifically, this economist said that implementing the 2016 Overtime Rule would encourage firms to hire more workers, provide employees with more bargaining power, and help prevent worker exploitation. One economist said implementing the 2016 Overtime Rule only in the states could increase the wage differential between Puerto Rico and in turn increase outmigration from Puerto Rico. One member of the Puerto Rico Economic Administration said that if Puerto Rico were to have a lower threshold than the U.S. mainland the effects might be worse than those caused by the increased labor costs of implementing the higher threshold. Another economist said that while hours may be adjusted or layoffs may occur immediately following implementation of the Overtime Rule, these impacts would not be a major concern within 3 to 4 years. Conclusions Puerto Rico has long been experiencing severe economic challenges and its default on over a billion dollars of debt payments since 2015 has focused the need for attention to this territory. As Puerto Rico officials were in the process of taking action to update their methodology for reporting inflation-adjusted GDP, Hurricane Maria exacerbated the territory’s economic challenges. FEMA estimates that it may take years for Puerto Rico to recover. The lack of accurate economic and comprehensive labor data hinders policymaking, including a determination of the potential impact of changes to the overtime regulations. To help address Puerto Rico’s economic challenges now and in the future, the federal government and investors need updated and reliable data. Going forward, having BEA include Puerto Rico in its calculation of GDP would provide federal and local authorities, as well as businesses and investors, with reliable data on Puerto Rico’s economic condition that can be directly compared with the United States and other territories. BEA’s long-term goals include measuring Puerto Rico’s GDP, but having reliable data now would help address significant economic challenges in the short term. Likewise, studying whether including Puerto Rico in the Current Population Survey is feasible would allow DOL and other policymakers to be better positioned to fully consider the cost of including the territory against the implications of exclusion. Recommendations for Executive Action We are making a total of three recommendations, including two to Commerce and one to DOL. Specifically: The Secretary of Commerce should ensure that the Bureau of Economic Analysis includes Puerto Rico in its reporting on gross domestic product, as it does for four other U.S. territories. (Recommendation 1) The Secretary of Commerce, in cooperation with DOL’s Bureau of Labor Statistics, should conduct a study on the feasibility of including Puerto Rico in its reporting of the Current Population Survey. (Recommendation 2) The Secretary of Labor, in cooperation with the Commerce’s Census Bureau, should conduct a study on the feasibility of including Puerto Rico in its reporting of the Current Population Survey. (Recommendation 3) Agency Comments and Our Evaluation We provided a draft of the report to the Government of Puerto Rico, the Department of Commerce (Commerce), and the Department of Labor (DOL) for review and comment. In written comments that are reproduced in appendix IV, Commerce agreed with the recommendations made to it. In an email, DOL’s Deputy Assistant Secretary for Policy stated that the agency did not have any comments on the report. In addition, Commerce and the Government of Puerto Rico provided technical comments, which we incorporated into the report as appropriate. We are sending copies of this report to the appropriate congressional committees, the Government of Puerto Rico, the Secretary of the Department of Commerce, the Secretary of the Department of Labor, and other interested parties. In addition, this report is available at no charge on the GAO website at http://gao.gov. If you or your staff have any questions about this report, please contact Cindy Brown Barnes at (202) 512-7215 or Oliver Richard at (202) 512- 8424.You may also reach us by e-mail at brownbarnesc@gao.gov or richardo@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix V. Appendix I: Objectives, Scope, and Methodology Our review addressed: (1) the economic conditions in Puerto Rico as of the end of 2016; and (2) the potential effects of implementing the 2016 Overtime Rule on Puerto Rico’s economy. To evaluate the current economic conditions in Puerto Rico quantitatively, we analyzed (1) data from the Puerto Rico Planning Board’s (Planning Board) Statistical Appendix for the Governor that includes data on Puerto Rico’s gross domestic product (GDP) from 1990 through 2016 and (2) passenger data for Puerto Rican airports for years 2016 and 2017 from the Bureau of Transportation Statistics (BTS). To analyze the real GDP of Puerto Rico from 1990 through 2016 and Puerto Rico GDP by industry for 2016, we relied on data from the Planning Board’s Statistical Appendix for the Governor. We interviewed Planning Board officials responsible for producing the annual GDP estimates to understand how the data were prepared and any limitations to the data, and concluded that the while we have concerns over the precision of real GDP data, they were sufficiently reliable for our purposes of determining the direction of growth. To analyze the flow of passengers through Puerto Rican airports before and after Hurricane Maria, we relied on the BTS’ monthly passenger data for 2016 and 2017 as accessed through Diio Mi: Market Intelligence for the Aviation Industry. Diio Mi is a private contractor that provides online access to U.S. airline financial, operation, and passenger data. We reviewed the relevant documentation of the dataset and previous GAO reports and found the dataset sufficiently reliable for our purposes. To assess the potential effects of the 2016 Overtime Rule on Puerto Rico quantitatively, we analyzed and reported data from the American Community Survey (ACS) for calendar year 2015, because DOL used 2015 data in its impact analysis of the Overtime Rule on the states. The ACS is a national survey designed and administered by the Census Bureau (Census), and it contains data on individual earnings. Since 2005, the ACS has also included the Puerto Rico Community Survey (PRCS) which extends the survey through Puerto Rico. Our data set selection process included interviews with current and former agency officials as well as review of dataset documentation such as data handbooks, data dictionaries, and guidance on the different versions of data available. The ACS is conducted annually with estimates based on 1-year and 5-year data with benefits and drawbacks for each version. The 1-year ACS is updated the earliest and gives the most current data; however, according to Census’ guidance on the different ACS samples, it also contains the smallest sample size and is more appropriate for analyzing large populations than small ones. The 5-year ACS is the most reliable data, according to Census’ ACS guidance, as it includes the largest population size, which can be used to analyze small populations; however, the 5- year ACS is the least current version of the data. Additionally, while the 2015 5-year ACS data were not the most recent available, we used them because the Department of Labor (DOL) used 2015 data in its impact analysis of the 2016 Overtime Rule for the states. Based on these benefits and limitations, we chose to use the 2015 ACS 5-year estimates. Estimates produced from ACS data are subject to sampling error. For all of our estimates we weighted observations based on the individual weight. We compared our estimates of values derived from our weighting procedures to those published by the DOL and found them to be consistent. In addition to estimates, we generated standard errors or the margin of error for the 95 percent confidence interval, and report them with estimates in figures and tables. Based on our data checks, reviews of documentation and interviews with agency officials, we found the ACS data to be sufficiently reliable for our purposes. In addition to the quantitative data collected, we reviewed relevant federal laws, regulations, court documents, agency guidance, and internal controls related to the 2016 Overtime Rule, labor in Puerto Rico, and federal statistical measures for Puerto Rico. Additionally, we reviewed previous GAO reports on Puerto Rico and its economy. We interviewed DOL and Commerce officials at the national level. We interviewed Puerto Rican government officials to better understand current economic conditions and the statistical measure used to reflect the economic conditions. We interviewed representatives of national and Puerto Rican employer and labor organizations to gain their perspectives on the impact of the 2016 Overtime Rule and the condition of the economy. We conducted 10 facilitated group discussions with Puerto Rican employers in the manufacturing, restaurant, hotel, hospital, and professional services industries. These are some of the industries that employ the largest number of people in Puerto Rico and are among the most likely to be impacted by the 2016 Overtime Rule. Employers were selected to represent both large and small business perspectives in each industry. Views reported by participants in these groups are not representative of those of all Puerto Rico employers and for that reason are not generalizable. We also interviewed four economists, identified from prior work and interviews with agency officials, industry groups, and labor groups as having expertise relating to the 2016 Overtime Rule or the Puerto Rico economy, regarding the economic conditions of Puerto Rico and the potential economic impacts of the 2016 Overtime Rule. Analysis of American Community Survey Data For our analysis of the effects of the 2016 Overtime Rule on Puerto Rico, we mirrored the analysis conducted by DOL for the impacts of the rule on the United States. While the methodologies are similar, the DOL analysis used the Current Population Survey (CPS) data that do not include Puerto Rico. The ACS data we use serves a similar role; however, we made a few adjustments to the analysis in light of available data. Specifically, in the ACS, there is no variable identifying whether the individual works an hourly job. In order to simulate the removal of hourly workers from the sample, we randomly assign hourly worker designation to the same proportion of the population that are classified as hourly workers in DOL’s analysis (41.01 percent). A few assumptions are associated with this manipulation: 1) We are assuming that the proportion of the population that works an hourly job in Puerto Rico is similar to that of the 50 states, and 2) when we perform industry and region analyses, we assume the same hourly worker proportion across all industries and regions. Additionally, in its final rule, DOL analyzes the impact of the 2016 Overtime Rule on individual worker’s propensity to work multiple jobs; the ACS does not identify workers employed in multiple jobs and we could not perform this analysis. Also, the CPS data contain several variables on the number of hours worked that allowed DOL to analyze who likely works overtime on a regular basis. While the ACS includes a variable indicating the usual number of hours worked per week over the past year, we found it does not capture the schedule fluctuations as accurately as the CPS variables. As such, we do not calculate the dollar amount transfers from employers to employees and dead weight losses, or the loss in economic efficiency from the rule that DOL shows in its analyses. To the extent possible, we used DOL’s methodology to determine the potential effect of the 2016 Overtime Rule in Puerto Rico. However, due to data limitations there were some ways in which our approach differed from the approach used by DOL. In DOL’s analysis, the sample includes only the workers covered by its regulation. We adapt the sample to match the DOL’s analysis as follows: 1) remove military personnel, unpaid volunteers, self-employed individuals, clergy and other religious workers, and federal employees, 2) remove blue-collar workers and workers paid hourly; and 3) remove workers who are categorized under occupation and industry codes that are generally exempt under other exemptions. For consistency, our analysis makes the same adjustments to our sample that DOL makes in its analysis. Since we use the 2015, 5-year ACS data, we first put all wages into 2015 levels using ACS defined variables. Next, we inflate from 2015 to 2017 levels using the calendar year consumer price index (CPI-U). Just as in DOL’s analysis, we do not know whether any specific worker satisfies the duties test of the 2016 Overtime Rule, so we follow the same steps DOL took in its final rule. These steps include using DOL’s probabilities that specific job codes meet the duties test and assigning the probability to individual workers using the gamma distribution with the shape parameter alpha was set to the squared quotient of the sample mean divided by the sample standard deviation, and the scale parameter beta was set to the sample variance divided by the sample mean. Additionally, DOL explicitly removes certain industries and occupations from the sample and we follow its methodology exactly to remove these. Finally, to estimate the population that would have been affected by the 2016 Overtime Rule, we limit the sample to only those above the 2004 overtime salary threshold ($23,660) and below the 2016 salary threshold ($47,476) just as DOL did in its analysis. DOL’s estimates for industries and regions use different groupings than those provided in the ACS; however, since the ACS variables use the same coding but, a finer level, we can recreate the variables used in DOL’s analysis. For example, we used Census guidance on converting 2012 industry codes to create a major industry variable from the ACS industry codes. Similarly, to compare our estimates to DOL estimates by region, we take the ACS data, which is reported state-by-state, and put it into larger regions, such as “Northeast.” We conducted this performance audit from September 2016 to June 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Effects of Implementing Alternative Overtime Rule Thresholds on Puerto Rico Table 3 shows how different populations of workers would be affected by applying alternative overtime thresholds in Puerto Rico. For our analysis, in order to more closely replicate Department of Labor’s (DOL) design, we use the same alternative salary thresholds DOL analyzed in its final rule, which are defined as follows: Alternative 1: Inflate the 2004 Level - takes the 2004 overtime threshold and inflates it to fiscal year 2015 dollars using the consumer price index. This method leads to an overtime threshold of $570 per week or $29,640 per year. Alternative 2: 2004 Methodology - uses the 2004 final rule and updates it with data from the third quarter of 2015. This method leads to an overtime threshold of $596 per week or $31,015 per year. Alternative 3: Kantor Long Test - is based on a 1958 Report and Recommendations on Proposed Revision of Regulations, Part 541, by Harry S. Kantor. This methodology uses data collected on actual salaries paid to executive, administrative, or professional (EAP) employees grouped by geographic region, industry group, number of employees, and city size. DOL then used the long-duties test such that no more than about 10 percent of exempt EAP employees in the lowest-wage region, lowest-wage industry, smallest establishment group, or smallest city group will fail to meet the test. This method leads to a threshold of $684 per week or $35,568 per year. Alternative 4: 40th Percentile of Full-time Salaried Workers (Nationally) - takes all full-time salaried workers in the United States and calculates the 40th percentile of their wages. This method leads to a threshold of $972 per week or $50,544 per year. Alternative 5: Kantor Short Test - is also based on the Kantor method described in the third alternative, but uses the methodology associated with the short-duties test instead of the long-duties test. To do this, DOL took the $684 per week of the Kantor Long Test and inflated it by the average percent wage difference between the long and short test from 1949 through 1975 (149 percent). Multiplying $684 per week by 149 percent yields a Kantor Short Test threshold of $1,019 per week or $52,984 per year. Alternative 6: Inflate 1975 Short Test Level - takes the 1975 short- duties test salary level and inflates it to fiscal year 2015 dollars. This leads to a threshold level of $1,100 per week or $57,205 per year. Appendix III: Data Checks for American Community Survey The following tables show how the Department of Labor’s (DOL) analysis of the 2016 Overtime Rule in the United States using the analysis of Current Population Survey (CPS) data compares to our analysis using the American Community Survey (ACS) data. This analysis was conducted to support using ACS data as an alternative to CPS data. The methodologies used in our analysis and the DOL analysis are similar, but some adjustments were made to our analysis to attempt to replicate the DOL’s analysis of CPS data because variables were missing from the ACS data. Between the slightly different methodologies and different data sets, we expect the estimates to be of similar magnitudes, but not necessarily identical. The results indicate that our adjustments to the methodology and use of a different data yield similar results and add validity to our estimates for Puerto Rico. Appendix IV: Comments from the Department of Commerce Appendix V: GAO Contacts and Staff Acknowledgments GAO Contacts GAO Acknowledgments In addition to those named above, Kimberley Granger and Seyda Wentworth, Assistant Directors; Amber Yancey-Carroll, Analyst-in- Charge; Pedro Almoguera and Michael Naretta made key contributions to this report. Also contributing to this report were Jeffrey Arkin, David Blanding, David Chrisinger, Sarah Gilliland, Robin Marion, Jonathan S. McMurray, Sheila R. McCoy, Thomas Moscovitch, Dominic Nadarski, Mimi Nguyen, Karissa Robie, Benjamin Sinoff, Almeta Spencer, Amy Sweet, Anjali Tekchandani, Rosemary Torres Lerma, and Kathleen van Gelder.

Why GAO Did This Study Puerto Rico, the largest and most populous territory of the United States, is subject to congressional authority, although it has broad authority over matters of internal governance. After it defaulted on over $1.5 billion in public debt since 2015, Congress passed PROMESA to establish federal oversight of fiscal affairs. This debt crisis coincided with DOL finalizing the 2016 Overtime Rule, which was invalidated in federal court and is being appealed. PROMESA included a provision for GAO to assess the rule's impact on Puerto Rico and examine its economic condition. This report (1) examines the economic conditions in Puerto Rico as of the end of 2016, and (2) assesses the potential effects of applying the 2016 Overtime Rule to Puerto Rico. GAO analyzed 1990-2016 economic data and replicated DOL's impact analysis of the 2016 Overtime Rule using 2015 ACS data, the same year used by DOL in its analysis. GAO also reviewed federal laws, regulations, court documents, agency guidance, and criteria related to the federal overtime rule; facilitated group discussions with employers in Puerto Rico from industries most likely to be impacted by the rule; and interviewed relevant stakeholders and labor groups. What GAO Found Unreliable economic and limited labor data make conditions in Puerto Rico difficult to evaluate. Puerto Rico Planning Board data show that from 2005 to 2016 Puerto Rico's gross domestic product (GDP), a principal economic indicator, decreased by over 9 percent, after adjusting for inflation, and the devastation brought by Hurricane Maria in 2017 has worsened economic conditions. While the overall downward trend is reliable, GAO found that the Planning Board uses outdated methods to calculate GDP, which results in unreliable data from year to year and can make it difficult for policymakers to fully analyze specific economic needs and develop long-range plans. The Bureau of Economic Analysis (BEA), within the U.S. Department of Commerce (Commerce), does not calculate GDP for Puerto Rico, as it does for the other U.S. territories. For 6 years, BEA has provided technical support to the Planning Board to update its methods and Planning Board officials described plans to do so, but its methods remain outdated. A 2016 Congressional Task Force recommended that BEA calculate Puerto Rico's GDP, and BEA considers it a long-term goal; however, BEA has not taken steps to do so. Further, Puert Rico has limited labor statistics because it is not included in the Current Population Survey (CPS), which is produced by Commerce's Census Bureau (Census) and Department of Labor's (DOL) Bureau of Labor Statistics (BLS). CPS provides detailed information about employment, such as hours of work and earnings. The Puerto Rico Oversight, Management, and Economic Stability Act (PROMESA) suggested that Census conduct a study to determine the feasibility of expanding data collection to include Puerto Rico. Census officials said that they estimated the cost of such a study but have not yet conducted it. Census officials also cited concerns with data collection burdens. However, without CPS data on Puerto Rico, policymakers are limited in estimating the full economic impact of different policy changes. For example, DOL did not have the data needed to include Puerto Rico in its assessment of the economic impact of DOL's 2016 Overtime Rule. Conducting such a study would help policymakers consider the tradeoffs of including Puerto Rico in the CPS. GAO used a different dataset—American Community Survey (ACS)—to assess the potential effects of applying the 2016 Overtime Rule, which would have increased the salary level threshold from $23,660 to $47,476 at which executive, administrative, and professional workers would not be eligible for overtime pay. GAO estimated that about 47,250 of 1.06 million workers in Puerto Rico would be affected—that is, they would become eligible for overtime pay. In response to a salary level threshold increase, employers from selected industries in Puerto Rico told GAO that they might increase certain workers' salaries, but cut overtime hours for other workers, and adjust the number of staff. An economist and a labor group official said that employers could respond by adjusting the number of staff or their hours, but the impacts to employers may be limited and the workforce could benefit. In 2017, a federal district court invalidated the 2016 Overtime Rule and the overtime salary threshold remains at $23,660, but that decision is currently on appeal. What GAO Recommends GAO recommends that BEA include Puerto Rico in its reporting on GDP and that Census and BLS study the feasibility of including Puerto Rico in the CPS. Commerce agreed with our recommendations and DOL did not have any comments on the report.

gao_GAO-18-559

gao_GAO-18-559_0

Background A consumer reporting agency is a person or entity that assembles or evaluates consumer credit information or other consumer information for the purpose of furnishing consumer reports to others. This includes companies that compile and store electronic files of consumer information, which they then sell to other businesses and organizations that use the information to assess or evaluate creditworthiness. Furnishing of information by creditors and others to CRAs is voluntary, as federal law generally does not require such reporting, and information compiled on individual consumers can vary among the CRAs. A lender uses the information provided to determine whether to offer credit to an individual, the rate of interest to be assigned to the loan, and other terms of the contract. In addition, a growing number of entities use information provided by CRAs to help make decisions about individuals’ credit worthiness when determining eligibility for insurance, housing, or employment, among other things. Information from CRAs can also be used for other purposes, such as to identify potential customers with specific characteristics for new credit card accounts. CRAs may provide a variety of verification services to government and private sector organizations. For example, Equifax provides income and employment verification services using information collected from employers. Equifax, TransUnion, and Experian—the three major CRAs—also leverage information they collect from organizations, such as financial institutions, utilities, cell phone service providers, public records, and government sources, to offer identity verification services. Other entities, including federal agencies, use identity verification when they enroll new applicants for benefits and services. In addition, the IRS uses identity verification to ensure that individuals who want to access prior year tax returns are the legitimate filers of those returns. With regard to identity verification, CRAs typically use information they collect to generate questions that federal agencies and other entities can use to test applicants’ knowledge of information in their credit file. These questions and answers are typically the basis for identity proofing—the process of comparing evidence from an individual with a trusted source of data to verify that the individual is who they claim to be. The evidence generally consists of information or documentation that only the legitimate individual should know or have access to. For example, a driver’s license, passport, knowledge of recent financial transactions, and biometric information are all considered relatively strong evidence that the individual is who they say they are. A Data Breach Can Have Harmful Results Although there is no commonly agreed-upon definition, the term “data breach” generally refers to an unauthorized or unintentional exposure, disclosure, or loss of an organization’s sensitive information. This information can include PII, such as Social Security numbers, or financial information, such as credit card numbers. A data breach can occur under many circumstances and for many reasons. It can be inadvertent, such as from the loss of an electronic device, or deliberate, such as from the theft of a device. A breach can also occur as a result of a cyber-based attack by a malicious individual or group, agency insiders, foreign nation, terrorist, or other adversary. Data breaches have occurred at all types of organizations, including private, nonprofit, and federal and state entities. The loss or unauthorized disclosure of information in a data breach can lead to serious consequences and can result in substantial harm to individuals, private sector organizations, and the federal government. Examples of harmful results include: loss or theft of resources, including money and intellectual property, and identity theft; inappropriate access to and disclosure, modification, or destruction of sensitive information; harm to national security; use of computer services for unauthorized purposes or to launch an attack on other computer systems; damage to networks and equipment; loss of privacy, emotional distress, or reputational harm; loss of public confidence; and high costs to remediate the effects of the breach. Attackers Use a Variety of Tools and Techniques Cyber criminals seeking access to sensitive information, such as PII, typically use a variety of readily available software tools to carry out attacks. These tools can be used to intercept and capture data as they are transmitted, exploit known vulnerabilities in commercially available software, and facilitate e-mail phishing techniques for gaining unauthorized access to systems and information. Attackers often use similar techniques and tools, making it difficult to distinguish one attacker from another. When custom-built tools are used, an attacker may rely on unique methods or display other telltale signs that can be used for identification; such tools are usually used when a target’s defenses justify them. Off-the-shelf tools are usually enough to conduct a successful attack that allows an attacker to steal data, bring systems down, or gain further access to systems and resources. Attackers often begin with network-scanning programs, which are used to map the layout of a targeted network and determine the location of data repositories that may contain information of interest. Some scanners are designed to scan only a single networked computer, extracting as much data about that system as possible. Others can scan Internet addresses across the web to identify potential targets by determining whether they are using a version of software that is vulnerable to an attack. Once a target has been identified, the attacker will generally attempt to gain access to the system or network without leaving any indication of who they are or from where they launched their attack. This is commonly accomplished using tools that mask the attacker’s origin by using the Internet address of another computer from another location. While an investigator can sometimes use forensic tools to trace the original Internet address, often this leads to misleading information. Attackers use additional tools and techniques to gain unauthorized access to systems and data on the target network and to transfer stolen data back to the attacker’s own computer system. One such technique is to leverage the access rights gained on the originally compromised system to get further access into other servers on the network. To do this, an attacker can use standard, off-the-shelf tools for navigating systems and managing information that blend in with normal network activity. For example, encryption can be used to hide the transfer of sensitive information from one server to another or out of the network entirely. This enables the attacker to continue probing for more repositories of information and stealing copies of that information without being detected by the targeted network’s system administrators. Federal Agencies Oversee CRA Activities, Including Protection of Personally Identifiable Information CRAs have been subject to federal regulation since the passage of the Fair Credit Reporting Act in 1970. Currently, FTC and BCFP are the two federal agencies with primary oversight responsibilities for CRAs. FTC was given responsibility for administratively enforcing CRAs’ compliance with the Fair Credit Reporting Act at the time of enactment. As part of the Dodd-Frank Wall Street Reform and Consumer Protection Act (Dodd- Frank Act), BCFP was given authority to enforce a number of federal consumer financial laws, including the Fair Credit Reporting Act. BCFP also has begun exercising supervisory authority over certain larger participants in the credit reporting market. FTC Has Enforcement Authority over CRAs FTC has authority, subject to certain exceptions, to investigate any organization that maintains consumer data and to bring enforcement actions for violations of laws that concern the protection of consumer information. FTC also exercises enforcement authority over CRAs through the Gramm-Leach-Bliley Act and the related “Safeguards” and “Privacy Rules.” The Fair Credit Reporting Act promotes the accuracy, fairness, and privacy of information collected or used to help make decisions about individuals’ eligibility for credit, insurance, employment, housing, or other benefits. CRAs that compile credit histories and other personal information into consumer reports must adhere to the act’s provisions for ensuring the accuracy and permissible uses of such information. The Gramm-Leach-Bliley Act requires that federal financial regulators and FTC establish standards and protections to ensure the security and confidentiality of customer information. These standards and protections must be implemented by companies of all sizes that are engaged in financial activities, including Equifax and all other CRAs. Further, the act requires financial institutions to protect the security of customers’ personal information. As part of its implementation of the Gramm-Leach-Bliley Act, FTC issued the “Safeguards Rule”, which requires financial institutions develop, implement, and maintain a comprehensive information security program to keep information about a customer of a financial institution secure and confidential. In addition to developing their own safeguards, companies covered by the rule are responsible for requiring their affiliates and service providers to implement and maintain safeguards to protect customer information in their care. In determining whether it should take enforcement action against a company for a violation of data security provisions, FTC considers a number of factors, including whether a company’s data security measures are commensurate with the company’s size. FTC does not have supervisory authority to examine CRAs for compliance with the Federal Trade Commission Act; therefore, the agency typically must rely on its enforcement authority after an incident has occurred. Finally, FTC enforces Section 5 of the Federal Trade Commission Act, which prohibits “unfair or deceptive acts or practices in or affecting commerce.” FTC officials told us that failing to properly protect consumer data can be considered an unfair or deceptive act or practice. BCFP Has Enforcement and Supervisory Authorities over CRAs In 2010, the Dodd-Frank Act gave BCFP enforcement authority over all CRAs and certain other persons for violations of most provisions of the Fair Credit Reporting Act; certain provisions of the Gramm-Leach-Bliley Act; and for unfair, deceptive, or abusive acts or practices under sections 1031 and 1036 of the Dodd-Frank Act. BCFP has taken enforcement actions against CRAs for violations of the Fair Credit Reporting Act and for deceptive practices. In 2012, BCFP also extended its supervisory authority to include larger CRAs—that is, those with more than $7 million in annual receipts from consumer reporting activities. BCFP staff review certain of these larger CRAs on an ongoing basis, and BCFP staff said that their recent examinations of CRAs have focused on compliance with Fair Credit Reporting Act requirements related to accuracy and resolving consumer disputes. BCFP has also examined CRAs subject to the BCFP’s supervisory authority for compliance with other Fair Credit Reporting Act requirements, including those related to ensuring the accuracy of information in consumer reports, furnishing information only to those with a permissible purpose, and compliance with the consumer dispute process. BCFP also has supervisory authority over some aspects of the Gramm- Leach-Bliley Act. For example, BCFP examines larger CRAs for whether they restrict the sharing and disclosure of nonpublic personal information to third parties. BCFP does not have supervisory or enforcement authority over the “Safeguards Rule” enacted by FTC as part of the agency’s implementation of the Gramm-Leach-Bliley Act. Finally, BCFP has authority to examine larger CRAs for any unfair, deceptive, or abusive acts or practices and to bring enforcement actions against CRAs of all sizes for such acts or practices. According to BCFP staff, in some cases, a CRA could commit an unfair, deceptive, or abusive act or practice or violation of other applicable law in connection with its data security practices. GAO Has Previously Reported on Data Protection Issues We have previously made recommendations to agencies regarding the protection of PII, and proposed Matters for Congressional Consideration in areas where laws could be enhanced. For example, in our recent report on data oversight at the Centers for Medicare and Medicaid Services (CMS), we recommended that the agency ensure that all third parties that receive CMS data have clear requirements for the protection of that data, that CMS properly oversee the implementation of those requirements, and that the agency ensure identified issues are remediated. Additionally, our recent report on the oversight of students’ PII at the Department of Education included seven recommendations for better protection of student PII and for improving department policies to meet federal privacy guidelines. All of these recommendations currently remain open while the agencies take actions to address them. In addition to recommendations for agencies, we have proposed two Matters for Congressional Consideration related to data protection. In 2008, we reported that the Privacy Act and E-Government Act of 2002 may not adequately ensure that consumers are notified in the event of a breach by federal agencies and that existing laws could better ensure that consumers are aware of what PII federal agencies collect and how they use it. Based on this finding, we suggested that Congress consider amending applicable laws to ensure that all PII collected by federal agencies is protected and that its use is limited to the stated purpose of the collection. With regard to data collected by private entities, in 2013, we reported that existing federal laws provide consumers with only limited protection for data that is collected and used for marketing purposes. Consequently, we asked Congress to consider strengthening the current consumer privacy framework to reflect the effects of changes in technology and the marketplace while also ensuring that any limitations on data collection and sharing do not unduly inhibit the economic and other benefits to industry and consumers that data sharing can accord. Attackers Exploited Vulnerabilities That Equifax Subsequently Reported Taking Actions to Address In March 2017, unidentified individuals discovered the presence of a known vulnerability in software running on Equifax’s online dispute portal that could be used to obtain access to the system. In May of that year, attackers exploited the vulnerability and began to extract data containing PII from Equifax’s information systems. According to Equifax, the attackers used a number of techniques to disguise their exploit of the Equifax systems and the database queries they conducted. On July 29, 2017, Equifax discovered the breach and reported that it took actions to address the factors that allowed the attackers to successfully gain access to its network. Further, the company reported that it took steps to identify, notify, and provide support to individuals who were potentially impacted by the breach. Attackers Identified and Exploited Vulnerabilities to Steal Data Equifax has stated that, on March 10, 2017, unidentified individuals scanned the company’s systems to determine if the systems were susceptible to a specific vulnerability that the United States Computer Emergency Readiness Team had publicly identified just 2 days earlier. The vulnerability involved the Apache Struts Web Framework and would allow an attacker to execute commands on affected systems. Equifax officials stated that, as a result of this scanning, the unidentified individuals discovered a server housing Equifax’s online dispute portal that was running a version of the software that contained the vulnerability. Using software they obtained from an unknown source and that was designed to exploit the vulnerability, the unidentified individuals subsequently gained unauthorized access to the Equifax portal and confirmed that they could run commands. No data was taken at this time. According to Equifax officials, beginning on May 13, 2017, in a separate incident following the initial unauthorized access, attackers gained access to the online dispute portal and used a number of techniques to disguise their activity. For example, the attackers leveraged existing encrypted communication channels connected to the online dispute portal to send queries and commands to other systems and to retrieve the PII residing on the systems. The use of encryption allowed the attackers to blend in their malicious actions with regular activity on the Equifax network and, thus, secretly maintain a presence on that network as they launched further attacks without being detected by Equifax’s scanning software. Equifax officials added that, after gaining the ability to issue system-level commands on the online dispute portal that was originally compromised, the attackers issued queries to other databases to search for sensitive data. This search led to a data repository containing PII, as well as unencrypted usernames and passwords that could provide the attackers access to several other Equifax databases. According to Equifax’s interim Chief Security Officer, the attackers were able to leverage these credentials to expand their access beyond the 3 databases associated with the online dispute portal, to include an additional 48 unrelated databases. After reviewing system log files that recorded the attackers’ actions, Equifax officials determined that the attackers then ran a series of queries in an effort to try to extract PII from the databases they had located. Altogether, the attackers ran approximately 9,000 queries, a portion of which successfully returned data containing PII. As before, Equifax officials stated that the attackers were able to disguise their presence by blending in with regular activity on the network. After successfully extracting PII from Equifax databases, the attackers removed the data in small increments, using standard encrypted web protocols to disguise the exchanges as normal network traffic. The attack lasted for about 76 days before it was discovered. Figure 1 depicts an analysis of how the attackers gained access into Equifax’s systems and exploited vulnerabilities. After Becoming Aware of the Attack, Equifax Took Steps to Block the Attackers Equifax’s assessment of the data breach began with actions it took to identify that it was being attacked as well as subsequent actions to block the intrusion. Equifax officials stated that, on July 29, 2017— approximately 2.5 months after the attackers began extracting sensitive information on May 13, 2017—security personnel conducting routine checks of the operating status and configuration of IT systems detected the intrusion on the online dispute portal. As reported by Equifax, a network administrator conducting routine checks of the operating status and configuration of IT systems discovered that a misconfigured piece of equipment allowed attackers to communicate with compromised servers and steal data without detection. Specifically, while Equifax had installed a device to inspect network traffic for evidence of malicious activity, a misconfiguration allowed encrypted traffic to pass through the network without being inspected. According to Equifax officials, the misconfiguration was due to an expired digital certificate. The certificate had expired about 10 months before the breach occurred, meaning that encrypted traffic was not being inspected throughout that period. As a result, during that period, the attacker was able to run commands and remove stolen data over an encrypted connection without detection. Equifax officials stated that, after the misconfiguration was corrected by updating the expired digital certificate and the inspection of network traffic had restarted, the administrator recognized signs of an intrusion, such as system commands being executed in ways that were not part of normal operations. Equifax then blocked several Internet addresses from which the requests were being executed to try to stop the attack. Equifax reported that, on July 30, 2017, after its information security department observed additional suspicious activity continuing to occur, the online dispute portal was taken offline. The next day, the Chief Security Officer, in coordination with internal stakeholders, informed the Chief Executive Officer of the attack on the portal. Equifax Identified Several Factors That the Attacker Exploited During the Breach To further assess the scope of the breach and identify its causes, Equifax began an investigation to identify the vulnerabilities that had been exploited to steal PII from its systems. Concurrent with this effort, company officials stated that they also began examining the data repositories that had been accessed to try to determine how much data had been taken and how many individuals were potentially impacted. According to Equifax officials, the investigation took place between August 2 and October 2, 2017, with the help of an external cybersecurity consultant. Equifax officials stated that the company’s investigation was facilitated by the use of electronic logs that had not been damaged or erased by the attackers on the affected systems. These logs recorded commands that were issued by the attackers throughout the attack, such as commands to retrieve or display the contents of data repositories. By examining the logs, Equifax worked to reconstruct the sequence of specific actions that the attackers had taken and, consequently, determine what specific data had been compromised. In addition to initiating its internal investigation, on August 2, 2017, the company notified the Federal Bureau of Investigation of the breach. Based on its cybersecurity consultant’s analysis and recommendations following the breach, Equifax determined that several major factors had facilitated the attackers’ ability to successfully gain access to its network and extract information from databases containing PII. Specifically, Equifax officials told us that key factors that led to the breach were in the areas of identification, detection, segmentation, and data governance: Identification. According to Equifax officials, the Apache Struts vulnerability was not properly identified as being present on the online dispute portal when patches for the vulnerability were being installed throughout the company. After receiving a notice of the vulnerability from the United States Computer Emergency Readiness Team in March 2017, Equifax officials stated that they circulated the notice among their systems administrators. However, the recipient list for the notice was out-of-date and, as a result, the notice was not received by the individuals who would have been responsible for installing the necessary patch. In addition, Equifax officials stated that although the company scanned the network a week after the Apache Struts vulnerability was identified, the scan did not detect the vulnerability on the online dispute portal. Detection. As reported by Equifax officials, an expired digital certificate contributed to the attackers’ ability to communicate with compromised servers and steal data without detection. Specifically, while Equifax had installed a tool to inspect network traffic for evidence of malicious activity, the expired certificate prevented that tool from performing its intended function of detecting malicious traffic. The certificate had expired before May 2017, meaning that traffic was not being inspected throughout the breach. Segmentation. Because individual databases were not isolated or “segmented” from each other, the attackers were able to access additional databases beyond the ones related to the online dispute portal, according to Equifax officials. The lack of segmentation allowed the attackers to gain access to additional databases containing PII, and, in addition to an expired certificate, allowed the attackers to successfully remove large amounts of PII without triggering an alarm. Data Governance. Data governance includes setting limits on access to sensitive information, including credentials such as usernames and passwords. According to Equifax officials, the attackers gained access to a database that contained unencrypted credentials for accessing additional databases, such as usernames and passwords. This enabled the intruders to run queries on those additional databases. In addition to these four broad categories, Equifax officials noted one other factor that also facilitated the breach. Specifically, the lack of restrictions on the frequency of database queries allowed the attackers to execute approximately 9,000 such queries—many more than would be needed for normal operations. Equifax Reported Taking Steps to Strengthen its Cybersecurity Controls According to Equifax’s public filings, including its annual 10-K filing submitted to the Securities and Exchange Commission in March 2018 and its notice of 2018 annual meeting and proxy statement, following the 2017 incident, Equifax undertook a variety of remediation efforts to address the factors identified in their investigation. Equifax officials responsible for coordinating the response to the incident stated that, once the company identified how the attackers were able to gain unauthorized access to company systems and remove sensitive data, it took measures to address the internal factors that led to the breach. The measures were intended to better protect the company’s infrastructure from future disruptions, compromises, or failures. We did not independently assess Equifax’s efforts to address the identified factors. Specifically, Equifax officials stated that system-level remediation measures were implemented to address the factors that led to the breach. For example, to work toward addressing concerns about identifying vulnerable servers, Equifax reportedly is implementing a new management process to identify and patch software vulnerabilities and confirm that vulnerabilities have been addressed. Also, to help ensure that detection of malicious activity is not hindered in the future, Equifax officials said they have developed new policies to protect data and applications and implemented new tools for continuous monitoring of network traffic. Further, in an effort to improve segmentation between devices that do not need to communicate, Equifax officials stated that they have implemented additional controls to monitor communications at the external boundary of the company’s networks and added restrictions on traffic between internal servers. Finally, to help address data governance issues, the officials said they were implementing a new security controls framework and tighter controls for accessing specific systems, applications, and networks. In addition to these measures, Equifax stated that they implemented a new endpoint security tool to detect misconfigurations, evaluate potential indications of compromise, and automatically notify system administrators of identified vulnerabilities. Further, Equifax officials reported that the company has implemented a new governance structure to regularly communicate risk awareness to Equifax’s board of directors and senior management. The new structure requires the company’s Chief Information Security Officer to report directly to the Chief Executive Officer. Officials said this should allow for greater visibility of cybersecurity risks at top management levels. Equifax Reported Taking Steps to Identify Affected Individuals Following the shutdown of its online dispute portal, Equifax took steps to identify what data had been lost and the number of individuals affected so that it could fulfill its responsibility to notify affected individuals. To develop its estimate of the number of individuals affected by the data breach, Equifax stated that it recreated the attackers’ database queries on a separate system that could run the queries at high speed, allowing Equifax to generate its estimate in a relatively short period of time. Equifax staff then worked to reconstruct queries against the data tables to identify which queries had successfully extracted data and which individuals were associated with that data. However, as is commonly experienced with large breaches, Equifax faced challenges in determining exactly how many individuals were affected. According to Equifax officials, much of the stolen data consisted of incomplete records without full sets of identifying information. Some data sets included information that could be matched to more than one known individual. Subsequently, Equifax officials stated that they compared these data sets with information in the company’s internal databases that were not impacted by the data breach to make matches with known identities. For example, Equifax took partial records that did not include all fields and ran an analysis to determine whether Social Security numbers and names included in the records could be matched with those in Equifax’s core credit reporting databases. In addition, Equifax performed analyses to remove duplicates and to determine whether a person could be linked to incomplete records based on Social Security numbers. After Equifax completed its initial analysis of the datasets, it estimated that approximately 143 million U.S. consumers had been affected by the breach. Moreover, Equifax’s initial analysis, reported on September 7, 2017, indicated that multiple types of PII had been compromised, including individuals’ names, Social Security numbers, birth dates, addresses, and driver’s license numbers. Because many of the records were incomplete, not all of the types of PII had been compromised for all affected individuals. In addition, Equifax determined that credit card numbers for approximately 209,000 consumers and certain dispute documents, which had included PII for approximately 182,000 consumers, had been accessed. These documents contained PII associated with specific items from dispute cases that were submitted to Equifax as evidence supporting disputes they filed about the accuracy of their credit reports, such as utility bills. Equifax made two revisions over time to its estimate of affected individuals. First, in late September 2017, Equifax determined that it had incorrectly concluded that one of the attackers’ queries had not returned any data. After additional analysis, including a determination that the query had, in fact, allowed the attackers to access PII from approximately 2.5 million additional U.S. consumers, Equifax revised the number of affected individuals from 143 million to 145.5 million on October 2, 2017. Second, on March 1, 2018, Equifax stated that it had identified approximately 2.4 million U.S. consumers whose names and partial driver’s license information were stolen. The newly identified individuals were based on names and partial driver’s license information contained in a data table that Equifax had not previously identified as including individuals compromised in the breach. According to Equifax officials, Equifax’s original investigation had not identified these individuals because their names and partial driver’s license information were not stolen together with their Social Security numbers. To identify as many potentially affected individuals as possible, Equifax contracted with a third-party data source that had access to a driver’s license database and mapped the partial driver’s licenses to an Equifax database containing Social Security numbers. According to Equifax officials, some of the individuals within this group of 2.4 million were already included in the previous total of 145.5 million affected individuals, while others were not. As of August 2018, Equifax had not determined exactly how many of the 2.4 million individuals were included in the previous total of 145.5 million. Equifax Notified Affected Individuals and Offered Monitoring Services On September 7, 2017, after Equifax had determined the extent of the breach and developed a remediation plan for potentially impacted consumers, the company provided written notification to all U.S. state attorneys general regarding the approximate number of potentially affected residents in each state and its plans for consumer remediation. The notification included steps individuals could take to determine if they were affected by the breach and to help protect against misuse of their personal information. The company also issued a press release to the public providing information about the breach and the types of PII that had been compromised. Further, the press release issued on September 7, 2017, stated that the company had set up a dedicated website to help individuals determine if their information might have been stolen in the breach. In addition, Equifax improved the search tool it had developed to help U.S. consumers determine if they were impacted and expanded its call center operations. However, the website experienced several technical issues, including excessive downtime and inaccurate data. Equifax officials acknowledged these shortcomings and said they took measures to address them, including improving the stability of the website and accuracy of the information it provided. Additionally, Equifax reported that it would provide several services to all U.S. consumers, regardless of whether their information had been compromised, free of charge for one year. Those services included credit monitoring, individual copies of Equifax credit reports, notification of changes to credit reports, a credit “lock” allowing individuals to prevent third-parties from accessing their Equifax credit report, identity theft insurance covering certain expenses related to the process of recovering from identity theft, and a Social Security number monitoring service that would scan suspicious websites for an individual’s Social Security number. These services were offered to consumers from September 7, 2017, until January 31, 2018, when Equifax announced a new service called “Lock & Alert.” This new service allows consumers to use their smartphone or computer to lock and unlock their Equifax credit report. Equifax announced that it was making this service available to all consumers at no cost. Federal Agencies Took a Variety of Actions in Response to the Equifax Breach After Equifax announced the data breach, federal customer agencies took a variety of actions based on their responsibilities and how the breach affected their operations. Specifically, the agencies that were customers of the company’s services conducted independent assessments of the company’s security controls, revised their own identity proofing processes, and made changes to their contracts with Equifax, among other activities. Equifax did not ask the Department of Homeland Security (DHS), which is the central agency that responds to cyber incidents across the federal government, to assist in responding to the breach. Nevertheless, the department took the step of reminding federal agencies of the importance of correcting the software vulnerability that led to the breach. In addition, the oversight agencies, BCFP and FTC, began taking actions to investigate the breach and inform the public. Major Federal Customers of Equifax Took Steps to Ensure Their Activities Were Not Adversely Affected by the Breach IRS, SSA, and USPS—large agencies that were major customers of Equifax at the time of the breach—assessed the potential impact of the breach to their own operations as well as to the operations of their consumers. For example, these agencies assessed the technical impact of the breach on their own systems that rely on Equifax services to determine whether the breach could have compromised the integrity of their identity proofing processes. While there was no breach of agency systems or information, they also sought to determine which of their customers were directly affected by the breach, recognizing that those individuals could be at heightened risk of identity fraud. Information security officials we spoke to at IRS, SSA, USPS, and DHS expressed concern about how the breached data could be used to compromise sensitive information or fraudulently procure government services, even from agencies that are not direct customers of Equifax. Representatives of IRS, SSA, and USPS noted that they responded to the breach independently of other agencies, because they said it was unclear whether any single federal agency had responsibility for coordinating government actions in response to a breach of this type in the private sector. According to the three agencies, their actions included the following: Identified affected individuals. Due to concerns about the potential for fraud using the stolen data, IRS and SSA both obtained from Equifax a list of the individuals affected by the Equifax breach. The agencies then used these lists to identify which of their own customers were affected and to look for potential instances of identity fraud affecting those customers. Performed independent assessments of Equifax security controls. According to information security officials at IRS, SSA, and USPS, the agencies independently conducted site visits at Equifax’s data center in Alpharetta, Georgia, where they reviewed the company’s security controls. According to SSA officials, their agency’s review assessed compliance with the baseline set of controls required by the National Institute of Standards and Technology for systems determined to pose a moderate level of risk. SSA officials stated that they shared the results of their assessment with IRS, the Office of Management and Budget, House Ways and Means Social Security Subcommittee, and the Senate Committee on Finance. USPS officials said they reviewed both physical security and cybersecurity controls at Equifax’s data centers in Alpharetta, Georgia and St. Louis, Missouri locations. IRS officials said they also conducted a security assessment at Equifax’s Alpharetta data center, as well as a separate review of physical security and cybersecurity controls at the company’s St. Louis, Missouri site. The officials of all three agencies said that their reviews did not uncover any major new problems, but did identify a number of lower-level technical concerns that they required Equifax to address. Modified contracts with Equifax. IRS and SSA made changes to contracts they had with Equifax to require prompt notification of any future breach, among other things. According to officials from both agencies, Equifax did not directly notify major federal customers of the 2017 breach prior to its public announcement because its contracts with these agencies required notification only of breaches directly involving the systems that provided services to the federal government. SSA officials stated that it was important to update the agency’s contract to require Equifax to promptly notify SSA of any data breach, regardless of which of the company’s systems it may affect. IRS officials stated that a similar change was made to their contract with Equifax for credit reporting services. The contract change also required the company to notify IRS within one hour after a breach is discovered, rather than within the previous time frame of 24 hours. In addition, according to the officials, cybersecurity language in the IRS’s contract was modified to ensure better implementation and oversight of technical security controls. Communicated with the public and affected individuals. IRS made public announcements about the impact of the breach, noting that the agency did not expect the breach to have any impact on taxpayers’ ability to securely file tax returns. SSA issued a public blog post reminding consumers about steps they could take to protect their Social Security numbers. Made changes to agency identity-proofing procedures. Following its assessment, IRS updated its internal cybersecurity contractor requirements and controls related to incident handling. Further, upon completing its assessment, USPS initiated discussions with the National Institute of Standards and Technology to determine risks associated with the knowledge-based verification questions it had been using with Equifax’s identity-proofing service. USPS subsequently changed its process, removing certain knowledge- based verification questions and adding a procedure whereby customers receive a code in the mail that they can use to verify their mailing addresses. Canceled a short-term contract with Equifax. Before the Equifax breach, Equifax was the incumbent contractor at IRS for taxpayer identity and verification services. In June 2017, prior to the discovery of the breach, IRS began a new acquisition for these services by issuing a request for quotations to three CRA vendors (including Equifax and Experian) holding contracts under the federal supply schedule. IRS selected Experian as offering the lowest-priced, technically acceptable quotation, for issuance of a fixed-price task order and establishment of a blanket purchase agreement. Equifax filed a bid protest on July 5, 2017 with GAO challenging the IRS’s evaluation of Experian’s quotation. As described elsewhere in this report, Equifax discovered the breach on July 29 and, after investigating it, announced the breach on September 7. On September 29, during GAO’s consideration of the protest, IRS awarded Equifax a short-term, sole-source contract for $7.25 million to cover its need for the identity and verification services during the time frame needed to resolve the protest. IRS considered these services “critical” that “cannot lapse.” However, following the completion of its breach-related security assessments, IRS issued Equifax a stop-work order to suspend its performance under the short- term, sole-source order. GAO denied Equifax’s protest on October 16, 2017 and IRS proceeded with the task order issued to Experian for the taxpayer identity and verification services. DHS Offered Breach Response Services to Equifax In its role as the center for federal information security incident prevention and response, DHS offers services to assist federal agencies in preparing for potential cyber incidents, maintaining awareness of the current threat environment, and dealing with ongoing breaches. Under a Presidential directive, DHS is also responsible for assisting public- and private- sector critical infrastructure owners and operators in preparing for, preventing, protecting against, mitigating, responding to, and recovering from a cyber incident. In September 2017, shortly after the Equifax breach was publicly announced, DHS contacted the company to offer its professional services related to forensic analysis and breach response. However, according to officials at both organizations, Equifax notified DHS officials that the company had already retained professional services from a private cybersecurity consultant and, thus, declined assistance from DHS. Oversight Agencies Opened Investigations and Provided Information and General Advice to Consumers According to Equifax officials, the company informed regulators about the data breach on September 7, 2017—when the general public was notified. FTC announced that it was investigating the Equifax breach, and Equifax stated in its annual report that several governmental agencies, including FTC and BCFP, were continuing to investigate events related to the breach. BCFP staff told us that, immediately following notification of the breach, they participated in conference calls with Equifax to learn more about the breach. According to the officials, their calls with Equifax focused on ensuring consumers were provided with accurate information about the breach and what they could do to protect themselves. Equifax officials told us that they also informed FTC, the Securities and Exchange Commission, various states’ attorneys’ general, and the Financial Services Information Sharing and Analysis Center, that it had suffered a breach. Shortly after Equifax’s public announcement of the breach, BCFP released a blog post on the top 10 ways that consumers could protect themselves in the wake of the breach. Suggestions included regularly reviewing credit reports, checking credit card statements, and changing passwords for all financial accounts. In addition, BCFP posted on its website actions consumers could take to protect themselves against fraud or identity theft, including freezing credit and placing fraud alerts. BCFP staff told us that, while the agency posts information to its website, it does not provide individual legal assistance to consumers. Nevertheless, the staff said that consumers can file a complaint with BCFP if they are experiencing issues related to a CRA. BCFP staff added that they received a large volume of consumer complaints following the Equifax breach. BCFP staff said they use such complaints as one factor to prioritize future supervisory examinations, as well as investigations and enforcement actions. In October 2017, BCFP also began conducting targeted data security and cybersecurity examinations. Specifically, in addition to assessing whether the CRAs’ data security practices and policies constitute violations of federal consumer financial law, BCFP began assessing risks to consumers posed by potential cybersecurity lapses and to markets for consumer financial products and services. BCFP staff said that whether BCFP continues to conduct CRA cybersecurity examinations will depend on whether they identify the issue as a priority through future examination prioritization processes. Similarly, FTC released a statement to consumers with information about the breach, such as when it occurred and the types of data compromised. The statement also included guidance on steps consumers could take to help protect their information from being misused. For example, FTC encouraged individuals to visit Equifax’s website to find out whether their information may have been exposed, provided links to obtain a free credit report, and offered other information about credit freezes and fraud alerts. On June 25, 2018, eight state banking regulators issued a consent order requiring Equifax to address various data security issues. The order included several areas of concern, including general information security, internal audits, and board and management oversight. More specifically, the order required Equifax, the board, or its audit committee to, among other things: provide a written risk assessment that identifies foreseeable threats and vulnerabilities to the confidentiality of PII; establish a formal and documented internal audit program that is capable of effectively evaluating information technology controls; improve the oversight of its information security program; improve oversight and documentation of its critical vendors; improve standards and controls for supporting the patch management function; and enhance oversight of IT operations as it relates to disaster recovery and business continuity functions. Under the consent order, Equifax was required to submit a list of all remediation projects planned, in process, or implemented to the state regulatory agencies by July 31, 2018. Agency Comments and Third-Party Views We provided a draft of this report to BCFP, DHS, FTC, IRS, SSA, USPS, and Equifax for comment. SSA and USPS provided written responses expressing appreciation for the opportunity to review the draft report. The SSA and USPS responses are reprinted in appendices II and III, respectively. In addition, BCFP, DHS, FTC, IRS, SSA, USPS, and Equifax provided technical comments orally and via email, which we have incorporated, as appropriate. As agreed with your offices, unless you publicly announce the contents of this report earlier, we plan no further distribution until 29 days from the report date. At that time, we will send copies to the appropriate congressional committees, Equifax, and to the Acting Director of the Bureau of Consumer Financial Protection; the Chairman of the Federal Trade Commission; the Secretary of the Department of Homeland Security; the Commissioners of the Internal Revenue Service and Social Security Administration; and the Postmaster General of the United States. In addition, the report will be available at no charge on the GAO website at http://www.gao.gov.We are sending copies of this report to In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact Nick Marinos at (202) 512-9342 or marinosn@gao.gov, or Michael Clements at (202) 512-8678 or clementsm@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made major contributions to this report are listed in appendix IV. Appendix I: Objectives, Scope, and Methodology Our objectives were to (1) summarize the events regarding the 2017 Equifax breach and the steps taken by the company to assess, respond to, and recover from the incident and (2) describe the actions that federal customers and oversight agencies took in response to the breach. To address the first objective, we obtained and assessed documentation generated in response to the breach. Specifically, we analyzed the results of security assessments conducted by Equifax and its cybersecurity consultant following the breach, which included information about how the attacker gained access to Equifax’s systems and the specific vulnerabilities that were exploited. This documentation included the report summarizing the results of the consultant’s forensic analysis of Equifax systems and the consultant’s recommendations to Equifax to address the factors that led to the breach. We also reviewed Equifax’s relevant public filings to the Securities and Exchange Commission and statements it provided to the public and shareholders, which included information about the data breach and the company’s efforts for remediation. Further, we conducted a site visit to the Equifax data center in Alpharetta, Georgia, to interview knowledgeable officials, such as the interim Chief Security Officer and other officials knowledgeable about how Equifax stores and processes data, and observed physical security measures. In addition, to clarify details of the breach and the steps that Equifax took, we interviewed officials at Equifax who were responsible for coordinating reviews conducted following the breach. Specifically, we interviewed the interim Chief Security Officer and government relations employees, who were responsible for coordinating Equifax’s interaction with federal agencies in response to the incident. We did not independently assess Equifax’s information security controls or the steps the company took to address identified factors that contributed to the ineffective implementation of those controls. Specifically, the scope of our report was to report on actions taken by Equifax and agencies in response to the breach. Consequently, the information in this report is based on public filings and announcements as well as information provided to us by the company. We did not reach conclusions regarding the adequacy or efficacy of Equifax’s security measures. To address the second objective, we selected three major federal agencies, Internal Revenue Service (IRS), Social Security Administration (SSA), and United States Postal Service (USPS), which were Equifax’s largest federal customers at the time of the breach. We initially identified these customer agencies by reviewing public reports following the breach that identified federal agencies that were major Equifax customers at the time. We also interviewed Equifax officials responsible for managing government accounts to confirm that these three agencies were the only large-scale federal customer agencies that interacted with Equifax following the breach. Other federal agencies also have contracts with Equifax for a variety of services; we did not conduct audit work for this engagement at any other agencies because we narrowed our selection criteria to the largest federal agencies that used Equifax’s services to conduct their identity-proofing processes. Subsequently, we analyzed documentation from IRS, SSA, and USPS to describe the relevant actions these agencies took in response to the breach, as well as documentation regarding oversight by BCFP and FTC, which are the federal agencies with primary oversight responsibilities over CRAs. Specifically, we reviewed relevant laws and BCFP guidance on data security examinations. In addition, we spoke with BCFP and FTC officials about their actions in response to the data breach and reviewed their websites for information provided to consumers. We also selected and reviewed contracts between Equifax and each of the three selected agencies—IRS, SSA, and USPS—to determine what changes were made to services, such as identity-proofing solutions, provided by Equifax to federal agencies as a result of the breach. The contracts we reviewed were the ones identified by IRS, SSA, and USPS as contracts with Equifax for credit reporting or identity-proofing services. Further, we conducted interviews with agency officials at BCFP, FTC, DHS, IRS, SSA, and USPS to determine what actions customer and oversight agencies took in response to the breach. The officials we interviewed were responsible for conducting their agencies’ security assessment of Equifax at the time of the data breach. These included officials at each agency that had a role in responding to the Equifax breach, such as investigators at the oversight agencies and information security officials at the federal customer agencies. To address both objectives, and to identify how federal requirements apply to credit reporting agencies, we analyzed relevant federal laws to determine the responsibilities of agencies and their contractors. Specifically, we reviewed the following laws: Dodd-Frank Wall Street Reform and Consumer Protection Act; Fair Credit Reporting Act; Privacy Act of 1974; and E-Government Act of 2002. We conducted this performance audit from November 2017 to August 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Comments from the Social Security Administration Appendix III: Comments from the United States Postal Service Appendix IV: GAO Contacts and Staff Acknowledgments GAO Contacts Staff Acknowledgments In addition to the individuals named above, John de Ferrari and John Forrester (assistant directors); Tina Torabi (analyst-in-charge); Bethany Benitez, Chris Businsky, Kavita Daitnarayan, Nancy Glover, Andrea Harvey, Thomas Johnson, David Plocher, Tovah Rom, Rachel Siegel, and Winnie Tsen made key contributions to this report.

Why GAO Did This Study CRAs such as Equifax assemble information about consumers to produce credit reports and may provide other services, such as identity verification to federal agencies and other organizations. Data breaches at Equifax and other large organizations have highlighted the need to better protect sensitive personal information. GAO was asked to report on the major breach that occurred at Equifax in 2017. This report (1) summarizes the events regarding the breach and the steps taken by Equifax to assess, respond to, and recover from the incident and (2) describes actions by federal agencies to respond to the breach. To do so, GAO reviewed documents from Equifax and its cybersecurity consultant related to the breach and visited the Equifax data center in Alpharetta, Georgia, to interview officials and observe physical security measures. GAO also reviewed relevant public statements filed by Equifax. Further, GAO analyzed documents from the IRS, SSA, and USPS, which are Equifax's largest federal customers for identity-proofing services, and interviewed federal officials related to their oversight activities and response to the breach. What GAO Found In July 2017, Equifax system administrators discovered that attackers had gained unauthorized access via the Internet to the online dispute portal that maintained documents used to resolve consumer disputes (see fig.). The Equifax breach resulted in the attackers accessing personal information of at least 145.5 million individuals. Equifax's investigation of the breach identified four major factors including identification, detection, segmenting of access to databases, and data governance that allowed the attacker to successfully gain access to its network and extract information from databases containing personally identifiable information. Equifax reported that it took steps to mitigate these factors and attempted to identify and notify individuals whose information was accessed. The company's public filings since the breach occurred reiterate that the company took steps to improve security and notify affected individuals. The Internal Revenue Service (IRS), Social Security Administration (SSA), and U.S. Postal Service (USPS)—three of the major federal customer agencies that use Equifax's identity verification services—conducted assessments of the company's security controls, which identified a number of lower-level technical concerns that Equifax was directed to address. The agencies also made adjustments to their contracts with Equifax, such as modifying notification requirements for future data breaches. In the case of IRS, one of its contracts with Equifax was terminated. The Department of Homeland Security offered assistance in responding to the breach; however, Equifax reportedly declined the assistance because it had already retained professional services from an external cybersecurity consultant. In addition, the Bureau of Consumer Financial Protection and the Federal Trade Commission, which have regulatory and enforcement authority over consumer reporting agencies (CRAs) such as Equifax, initiated an investigation into the breach and Equifax's response in September 2017. The investigation is ongoing. What GAO Recommends GAO is not making recommendations in this report. GAO plans to issue separate reports on federal oversight of CRAs and consumer rights regarding the protection of personally identifiable information collected by such entities. A number of federal agencies and Equifax provided technical comments which we incorporated as appropriate.

gao_GAO-19-136

gao_GAO-19-136_0

Background Software development approaches have evolved over time. DOD weapon system acquisition programs have traditionally developed software using what is known as the waterfall development approach, first conceived in 1970 as linear and sequential phases of development over several years that result in a single delivery of capability. Figure 1 depicts an overview of the waterfall approach. Within industry, software development has evolved with the adoption of newer approaches and tools. For example, while a traditional waterfall approach usually is often broadly scoped, multiyear, and produces a product at the end of a sequence of phases, an incremental approach delivers software in smaller parts, or increments, in order to deliver capabilities more quickly. This development technique has been preferred for acquiring major federal IT systems, to the maximum extent practicable, and in OMB guidance since at least 2000. In addition, iterative development promotes continual user engagement with more frequent software releases to users. Figure 2 shows an overview of incremental and iterative development. DevOps is a more recent type of software development first used by industry around 2009. According to the Defense Innovation Board, DevOps represents the integration of software development and software operations, along with the tools and culture that support rapid prototyping and deployment, early engagement with the end user, and automation and monitoring of software. Figure 3 shows a notional representation of the DevOps approach based on DOD and industry information. There are also a variety of other software development approaches. Incremental, Iterative, and DevOps approaches are further described as follows: Incremental development sets high level requirements early in the effort, and functionality is delivered in stages. Multiple increments deliver a part of the overall required program capability. Several builds and deployments are typically necessary to satisfy approved requirements. DOD guidance for incremental development for software-intensive programs states that each increment should be delivered within 2 years, and OMB guidance issued pursuant to FITARA requires delivery of software for information technology investments in 6-month increments. Iterative development takes a flexible approach to requirements setting. In this approach, requirements are refined in iterations based on user feedback. We include Agile development approaches in this category of development; although most Agile approaches include aspects of both iterative and incremental development, as shown in figure 4. The Agile approach was first articulated in 2001 in what is known as the Agile Manifesto. The Agile Manifesto states the importance of four values: (1) individuals and interactions over processes and tools, (2) working software over comprehensive documentation, (3) customer collaboration over contract negotiation, and (4) responding to change as opposed to following a pre-set plan. Approaches that share common Agile principles include: Scrum, Extreme Programming, and Scaled Agile Framework, among others. These approaches stress delivering the most value as early as possible and constantly improving it throughout the project lifecycle based on user feedback. Within industry, Agile development approaches typically complete iterations within 6 weeks, and deliver working software to the user at the end of each iteration. According to DOD and industry, iterative development approaches have led to quicker development at lower costs and have provided strategic benefit through rapid response to changing user needs. DevOps is a variation of Agile that combines “development” and “operations,” emphasizing communication, collaboration, and continuous integration between both software developers and users. According to the Software Engineering Institute, DevOps is commonly seen as an extension of Agile into the operations side of the process, implementing continuous delivery through automated pipelines. In general, all stakeholders—including operations staff, testers, developers, and users—are embedded on the same team from the project’s inception to its end, ensuring constant communication. Automated deployment and testing is used instead of a manual approach, and the developer’s working copies of software are synchronized with the users. Software code is continuously integrated and delivered into production or a production-like environment. According to industry reports, the use of DevOps may lower costs due to immediate detection of problems as well as result in a greater confidence in the software because the users have continuous visibility into development, testing, and deployment. According to DOD officials from the Undersecretary of Defense, Research and Engineering, adopting Agile and DevOps within DOD weapon system acquisitions—which includes DOD space programs—is challenging and requires programs to adopt comprehensive strategies that cover broad topics. Officials said these strategies should include plans for cultural adoption by the program office and contractor; training and certification for program office and contractor personnel; and tools, metrics, and processes that support continuous integration and delivery, among others. Collaboration between Developers and Users Is Key to Reducing Program Risk While there are a variety of approaches to developing software, involving users in early stages and throughout software development helps detect deficiencies early. Industry studies have shown it becomes more expensive to remove conceptual flaws the later they are found. Previous GAO reports as well as other DOD and industry studies have also found that user involvement is critical to the success of any software development effort. For example, we previously reported that obtaining frequent feedback is linked to reducing risk, improving customer commitment, and improving technical staff motivation. We also previously reported that two factors critical to success in incremental development were involving users early in the development of requirements and prior to formal end-user testing. In the Fiscal Year 2010 NDAA, Congress directed DOD to develop and implement a new acquisition process for information technology systems that, among other things, include early and continuous involvement of the user. This statute, in addition to DOD’s 2010 report to Congress in response to the statute, and DODI 5000.02 identify characteristics of effective user engagement for DOD acquisitions, including: Early engagement: Users are involved early during development to ensure that efforts are aligned with user priorities. Continual engagement: Users are involved on a regular, recurring basis throughout development to stay informed about the system’s technical possibilities, limitations, and development challenges. Feedback based on actual working software: User feedback during development is based on usable software increments to provide early insight into the actual implementation of the solution and to test whether the design works as intended. Feedback incorporated into subsequent development: User feedback is incorporated into the next build or increment. Software Enables Operational Capability in All Segments of Space Systems Defense space systems typically consist of multiple segments: one or more satellites, ground control systems, and, in some cases, terminals for end-users. Each segment depends on software to enable critical functionality, such as embedded software in satellite vehicles, in applications installed on computer terminals in ground control stations, or embedded signal processing software in user terminals to communicate with satellites, shown in figure 5. Selected Software- Intensive Space Systems Have a History of Cost Growth and Schedule Delays We have previously reported on significant cost growth and schedule delays in numerous DOD space systems, with some space program costs rising as much as 300 percent, and delays so lengthy that some satellites spend years in orbit before key capabilities are able to be fully utilized. In particular, the programs described below have experienced significant software challenges, including addressing cybersecurity requirements, which have contributed to cost growth and schedule delays. Joint Space Operations Center (JSpOC) Mission System Increment 2 (JMS) The Air Force's JMS program aims to replace an aging space situational awareness and command and control system with improved functionality to better track and catalogue objects in the earth's orbit to support decision making for space forces. Increment 2 is to replace existing systems and deliver additional mission functionality. The Air Force is providing this functionality in three deliveries: the first delivery—Service Pack 7—provided hardware and software updates and was delivered in September 2014; the second delivery—Service Pack 9—aims to improve functions currently being performed, such as determining space object orbits and risks of collision; and the final delivery—Service Pack 11— aims to provide classified functionality. The government is serving as the system integrator directly managing the integration of government and commercially developed software onto commercial, off-the-shelf hardware, so there is no prime contractor. Historical software development challenges include: In 2015, we found that inconsistencies in the program’s software development schedule made it unclear whether the program would be able to meet its remaining milestones. The same year, the program declared a schedule breach against its baseline due, in part, to delays in resolving deficiencies identified during software testing. In 2016, DOD noted that the revised schedule was still highly aggressive with a high degree of risk because the program was concurrently developing and testing software. In 2017, developmental tests found a number of mission critical software deficiencies, which delayed operational testing. The Director of Operational Test and Evaluation also noted that additional work remained to help provide adequate cyber defense for JMS. During operational testing in 2018, JMS was found not operationally effective and not operationally suitable due, in part, to missing software requirements, urgent deficiencies that affected system performance, and negative user feedback. Mobile User Objective System (MUOS) The Navy’s MUOS program aims to provide satellite communications to fixed and mobile terminal users with availability worldwide. MUOS includes a satellite constellation, a ground control and network management system, and a new waveform for user terminals. The ground system includes the ground transport, network management, satellite control, and associated infrastructure to both operate the satellites and manage the users’ communications. The MUOS constellation is complete, and, according to program officials, software development officially ended in 2012 with the delivery of the waveform software. However, the user community still cannot monitor and manage MUOS. MUOS has two types of users: ground operators responsible for managing the MUOS communications network, and the military users of radios. Space and Missile Defense Command / Army Forces Strategic Command (SMDC/ARSTRAT) was the user representative while MUOS was developed. While DOD allowed the program to move into sustainment—the phase after development is formally completed—the program continues to resolve challenges with the ground segment, and the contractor continues to deliver software updates to address deficiencies. In 2017, the program transitioned its software sustainment efforts to an Agile development approach in preparation for a follow-on operational test currently scheduled to begin in June 2019. While Lockheed Martin Space Systems is the prime contractor for MUOS, we evaluated software efforts conducted by General Dynamics, the subcontractor performing software development. Historical software development challenges include: In 2014, DOD found that 72 percent of the software was obsolete. Also in 2014, operational testing was delayed due to software reliability issues in the ground system and waveform. In 2015, we found that over 90 percent of MUOS’ planned capability was dependent on resolving issues related to integrating the MUOS waveform, terminals, and ground systems. Also in 2015, operational tests determined MUOS was not operationally effective, suitable, or survivable due in part to cybersecurity concerns in the ground system. As of 2016, there were still existing and emerging cybersecurity vulnerabilities to be addressed. Lockheed Martin Space Systems (Prime) General Dynamics (Software development subcontractor) Contract Type: Cost Plus Incentive and Award Fee/Fixed Price Incentive (Firm Target) and Award Fee Naval Computer and Telecommunications Area Master Station Pacific (NCTAMS PAC) Space and Missile Defense Command / Army Forces Strategic Command (SMDC/ARSTRAT) Next Generation Operational Control System (OCX) The Air Force’s OCX program is designed to replace the current ground control system for legacy and new GPS satellites. OCX software is being developed in a series of blocks: Block 0 is planned to provide the launch and checkout system and support initial testing of GPS III satellites and cybersecurity advancements. Blocks 1 and 2 are planned to provide command and control for previous generations of satellites and GPS III satellites as well as monitoring and control for current and modernized signals. The OCX contractor delivered Block 0 in September 2017. The Air Force took possession of Block 0 in October 2017 by signing a certificate of conformance, and will accept it at a later date after Block 1 is delivered. Historical software development challenges include: In 2013, DOD paused OCX development due to incomplete systems engineering, which led to continuous rework and deferred requirements. In 2015, we reported that, among other things, OCX had significant difficulties related to cybersecurity implementation. In 2016, the program declared a Nunn-McCurdy unit cost breach. Also in 2016, the contractor began implementing DevOps at the recommendation of Defense Digital Service but, according to the program office and contractor, only planned to automate development without the operations component of DevOps. The contractor did not achieve initial planned schedule efficiencies. In 2017, the Air Force accepted Block 0 despite over 200 open software defects. According to the program, when Block 0 was accepted there was also a plan to resolve the open software defects by the time of the first launch. Since then, according to the program office, all necessary defects related to launch have been addressed. In 2018, DOD noted that the schedule was at risk since the program made aggressive assumptions in its plan to develop, integrate, test software, and resolve defects. Space-Based Infrared System (SBIRS) The Air Force’s SBIRS program is an integrated system of both space and ground elements that aim to detect and track missile launches. SBIRS is designed to replace or incorporate existing defense support ground stations and satellites to improve upon legacy system timeliness, accuracy, and threat detection sensitivity. The Air Force is delivering the SBIRS ground system in one program with two increments: the first increment became operational in 2001 and supports functionality of existing satellites. The second increment, which is still in development, is designed to provide new space segments, mission control software and hardware, and mobile ground capability. The Air Force is delivering these capabilities in multiple blocks: Block 10 was accepted in 2016 and introduced new ground station software and hardware. Block 20 is expected to be complete by late 2019 and is planned to further improve ground station software. Historical software development challenges include: In 2001, 2002, and 2005, cost increases and schedule delays due, in part, to software complexity problems led to four separate Nunn- McCurdy unit cost breaches. In September 2007, we found that the amount of rework resulting from unresolved software discrepancies was contributing to cost growth and schedule delays. In addition, the program had software algorithms that were not yet completed or demonstrated, hundreds of open deficiency reports, and a lack of coordination between space and ground system software databases. In 2016, DOD said that software deficiencies were contributing to delays in delivering the ground architecture. In 2018, DOD noted that flight software development remained a concern to the overall program schedule. According to SBIRS users and the program office, cybersecurity issues found during Block 10 testing are still being addressed as a part of the Block 20 effort. DOD programs we reviewed frequently did not involve users early or continually during development, base user feedback on actual working software, or incorporate user feedback into subsequent software deliveries. Most programs had plans to incorporate these elements of user engagement throughout their software development efforts, but they often did not follow those plans due, in part, to the lack of specific guidance on user involvement and feedback. Regarding frequency of software delivery, while DODI 5000.02 suggests that programs deliver incremental software deliveries every 1 to 2 years, the programs we reviewed often continued to deliver software consistent with the long delivery schedules common to waterfall development. DOD is taking steps to address this issue. Selected DOD Programs Often Did Not Effectively Engage Users The four programs we reviewed often did not demonstrate key characteristics of effective user engagement as summarized below: Early engagement. OCX involved users early and JMS planned to involve users early but, in practice, did not do so; SBIRS and MUOS did not plan to involve users early in software development. Continual engagement. JMS, OCX, and SBIRS all planned to continually involve users but, in practice, did not fully do so; MUOS did not plan to do so. Feedback based on actual working software. OCX and SBIRS have provided users opportunities to provide such feedback but only years into software development; JMS and MUOS did not provide opportunities for feedback. Feedback incorporated into subsequent development. JMS, OCX, and SBIRS all planned to incorporate user feedback but, in practice, have not done so throughout development; MUOS did not plan to do so during software development. Program efforts to involve users often did not match what their planning documentation described. In addition, when user input was collected, program officials did not capture documentation of how user feedback was addressed. Further, we found that, in practice, none of the programs we reviewed had users providing feedback on actual working software until years after system development began. This was the case even for programs utilizing Agile or iterative-incremental software development approaches, where user involvement and feedback from using functional systems early in the development cycle is foundational. These shortcomings were due, in part, to the lack of specific guidance on user involvement and feedback. Both DODI 5000.02 and DOD’s guiding principles for delivering information technology acquisitions note that software should be developed via usable software deliveries to obtain user acceptance and feedback for the next segment of work, but this guidance lacks specificity. In particular, DOD does not specify when to involve users and request their feedback, how frequently to seek user involvement and feedback on software deliverables, how to report back to users on how that feedback was addressed, and how to document the results of user involvement and feedback. As a result of programs’ shortcomings with user involvement and feedback, programs risk delivering systems that do not meet user needs. In selected cases, delivered software was deemed operationally unsuitable by DOD testers and required substantial rework. Further details on the extent to which programs implemented the four key characteristics are described below. JMS: Program documents created at the start of JMS system development contain specific operating procedures for conducting interactions with the user community—Air Force personnel who track and catalogue objects in orbit—during acquisition and fielding. However, the program has not followed these operating procedures during system development. Early Engagement. The JMS program office planned to involve users early in development but, in practice, did not do so. JMS program documentation states that users were to be involved in user engagement sessions within the first 4 weeks of iterative development. However, the first documented user engagement session was held more than a year after development start. Continual Engagement. The JMS program office planned to engage users throughout development but, in practice, did not do so. JMS program documentation states that user engagement sessions are to be held regularly during development—roughly every 2 to 4 weeks. However, in practice, program officials told us they only involved users as needed during software development. We found that the frequency of user engagement events varied from several weeks to more than 6 months. According to program officials, there were limited users available, and their operational mission duties were prioritized over assisting with system development. Feedback Based on Actual Working Software. The JMS program office did not provide users an opportunity to give feedback based on actual working software during development. According to program documentation, designs and notional drawings, not working software, were to be used for user engagement sessions. While JMS did provide users opportunities to provide feedback, this feedback was not on actual working software. Program officials said the goal of these events was never intended to include user feedback on actual working software. However, users told us that when they were finally able to use the system for the first time, 4 years after development started, it did not function as needed. The software did not execute what it had been designed to do, and earlier user engagement on actual working software may have identified these issues. Feedback Incorporated Into Subsequent Development. The JMS program office planned to incorporate user feedback into development but, in practice, did not do so. JMS program documentation states that the program will document user feedback from user engagement events using summary notes communicated back to the user. However, JMS users said it was often unclear if their feedback was incorporated. For example, in March 2016, a user engagement event was held to discuss any questions and concerns relating to the planned system’s conjunction assessment—a key feature that predicts orbit intersection and potential collision of space objects— that resulted in 8 user-identified issues. When we met with the users in 2018, they told us that conjunction assessment issues remained unaddressed, and they would still be reliant on the legacy system to fully execute the mission and perform their duties. The legacy system is still needed, they said, because the program deferred critical functions, and the most recent operational test found the system to be operationally unsuitable. MUOS: The MUOS program office did not engage users—Army Forces Strategic Command personnel who support the narrowband and wideband communications across the Air Force, Marines, Navy, and Army—during software development but are engaging users while developing software during sustainment, the acquisition phase after development when the program mainly supports and monitors performance. Following the end of development, at an operational test event in 2015, DOD testers deemed the system was operationally unsuitable. The MUOS program office moved to an Agile development approach in 2017 to address software deficiencies in preparation for the next operational test event. Early Engagement. The MUOS program office did not engage users early in development. Program documentation does not describe any plans for user engagement or involvement during development and, according to program officials, no users evaluated the actual system during development. Continual Engagement. The MUOS program office did not continually engage with users. Program documentation does not describe any plans for user engagement or involvement during development. Program officials said no users evaluated the system during development because there were no users with real world experience on a system like MUOS. However, as previously noted, SMDC/ARSTRAT represented end users’ interests during MUOS development. Feedback Based on Actual Working Software. The MUOS program office did not provide users an opportunity to give feedback based on actual working software. Program documentation does not describe a process for obtaining user feedback based on actual working software. The first time users had a chance to fully operate the system was after development ended, in preparation for operational testing in 2014, which identified numerous defects. Additionally, MUOS users said that they have since identified 128 functions in 11 critical areas that must be addressed or they will not accept the system. Users also said that some of the vulnerabilities found during operational testing, including cybersecurity vulnerabilities, have been deferred. Feedback Incorporated Into Subsequent Development. The MUOS program office did not incorporate user feedback into development. Program documentation did not describe plans to gain user feedback or acceptance into the development of the MUOS system. In addition, users and the contractor told us that program officials did not allow direct interaction during development due to a concern that such interactions could lead to changes in system requirements. The program office said that user involvement to-date has not caused delays to testing or software delivery. OCX: The OCX program had limited user engagement, but has recently held user engagement events based on releases of actual working software. The program has made efforts to obtain feedback from users, but users have noted there is no time in the schedule to address much of their feedback prior to delivering the system. Early Engagement. The OCX program office involved users early in development in accordance with its plans. From 2011, OCX users were involved in technical meetings where they provided feedback on the concept of operations and the design of the system. Continual Engagement. The OCX program office planned to engage users throughout development but, in practice, did not fully do so. OCX planning documentation includes multiple opportunities for user engagement at various stages of system development, including operational suitability and “hands-on” interaction with an integrated system. According to the program office, numerous events were held for users to give feedback on the system. However, since 2012, the program has only held one of its planned events to address operational suitability. In addition, other opportunities for users to operate the system have been removed to accommodate the program’s schedule, such as “day in the life” events that allowed users to validate the system as they would actually operate it. Users said that removing events like these created fewer opportunities to identify and resolve new deficiencies. Feedback Based on Actual Working Software. OCX did not plan to provide users an opportunity to give feedback based on actual working software but, in practice, did so years into development. OCX planning documents rely on simulations and mock-ups for evaluating system usability. However, users told us that mock-ups do not allow them to test functionality and may not be representative of the final delivered product. Starting in 2014—2 years after development started—users had opportunities to review the limited functionality available at the time. Since 2017, users said they were able to test working software. Feedback Incorporated Into Subsequent Development. The OCX program office planned to incorporate user feedback into development but, in practice, did not do so throughout development. OCX planning documentation includes a user comment response process that would collect and validate user comments and communicate results back to the users. According to the program office, for OCX Block 0, users provided feedback that was incorporated prior to the first launch. While OCX users said that they have the opportunity to provide feedback, there is a growing list of unaddressed Block 1 issues to be resolved. Some of these feedback points, if left unresolved, may result in operational suitability concerns and a delayed delivery to operations. According to the program office, critiques from the users have either been closed, incorporated into the OCX design, or are still under assessment between the contractor and users. A majority of user feedback points for the OCX iteration currently in development remain unresolved, as depicted in figure 6. In 2016, DOD told the Air Force and the contractor to utilize DevOps. As previously noted, DevOps is intended to release automated software builds to users in order to unify development and operations and increase efficiency. The contractor stated it implemented DevOps in 2016. However, both the Air Force and the contractor admitted in 2018 they never had plans to implement the “Ops” side of DevOps, meaning they didn’t plan to automatically deliver software builds to the users. Without incorporating the users and experts in maintainability and deployment, the program is not benefiting from continuous user feedback. SBIRS: SBIRS users—Air Force personnel who operate, command, and control SBIRS satellites to detect and track missile launches—were not involved during early system development and the program only recently increased the frequency of user events. SBIRS users have been able to provide feedback on working software but are unaware how this feedback is incorporated into software development. Early Engagement. The SBIRS program office did not engage users early in development because users were not in place and user groups were not defined. The program planning documentation that instituted the framework for user involvement was not in place until 2004. According to SBIRS users and test officials, this resulted in a poor interface design and users being unable to respond adequately to critical system alerts when using the system. Though the program contractor told us that user involvement is critical for ensuring the developers deliver a system that users need and will accept, DOD officials said that users were not integrated with the development approach until the software was ready to be integrated into a final product. Continual Engagement. The SBIRS program office planned to engage users throughout development but, in practice, did not do so. SBIRS planning documentation includes users involved in regular working groups throughout development. SBIRS users began to be involved with system development in 2013 on a weekly basis. Users were not involved during the 17 years of system development prior to this time. Feedback Based on Actual Working Software. The SBIRS program did not plan to provide users an opportunity to give feedback based on actual working software during development but, in practice, did so years into development. SBIRS documentation only outlines user engagement as reviewing and commenting on design plans. While users were able to provide feedback on working software in 2017, these events did not occur until 21 years after the start of development when the software was ready to be integrated. When users were able to provide feedback, they identified issues with the training system and cybersecurity. Feedback Incorporated Into Subsequent Development. The SBIRS program planned to incorporate user feedback into development but, in practice, did not do so. SBIRS planning documentation includes methods for users to provide feedback, but users said there is no feedback loop between them and the developers; therefore, users are unaware if their comments and concerns are addressed or ignored. Selected Programs Have Generally Not Delivered Software Frequently, but DOD Is Taking Steps to Improve Efforts DOD officials and DODI 5000.02 point to the benefits of delivering smaller packages of software more frequently, but the four programs we examined have generally delivered them infrequently. DOD is beginning to take steps to address these issues, such as establishing an independent advisory panel and considering recommendations issued by the Defense Science Board on the design and acquisition of DOD software. Selected programs continue to focus on infrequent deliveries. According to industry practices, short, quick deliveries allow a program to deliver useful, improved capabilities to the user frequently and continually throughout development. Within industry, iterations for Agile development approaches are typically up to 6 weeks, and working software is delivered to the user at the end of each iteration. In addition, DODI 5000.02 states that for incremental development increments should be delivered within 2 years. While two programs in our review—JMS and MUOS—say they have undertaken elements of Agile development, which emphasize smaller deliveries of frequent software to users, they still struggled to move away from the long delivery schedules common to waterfall development. In addition, the two programs with incremental development—OCX and SBIRS—have not delivered within suggested DOD time frames. See figure 7 below for program software deliveries. Further observations on each of the four programs follow: JMS program officials and documentation indicate that the program is using an Agile development approach to deliver smaller, rapid deliveries to minimize risk. According to JMS program documentation, software releases were to be delivered in 6-month intervals. However, the program only delivered actual working software once during development—a delivery of capability in 2014. The program was operationally accepted in late 2018. However, only 3 of 12 planned capabilities were accepted for operational use. The MUOS program used a traditional waterfall approach during development from 2004 to 2012 and has only had one overall software product delivery during that time. The program completed the software in 2012, yet continued to make changes during sustainment using the waterfall methodology and adopted an Agile approach in 2017 to address deficiencies. Since this adoption, it has delivered software more frequently—about every 3 months. This is a significant improvement over the delivery time frames during the MUOS waterfall development approach. The OCX program is using an “iterative-incremental” development approach. According to OCX software development plans, this approach was to enable early and frequent deliveries of capabilities. Specifically, the program plans for iterations to be completed every 22 weeks. However, since software development began in 2012, OCX has delivered just one increment of software, referred to by the OCX program as a block. The SBIRS program began in 1996, using a waterfall approach, and has had two deliveries of software. SBIRS Increment 1 was delivered in 2001, and the next increment, SBIRS Increment 2, Block 10, was delivered 15 years later, in 2016. The next increment, SBIRS Increment 2, Block 20, is expected to be delivered in 2019. Part of the reason programs delivered larger software packages less frequently was the adherence to the process steps in the DODI 5000.02 that were designed under the waterfall approach. While DODI 5000.02 authorizes programs to tailor their acquisition procedures to more efficiently achieve program objectives, none of the programs that were trying to employ a newer development approach took steps to tailor procedures in order to facilitate development. For example, the OCX contractor said it was delayed by complying with technical reviews under a military standard for traditional waterfall approaches, such as the Preliminary Design Review, Critical Design Review, and others, but the OCX program did not alter these reviews, despite having flexibility to do so. The contractor told us a more tailored approach would enable execution of smaller iterations of software deliverables. Similarly, the JMS program office noted that it was not fully able to integrate Agile development practices because of all the different technical reviews, but JMS did not tailor these requirements to more efficiently achieve outcomes, despite flexibility to do so. DOD officials have acknowledged these challenges and have recently begun recommending steps to address them. Officials we spoke with from Defense Digital Service, Director of Operational Test and Evaluation, and DOD leadership said that rapid development of software using newer software practices does not fit with the requirements of the DOD acquisition process. Further, DOD’s Special Assistant for Software Acquisition said that DOD software development should be iterative, providing the critical capabilities in smaller, more frequent deliveries rather than delivering capabilities in a single delivery via traditional waterfall software development. In addition, other DOD officials we interviewed agreed that since DOD programs may not always know the full definition of a system’s requirements until late in development, additional flexibility to tailor acquisition approaches could improve software acquisitions. In acknowledging the challenges in moving from a waterfall model to a more incremental approach, various DOD groups have made recommendations to support delivery of smaller, more timely software deliverables: In February 2018, the Defense Science Board issued a series of recommendations to support rapid, iterative software development. The recommendations included requiring all programs entering system development to implement iterative approaches and providing authority to the program manager to work with users. In April 2018, the Defense Innovation Board made recommendations to improve DOD software acquisitions, such as moving to more iterative development approaches that would deliver functionality more quickly. In June 2018, the DOD Section 809 Panel recommended eliminating the requirements for Earned Value Management (EVM)—one of DOD’s primary program planning and management tools—in Agile programs. However, other DOD and industry guides state that Agile programs can still report EVM if certain considerations are made, such as an Agile work structure that provides a process for defining work and tracking progress of this work against planned cost and schedule. Pursuant to the Fiscal Year 2019 National Defense Authorization Act, DOD is required, subject to authorized exceptions, to begin implementation of each recommendation submitted in the final report of the Defense Science Board Task Force on the Design and Acquisition of Software for Defense Systems by February 2020. For each recommendation that DOD is implementing, it is to submit to the congressional defense committees a summary of actions taken; and a schedule, with specific milestones, for completing implementation of the recommendation. We intend to monitor DOD’s progress in implementing the recommendations. Selected Program Offices Have Had Software-Specific Management Challenges but Are Taking Steps to Address Weaknesses The programs we reviewed faced management challenges using commercial software, applying outdated software tools and metrics, and having limited knowledge and training in newer software development. DOD is taking steps to address these challenges. Selected DOD Programs Face Difficulties Identifying the Effort Required by and Mitigating the Risk Associated with Commercial Software DOD has previously encouraged DOD acquisition programs to use commercial software where appropriate. For example, in 2000 and in 2003, DOD policy encouraged considering the use of commercial software. In addition, regulations continue to emphasize consideration of commercial software suitable to meet the agency’s needs in acquiring information technology. DOD officials said that, although the effort to maintain commercial software may be equivalent to developing such capabilities in-house, programs should still consider the use of commercial software because DOD and its contractors may lack the technical skillsets to develop a similar product. However, three of the programs we reviewed had difficulty integrating and maintaining modified commercial software during development: The JMS acquisition approach was to only use commercial and government-provided software with no new software development planned, but the commercial products selected were not mature and required additional development, contributing to schedule delays. The MUOS program underestimated the level of effort to modify commercial software, which increased cost and introduced schedule delays in completing both the ground system and the waveform. According to an Aerospace official who advised the program on software issues, the MUOS software development approach was to use a commercial software solution but with substantial modifications. In particular, the MUOS contractor planned to take a commercial cellular system and substantially modify it for MUOS. This official, along with the MUOS program office, said that underestimating the level of effort to modify and integrate the commercial software has been the program’s biggest challenge. In September 2015, we found that the OCX contractor was overly optimistic in its initial estimates of the work associated with incorporating open source and reused software. Further, according to the Air Force, OCX program managers and contractors did not appear to follow cybersecurity screening or software assurance processes as required. For example, open source software was incorporated without ensuring that it was cybersecurity-compliant. These problems led to significant rework and added cost growth and schedule delays to address the cybersecurity vulnerabilities and meet cybersecurity standards. In addition, in an independent assessment of OCX, officials from the MITRE Corporation said that there is a lack of appreciation for the effort required for commercial software integration, stating that the level of effort is “categorically underestimated.” Some program officials noted that commercial software updates led to system instability and increased costs. For example, OCX program officials said that updating an operating system version led to 38 other commercial software changes. Each of these changes had to be configured, which took considerable time and added cost to the program. Similarly, the SBIRS contractor said they have been concerned that updates to commercial software could create a domino effect of instability, and the risks could outweigh the benefits of the update. For example, if one commercial software product is updated and becomes unstable, instability may be introduced to other commercial software products and software components. On the other hand, not updating software products could lead to cybersecurity concerns. As we previously noted, developers of commercial software generally update software to address identified flaws and cybersecurity vulnerabilities. We also reported in a review of weapon systems cybersecurity that, although there are valid reasons for delaying or forgoing weapon systems patches, this means some weapon systems are operating, possibly for extended periods, with known vulnerabilities. In addition, the lifecycles of commercial software can contribute to management challenges when these products become obsolete. For example, in 2014, a MUOS Ground System Deep Dive review identified that 72 percent of the MUOS software was considered to be obsolete. According to program officials, commercial software became obsolete before or soon after it was fielded, especially for operating systems and browsers, due to the long MUOS development cycle. Software obsolescence is also among the top risks of the OCX program and has contributed to additional costs during development. DOD officials and others have started to acknowledge challenges in using commercial software. For example, as we previously reported in 2018, DOD has stated that many weapon systems rely on commercial and open source software and are subject to any cyber vulnerabilities that come with them. While DOD states that using commercial software is a preferred approach to meet system requirements, some program officials we interviewed told us that the effort to modify and update commercial software is underestimated. DOD is working on helping programs understand commercial software risks. For example, in January 2018, DOD published a Guidebook for Acquiring Commercial Items. In addition, Defense Acquisition University offers several modules designed to address challenges in integrating commercial solutions. Selected DOD Programs Are Using Outdated Software Tools and Metrics but Are Updating Them Three of the DOD programs we reviewed have experienced challenges in using outdated software tools or identifying appropriate performance metrics as they transition to newer software development approaches. Contractors continue to rely upon outdated software tools and experience challenges. We found that three of the programs we reviewed used tools that are considered outdated and lack the flexibility needed for iterative development. Contractors for three of the four programs we reviewed have experienced software development challenges due to outdated tools: The SBIRS contractor uses a suite of tools that is considered outdated for newer commercial approaches. For example, one of these tools relies on a central database that, if corrupted, will stop development work and could take days or weeks to fix. According to the contractor, fixing this database has led to multiple periods of downtime and schedule delays. The MUOS contractor also uses a toolset that is considered outdated by commercial software development experts. The program moved to a newer Agile development approach in 2017 but has retained an older software development toolset. The MUOS contractor said they are heavily reliant on these tools for development and do not anticipate changing the toolset. The OCX contractor also uses tools that are considered outdated by commercial approaches. According to the contractor, these tools have been in place for many years, and switching over to a new set of tools would not be in the best interest of the program because it could be disruptive to ongoing development. Defense Digital Service experts said that a particular suite of tools used by the OCX contractor is outdated because the tools lack the flexibility needed for iterative development. Both MUOS and SBIRS contractors said that they have had to train new employees to use their outdated tools. For example, the SBIRS contractor told us that when new employees begin work on the SBIRS program, they already know how to use newer tools but have to be trained on the outdated tools used for SBIRS development. The SBIRS contractor said this has affected retention of its workforce in some cases, and the program has allocated funding to transition to newer tools in order to better recruit and retain personnel. What is Cloud-Based Testing? Cloud-based testing uses cloud computing environments to simulate an application’s real-world usage. According to international standards, cloud testing can lead to cost savings, improved testing efficiency, and more realistic testing environments. Two contractors have taken steps to update their software tools to increase automation and cloud-based testing but have not yet experienced the anticipated efficiencies: The OCX contractor is attempting to employ cloud-based testing and a DevOps approach. The contractor said it had to gain approval from the DOD Chief Information Office to employ commercial cloud-based testing for the unclassified portions of OCX but it has not gained similar approval for the classified portion. The SBIRS contractor is using a software testing tool that would allow for faster automated testing but is not yet realizing the full benefit of its use. The SBIRS testers did not use this tool in the way it was intended. Specifically, the contractor said that when the software was deployed to the testing environment, testers deactivated the software at the end of their shifts instead of allowing it to run continuously until the tests were complete. The contractor said the testers did this because there were concerns over unauthorized access to the system if no one was present. As a result, the contractor separated the tests into 8-hour segments rather than allowing the tests to run continuously, reducing the effectiveness and value of automated testing. The Defense Science Board, Defense Innovation Board, and others have recommended DOD use tools that enable the developers, users, and management to work together daily. As noted, DOD is required to begin implementation of the recommendations made in the Defense Science Board report. Software metrics are measurements which provide insight to the status and quality of software development. Metrics may not support newer development approaches. We have previously found that leading developers track software-specific metrics to gauge a program’s progress, and that traditional cost and schedule metrics alone may not provide suitable awareness for managing iterative software development performance. Three programs have faced challenges in identifying and collecting metrics that provide meaningful insight into software development progress: JMS planned to collect traditional software development metrics to measure software size and quality, as well as Agile metrics that provide insight into development speed and efficiency. However, officials from the JMS government integrator managing sub-contracts said they lack regular reporting of metrics and access to data from subcontractors that would allow them to identify defects early. These officials said this was a challenge because the program has to run its own quality scans at the end of each sprint instead of being able to identify defects on a daily basis. MUOS program officials were able to receive Agile metrics from the contractor when they transitioned to Agile development, but they lacked access to the source data, which they said hindered their ability to oversee development. OCX program officials said they plan to use performance-based metrics throughout the remainder of the program. However, the metrics may not adequately track performance as intended. The Defense Contract Management Agency reviewed OCX metrics, particularly those related to DevOps, and expressed concern that program metrics may only measure total defects that were identified and corrected but may not provide insight into the complexity of those defects. DOD is taking steps to identify useful software development metrics and ways to include them in new contracts. DOD is aware of challenges with metrics and is taking actions to address the issues. For example, the Defense Innovation Board is consulting with commercial companies to determine what metrics DOD should collect; and the Air Force’s Space and Missile Systems Center has tasked The Aerospace Corporation with examining how to apply software performance metrics in contracts for DOD space programs. DOD offices such as the Defense Science Board and DOD Systems Engineering, as well as several Federally Funded Research and Development Centers including the Software Engineering Institute and The Aerospace Corporation, have also attempted to identify new metrics in correlation with advances in software development approaches. Two Program Offices Lacked Newer Software Development Knowledge, but DOD Is Working to Improve Training Two program offices we reviewed experienced challenges due to limited software development knowledge: OCX experienced an extended period of inefficient processes because it lacked an understanding of newer approaches. According to Defense Digital Service, when the Office of Secretary of Defense advised the OCX program in May 2016, it discovered that neither the program office nor contractor had been aware of the benefits of automated testing. Defense Digital Service helped the OCX contractor automate a process that had been taking as long as 18 months to one in which the same process takes less than a day. If the program office had been aware of newer software approaches, it could have recognized these inefficiencies much earlier and avoided unnecessary schedule delays. The MUOS contractor lacked an “Agile advocate” in the program office, which undermined its ability to fully employ an Agile development approach. For example, even after the contractor adopted an Agile approach, the program office directed the contractor to plan out all work across software builds in order to maintain control over requirements—similar to a waterfall approach but inefficient in Agile. According to the Software Engineering Institute, without an Agile advocate in a program’s leadership, organizations tend to do a partial Agile or “Agile-like” approach. Program officials from the programs we reviewed said that while they have taken some software development training, more would be beneficial. The JMS program office said that there are external training courses available locally as well as trainings at Air Force’s Space and Missile Systems Center, but neither are required. JMS program officials said that, while specific software training has not been required for the program outside of Defense Acquisition University certifications, courses on managing software-intensive programs would have been beneficial. Similarly, Defense Contract Management Agency officials told us that OCX program officials would have benefited from more software development training. The MUOS program office said its training on software acquisition, software and systems measurement, software planning supportability and cost estimating, and software policies and best practices was sufficient, but the program office did not have newer software development training prior to transitioning to an Agile development approach. DOD is working to improve software acquisition training requirements and update them to reflect changes in the software development industry. For example, in 2017, the Defense Acquisition University introduced a course on Agile software development that includes how Agile fits into the overall Defense Acquisition System and how to manage an Agile software development contract. DOD told us it is also working with the Defense Acquisition University to help inform a course on DevOps automation. Conclusions Software is an increasingly important enabler of DOD space systems. However, DOD has struggled to deliver software-intensive space programs that meet operational requirements within expected time frames. Although user involvement is critical to the success of any software development effort, key programs often did not effectively engage users. Program efforts to involve users and incorporate feedback frequently did not match plans. This was due, in part, to the lack of specific guidance on the timing, frequency, and documentation for user involvement and feedback. The lack of user engagement has contributed to systems that were later found to be operationally unsuitable. Selected programs have also faced challenges in delivering software in shorter time frames, and in using commercial software, applying outdated software tools and metrics, and having limited knowledge and training in newer software development techniques. DOD acknowledges these challenges and is taking steps to address them. Recommendations for Executive Action We are making the following two recommendations to DOD: The Secretary of Defense should ensure the department’s guidance that addresses software development provides specific, required direction on when and how often to involve users so that such involvement is early and continues through the development of the software and related program components. (Recommendation 1) The Secretary of Defense should ensure the department’s guidance that addresses software development provides specific, required direction on documenting and communicating user feedback to stakeholders during software system development. (Recommendation 2) Agency Comments and Our Evaluation We provided a draft of this product to the Department of Defense for comment. In its comments, reproduced in appendix II, DOD concurred. DOD also provided technical comments, which we incorporated as appropriate. We are sending copies of the report to the Acting Secretary of Defense; the Secretaries of the Army, Navy, and Air Force; and interested congressional committees. In addition, the report will be available at no charge on GAO’s website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-4841 or ludwigsonj@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made major contributions to this report are listed in appendix III. Appendix I: Objectives, Scope, and Methodology Senate and House reports accompanying the National Defense Authorization Act for Fiscal Year 2017 contained provisions for GAO to review challenges in software-intensive Department of Defense (DOD) space systems, among other things. This report addresses, for selected software-intensive space programs, (1) the extent to which these programs have involved users and delivered software using newer development approaches; and (2) what software-specific management challenges, if any, these programs have faced. To select the programs, we identified a non-generalizable, purposeful sample of four major defense programs representing different space military services where software is an essential component and where each program has experienced cost growth or schedule delays attributed, in part, to software challenges. We began our selection process with 49 DOD space programs from the U.S. Air Force and Navy services as identified by the Office of the Assistant Secretary of the Air Force for Space Acquisition and a GAO subject matter expert. We then narrowed our selection to 19 Major Defense Acquisition Programs (MDAP) and Major Acquisition Information System (MAIS) programs identified by DOD. Next, using information from prior GAO Annual Weapons Assessments, DOD Selected Acquisition Reports, DOD Defense Acquisition Executive Summary Reports, and the Defense Acquisition Management Information Retrieval system, we identified 15 programs that were software-intensive systems as defined in the international standard ISO/IEC/IEEE 42207. This standard states that a software- intensive system is one where software contributes essential influences to the design, construction, deployment, and evolution of the system as a whole. From these 15 programs, 8 were found to have had cost growth or schedule delays attributed, in some part, to software development. We further analyzed these 8 programs for unit cost or schedule breaches as defined in 10 U.S.C. § 2433 and 10 U.S.C. § 2366b, ultimately resulting in 7 programs. Finally, from these 7 programs, we chose a purposeful sample of 5 programs, ensuring representation from different DOD services and Acquisition Categories. Family of Advanced Beyond Line-of-Sight Terminals (FAB-T); Air Next Generation Operational Control System (OCX); Air Force MDAP Joint Space Operations Center Mission System Increment 2 (JMS); Air Force MAIS Mobile User Objective System (MUOS); Navy MDAP Space-Based Infrared System (SBIRS); Air Force MDAP We were unable to assess FAB-T software issues with the same level of detail as the other programs we reviewed because, despite prior software challenges, the program stated it does not have documentation that separately tracks software-related requirements or efforts. This brought our total to 4 selected programs. To address the objectives, we interviewed officials from the Undersecretary of Defense for Acquisition and Sustainment, Office of the Deputy Assistant Secretary of Defense for Systems Engineering, Office of Cost Assessment and Program Evaluation, Office of the Director of Operational Test and Evaluation, Defense Digital Service, Defense Innovation Board, and the Office of the Assistant Secretary of the Air Force for Space Acquisition. We also interviewed officials from the selected program offices and their respective contractors, subcontractor, integrator, space systems users, a DOD test organization, and Federally Funded Research and Development Centers. In addition, we conducted a literature search using a number of bibliographic databases, including ProQuest, Scopus, DIALOG, and WorldCat. We reviewed documentation that focused on software-intensive major military acquisitions. We conducted our search in March 2018. To determine how effectively selected DOD software-intensive space programs have involved users and adopted newer software development approaches, we reviewed applicable DOD policies, guidance, and federal statute that identify characteristics of user engagement. These sources were the Department of Defense Instruction (DODI) 5000.02; Office of the Secretary of Defense Report to Congress, A New Approach for Delivering Information Technology in the Department of Defense; and National Defense Authorization Act for Fiscal Year 2010. We supplemented this with Defense Science Board and Defense Innovation Board documentation, and other industry analyses. We then reviewed relevant program plans and documentation, such as human engineering and human systems integration plans, standard operating procedures, acquisition strategies, software development plans, and other program user engagement guidance to identify plans for user engagement. We then conducted interviews with space system users and analyzed software development documentation to evaluate the extent to which programs met these DOD user engagement characteristics. We also analyzed user feedback reports to identify trends in user feedback. We also examined DOD and OMB guidance and applicable leading practices to identify time frames for delivering software under incremental and iterative software development approaches, and we compared these time frames to program performance. To determine what software-specific management challenges, if any, selected programs faced, we reviewed reports and studies on software tools and metrics used to manage software programs, including GAO reports, DOD policies and guidance, and studies from the Software Engineering Institute. We then reviewed program documents, such as Software Development Plans, System Engineering Plans, System Engineering Management Plans, Software Resource Data Reports, Test and Evaluation Master Plans, Master Software Build Plans, and Obsolescence Plans, as applicable, as well as contracts and Statements of Work. We reviewed defect metrics and reports on amounts of new, reused, inherited, and commercial software; test and evaluation reports; program management reports; and external program assessments. We also evaluated program retrospectives and DOD reports on leading practices to understand how programs are making efforts to address challenges in these areas. We spoke with contractors and an applicable subcontractor and government integrator, program officials, and officials from Federally Funded Research and Development Centers to understand program issues, including program office and contractor training requirements. We conducted this performance audit from November 2017 to March 2019 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Comments from the Department of Defense Appendix III GAO Contacts and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Raj Chitikila, Assistant Director; Pete Anderson, Erin Carson, Jordan Kudrna, Matthew Metz, Roxanna Sun, and Jay Tallon made key contributions to this report. Assistance was also provided by Mathew Bader, Virginia Chanley, Susan Ditto, Sarah Gilliland, Carol Harris, Harold Podell, Andrea Starosciak, Anne Louise Taylor, and Alyssa Weir. Related GAO Products Weapon Systems Cybersecurity: DOD Just Beginning to Grapple with Scale of Vulnerabilities. GAO-19-128. Washington, D.C.: October 9, 2018. Weapon Systems Annual Assessment: Knowledge Gaps Pose Risks to Sustaining Recent Positive Trends. GAO-18-360SP. Washington, D.C.: April 25, 2018. Information Technology: Agencies Need to Involve Chief Information Officers in Reviewing Billions of Dollars in Acquisitions. GAO-18-42. Washington, D.C.: January 10, 2018. Global Positioning System: Better Planning and Coordination Needed to Improve Prospects for Fielding Modernized Capability. GAO-18-74. Washington, D.C.: December 12, 2017. Information Technology Reform: Agencies Need to Improve Certification of Incremental Development. GAO-18-148. Washington, D.C.: November 7, 2017 Space Acquisitions: DOD Continues to Face Challenges of Delayed Delivery of Critical Space Capabilities and Fragmented Leadership. GAO-17-619T. Washington, D.C.: May 17, 2017. Defense Acquisitions: Assessment of Selected Weapon Programs. GAO-17-333SP. Washington, D.C.: March 30, 2017. Immigration Benefits System: U.S. Immigration Services Can Improve Program Management. GAO-16-467. Washington, D.C.: July 7, 2016. GPS: Actions Needed to Address Ground System Development Problems and User Equipment Production Readiness. GAO-15-657. Washington, D.C.: September 9, 2015. Defense Acquisitions: Assessments of Selected Weapon Programs. GAO-15-342SP. Washington, D.C.: March 12, 2015. Defense Major Automated Information Systems: Cost and Schedule Commitments Need to Be Established Earlier. GAO-15-282. Washington, D.C.: February 26, 2015. Standards for Internal Control in the Federal Government. GAO-14-704G. Washington, D.C.: September 2014. Software Development: Effective Practices and Federal Challenges in Applying Agile Methods. GAO-12-681. Washington, D.C.: July 27, 2012. Information Technology: Critical Factors Underlying Successful Major Acquisitions. GAO-12-7. Washington, D.C.: October 21, 2011. Space Acquisitions: Development and Oversight Challenges in Delivering Improved Space Situational Awareness Capabilities. GAO-11-545. Washington, D.C.: May 27, 2011. Significant Challenges Ahead in Developing and Demonstrating Future Combat System’s Network and Software. GAO-08-409. Washington, D.C.: March 7, 2008. Space Based Infrared System High Program and its Alternative. GAO-07-1088R. Washington, D.C.: September 12, 2007. Defense Acquisitions: Stronger Management Practices Are Needed to Improve DOD’s Software-Intensive Weapon Acquisitions. GAO-04-393. Washington, D.C.: March 1, 2004. Information Security: Effective Patch Management is Critical to Mitigating Software Vulnerabilities. GAO-03-1138T. Washington, D.C.: September 10, 2003. Test and Evaluation: DOD Has Been Slow in Improving Testing of Software-Intensive Systems. GAO/NSIAD-93-198. Washington, D.C.: September 29, 1993. Mission-Critical Systems: Defense Attempting to Address Major Software Challenges. GAO/NSAID-93-13. Washington, D.C.: December 24, 1992. Space Defense: Management and Technical Problems Delay Operations Center Acquisition. GAO/IMTEC-89-18. Washington, D.C.: April 20, 1989.

Why GAO Did This Study Over the next 5 years, DOD plans to spend over $65 billion on its space system acquisitions portfolio, including many systems that rely on software for key capabilities. However, software-intensive space systems have had a history of significant schedule delays and billions of dollars in cost growth. Senate and House reports accompanying the National Defense Authorization Act for Fiscal Year 2017 contained provisions for GAO to review challenges in software-intensive DOD space programs. This report addresses, among other things, (1) the extent to which these programs have involved users; and (2) what software-specific management challenges, if any, programs faced. To do this work, GAO reviewed four major space defense programs with cost growth or schedule delays caused, in part, by software. GAO reviewed applicable statutes and DOD policies and guidance that identified four characteristics of effective user engagement. GAO reviewed program documentation; and interviewed program officials, contractors, and space systems users. GAO also analyzed program metrics, test and evaluation reports, and external program assessments. What GAO Found The four major Department of Defense (DOD) software-intensive space programs that GAO reviewed struggled to effectively engage system users. These programs are the Air Force's Joint Space Operations Center Mission System Increment 2 (JMS), Next Generation Operational Control System (OCX), Space-Based Infrared System (SBIRS); and the Navy's Mobile User Objective System (MUOS). These ongoing programs are estimated to cost billions of dollars, have experienced overruns of up to three times originally estimated cost, and have been in development for periods ranging from 5 to over 20 years. Previous GAO reports, as well as DOD and industry studies, have found that user involvement is critical to the success of any software development effort. For example, GAO previously reported that obtaining frequent feedback is linked to reducing risk, improving customer commitment, and improving technical staff motivation. However, the programs GAO reviewed often did not demonstrate characteristics of effective user engagement that are identified in DOD policy and statute: Early engagement. OCX involved users early; JMS planned to but, in practice, did not; SBIRS and MUOS did not plan to involve users early. Continual engagement. JMS, OCX, and SBIRS all planned to continually involve users but, in practice, did not fully do so; MUOS did not plan to do so. Feedback based on actual working software. OCX and SBIRS provided users opportunities to give such feedback but only years into software development; JMS and MUOS did not provide opportunities for feedback. Feedback incorporated into subsequent development. JMS, OCX, and SBIRS all planned to incorporate user feedback but, in practice, have not done so throughout development; MUOS did not plan to do so. As reflected above, actual program efforts to involve users and obtain and incorporate feedback were often unsuccessful. This was due, in part, to the lack of specific guidance on user involvement and feedback. Although DOD policies state that users should be involved and provide feedback on software development projects, they do not provide specific guidance on the timing, frequency, and documentation of such efforts. Without obtaining user feedback and acceptance, programs risk delivering systems that do not meet users' needs. In selected instances, the lack of user involvement has contributed to systems that were later found to be operationally unsuitable. The programs GAO reviewed also faced software-specific challenges in using commercial software, applying outdated software tools, and having limited knowledge and training in newer software development techniques. For example, programs using commercial software often underestimated the effort required to integrate such software into an overall system. Secondly, selected programs relied on obsolete software tools that they were accustomed to using but which industry had since replaced. Finally, GAO found that two of the reviewed programs lacked knowledge of more modern software development approaches. DOD has acknowledged these challenges and has efforts underway to address each of them. What GAO Recommends GAO is making two recommendations that DOD ensure its guidance that addresses software development provides specific, required direction on the timing, frequency, and documentation of user involvement and feedback. DOD concurred with the recommendations.

gao_GAO-18-702T

gao_GAO-18-702T_0

IRS Has Broad Efforts Underway to Address Authentication Challenges Our report noted that IRS has established organizational structures essential to supporting its taxpayer authentication efforts. Specifically, IRS created an Identity Assurance Office (IAO) in 2015 to work with stakeholders across IRS to review and assess the agency’s various authentication programs and efforts. In 2016, IAO led an effort that identified over 100 interactions between IRS and taxpayers that require authentication and categorized these interactions based on potential risks to the agency and taxpayers. Further, in December 2016, IAO released its Roadmap for developing a modern and secure authentication environment for all taxpayers regardless of how they interact with IRS— online, over the telephone, in person, or via correspondence. We also found that IRS is working to address its authentication challenges by collaborating with industry members and state partners via the Security Summit. The Security Summit was established in 2015 as an ongoing effort between industry experts from tax software companies, paid preparers, financial institutions, and states to improve information sharing and fraud detection and to address common IDT challenges. The Security Summit’s authentication workgroup leads several initiatives aimed at verifying the authenticity of the taxpayer and the tax return at the time of filing. One initiative involves analyzing data elements—such as trusted customer requirements and other characteristics of the return— that are collected during the tax return preparation and electronic filing process. In addition, in 2016 the authentication workgroup recommended improved account password standards to help protect taxpayers’ accounts from being taken over by criminals. Overall, we found that officials—representing IRS, industry, and states— expressed positive views about the level of commitment and cooperation guiding the group’s authentication efforts. Officials with whom we spoke stated that they are dedicated to continuing to address authentication issues collaboratively because they have a mutual interest in improving authentication to reduce tax refund fraud. IRS Has Begun to Implement Its Authentication Strategy, but Has Not Articulated Priorities and Resource Needs In its Roadmap, IRS outlined six core authentication objectives, 10 high- level strategic efforts, and 14 foundational initiatives to help it address authentication challenges and identify opportunities for future investment. While we found that IRS has made progress on some efforts identified in its Roadmap, it has not prioritized the initiatives supporting its strategy nor identified the resources required to complete them, consistent with program management leading practices. For example, one of IRS’s foundational initiatives is to send event-driven notifications to taxpayers, such as when they file a return or request a tax transcript. Such notifications could help IRS and taxpayers detect potentially fraudulent activity at the earliest stage and help improve authentication of tax returns. The Roadmap identifies seven supporting activities for this foundational initiative. One is to provide taxpayers with greater control over their online accounts. Another supporting activity is to determine methods for sending notifications to taxpayers about activity on their account. However, IRS has not identified the resources required to complete these activities, and the Roadmap notes that six of the seven activities will take between 6 months to 3 years to complete. In December 2017, IRS officials stated that they had developed business requirements for the foundational initiative to give taxpayers greater control over their online accounts. However, IRS has not identified funding for the initiative’s other supporting activities—such as developing requirements to send push notifications to taxpayers—and implementation will depend on the availability of future resources. In December 2017, IRS officials stated that each of the strategic efforts and foundational initiatives identified in the Roadmap are a high priority, and they are working to address them concurrently while balancing the availability of resources against the greatest threats to the tax environment. As noted in our report, we recognize that a strategy is necessarily high-level and that IRS must remain flexible and use available resources to respond to unexpected threats. Identifying resources and prioritizing activities in its Roadmap will help IRS clarify tradeoffs between costs, benefits, and risks and aid in decision making. Further, such efforts may also help IRS establish clearer timelines and better respond to unexpected events. As such, we recommended that IRS estimate the resources (i.e., financial and human) required for the foundational initiatives and supporting activities identified in its Roadmap and prioritize its foundational initiatives. IRS agreed with our recommendations and is currently working to finalize its overall authentication approach. Additional Actions Could Help IRS Enhance Security and Stay Ahead of Fraudsters Given the widespread availability of personally identifiable information that fraudsters can use to perpetrate tax fraud, it is essential for IRS to further strengthen taxpayer authentication to stay ahead of fraudsters’ schemes. In our report, we identified two additional areas that IRS must address to better position the agency and protect taxpayers against future threats. First, we found that IRS has taken preliminary steps to implement NIST’s June 2017 guidance for secure online authentication, however it had not yet established detailed plans, including timelines, milestone dates, and resource needs to fully implement it. Among other things, NIST’s new guidance directs agencies to assess the risk for each component of identity assurance—identity proofing, authentication, and federation— rather than conducting a single risk assessment for the entire process. According to NIST officials, this approach gives agencies flexibility in choosing technical solutions; aligns with existing, standards-based market offerings; is modular and cost-effective; and enhances individual privacy. In short, following NIST’s new guidance will help provide IRS with better risk-based assurance that the person trying to access IRS’s online services is who they claim to be. As noted in our report, IRS has taken preliminary steps to implement the new NIST guidance. These efforts include forming a task force to guide IRS’s implementation of NIST guidance and working with the Security Summit to develop an implementation framework for state and industry partners. IRS has also begun analyzing gaps between IRS’s current authentication procedures and the new guidance. In addition, in December 2017, IRS implemented a more secure online authentication option consistent with the new guidance through its mobile application, IRS2Go. After taxpayers link their IRS online account with the mobile app, they can use it to generate a security code to log into their account. This option provides taxpayers with an alternative to receiving the security code via a text message, which NIST considers to be less secure. We recommended that IRS develop a plan—including a timeline, milestone dates, and resources needed—for implementing changes to its online authentication programs consistent with new NIST guidance, and also implement these improvements. IRS agreed with our recommendations, but noted that its ability to complete these efforts will depend on the availability of resources. Second, we found that IRS lacks a comprehensive, repeatable process to identify and evaluate potential new authentication technologies and approaches. Our discussions with representatives from industry and financial institutions and with government officials indicate that there is no single, ideal online authentication solution that will solve IRS’s challenges related to IDT refund fraud. These representatives advocate an approach to authentication that relies on multiple strategies and sources of information, while giving taxpayers options for further protecting their information. We identified several authentication options in our report that IRS could consider, including the following: Possession-based authentication. This type of authentication offers users a convenient, added layer of security when used as a second factor for accessing websites or systems that would otherwise rely on a username and password for single-factor authentication. For example, as noted in our report, according to an industry official, authentication using a trusted device or “security key” based on Universal Second Factor standards complies with NIST’s new guidance for digital authentication. While IRS is not likely to provide the devices to taxpayers, it could enable its systems to accept these trusted devices as authenticators for taxpayers who elect to use them. Working with trusted partners. IRS could partner with organizations it trusts that are accessible to taxpayers and enable the partners to identity-proof and authenticate taxpayers. Trusted partners could include tax preparers, financial institutions, or other federal or state agencies. In the course of our work, IRS officials stated that they had been exploring such options with both the Social Security Administration and the U.S. Postal Service; however, at the time of our report, the agencies had not yet made decisions about next steps. Expanding existing IRS services to further protect taxpayers. IRS could expand the functionality of its online account to further protect taxpayers from IDT refund fraud. For example, IRS could develop additional functionality that allows the taxpayer to designate a bank account or a preference for a paper check for receiving a tax refund. If a fraudster filed a return with different information, the return would be automatically rejected. IRS officials told us the agency continually researches new identity assurance processes and technologies and has talked with other agencies, industry groups, and vendors to better understand how particular technology solutions could apply to IRS’s environment. However, during the course of our work, IRS could not provide us evidence of a repeatable, comprehensive process to identify and evaluate available authentication technologies and services. Such a process could compare options for in-house authentication solutions with off-the-shelf solutions based on estimates of cost, schedule, and benefits, as applicable. To this end, we recommended that IRS develop a process to identify and evaluate alternative options for improving taxpayer authentication, including technologies in use by industry, states, or other trusted partners; and based on this approach, include and prioritize these options, as appropriate, in its Roadmap. IRS agreed with these recommendations, but did not provide additional details on how it plans to address them. In conclusion, IRS’s authentication environment is one component of a broad, complex information technology infrastructure, and we have previously reported on the many challenges the agency faces as it modernizes its tax systems. Taxpayer authentication has become more difficult with the wide availability of personally identifiable information and fraudsters’ ability to develop more complex and sophisticated methods to commit fraud undetected. Addressing the issues we describe above could better position IRS to identify and mitigate vulnerabilities in its authentication efforts and better protect taxpayers and the Treasury. Chairman Jenkins, Ranking Member Lewis, and members of the Subcommittee, this concludes my prepared remarks. I look forward to answering any questions that you may have at this time. GAO Contacts and Staff Acknowledgments If you or your staff have any questions about this testimony, please contact James R. McTigue, Jr. at (202) 512-9110 or mctiguej@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this statement. Key contributors to this testimony include Neil Pinney, Assistant Director; Heather A. Collins, Analyst-in-Charge; Dawn Bidne; and Bryan Sakakeeny. This is a work of the U.S. government and is not subject to copyright protection in the United States. The published product may be reproduced and distributed in its entirety without further permission from GAO. However, because this work may contain copyrighted images or other material, permission from the copyright holder may be necessary if you wish to reproduce this material separately.

Why GAO Did This Study This testimony summarizes the information contained in GAO's June 2018 report, entitled Identity Theft: IRS Needs to Strengthen Taxpayer Authentication Efforts ( GAO-18-418 ). What GAO Found The Internal Revenue Service (IRS) has identified over 100 interactions requiring taxpayer authentication based on potential risks to IRS and individuals. IRS authenticates millions of taxpayers each year via telephone, online, in person, and correspondence to ensure that it is interacting with legitimate taxpayers. IRS's estimated costs to authenticate taxpayers vary by channel. IRS has made progress on monitoring and improving authentication, including developing an authentication strategy with high-level strategic efforts. However, it has not prioritized the initiatives supporting its strategy nor identified the resources required to complete them, consistent with program management leading practices. Doing so would help IRS clarify relationships between its authentication efforts and articulate resource needs relative to expected benefits. Further, while IRS regularly assesses risks to and monitors its online authentication applications, it has not established equally rigorous internal controls for its telephone, in-person, and correspondence channels, including mechanisms to collect reliable, useful data to monitor authentication outcomes. As a result, IRS may not identify current or emerging threats to the tax system. IRS can further strengthen authentication to stay ahead of fraudsters. While IRS has taken preliminary steps to implement National Institute of Standards and Technology's (NIST) new guidance for secure digital authentication, it does not have clear plans and timelines to fully implement it by June 2018, as required by the Office of Management and Budget. As a result, IRS may not be positioned to address its most vulnerable authentication areas in a timely manner. Further, IRS lacks a comprehensive process to evaluate potential new authentication technologies. Industry representatives, financial institutions, and government officials told GAO that the best authentication approach relies on multiple strategies and sources of information, while giving taxpayers options for actively protecting their identity. Evaluating alternatives for taxpayer authentication will help IRS avoid missing opportunities for improving authentication.

gao_GAO-18-658

gao_GAO-18-658_0

Background General Process for Managing Pharmacy Inventories in Hospital Settings In general, the process for managing inventories of medications at VAMCs and non-VA pharmacies in hospital settings is similar. The steps of the process are (1) procuring medications from vendors or other suppliers, (2) receiving and storing medications, (3) tracking medications to account for all items and prevent diversion, (4) dispensing medications to patients, and (5) disposing of expired or wasted medications. Hospital settings include both inpatient and outpatient pharmacies. Procurement. Pharmacies use a procurement process to order medications for pharmacy inventory, which includes activities such as medication selection, cost analysis, purchasing procedures, and record keeping. As part of medication selection, pharmacies may use a formulary, which is a list of medications that have been approved for prescription within a hospital or health care system. A prime vendor or wholesaler is one of the most commonly used sources to obtain medications for the pharmacy. Prime vendors order large quantities of medications from manufacturers, allowing pharmacies to purchase various products from many drug manufacturers at once. Orders for products that are not carried by the prime vendor may need to be ordered through another source, such as directly from the manufacturer. Receipt and storage. When medications are delivered to the pharmacy, staff are to take several steps to properly receive and store the shipment. For example, to ensure there is segregation of duties, the person responsible for ordering and purchasing the medications is supposed to be different than the person receiving and stocking pharmacy inventory. Additionally, any delivered products that require special storage conditions, such as freezing or refrigeration, are to be checked in first to maintain the stability of the medication. Tracking. Once in storage, pharmacies use a variety of tools to account for the filling, dispensing, and removal of medications in both inpatient and outpatient settings. Some pharmacies have software that allows them to track inventory in real time, an ability known as maintaining perpetual inventory. A perpetual inventory system is a method of recording the quantity of a particular medication continuously as prescriptions are filled and dispensed. After each prescription is filled and dispensed to the patient, the amount of medication used for the prescription is removed from the inventory to ensure the quantity on hand recorded by the software is always current. Many medications have barcodes on their packaging to allow for easy identification of the medication in a computer system. The barcode generally includes the product’s National Drug Code, which indicates the name and package size of the medication. In the hospital setting, medications can be scanned out of the pharmacy and into machines for storage on hospital wards. Dispensing. In both inpatient wards and outpatient pharmacies, automated dispensing machines and barcode technology can assist staff in maintaining and dispensing medications to patients. Automated dispensing machines generally include several drawers and cabinets that have pockets or trays that hold preset levels of a variety of common medications. They may also be used to hold controlled substances, generally in locked boxes or cubes within the machine. On hospital wards medication in automated dispensing machines is often packaged in unit doses—individually packaged medications for patient use. Barcodes can help verify a prescription before nurses give medication to a patient. Hospitals that do not have automatic dispensing machines use carts with drawers filled with each patient’s medication. Outpatient pharmacies use automated dispensing machines to assist with filling prescriptions. Depending on the type of automated dispensing machine, the capabilities can include label printing, pill counting, pouring pills into prescription bottles, and applying the label to the prescription bottle. Return or disposal. Medication waste and expired medications are to be pulled from pharmacy inventory and either returned to a reverse distributor or manufacturer for credit or, if not eligible for return, disposed of by the pharmacy or sent to an outside company for destruction. Reverse distributors charge a fee, which is generally a percentage of the refund that is automatically deducted from the final refund amount. Figure 1 provides an overview of the steps of the pharmacy inventory management process. VA Organizational Structure and Pharmacy Policies VA’s health care system is organized into entities at the headquarters, regional, and local levels. At the headquarters level, PBM is responsible for supporting VISNs and VAMCs with a broad range of pharmacy services, such as promoting appropriate drug therapy, ensuring medication safety, providing clinical guidance to pharmacists and other clinicians, and maintaining VA’s formulary of medications and supplies VAMCs use to deliver pharmacy benefits. VA’s OIT is responsible for providing technology services across the department, including the development and management of all IT assets and resources. As such, the office supports VA’s health care system in planning for and acquiring IT capabilities within VA’s health care system network of hospitals, outpatient facilities, and pharmacies. VA’s NAC is responsible for administering various health care-related acquisition and logistics programs across VA. At the regional level, VAMCs are located in one of 18 VISNs. Each VISN is responsible for overseeing VAMC pharmacies within a defined geographic region. At the local level, there are approximately 170 VAMCs. Each VAMC is responsible for implementing VA’s pharmacy policies and programming. VA policy establishes parameters for VAMCs to follow when managing their pharmacy inventories. These policies address various aspects of pharmacy services, including inpatient and outpatient pharmacy services, general pharmacy requirements, supply chain management, controlled substances management, and the formulary management process. For example, the Supply Chain Inventory Management directive states that all VAMC pharmacies should use the prime vendor inventory management software to calculate the amount of each inventory item they need to reorder. However, the directive also states that there are additional pharmacy inventory tools available to VAMC pharmacies and that each pharmacy has the option to use its own automated inventory management systems to generate orders for its prime vendor. VA policy does not specify minimum quantities to order; instead, VAMC procurement staff is authorized to use their expertise to determine the appropriate quantity to order. Selected VAMCs Have Generally Similar Approaches for Managing Their Pharmacy Inventories, with Some Unique Approaches In general, all five of the selected VAMCs we reviewed take similar approaches for the various steps included in the pharmacy inventory management process—that is, procuring medications from vendors or other suppliers, receiving and storing these medications, tracking medications at the pharmacy to account for all items and prevent diversion, dispensing medications to patients, and disposing of expired medications. (See fig. 2). We found that while the five selected VAMCs have similar approaches for receiving and storing, dispensing, and disposing of medications, some VAMCs have also taken unique approaches in implementing two steps of the pharmacy inventory management process: procurement and tracking. VA policy outlines parameters for VAMCs to manage their pharmacy inventories, and VA officials told us that VAMC pharmacy staff can use discretion to implement their own approaches for managing their pharmacy inventories. Procurement All five of the selected VAMC pharmacies we reviewed use several sources of information to inform future orders—including past purchase order history reports from VA’s prime vendor, manual inventory counts by pharmacy staff, and automated dispensing machine inventory information. VA officials told us that all VAMCs also track procurement spending and its impact on the VAMCs’ budget and spending. However, pharmacy officials at one of the selected VAMCs we visited told us they use VA’s health information system—Veterans Health Information Systems and Technology Architecture (VistA)—and additional prime vendor reports to identify specific information regarding 1) expiring medications that may need to be re-purchased, 2) medications that account for the top 80 percent of pharmacy costs, and 3) all medications that are purchased daily. VAMC officials told us these reports help them to better manage pharmacy inventory and track pharmacy spending. To better anticipate and address potential medication shortages, officials at another selected VAMC pharmacy told us they established a shortage committee that meets on a weekly basis. Established in September 2017, the committee includes the Director of Pharmacy and other pharmacy staff. Our review of meeting notes shows that the committee discusses which medications could experience or are experiencing shortages and how the VAMC could adjust to these shortages by, for example, developing clinical and logistical solutions to help maintain optimal patient care. According to the officials at the selected VAMC pharmacy, the committee has been an effective resource to help manage pharmacy inventory problems should they occur. Several VAMC officials also told us that the procurement technicians, who are responsible for ordering pharmacy inventory, are very important because they possess valuable institutional knowledge based on many years of experience and training. However, VAMC officials told us the salaries and potential career advancement opportunities for procurement technicians can be limited, and the officials expressed concern that these technicians could find better opportunities within the VAMC or with external employers. To help retain procurement technicians, two of the selected VAMC pharmacies we visited have created higher paying procurement technician positions (General Schedule level 8 positions, instead of GS-6 or GS-7). Tracking To better identify potential instances of diversion, two of the selected VAMC pharmacies use enhanced analytics software on the automated dispensing machines in their inpatient wards to track how frequently controlled substances and other frequently utilized medications are prescribed. For example, one of the pharmacies uses data from these reports to identify how often individual staff members are accessing automated dispensing machines. Additionally, officials at a third VAMC recently deployed automated dispensing machines that are equipped with an enhanced analytics program that can identify trends associated with diversion. The remaining two VAMCs we visited do not have enhanced analytic software that could help them to identify instances of potential diversion. Across all 5 selected VAMCs, we observed several different IT systems used to help manage non-controlled inpatient inventory. One of the selected VAMC pharmacies uses a modular automated dispensing machine together with inventory management software that maintains a perpetual inventory for most non-controlled substances stored in its inpatient pharmacy. (See fig. 3). According to officials, this software has allowed the pharmacy to reduce waste and improve staff workflow, as staff do not have to spend time tracking down inventory. None of the other VAMC pharmacies we visited have the capability to track non- controlled substances in real time. Additionally, to more efficiently identify medication lot numbers during recalls, one VAMC pharmacy we visited was in the process of implementing a technology that allows pharmacy staff to scan a case of medication with the same national drug code, lot number, and expiration date and then print and attach a radio frequency identification tag to each medication bottle. The tag allows for quick electronic identification of the medication for disposal. Other selected VAMC pharmacies manually identify recalled medications from inventory based on the name of the medication and lot number. Oversight of Pharmacy Inventory Management Is Limited, as VA Lacks a Comprehensive Inventory Management System or a Focal Point for System-Wide Oversight VA does not yet have a VA-wide pharmacy inventory management system in place that would allow it to monitor VAMC pharmacy inventory in real time and provide better oversight of how VAMC pharmacies manage their inventories. We found that VACO and the five VISNs we reviewed provide some oversight related to VAMC pharmacy inventory management. However, that oversight is limited, as no entity has been assigned responsibility for overseeing system-wide performance of VAMC pharmacies in managing their inventories. VA’s Oversight of VAMCs’ Pharmacy Inventory Management Is Limited, in Part Because VA Has Not Yet Implemented a Comprehensive Inventory Management System VA’s oversight of VAMC pharmacy inventory management is limited in part because VA currently lacks a comprehensive system that would allow the department and its VAMCs to monitor pharmacy inventory in real time. According to PBM officials, the lack of a VA-wide system makes it difficult to oversee VAMC pharmacy inventory management, and PBM has recognized the lack of such a system as a material weakness for several years. PBM officials said that implementation of a VA-wide pharmacy inventory management system would allow them to monitor each VAMC’s pharmacy inventory in real time, which would, in turn, allow them to better manage inventory and help alleviate shortages at the national level by facilitating transfers of inventory between VAMCs as needed. Additionally, officials said that such a system would lead to better planning and projections for purchasing decisions, allow PBM to track medication expiration dates and lot numbers more effectively, and improve VAMC staff response to medication recalls. Although VA has acknowledged the need for a VA-wide pharmacy inventory management system, such a system may not be available for the foreseeable future. PBM officials told us they have requested this system since the early 2000s. However, despite the documented technological challenges VA faces in overseeing its VAMC pharmacies, changing IT priorities, funding challenges, and the narrowing of the scope of a Pharmacy Re-engineering Project have prevented the system’s development. In 2017, we reported that VA’s pharmacy systems could not maintain a real-time inventory across the VAMCs, and we recommended that VA assess the priority for establishing an inventory management system capable of monitoring medication inventory levels and indicating when medications needed to be reordered. VA concurred with our recommendation. In June 2017, VA announced its intention to replace VistA— VA’s health information system—with an off-the-shelf electronic health record system. VA officials told us that the new system will have the capability to monitor pharmacy inventory in real time across VA. VA signed the contract for this new system in May 2018; however, full implementation is expected to take up to 10 years. In the interim, VA officials told us that while they will maintain current pharmacy systems, they do not plan to build any new systems—including a VA-wide pharmacy inventory management system—so they can efficiently manage resources in preparation for the transition to the new system. VACO and VISNs Provide Some Limited Oversight, but VA Lacks a Focal Point for System-Wide Oversight of Pharmacy Inventory Management VACO and the five VISNs we spoke with provide some limited oversight related to VAMC pharmacy inventory management, but no entity has system-wide responsibility for overseeing the performance of VAMC pharmacies in managing their inventories. Instead, responsibility for overseeing pharmacy inventory management is largely delegated to each VAMC’s leadership. (See fig. 4 for a description of VACO headquarters, VISN, and VAMCs’ roles and responsibilities in managing pharmacy inventory.) In absence of a VA-wide inventory management system, PBM officials told us that they have employed manual workaround mechanisms to oversee pharmacy management processes. Specifically, PBM requires VAMC pharmacies to conduct an annual inventory of all medications and a quarterly inventory of 5 selected high-value non-controlled medications at risk of diversion. PBM officials told us they remind VAMCs of the requirement to conduct these inventories, collect and aggregate the data from these inventories, and make summary reports from these data available as a resource to the VPEs and VAMC Chiefs of Pharmacy. PBM officials acknowledged that these manual workarounds are inefficient, increase labor costs, and leave the agency with an inability to see on- hand inventory across the system in real time. Additionally, the manual workarounds may be implemented differently at each VAMC, resulting in varying degrees of data reliability and limited opportunities for high-level oversight and data consolidation. PBM officials said that they do not independently analyze these data to identify trends, and they acknowledged that both the quarterly and annual inventories have limited usefulness for overseeing inventory management system-wide. Additionally, officials at some of the selected VAMCs told us they found the quarterly and annual inventories to have limited usefulness for managing their pharmacy inventories. PBM officials told us they also hold regular meetings with VPEs and VAMCs, which provide the opportunity for discussion of pharmacy inventory management issues. However, our review of the minutes of the meetings between PBM and VPEs found that, over the past 3 years, pharmacy inventory management was rarely a topic of discussion. PBM officials noted that there is always an opportunity for open discussion at these meetings for VPEs to raise any issues, including issues related to pharmacy inventory management, but these discussions may or may not be captured in the meeting minutes. PBM officials said they also regularly discuss various topics with the VAMC Chiefs of Pharmacy and other staff, but none of these calls are directly related to pharmacy inventory management. Officials from VACO’s NAC and OIT told us that they provide some assistance related to pharmacy inventory management but do not take part in the day-to-day management at the VAMC level and also do not have any oversight responsibilities. For example, a NAC official said the office coordinates with PBM on medication shortage issues and establishes national contracts for medications. NAC also sends out a weekly shortages report to various pharmacy groups as a tool to help them with known or expected shortages. Additionally, NAC’s Pharmaceutical Prime Vendor team is responsible for administering the contract with the prime vendor through daily monitoring of issues and quarterly reviews with the prime vendor and PBM. OIT develops pharmacy-related applications for VistA based on requirements from PBM, and officials said that the majority of OIT’s support to VAMCs consists of assisting them with issues related to VistA. At the VISN level, VPEs we interviewed also said they conduct some pharmacy inventory management oversight activities for the VAMCs within their network. While in general VA policy does not outline any specific roles for VPEs related to oversight of pharmacy inventory management, all five VPEs told us that they review the results of their VAMCs’ annual inventories and discuss any issues that arise from this exercise with VAMCs as needed. VPEs told us that they also review the results of the quarterly inventory of five selected high-value, non- controlled substances and may follow-up with the VAMCs if their actual inventory of the medications is inconsistent with expected levels. Additionally, some VPEs reported that they have undertaken additional oversight activities apart from reviewing results of the mandatory inventories. For example, one VPE told us he has developed a dashboard with 53 measures that, while focused on formulary management, also have inventory management implications. Additionally, this VPE said that a VISN-wide procurement work group meets on a monthly basis and serves as a venue for procurement technicians to share inventory management best practices. Such additional activities may be helpful, but since VPEs only have responsibility for VAMC pharmacies within their network, they may not be aware of pharmacy inventory management approaches being used at other VAMCs across VA. Although VA offices at the headquarters and regional levels provide some assistance and oversight of how VAMCs manage pharmacy inventory at the local level, VA has not designated a focal point with defined responsibilities for system-wide oversight; instead they rely on local leadership to oversee pharmacy inventory management at the VAMCs. As a result, VA cannot assess the overall performance of VAMCs’ management of their pharmacy inventories. The lack of a focal point with defined oversight responsibilities is inconsistent with federal internal control standards for establishing structure and authority to achieve the entity’s objectives and internal controls related to monitoring. Specifically, internal controls state that management should establish an organizational structure, assign responsibility, and delegate authority to achieve the entity’s objectives. Also, internal controls state that management should establish and operate monitoring activities to monitor the internal control system and evaluate the results. VA’s actions are also inconsistent with the Office of Management and Budget’s guidance for enterprise risk management and internal control in managing an agency. Enterprise risk management is intended to yield an “enterprise- wide,” strategically aligned portfolio view of organizational challenges that provides better insight about how to most effectively prioritize resource allocations to ensure successful mission delivery. Without a focal point for system-wide oversight of VAMC pharmacy inventory management, VA has limited awareness of the unique approaches that VAMCs use to manage their inventories and is missing an opportunity to evaluate these approaches. Additionally, VA cannot effectively share and standardize pharmacy inventory management best practices as appropriate. Having a focal point for system-wide oversight could allow VA to identify potential best practices that could be disseminated more widely across its facilities. Conclusions Due to the decentralized nature of VA’s organization, VA policy gives VAMC pharmacies latitude in managing their pharmacy inventories. Several of the VAMCs we visited have taken unique approaches to procuring or tracking their inventory. However, because VA does not have a focal point to systematically oversee VAMCs’ pharmacy management efforts, VA is missing opportunities to evaluate the effectiveness of these efforts, as well as share best practices and standardize them across VA as appropriate. PBM officials told us that the lack of a VA-wide pharmacy inventory management system limits their ability to oversee VAMC pharmacy inventory management. However, our review shows that even without this system there are existing mechanisms that a focal point could leverage to more systematically oversee how VAMC pharmacies manage their inventories. For example, a focal point could ensure that PBM officials, the VPEs, and VAMC pharmacy staff devote time to discussing pharmacy inventory management approaches and related issues during regularly scheduled telephone meetings. Leveraging these existing mechanisms is especially important given that VAMCs have historically had challenges in managing their inventories, and also because a VA- wide pharmacy inventory management system may not be available for the foreseeable future. Recommendation for Executive Action We are making the following recommendation to the Department of Veterans Affairs: The Secretary of the VA should direct the Undersecretary for Health to designate a focal point for overseeing VAMCs’ pharmacy inventory management system-wide and define the focal point’s responsibilities. (Recommendation 1) Agency Comments and Our Evaluation We provided a draft of this report to VA for review and comment. In its written comments, reproduced in appendix I, VA stated that it concurred in principle with our recommendation. VA also provided technical comments, which we incorporated as appropriate. In response to our recommendation, VA stated it plans to establish by December 31, 2018, a committee of internal stakeholders and subject matter experts to provide options for overseeing VAMCs’ pharmacy inventory management. However, it was unclear from VA’s response whether the planned committee will recommend or designate an entity or focal point with system-wide oversight responsibilities. VA noted in its general comments that it does have entities or individuals—referred to as focal points by VA—responsible for specific functions. However, these entities do not provide system-wide oversight that could allow the department to better understand VAMCs’ approaches to pharmacy inventory management. As we noted in our report, without a focal point for system-wide oversight, VA has limited awareness of the unique approaches that VAMCs use to manage their inventories and is missing an opportunity to evaluate these approaches and standardize them across VA as appropriate. Additionally, in its general comments, VA raised concerns regarding our characterization in the draft report of medication shortages and the use of automated dispensing units in the context of controlled substances. In response, we updated the report to include more information about one VAMC’s use of a committee to address medication shortages. We also clarified that three VAMCs are using (or will soon have the capability to use) enhanced analytic software to better leverage data generated through their automated dispensing machines, which allows them to more easily identify potential diversion. Finally, VA noted that we did not discuss PBM’s multiple requests for an enterprise-management system since the early 2000s; however, this information was included as part of the draft report sent to VA for review and remains in our final report on page 14 as part of our finding on the lack of a VA-wide pharmacy inventory management system. We are sending copies of this report to the Secretary of the Department of Veterans Affairs and appropriate congressional committees. The report is also available at no charge on GAO’s website at http://www.gao.gov. If you or your staff has any questions regarding this report, please contact Sharon M. Silas at (202) 512-7114 or silass@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix II. Appendix I: Comments from the Department of Veterans Affairs Appendix II: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Rashmi Agarwal, Assistant Director; Nick Bartine, Analyst-in-Charge; Muriel Brown; Kaitlin Farquharson; Krister Friday; Sandra George; Courtney Liesener; Diona Martyn; and Michelle Paluga made key contributions to this report.

Why GAO Did This Study VA provides health care services, including pharmacy services, to approximately 9 million veterans each year. Since 2000, VAMCs have faced recurring challenges in managing their pharmacy inventories, including difficulties with accurately accounting for and updating inventory totals through their pharmacy systems. GAO was asked to review VA pharmacy inventory management. This report (1) describes approaches selected VAMCs use to manage their pharmacy inventories and (2) assesses the extent to which VA oversees VAMCs' efforts to manage their pharmacy inventories. To conduct this work, GAO visited a non-generalizable selection of five VAMCs chosen for the complexity of services offered and variation in location. GAO also reviewed VA national policies and local polices for the selected VAMCs and interviewed VA officials at the headquarters, regional, and local levels. GAO assessed VA's oversight of pharmacy management in the context of federal internal control standards. What GAO Found Selected Department of Veterans Affairs' (VA) medical centers (VAMC) use generally similar approaches for managing their pharmacy inventories. For example, all VAMCs store certain medications in secured areas. However, GAO found that VAMCs have also taken unique approaches for procuring and tracking medications, as allowed under VA policy. For example, to better address medication shortages, one VAMC pharmacy GAO visited established a shortage committee that meets on a weekly basis. Another VAMC pharmacy uses an automated dispensing machine together with compatible software that allows the pharmacy to track the location of most inpatient medications in real-time (see figure). GAO also found that VA's oversight of VAMCs' pharmacy inventory management is limited as VA lacks a comprehensive inventory management system or a focal point for system-wide oversight. In May 2018, VA signed a contract for a new electronic health records system that should allow VA to monitor VAMCs' inventories; however, VA officials expect implementation of this system to take up to 10 years. Based on a review of VA policies and interviews with VA officials, GAO found that VA has not designated a focal point with defined responsibilities for system-wide oversight of VAMCs' pharmacy inventory management. This is inconsistent with federal internal control standards for monitoring and establishing structure and authority to achieve an entity's objectives. Without a focal point for system-wide oversight, VA has limited awareness of the unique approaches that VAMCs use to manage their inventories and is missing an opportunity to evaluate these approaches. Additionally, VA cannot effectively share and standardize inventory management best practices as appropriate. Having a focal point is especially important given that VAMCs have historically had challenges in managing their inventories and a comprehensive pharmacy inventory management system may not be available for the foreseeable future. What GAO Recommends GAO recommends that VA designate a focal point for overseeing VAMCs' pharmacy inventory management efforts system-wide and define the focal point's responsibilities. VA concurred in principle with the recommendation.

gao_GAO-19-75

gao_GAO-19-75_0

Background The federal government has recognized 573 Indian tribes as distinct, independent political communities with certain powers of sovereignty and self-government, including some power to manage the use of their territory and resources and control economic activity within their jurisdiction. Some tribal lands include reservations—land set aside by treaty or other agreement with the United States, executive order, or federal statute or administrative action for the residence or use of an Indian tribe. Some tribal lands include parcels with different ownership; for example, parcels may be held in trust by the federal government for the benefit of a tribe or an individual tribal citizen. Trust and restricted lands can affect a tribe’s ability to use their land as collateral to obtain a loan. Tribal lands vary in size, demographics, and location. For example, the smallest in size are less than one square mile, and the largest, the Navajo Nation, is more than 24,000 square miles (the size of West Virginia). Tribal land locations can range from extremely remote, rural locations to urban areas. Indian tribes may form governments and subsidiaries to help manage tribal affairs including schools, housing, health, and economic enterprises. Internet access in the United States is generally privately financed. Broadband providers build infrastructure and sell broadband services to individual consumers. We previously reported that tribal lands can have conditions that increase the cost of broadband deployment, such as remote areas with challenging terrain, which increases construction costs, as well as relatively low population densities and incomes that make it difficult to recoup deployment costs. These conditions may make it less likely that a service provider will build or maintain a network. Some tribal governments provide Internet access to their members, through an information-technology or utility department, and others have created their own telecommunications companies to provide services. FCC has reported that in many instances, tribal governments must build and pay for their own communications infrastructure to ensure Internet access will be “delivered across Indian Country.” The term “broadband” commonly refers to Internet access that is high speed and provides an “always-on” connection, so users do not have to reestablish a connection each time they access the Internet. Telecommunications providers use a range of technologies to provide broadband service, including cable, fiber, satellite, and wireless. Wireless broadband connects users to the Internet using spectrum to transmit data between the customer’s location and the service provider’s facility, and can be transmitted using fixed wireless and mobile technologies, as shown in figure 1. Fixed wireless broadband technologies establish an Internet connection between fixed points, such as from a radio or antenna that may be mounted on a tower, to a stationary wireless device located at a home. This technology generally requires a direct line of sight, and can be delivered two ways: (1) as a point-to-point transmission—between two fixed points—or (2) as a point-to-multipoint transmission—from one point to multiple users. Mobile wireless broadband technologies also establish a connection to the Internet that requires the installation of antennas, but this technology provides connectivity to customers wherever they are covered by service, including while on the move, such as with a cell phone. Spectrum is the resource that makes wireless broadband connections possible. Spectrum frequency bands each have different characteristics that result in different levels of ability to cover distances, penetrate physical objects, and carry large amounts of information. For example, lower frequency bands are able to transmit signals that travel greater distances, thus requiring the use of fewer antennas, and are able to penetrate solid objects. Higher frequency bands are able to transmit more data, but are more easily obstructed. FCC administers spectrum for nonfederal users—such as state, local government, and commercial entities—through a system of frequency allocation and assignment. Allocation involves segmenting the radio spectrum into bands of frequencies designated for use by particular types of radio services or classes of users, such as commercial and nonfederal broadband services. Examples of some of the frequency bands that can be used by commercial and nonfederal entities for broadband services are shown in figure 2. Appendix II presents a full list of the auctioned licensed frequency bands that FCC told us could be used to provide broadband services. The frequency bands that can be used for broadband services are either licensed or unlicensed. For licensed spectrum, FCC can assign licenses through auctions, in which prospective users bid for the exclusive rights to transmit on a specific frequency band within geographic areas, ensuring that interference does not occur. License holders may sell or lease their license, in whole or in part, to another provider, a process that is known as a secondary market transaction, with FCC’s approval. FCC requires license holders to meet specified buildout requirements within a specified amount of time or face penalties, typically termination of all or part of the license. These buildout requirements are designed to ensure that licensees put spectrum to use within a specific period rather than let it sit idle and vary based on the type of license. FCC has also assigned licenses administratively in two frequency bands that can be used for broadband services. Specifically, prior to 1996 FCC assigned geographic licenses for exclusive use in the Educational Broadband Service (2496-2690 megahertz (MHz)), and from 2005 to 2015, FCC assigned non-exclusive nationwide licenses in the 3650-3700 MHz band, where use of the band may be shared by other license holders. FCC also authorizes the use of some spectrum for broadband services without a license on a non-exclusive basis. With unlicensed spectrum, an unlimited number of users can share frequencies using wireless equipment certified by FCC, such as wireless microphones, baby monitors, and garage door openers. In contrast to users of licensed spectrum, unlicensed users have no regulatory protection from interference by other licensed or unlicensed users in the bands. If multiple users are operating simultaneously on the same frequency band, the transmissions may be susceptible to interference, which reduces the quality of service. FCC’s rulemaking process includes multiple steps as outlined by law with opportunities for the public to participate during each step. In general, FCC initiates a rulemaking in response to statutes, petitions for rulemaking, or its own initiative, and releases a Notice of Proposed Rulemaking (NPRM) to propose new rules or to change existing rules. Any interested person may submit comments as part of the public record through electronic filings and meetings with FCC officials. Following internal analysis of the public record, FCC staff may propose actions for consideration for a vote, such as adopting final rules, amending existing rules, or stating that there will be no changes. All of FCC’s sitting commissioners vote on these items. The American Recovery and Reinvestment Act of 2009 directed FCC to develop a plan to ensure every American had access to high-speed Internet. In March 2010, an FCC task force issued the National Broadband Plan that included a centralized vision for achieving affordability and maximizing use of high-speed Internet. The plan made many recommendations to FCC, including that FCC should take into account the unique spectrum needs of tribal communities when implementing spectrum policies and evaluate its policies and rules to address obstacles to spectrum access by tribal communities. With regard to tribal lands, the plan recommended that FCC increase its commitment to government-to-government consultation with tribal leaders and consider increasing tribal representation in telecommunications planning. FCC established the Office of Native Affairs and Policy (ONAP) in July 2010 to promote the deployment and adoption of communication services and technologies to all native communities, by, among other things, ensuring consultation with tribal governments pursuant to FCC policy. Selected Tribal Entities Reported Barriers to Obtaining Licensed Spectrum Few Tribal Entities Have Obtained Spectrum Licenses, Although Representatives from Selected Tribal Entities Emphasized the Importance of Licensed Spectrum Through our analysis of FCC license data as of September 2018, we identified 18 tribal entities that held active spectrum licenses in bands that can be used to provide broadband services. Because tribal entities may hold licenses using entity names that do not include the search terms we identified in our review of the list of tribes in the Federal Register, there may be additional tribal entities that we have not identified. We found that most of the tribal entities obtained the licenses through FCC administrative assignment rather than through an FCC spectrum auction or secondary market transaction. Thirteen of the tribal entities we identified in FCC’s license data held administratively assigned licenses, and these licenses are subject to certain limitations and were only available to applicants for limited time periods. Eleven of these administratively assigned licenses are non- exclusive nationwide licenses in the 3.65 GHz frequency band (3550- 3700 MHz) and were available between 2005 and 2015, when FCC issued a new rule for this band and stopped accepting new applications for these licenses. Two of the tribal entities we identified held administratively assigned Educational Broadband Service licenses in the 2.5 GHz frequency band (2496-2690 MHz). These licenses allow for the transmission of educational materials by accredited educational institutions and government organizations, including tribes, engaged in formal education and require that licensees use the spectrum for educational purposes for a certain amount of time each week. Both of these tribal entities obtained these licenses after the last filing window closed in 1996 through a waiver and special temporary authority permit. Four tribal entities we identified in FCC’s license data held a total of 13 active licenses obtained through secondary market transactions, such as leases and sales of portions of partitioned licenses. Of these 13 secondary market transactions, 2 involved nationwide providers. Two of the tribal entities we identified held active licenses in bands available for broadband deployment that they obtained through an FCC spectrum auction. One of these tribal entities won with a winning bid of over $800,000 in a 2015 auction, and the other won two licenses with winning bids of under $50,000 in a 2002 auction. This second tribal entity also qualified for but did not win a 2003 auction. In addition to these two tribal entities, we identified the following four tribal entities that had applied to participate in auctions with varying results but did not hold active licenses in frequency bands available for broadband deployment as of September 2018: Two tribal entities each won a single spectrum license. The first won its license, which has since expired, in 2000, and the second won its license, which it has since been transferred to a nationwide provider through a secondary market transaction, in 2003. The first tribal entity also applied but did not qualify to participate in a 2001 auction. One tribal entity qualified to participate but did not win in a 2003 auction, and another tribal entity applied but did not qualify to participate in a 2008 auction. In addition, representatives from 2 of the 16 tribal entities we interviewed that were using wireless technologies told us that they use licensed spectrum that is owned by a private provider through a partnership relationship. We have previously reported that some tribes have formed partnership arrangements with other entities to increase broadband access on tribal lands. Most (14 of 16) of the tribal entities we contacted that were using wireless technologies told us that they are accessing various unlicensed bands, such as the 2.4 GHz and 5 GHz bands, to provide service. Representatives from eight of these tribal entities reported using only unlicensed spectrum for their fixed wireless networks. Representatives from 13 tribal entities told us that unlicensed spectrum had the advantage of being free, and representatives from one tribal entity told us that the equipment needed to access these spectrum bands is less expensive than equipment for accessing other spectrum bands. Representatives from some tribal entities reported success in using unlicensed spectrum in certain circumstances. For example, one tribal entity reported using unlicensed spectrum for homes in remote areas where the only potential signal degradation is from trees as well as to set up local hot spots that can serve 5 to 10 users at a time. Another tribal entity reported using primarily unlicensed spectrum to carry signals to end users together with non-exclusive licensed spectrum (3.65 GHz band) for locations where there is congestion in the unlicensed bands. However, representatives from the tribal entities we contacted that were using wireless technologies emphasized the advantages of licensed spectrum and discussed their experiences with the limitations of unlicensed spectrum. As described earlier, exclusive-use spectrum licenses protect license holders from interference from other users, whereas unlicensed spectrum provides no protection against interference. Representatives from 13 of 16 tribal entities identified the fact that unlicensed spectrum is available at no cost as an advantage of this type of spectrum. However, representatives from 15 of the 16 tribal entities identified limitations associated with unlicensed spectrum, such as interference, as described in table 1. Tribal associations, an academic group, a tribal consortium, and FCC have all highlighted the importance of exclusive-use licensed spectrum for tribal entities. Specifically, both a tribal association and an academic group we contacted discussed interference and other challenges of unlicensed spectrum. Representatives from one tribal association pointed out that unlicensed spectrum might not be available in the future if it is allocated for other purposes. Representatives from a tribal consortium we contacted told us that they are already using all of the available unlicensed spectrum for providing Internet access and that they cannot expand service without encountering interference and capacity limitations. Lastly, ONAP reported in 2012 that unlicensed spectrum is not an option across all tribal lands and that tribal access to robust licensed spectrum is a critical need. Representatives from the stakeholders we interviewed told us that there are also non-technological benefits for tribal entities to obtain greater access to licensed spectrum. For example: Enhanced ability to deliver additional Internet service. Representatives from one of the tribal associations, an academic group, and six of the tribal entities said that increased access to licensed spectrum would enable them to deliver their own Internet services and bridge service gaps, thus improving Internet access to their members. For example, representatives from three of these tribal entities said that such access would enable them to deploy in areas where providers that currently hold licenses were not willing to deliver services. In addition, representatives from another tribal entity said that having access to licensed spectrum is one factor that would enable the tribe to establish its own telecommunications company. Ability to sell or lease spectrum for profit. Representatives from one tribal association, an academic group, and two tribal entities told us that holding spectrum licenses would enable tribal governments to sell or lease their licenses. For example, we heard from one of these tribal entities that it was able to sell portions of its license that did not cover tribal lands and to use the profits from the sale to invest in its own network infrastructure. Opportunities for federal funding. Access to licensed spectrum may also provide tribal entities with more opportunity to obtain federal funding, specifically through two Universal Service Fund programs— the Mobility Fund and the Tribal Mobility Fund. These programs provide funding to broadband service providers to expand service in areas where it is not available, including tribal lands. However, service providers must hold, lease, or show they have access to licensed spectrum to participate in these programs, among other requirements. For example, the National Congress of American Indians stated that two tribal entities submitted applications to participate in the Mobility Fund program but were not eligible to participate in part because they did not hold a spectrum license. Moreover, representatives from two of the tribal entities we interviewed told us that they considered applying for one of these programs but realized they were ineligible because they did not have access to licensed spectrum. Furthermore, representatives from one of the tribal associations, an academic group, and seven of the tribal entities told us that having access to licensed spectrum would enable tribes to exercise their rights to sovereignty and self-determination. Representatives from three of the tribal entities we contacted said that they view spectrum as a natural resource that should be managed by the tribe. FCC officials, however, told us that spectrum is not considered a reserved right under treaties with Indian tribes, as it is not explicitly stated. In addition, representatives from four of the tribal entities told us that having access to licensed spectrum would ensure that spectrum is being used in a way that aligns with tribal goals and community needs, further supporting their rights to self-determination. Representatives from Selected Tribal Entities Reported Cost and Other Barriers to Accessing Licensed Spectrum Representatives from the tribal entities we contacted identified several barriers to accessing licensed spectrum through spectrum auctions and secondary market transactions. Regarding spectrum auctions, representatives from tribal entities that provide wireless Internet service most frequently (13 of 16) indicated that spectrum licenses are too expensive for tribal entities. For example, over 60 percent (983 of 1,611) of the winning bids from a 2015 spectrum auction, including bids for spectrum over non-tribal lands, were over $1 million. Representatives from one tribal entity explained that auction licenses are often too expensive for tribal entities because these licenses cover large geographic areas that may include non-tribal urban areas as well as rural tribal areas. Moreover, representatives from eight tribal entities stated that they are unable to obtain financing to participate in auctions because tribal governments cannot use tribal lands as collateral to obtain loans. In addition, representatives from eight tribal entities mentioned that participating in spectrum auctions requires auction-specific expertise that tribal entities may not have. Tribal entities also face barriers obtaining spectrum through secondary market transactions. Most of the spectrum allocated for commercial use has already been assigned through spectrum auctions and other mechanisms to private providers, including licensees that may not be providing service on tribal lands. In a single geographic area, several frequency bands could be used to deploy broadband services, as shown in figure 2, and licenses for these various frequency bands may be held by different providers. There may be tribal areas where providers hold licenses for bands but are not using the spectrum to provide Internet access. In other tribal areas, services may be offered using one or two of the spectrum licenses with the other licenses in the area remaining fallow and inaccessible to tribal entities. All three of the tribal associations we contacted confirmed that there are unused spectrum licenses over tribal lands, and representatives from a nationwide provider indicated that they only deploy services if there is a business case to support doing so. Accordingly, the secondary market is one of few avenues available to tribal entities that would like to access licensed spectrum. However, representatives from tribal entities we contacted identified the following challenges related to participating in the secondary market: Lack of willing sellers. Representatives from eight of the tribal entities, one of the tribal representative groups, and an industry association we contacted indicated that spectrum license holders are often unwilling to participate in secondary market transactions, citing a variety of reasons. For example, representatives from one tribal entity stated that large carriers have no business incentive to negotiate secondary market agreements with tribal entities and that tribal entities do not have the resources to make such transactions sufficiently lucrative for license holders. Representatives from another tribal entity stated that license holders may lack knowledge about the areas covered by their licenses, including tribal areas, and therefore may be unwilling to consider secondary market transactions. Representatives from a tribal representative group told us that license holders may be unwilling to consider secondary market transactions with tribal entities because spectrum is a valuable resource that may become even more valuable over time, and a representative from an industry association indicated that transaction costs such as legal fees outweigh any potential income from such transactions. None of the private providers we contacted reported entering into a secondary market transaction with tribal entities, but one of these providers stated that it had never been approached by a tribal entity interested in a secondary market transaction and was unaware of challenges that are unique to tribal entities. License holders unknown. Representatives from eight of the tribal entities we contacted stated that it is difficult to determine who holds spectrum licenses. For example, two tribal entities had to hire consultants to identify who held licenses for spectrum over the tribes’ lands, and another tribal entity relied on the expertise of its non-tribal partner to identify the license holders. Unaware of secondary market transactions. Representatives from six of the tribal entities we contacted were unaware of the possibility of accessing licensed spectrum through a secondary market transaction prior to our contacting them. Accordingly, secondary market transactions involving tribal entities are rare. As discussed above, our analysis of FCC license data identified four tribal entities that have successfully accessed licensed spectrum in this manner. Regarding one of these tribal entities’ experiences with the secondary market, the tribal representative we contacted stated that an Indian-owned telecommunications consulting company was pivotal in identifying the license holder and facilitating the transaction and that the transaction would not have happened without the consulting company. Representatives from this company told us that they conducted an analysis to identify unused spectrum licenses over the tribe’s land. The company identified three providers holding such licenses, but only one of those providers was willing to participate in a secondary market transaction. Representatives from another of the tribal entities that accessed licensed spectrum through the secondary market told us that they relied on the expertise of their non-tribal partner to facilitate these transactions. FCC Has Some Efforts to Enhance Tribal Access to Spectrum, but FCC Does Not Collect or Communicate Key Information to Tribal Entities FCC Has Taken Steps to Promote and Support Tribal Entities’ Ability to Obtain Spectrum, However These Efforts Are Not Likely to Address Tribal Spectrum Needs We found that FCC has taken the following actions to increase tribal access to and use of spectrum: (1) initiated proposed rulemakings on promoting tribal access to spectrum, (2) adopted rules to increase spectrum available for broadband use, and (3) conducted outreach and training for tribal entities on spectrum-related issues. Initiated Proposed Rulemakings on Promoting Tribal Access to Spectrum FCC issued two NPRMs—one in March 2011 and one in May 2018—that included policy options intended to enhance tribal access to spectrum. At the time of our report, FCC had not adopted new rules or taken further action on the 2011 rulemaking, and FCC had not taken further actions since the comments period ended on September 7, 2018, on the May 2018 rulemaking. According to FCC officials, the 2011 NPRM addressed several recommendations made in the National Broadband Plan to promote the greater use of spectrum over tribal lands. Among other things, the 2011 NPRM sought comments on three proposals to create new spectrum access opportunities for tribal entities (see fig. 3). FCC officials told us that they have reviewed public comments to the proposed rulemaking, but have no current plans to take further actions. We reviewed the public comments FCC received that pertained to the three proposals, which included comments from tribal associations, tribal governments, rural and nationwide industry associations, and tribal and private providers. Based on our analysis of the comments that included positions on the proposal for a tribal licensing priority, eight stakeholders—including industry associations, private providers, and a tribal government—were supportive of this proposal. However, we found that stakeholder views differed on implementing good faith negotiations and on the build-or-divest processes. In general, the tribal stakeholders indicated that they were supportive of these proposals, while the industry associations and private providers were not. In addition to reviewing the public comments, we asked representatives from the tribal and industry associations and private providers that we interviewed about their views of these proposals. Representatives from the three tribal associations and two rural industry associations were generally supportive of all three of the proposals, while representatives from one of the private providers that we interviewed told us they did not support any of the three proposals, because, for example, they said that there are more effective ways to increase broadband service over tribal lands. Representatives from another private provider said that they supported the tribal priority process but did not indicate their views on the other two proposals. Representatives from six tribal entities and a representative from a tribal consortium told us that these types of proposals would help them obtain spectrum. In May 2018, FCC issued an NPRM that sought comments on establishing a tribal priority window for tribal nations located in rural areas as part of a process to re-license the Educational Broadband Service spectrum band. As described above, FCC originally allocated this band to qualifying educational institutions and government organizations for the transmission of educational materials. While FCC permitted licensees to lease their excess capacity to commercial providers, FCC reported that significant portions of this band were not being used, primarily in rural areas. In an effort to make additional spectrum available for broadband use, FCC issued this NPRM seeking comments on options to promote the use of this spectrum over tribal lands. One of the options included implementing a local priority filing window so that tribal entities could get access to unassigned spectrum prior to an FCC auction. In a June 2018 order, FCC extended the comment deadline for the NPRM to August 8, 2018, partly in response to a request for a deadline extension. As a result, FCC also extended the deadline to respond to those comments to September 7, 2018. Because FCC was in the process of responding to these comments at the time of our review, we did not analyze these comments. Adopted Rules to Allocate Additional Spectrum for Broadband Use FCC has made additional unlicensed and licensed spectrum available for broadband use and has implemented rules that according to FCC, may make it easier for rural providers to obtain licenses. However, these efforts were not targeted to tribal entities, and according to ONAP’s 2012 report, allocating additional unlicensed spectrum may not be a technically feasible solution for all tribal entities, and such spectrum may not have the necessary capacity to handle an increase in users. In addition, representatives from the tribal associations and entities we contacted told us that there are limitations to the extent that these efforts can address the spectrum needs of tribal entities. In particular, they discussed the effect of FCC’s changes to the rules on the use of TV white space spectrum and the Citizens Broadband Radio Service spectrum: TV white space spectrum: In 2010, FCC made additional unlicensed spectrum available for broadband use by allowing providers to operate in the TV bands at locations where those frequencies were not in use, known as TV white space, but none of the tribal entities we interviewed was using this spectrum. A representative from a tribal consortium said that it used TV white space spectrum, and representatives from three of the tribal entities said that they were considering using it in the future because TV white space spectrum can better pass through some environmental barriers, such as trees, reaching more remote customers. However, representatives from five tribal entities, one tribal consortium, one academic group, and three companies that we interviewed told us about several limitations to the use of TV white space spectrum. For example: limited bandwidth capacity, which causes lower speeds, high latency, and limits the number of households that can be served; equipment needed to access TV whitespace spectrum is expensive and less available; the spectrum may not always be available; and similar to other unlicensed frequency bands, as described above, there is potential for interference and difficulty to pass through extreme terrain. Citizens Broadband Radio Service (CBRS) Spectrum: In 2015, FCC made additional licensed spectrum available for broadband use when it issued a new rule for the 3.65 GHz frequency band (3550-3700 MHz). However the tribal entities who held licenses in this band indicated there are limitations to their ability to use this band and their future use of this spectrum remains unknown. As described earlier, FCC had allocated non-exclusive nationwide licenses in this band. In the 2015 rule, FCC created the CBRS, increased the amount of spectrum allocated for commercial broadband use, and implemented a new licensing scheme. This three-tier priority licensing scheme for spectrum sharing included auctioning exclusive-use geographic licenses and allowing non-exclusive use of the band where a license holder is not operating, an approach that is intended to provide a low- cost entry point for users, but will have no protections from interference. Representatives from four of the five tribal entities that we contacted that held licenses in this band said that there were technical advantages to using it, such as the ability for a signal to pass through dense forests. However, representatives from two tribal entities said that the high cost of the equipment needed to access this spectrum prevented them from either using the frequency band extensively or at all. In addition, representatives from two tribal entities said that they were not sure about their ability to access this band in the future given the changes made in FCC’s 2015 rulemaking. FCC’s 2015 rule also created small-sized and shorter-termed licenses, which FCC stated would decrease the costs of obtaining a license and help rural providers access it. However, FCC issued an NPRM in 2017 that sought comments on suggested changes to CBRS, including increasing the geographic area covered by licenses and lengthening the license term. In October 2018, FCC adopted rules that, among other changes, increased the license area from census tracks to counties and extended the license term from 3 to 10 years, which FCC officials told us were modest changes made to accomplish FCC’s goals of creating incentives for investment, including in urban and rural areas, encouraging efficient spectrum use, and promoting robust network deployments. Conducted Outreach and Training for Tribal Entities on Spectrum-Related Issues FCC’s ONAP conducts training, consultation, and outreach to tribal entities on spectrum-related issues. For example, ONAP officials told us that they have conducted 21 training and consultation workshops for tribal entities on broadband and telecom since 2012, where spectrum has been discussed in general in the introduction and has been addressed specifically in separate sessions in some of the workshops. These officials also told us that they communicate with tribal entities prior to when FCC holds auctions or when implementing regulatory actions or policies that will affect tribal governments and spectrum over their lands. While representatives from 9 of the 16 tribal entities using wireless technologies told us that they had received some outreach on spectrum- related issues from FCC, representatives from 2 of these entities said that they had not. In addition, ONAP issued a report in 2012 to provide FCC with a review of its work with tribal governments and organizations, including information on its tribal broadband efforts, priorities, and tribal consultations. Among other things, the report included case-study information on tribal entities’ efforts to access spectrum. Although the report stated that this would be the first of such annual reporting, this is the only report that ONAP has issued on tribal issues. According to ONAP officials, ONAP has not published subsequent reports because it provides FCC with information on its work with tribal governments and organizations, including spectrum-related matters, through more frequent informal briefings and regular updates. FCC Does Not Collect Key Information Related to Spectrum over Tribal Lands or Communicate It to Tribal Entities FCC has not consistently collected information related to tribal access to spectrum. For example, FCC does not collect data on whether holders of spectrum licenses or auction applicants are tribal entities even though it collects self-reported data on licensee type, such as corporation and government entity. To obtain this information, FCC could include an option for the licensee type, along with the other options, in applications for future licenses and auctions that allows an applicant to identify as a tribal government or tribally owned entity. FCC officials told us that they use information on licensee type to determine eligibility for a license. Because eligibility is not based on whether the applicant is a tribal entity, FCC officials said this information is not needed. However, without this information, FCC does not have a comprehensive understanding of the extent that tribal entities are attempting to obtain or access licensed spectrum or have been successful at obtaining and accessing it. Additionally, FCC does not analyze information on unused licensed spectrum that exists over tribal lands, even though FCC has information—broadband availability data from providers and information on geographic areas covered by spectrum licenses—that could be used for such analysis. As we described earlier, representatives from all three of the tribal associations we contacted reported that there are unused spectrum licenses over tribal lands that could present opportunities through the secondary market for tribal entities to obtain spectrum. When we asked FCC officials why they do not analyze the extent that unused spectrum licenses exists over tribal lands, they told us that the spectrum data noted above is not specific enough to allow for a license by license analysis of unused spectrum. For example, they said that broadband availability data from providers is aggregated across wide spectrum bands to minimize reporting burdens on the wireless industry, and the data are not sufficiently detailed to identify which spectrum blocks and licenses are being used in particular areas. However, FCC could use this data to conduct, at a minimum, high-level analysis that would result in useful information on the extent to which unused spectrum exists over tribal lands. In addition, FCC officials told us that they evaluate the effectiveness of FCC’s secondary markets policies, which FCC views as a mechanism to promote the increased use of unused spectrum licenses, but this approach does not include an analysis of unused spectrum licenses as part of these efforts. As a result, FCC’s evaluations of the secondary market may not accurately reflect how these policies affect tribal access to spectrum. Because the secondary market is one of few ways for tribal entities to access licensed spectrum, an analysis of unused licensed spectrum that exists over tribal lands would enable FCC to better promote a robust secondary market that provides additional opportunities for tribes to access spectrum. FCC’s 2010 National Broadband Plan stated that ongoing measurement of spectrum utilization should be developed to better understand how spectrum resources are being used because some studies indicated that spectrum goes unused in many places much of the time. The plan also stated that any spectrum utilization studies that FCC conducts should identify tribal lands as distinct entities. In FCC’s February 2018 strategic plan, FCC stated that it will implement ongoing initiatives that will assist in spectrum policy planning and decision making, promote a robust secondary market in spectrum, and improve communications services in all areas of the United States, including tribal areas. Additionally, Standards for Internal Control in the Federal Government state that agencies should use quality information, including information that is complete, to inform the decision-making processes. FCC also does not make information on spectrum-license holders available in an easy or accessible manner; such information could be beneficial to the tribes in their efforts to obtain spectrum in the secondary market. As described earlier, the secondary market is a significant mechanism for tribal entities to obtain spectrum licenses, but representatives from the tribal entities we interviewed reported challenges related to participating in the secondary market, such as not knowing whom to contact should they wish to engage in a secondary market transaction to obtain a spectrum license. In July 2014, FCC stopped updating its spectrum dashboard, which provided the public with a way to identify who holds licenses in what areas, including features that allowed users to identify spectrum allocated and assigned in tribal lands. ONAP stated in its 2012 report that this feature represented the first step for individual tribal entities to reach out to licensees and seek leasing, partnership, or other arrangements that could ultimately result in the provision of service over tribal lands. FCC officials told us that the public may view electronic records of all wireless spectrum licenses in FCC’s Universal Licensing System, using a wide range of license and geographic parameters, such as licensee names, radio services, spectrum bands, and geographic locations. However, we attempted to navigate the Universal Licensing System to determine spectrum-license holders for specific tribal lands using geographic parameters, but we were unable to successfully do so because the system is so difficult to use. Furthermore, as described above, representatives from eight of the tribal entities that we contacted stated that it is difficult to determine who holds spectrum licenses. When we asked FCC officials why they do not communicate information to tribes about spectrum-related transactions over tribal lands, FCC officials also told us that they issue public notices on applications for all proposed spectrum transactions and on the winning bidders of all auctions, but they have not made it a practice to reach out directly to tribes to make them aware of when providers have obtained spectrum licenses that cover tribal lands. The National Broadband Plan stated that FCC should make data available that would promote a robust secondary market for spectrum licenses, such as information on how and to whom spectrum is allocated on tribal lands. Additionally, Standards for Internal Control in the Federal Government state the need for federal agencies to communicate with external entities and to enable these entities to provide quality information to the agency that will help it achieve its objectives. Tribal governments are an example of such external entities. The ability of tribal governments to make informed spectrum planning decisions and to participate in secondary market transactions is diminished without information from FCC on the spectrum transactions that occur over tribal lands. Providing this information in a manner that is accessible and easy for tribal entities to obtain could enable them to enter into leasing, partnership, or other arrangements to obtain spectrum. Conclusions Broadband service on tribal lands continues to lag behind the rest of the country, especially on rural tribal lands, which could hinder tribal efforts to promote self-governance, economic opportunity, education, public safety, and cultural preservation. FCC has reported that wireless technologies that access spectrum to deliver broadband services are cost-effective for remote and sparsely populated areas, such as tribal lands. However, FCC’s efforts to promote and support tribal entities’ access to spectrum have done little to increase tribal use of spectrum, as only very few tribes are accessing spectrum to be able to provide Internet service. Additionally, FCC lacks information that could help inform its decision- making processes related to spectrum policy planning, which is intended to improve communications services in all areas of the United States, including tribal lands. By collecting data on the extent that tribal entities are obtaining and accessing spectrum, FCC could better understand tribal spectrum issues and use this information as it implements ongoing spectrum initiatives. Furthermore, the secondary market is one of few ways for tribal entities to access licensed spectrum to be able to provide Internet service, and FCC has recognized the importance of promoting a robust secondary market. FCC could promote a more robust secondary market by analyzing data to better understand how much unused licensed spectrum exists over tribal lands and using that information to promulgate regulations and to evaluate how FCC policies affect tribal participation in the secondary market. Additionally, by making information on who holds spectrum licenses over tribal lands more accessible and easy to understand, FCC could remove a barrier tribes may face in attempting to obtain spectrum through the secondary market. Recommendations for Executive Action We are making the following three recommendations to the Chairman of FCC. The Chairman of FCC should collect data on the extent that tribal entities are obtaining and accessing spectrum and use this information as FCC implements ongoing spectrum initiatives. (Recommendation 1) The Chairman of FCC should analyze data to better understand the extent that unused spectrum licenses exist over tribal lands, such as by analyzing the data for a sample of tribal lands, and as appropriate use this information to inform its oversight of the secondary market. (Recommendation 2) The Chairman of FCC should make information on spectrum-license holders more accessible and easy to understand for interested parties, including tribal entities, to promote their ability to purchase or lease spectrum licenses from other providers. (Recommendation 3) Agency Comments We provided a draft of this report to FCC for comment. In its comments, reproduced in appendix III, FCC agreed with the recommendations. FCC also provided technical comments, which we incorporated as appropriate. We are sending copies of this report to the appropriate congressional committees, the Chairman of FCC, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-2834 or goldsteinm@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix IV. Appendix I: Objectives, Scope, and Methodology This report examines (1) what is known about the ability of tribal entities to obtain and access spectrum to provide broadband services on tribal lands and the reported barriers that may exist; and (2) the extent to which the Federal Communications Commission (FCC) promotes and supports tribal entities’ ability to obtain and access spectrum for broadband services. To address both objectives, we reviewed relevant statutes and regulations and FCC documents, including FCC’s Statement of Policy on Establishing a Government-to-Government Relationship with Indian Tribes, the National Broadband Plan, and FCC’s current strategic plan. We interviewed FCC officials and representatives from 3 tribal associations, 7 private providers that deliver Internet services over tribal lands, 3 industry associations that represent rural and urban telecommunications providers, 3 regional consortia, 3 companies that work with tribal entities, and 1 academic group. In addition, we selected 24 tribal entities—13 Indian tribes and nations and 11 tribally owned providers—to interview. To identify tribal entities that were using wireless technologies, we obtained recommendations from stakeholders, reviewed data on relevant federal grants, such as the Broadband Technology Opportunities Program, and conducted Internet research. We then selected 16 tribal entities considering (1) stakeholder suggestions, (2) population, (3) population density, and (4) urban or rural designation. We visited 7 of these tribes in Idaho, New Mexico, and Washington State. The views represented in our report are not generalizable to those of all stakeholders. See table 2 for a complete listing of the entities we interviewed. We also conducted a literature review to identify relevant academic, government, and media publications that were published between January 1, 2013, and January 11, 2018, that discuss the importance of and options to enhance tribal access to spectrum. To identify tribal entities that applied to participate in spectrum auctions or that held active spectrum licenses in bands that can be used to provide broadband service, we analyzed (1) FCC data on entities that applied to participate in auctions for spectrum in these bands and (2) FCC data on spectrum licenses in these bands that were active as of September 6, 2018. We also analyzed FCC license data, together with license information publicly available through FCC’s Universal Licensing System, to determine whether the tribal entities that held active licenses obtained those licenses through an FCC spectrum auction, administrative assignment, or the secondary market. We then reviewed the list of federally recognized tribes in the Federal Register and identified search terms related to these tribes. For example, we identified the following search terms based on the federally recognized tribe, Yurok Tribe of the Yurok Reservation, California, “Reservation, Tribe, and Yurok.” We then used the identified search terms to search for tribal entities in FCC’s data on auction participants and spectrum license holders. We manually reviewed these matches to identify tribal entities based on information from interviews, Internet research, and professional judgment. Because tribal entities may have applied to participate in spectrum auctions or may hold spectrum licenses under names not associated with their tribes, there may be additional tribal entities that we were unable to identify. Through interviews with FCC officials and review of related documentation, we determined that the license and auction participant data are sufficiently reliable for our purpose of identifying some tribal entities that have applied to participate in a spectrum auction or held active spectrum licenses as of September 2018. However, our analysis does not capture the extent that tribal entities may have obtained a license that is no longer active. To identify tribal entities that used unlicensed spectrum to deliver unlicensed service, we interviewed the tribal entities identified above. In addition, we obtained stakeholder views on the advantages and disadvantages of using unlicensed and licensed frequency bands and any barriers that tribal entities face in obtaining spectrum licenses by interviewing the selected stakeholders noted above. To determine the extent to which FCC promotes and supports tribal entities’ efforts to obtain and access spectrum, first, we reviewed FCC’s proposals in its ongoing 2011 Notice of Proposed Rulemaking In the Matter of Improving Communications Services for Native Nations by Promoting Greater Utilization of Spectrum over Tribal Lands and its ongoing 2018 Notice of Proposed Rulemaking In the Matter of Transforming the 2.5 GHz Band. We summarized public comments submitted, as of August 2018, by private and tribal providers, rural and nationwide industry associations, tribal associations, and tribal governments on FCC’s 2011 proposed rulemaking. We did not analyze comments on FCC’s 2018 Notice of Proposed Rulemaking because FCC was in the process of responding to these comments at the time of our review. Second, we reviewed rules that FCC officials identified that increased the availability of unlicensed and licensed frequency bands for broadband use and that may be particularly useful for tribal entities. These rules included FCC’s 2010 and 2012 rules related to TV white space spectrum and its 2015 rule and 2017 Notice of Proposed Rulemaking related to the Citizens Broadband Radio Services (CBRS) spectrum. We identified tribal entities that had been using CBRS frequency bands by reviewing FCC licensed data and TV white space frequency bands through interviews with tribal entities and regional consortia. Third, we identified FCC’s outreach activities to provide tribal entities with assistance on spectrum-related issues by interviewing FCC officials and reviewing documentation on the content of FCC-led trainings and workshops, e-mail correspondences, and related publications, such as public notices. Lastly, we interviewed FCC officials on the information that they collect, analyze, and report related to tribal use of spectrum and reviewed related documentation, including FCC’s Office of Native Affairs and Policy 2012 Annual Report and FCC’s license and auction data. We interviewed stakeholders, as noted above, and summarized their views of FCC efforts. We also compared FCC’s efforts against FCC’s 2018-2022 strategic plan, recommendations made in FCC’s National Broadband Plan, and Standards for Internal Control in the Federal Government related to using quality information. We conducted this performance audit from November 2017 to November 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Auctioned Licensed Spectrum Available for Commercial Broadband Services We identified the spectrum frequency bands that the Federal Communications Commission (FCC) has made available for commercial broadband services and that FCC assigns licenses through auctions. Table 3 describes these licenses, including the number and date of related auctions. Appendix III: Comments from the Federal Communications Commission Appendix IV: GAO Contact and Staff Acknowledgments GAO Contact Mark L. Goldstein, (202) 512-2834 or goldsteinm@gao.gov. Staff Acknowledgments In addition to the contact named above, Sally Moino (Assistant Director), Anne Doré (Analyst in Charge), Enyinnaya David Aja, Katherine Blair, Stephen Brown, Camilo Flores, Georgeann Higgins, John Mingus, Josh Ormond, Frank Rusco, Rebecca Rygg, Jay Spaan, Andrew Stavisky, James Sweetman, Jr., Hai Tran, and Michelle Weathers made key contributions to this report.

Why GAO Did This Study In 2018, FCC estimated that 35 percent of Americans living on tribal lands lack broadband service compared to 8 percent of Americans overall. Broadband service can be delivered through wireless technologies using radio frequency spectrum. According to FCC, increasing tribal access to spectrum would help expand broadband service on tribal lands. GAO was asked to review spectrum use by tribal entities—tribal governments and tribally owned telecommunications providers. This report examines (1) tribal entities' ability to obtain and access spectrum to provide broadband services and the reported barriers that may exist, and (2) the extent to which FCC promotes and supports tribal efforts to obtain and access spectrum. GAO interviewed 16 tribal entities that were using wireless technologies. Selected entities varied geographically, among other characteristics. GAO analyzed FCC's license and auction data as of September 2018, reviewed FCC's rulemakings on spectrum for broadband services, and interviewed other tribal and industry stakeholders and FCC officials. The information presented is not generalizable to all tribes or industry participants. What GAO Found The tribal entities GAO contacted cited various barriers to obtaining spectrum licenses in bands that can be used to provide broadband services. According to its analysis of data from the Federal Communications Commission (FCC), GAO identified 18 tribal entities that held active spectrum licenses in such bands. For example, of these 18 tribal entities, 4 obtained licenses through secondary market transactions—that is, they bought or leased the license from another provider, and 2 obtained a license through an FCC spectrum auction. Licensed spectrum is generally preferred because it offers better quality of service compared to unlicensed spectrum; however, almost all of the tribal entities GAO contacted said that they are accessing unlicensed spectrum to provide Internet service. They identified barriers to obtaining licensed spectrum through auctions and secondary market transactions, barriers such as high costs and, in the case of secondary market transactions, a lack of information on who holds licenses over tribal lands. Because most spectrum allocated for commercial use has already been assigned, the secondary market is one of very few avenues available to tribal entities that would like to access licensed spectrum. FCC has taken steps to promote and support tribal access to spectrum. For example, FCC issued proposed rulemakings in 2011 and 2018 that sought comment on tribal-specific proposals, such as establishing tribal-licensing priorities and initiating processes to transfer unused spectrum licenses to tribal entities. However, FCC has not finalized these rules and is in the process of responding to comments to the 2018 rulemaking. Also, while FCC has made additional spectrum available for broadband use in recent years, tribal stakeholders cited limitations with the spectrum FCC has made available. For example, FCC allows broadband providers to operate in unused television broadcast bands on an unlicensed basis. While stakeholders GAO interviewed cited some advantages of these bands, such as being useful to reach remote customers, they also noted technical and cost limitations that reduced the potential to improve tribal access to spectrum. FCC stated that it is implementing spectrum initiatives and recognizes the importance of promoting a robust secondary market to improve communications throughout the United States, including tribal lands. However, GAO found that FCC has not collected data related to tribal access to spectrum, analyzed unused licensed spectrum that exists over tribal lands, or made data available to tribal entities in an accessible and easy manner that could be beneficial in their efforts to obtain spectrum licenses from other providers. By collecting data on the extent that tribal entities are obtaining and accessing spectrum, FCC could better understand tribal spectrum issues and use this information as it implements ongoing spectrum initiatives. Further, given that the secondary market is one of few ways for tribal entities to access licensed spectrum to be able to provide Internet service, FCC could promote a more robust secondary market by analyzing unused licensed spectrum over tribal lands and using that information to inform FCC's oversight of the secondary market. Additionally, by making information available on who holds spectrum licenses over tribal lands, FCC could remove a barrier tribes may face in attempting to obtain spectrum through the secondary market. What GAO Recommends FCC should (1) collect data on tribal access to spectrum; (2) analyze unused licensed spectrum over tribal lands; and (3) make information available in a more accessible manner that would promote tribes' ability to purchase or lease spectrum licenses over their lands from other providers. FCC agreed with the recommendations.

gao_GAO-18-106

gao_GAO-18-106_0

Background CSPF is a defined benefit multiemployer pension plan. Multiemployer plans are often created and maintained through collective bargaining agreements between labor unions and two or more employers, so that workers who move from job to job and employer to employer within an industry can continue to accrue pension benefits within the same plan over the course of their careers. Multiemployer plans are typically found in industries with many small employers such as trucking, building and construction, and retail food sales. In 2017, there were about 1,400 defined benefit multiemployer plans nationwide covering more than 10 million participants. Multiemployer Plan Administration, Funding, and Benefits Administration Most multiemployer plans are jointly administered and governed by a board of trustees selected by labor and management. The labor union typically determines how the trustees representing labor are chosen and the contributing employers or an employer association typically determines how the trustees representing management are chosen. The trustees set the overall plan policy, direct plan activities, and set benefit levels (see fig. 1). Multiemployer plans are “prefunded,” or funded in advance, primarily by employer contributions. The employer contribution is generally negotiated through a collective bargaining agreement, and is often based on a dollar amount per hour worked by each employee covered by the agreement. Employer contributions are pooled in a trust fund for investment purposes, to pay benefits to retirees and their beneficiaries, and for administrative expenses. Multiemployer plan trustees typically decide how the trust fund should be invested to meet the plan’s objectives, but the trustees can use investment managers to determine how the trust fund should be invested. Multiemployer plan trust funds can be allocated among many different types of assets, any of which can generally be passively- or actively-managed, domestically or internationally based, or publicly or nonpublicly traded (see table 1). A plan’s funded percentage is its ratio of plan assets to plan liabilities. Because the amount needed to pay pension benefits for many years into the future cannot be known with certainty due to a variety of economic and demographic factors, including the potential volatility of asset values, estimates of a plan’s funded percentage may vary from year to year. Defined benefit pension plans use a “discount rate” to convert projected future benefits into their “present value.” The discount rate is the interest rate used to determine the current value of estimated future benefit payments and is an integral part of estimating a plan’s liabilities. The higher the discount rate, the lower the plan’s estimate of its liability. Multiemployer plans use an “assumed-return approach” that bases the discount rate on a long-term assumed average rate of return on the pension plan’s assets. Under this approach, the discount rate depends on the allocation of plan assets. For example, a reallocation of plan assets into more stocks and fewer bonds typically increases the discount rate, which reduces the estimated value of plan liabilities, and therefore, reduces the minimum amount of funding required. Looking at the entire “multiemployer system”—the aggregation of multiemployer plans governed by ERISA and insured by PBGC—shows that while the system was significantly underfunded around 2001 and 2009, its funded position has improved since 2009. Specifically, analyses published by the Center for Retirement Research at Boston College and the Society of Actuaries used plan regulatory filings to calculate the funded status for the system and determined that it was approaching 80 percent funded by 2014 after falling during the 2008 market downturn. However, some observers have noted that while many plans are making progress toward their minimum targets, a subset of plans face serious financial difficulties. Benefits Multiemployer retirement benefits are generally determined by the board of trustees. The bargaining parties negotiate a contribution rate and the trustees adopt or amend the plan’s benefit formulas and provisions. Decisions to increase benefits or change the plan are also typically made by the board of trustees. Benefit amounts are generally based on a worker’s years of service and either a flat dollar amount or the worker’s wage or salary history, subject to further adjustment based on the age of retirement. The Central States, Southeast and Southwest Areas Pension Fund (CSPF) CSPF was established in 1955 to provide pension benefits to International Brotherhood of Teamsters union members (Teamsters) in the trucking industry, and it is one of the largest multiemployer plans. In the late 1970s, CSPF was the subject of investigations by the IRS within the U.S. Department of the Treasury (Treasury), and by DOL and the U.S. Department of Justice (DOJ). The DOL investigation ultimately resulted in the establishment of a federal court-enforceable consent decree in 1982 that remains in force today. CSPF held more than $4.3 billion in Net Assets at the end of 1982 after the consent decree was established. The plan’s Net Assets peaked at nearly $26.8 billion at the end of 2007 and declined to about $15.3 billion at the end of 2016 (see fig. 2). As of 2016, CSPF reported that it had about 1,400 contributing employers and almost 385,000 participants. The number of active CSPF participants has declined over time. In 2016, 16 percent of about 385,000 participants were active, i.e., still working in covered employment that resulted in employer contributions to the plan. In comparison, CSPF reported in 1982 that 69 percent of more than 466,000 participants were active participants. Since the 1980s, CSPF’s ratio of active to nonworking participants has declined more dramatically than the average for multiemployer plans. By 2015, only three of the plan’s 50 largest employers from 1980 still paid into the plan, and for each full-time active employee there were over five nonworking participants, mainly retirees. As a result, benefit payments to CSPF retirees have exceeded employer contributions in every year since 1984. Thus, CSPF has generally drawn down its investment assets. In 2016, CSPF withdrew over $2 billion from investment assets (see fig. 3.). CSPF has historically had fewer plan assets than were needed to fully fund the accrued liability—the difference referred to as unfunded liability. In 1982, we reported that CSPF was “thinly funded”—as the January 1, 1980, actuarial valuation report showed the plan’s unfunded liability was about $6 billion—and suggested that IRS should closely monitor CSPF’s financial status. In 2015, the plan’s actuary certified that the plan was in “critical and declining” status. The plan has been operating under an ERISA-required rehabilitation plan since March 25, 2008, which is expected to last indefinitely. As of January 1, 2017, the plan was funded to about 38 percent of its accrued liability. In September 2015, CSPF filed an application with Treasury seeking approval to reduce benefits pursuant to provisions in the Multiemployer Pension Reform Act of 2014 (MPRA), which is fully discussed later in this section. The application was denied in May 2016 based, in part, on Treasury’s determination that the plan’s proposed benefit suspensions were not reasonably estimated to allow the plan to remain solvent. In 2017, CSPF announced it would no longer be able to avoid the projected insolvency. (See app. II for a timeline of key events affecting CSPF.) The Consent Decree As previously mentioned, CSPF was the subject of investigations in the 1970s by IRS, DOL, and DOJ. DOL’s investigation focused on numerous loan and investment practices alleged to constitute fiduciary breaches under ERISA, such as loans made to companies on the verge of bankruptcy, additional loans made to borrowers who had histories of delinquency, loans to borrowers to pay interest on outstanding loans that the fund recorded as interest income, and lack of controls over rental income. As a result of its investigation, DOL filed suit against the former trustees of CSPF and, in September 1982, the parties entered into a consent decree, which remains in force today. The consent decree provides measures intended to ensure that the plan complies with the requirements of ERISA, including providing for oversight by the court and DOL, and prescribes roles for multiple parties in its administration. For example, certain plan activities must be submitted to DOL for comment and to the court for approval, including new trustee approvals and some investment manager appointments. According to DOL, to prevent criminal influence from regaining a foothold of control over plan assets, the consent decree generally requires court-approved independent asset managers—called “named fiduciaries”—to manage CSPF’s investments. CSPF’s trustees are generally prohibited from managing assets; however, they remain responsible for selecting, subject to court approval, and overseeing named fiduciaries and monitoring plan performance. To focus attention on compliance with ERISA fiduciary responsibility provisions, the consent decree provides for a court-appointed independent special counsel with authority to observe plan activities and oversee and report on the plan. (See app. III for additional detail on the key provisions of the consent decree.) Legal Framework Employee Retirement Income Security Act of 1974 In 1974, Congress passed ERISA to protect the interests of participants and beneficiaries of private sector employee benefit plans. Among other things, ERISA requires plans to meet certain requirements and minimum standards. DOL, IRS, and PBGC are generally responsible for administering ERISA and related regulations. DOL has primary responsibility for administering and enforcing the fiduciary responsibility provisions under Part 4 of Title I of ERISA, which include the requirement that plan fiduciaries act prudently and in the sole interest of participants and beneficiaries. Treasury, specifically the IRS, is charged with determining whether a private sector pension plan qualifies for preferential tax treatment under the Internal Revenue Code. Additionally, the IRS is generally responsible for enforcing ERISA’s minimum funding requirements, among other things. ERISA generally requires that multiemployer plans meet minimum funding standards, which specify a funding target that must be met over a specified period of time. The funding target for such plans is measured based on assumptions as to future investment returns, rates of mortality, retirement ages, and other economic and demographic assumptions. Under the standards, a plan must collect a minimum level of contributions each year to show progress toward meeting its target, or the plan employers may be assessed excise taxes and owe the plan for missed contributions plus interest. Minimum contribution levels may vary from year to year due to a variety of economic and demographic factors, such as addressing differences between assumed investment returns and the plan’s actual investment returns. To protect retirees’ pension benefits in the event that plan sponsors are unable to pay plan benefits, PBGC was created by ERISA. PBGC is financed through mandatory insurance premiums paid by plans and plan sponsors, with premium rates set by law. PBGC operates two distinct insurance programs: one for multiemployer plans and another for single- employer plans. Each program has separate insurance funds and different benefit guarantee rules. The events that trigger PBGC intervention differ between multiemployer and single-employer plans. For multiemployer plans, the triggering event is plan insolvency, the point at which a plan begins to run out of money while not having sufficient assets to pay the full benefits that were originally promised when due. PBGC does not take over operations of an insolvent multiemployer plan; rather, it provides loan assistance to pay administrative expenses and benefits up to the PBGC-guaranteed level. According to PBGC, only once in its history has a financial assistance loan from the multiemployer pension insurance program been repaid. In 2017, PBGC provided financial assistance to 72 insolvent multiemployer plans for an aggregate amount of $141 million. For single-employer plans the triggering event is termination of an underfunded plan—generally, when the employer goes out of business or enters bankruptcy. When this happens, PBGC takes over the plan’s assets, administration, and payment of plan benefits (up to the statutory limit). The PBGC-guaranteed benefit amounts for multiemployer plans and the premiums assessed by PBGC to cover those benefit guarantees are significantly lower than those for single-employer plans. Each insured multiemployer plan pays flat-rate insurance premiums to PBGC based on the number of participants covered. The annual premium rate for plan years beginning in January 2017 was $28 per participant and it is adjusted annually based on the national average wage index. (See app. II for the PBGC premium rates that have been in effect since the consent decree was established in 1982.) When plans receive financial assistance, participants face a reduction in benefits. For example, using 2013 data, PBGC estimated 21 percent of more than 59,000 selected participants in insolvent multiemployer plans then receiving financial assistance from PBGC faced a benefit reduction. The proportion of participants facing reductions due to the statutory guarantee limits is expected to increase. About 51 percent of almost 20,000 selected participants in plans that PBGC believed would require future assistance were projected to face a benefit reduction. Since 2013, the deficit in PBGC’s multiemployer program has increased by nearly 700 percent, from a deficit of $8.3 billion at the end of fiscal year 2013 to $65.1 billion at the end of fiscal year 2017. PBGC estimated that at of the end of 2016, the present value of net new claims by multiemployer plans over the next 10 years would be about $24 billion, or approximately 20 percent higher than its 2015 projections. The program is projected to become insolvent within approximately 8 years. If that happens, participants who rely on PBGC guarantees will receive only a very small fraction of current statutory guarantees. According to PBGC, most participants would receive less than $2,000 a year and in many cases, much less. We have identified PBGC’s insurance programs as high-risk. This designation was made in part because multiemployer plans that are currently insolvent, or likely to become insolvent in the near future, represent a significant financial threat to the agency’s insurance program. We designated the single-employer program as high-risk in July 2003, and added the multiemployer program in January 2009. Both insurance programs remain on our high-risk list. Key Amendments to ERISA Affecting Multiemployer Plans Multiemployer Pension Plan Amendments Act of 1980 Among other things, the Multiemployer Pension Plan Amendments Act of 1980 (MPPAA) made employers liable for a share of unfunded plan benefits when they withdraw from a plan, unless otherwise relieved of their liability, and strengthened certain funding requirements. An employer that chooses to withdraw from a multiemployer plan may be required to continue to contribute if the plan does not have sufficient assets to cover the plan’s current and known future liabilities at the time the employer withdraws; however, these payments may not fully cover the withdrawing employer’s portion of the plan’s liabilities. In such cases, the employers remaining in the plan may effectively assume the remaining liability. The Pension Protection Act of 2006 The Pension Protection Act of 2006 (PPA) was intended to improve the funding of seriously underfunded multiemployer plans, among other things. It included provisions that require plans in poor financial health to take action to improve their financial condition over the long term and established two categories of troubled plans: (1) “endangered status” or “yellow zone” plans (this category also includes a sub-category of “seriously endangered”), and (2) more seriously troubled “critical status” or “red zone” plans. PPA further required plans in the endangered and critical zones to develop written plans to improve their financial condition, such as by revising benefit structures, increasing contributions, or both, within a prescribed time frame. Multiemployer plans in yellow or red zone status must document their remediation strategies in a written plan, notify plan participants, and report annually on whether scheduled progress has been made. Since the 2008 market decline, the number of participants in endangered and critical plans has generally been decreasing (see fig. 4). The Multiemployer Pension Reform Act of 2014 In response to the funding crisis facing PBGC and multiemployer pension plans, the Multiemployer Pension Reform Act of 2014 (MPRA) made changes to the multiemployer system that were intended to improve its financial condition. Key changes included: Creation of critical and declining status. MPRA created a new category, “critical and declining,” for plans in critical status projected to become insolvent during the current plan year or within any of the 14 succeeding plan years, or in certain circumstances, within any of the 19 succeeding plan years. In 2017, PBGC reported that more than 100 multiemployer plans (more than 7 percent of plans) representing approximately 1 million participants (about 10 percent of participants) have been determined to be “critical and declining.” Permitted reduction of accrued benefits. MPRA permits plans to reduce participants’ and beneficiaries’ accrued retirement benefits if the plan can demonstrate such action is necessary to remain solvent. Plans apply to Treasury for the authority to reduce benefits. Treasury, in consultation with PBGC and DOL, reviews the applications and determines whether the proposed changes would enable the plan to remain solvent. Increased PBGC premiums. MPRA also increased the PBGC premiums for multiemployer plans from $12 to $26 (per participant per plan year) in 2015 and from $26 to $28 in plan year 2017. The annual premium in subsequent years is indexed to changes in the national average wage index. Creation of new framework of rules for partition. Partition allows a multiemployer plan to split into two plans—the original and a successor. Partitions are intended to relieve stress on the original plan by transferring the benefits of some participants to a successor plan funded by PBGC and to help retain participant benefits in the plans at levels higher than the PBGC-guaranteed levels. CSPF’s Critical Financial Condition Is a Result of Factors That Reflect Challenges Experienced by the Multiemployer System CSPF Has Been Underfunded Since the Consent Decree Was Established At the time the consent decree was established in 1982, CSPF had less than half the estimated funds needed to cover plan liabilities (and to pay associated benefits over the lifetime of participants) and it has not attained 100 percent of its estimated funding need since then, according to regulatory filings. CSPF’s 1982 Form 5500 we reviewed shows that the plan was less than 40 percent funded prior to the consent decree becoming effective. Over the next two decades, the plan generally made progress toward achieving its targeted level of funding but was never more than 75 percent funded, and funding has generally deteriorated since its 2002 filing (see fig. 5). Overall, the plan’s unfunded liability increased by approximately $11.2 billion (in inflation-adjusted dollars) between January 1983 and January 2016. As a consequence, participant benefits were never fully secured by plan assets over this period, as measured by ERISA’s minimum funding standards, and the plan consistently needed to collect contributions in excess of those needed to fund new benefit accruals to try to make up for its underfunded status. Stakeholders Described Multiple Factors That Contributed to CSPF’s Critical Financial Condition, Many of Which Have Been Experienced by Other Multiemployer Plans CSPF officials and other stakeholders identified several factors that contributed to CSPF’s critical financial condition and reflect the challenges faced by many multiemployer plans. For example, like CSPF, many multiemployer plans have experienced financial difficulties due to a combination of investment losses and insufficient employer contributions. In addition to being underfunded prior to the consent decree going into effect, stakeholders identified other specific factors that contributed to CSPF’s critical financial condition, such as trends within the national trucking industry and its workforce, funding challenges and common investment practices of multiemployer plans, and the impact of market downturns on long-term investment performance. Stakeholders also described the effects of the 2007 withdrawal of a key employer, United Parcel Service (UPS), on CSPF’s critical financial condition. Key Industry Specific Workforce Trends Stakeholders we interviewed said changes to the workforce, such as declining union membership rates and changes resulting from industry deregulation, affected CSPF and some other multiemployer plans by reducing the number of workers able to participate in their plans. While the multiemployer structure distributes bankruptcy risk across many employers, for any particular multiemployer plan employers are often concentrated in the same industry, making the plans vulnerable to industry- specific trends and risks. For example, stakeholders noted the impact that the Motor Carrier Act of 1980 had on the trucking industry. Specifically, deregulation of the trucking industry reduced government oversight and regulation over interstate trucking shipping rates. The trucking industry became increasingly dominated by nonunion trucking companies resulting in the bankruptcy of many unionized trucking companies, according to stakeholders. New trucking companies typically did not join multiemployer plans because their labor force was not unionized and this, coupled with the bankruptcy of many contributing employers, contributed to a decrease in active participant populations for many plans serving the industry. As the total number of active participants in a plan declines, the resources from which to collect employer contributions declines proportionally. Stakeholders also said these changes were unforeseeable. Limitations on a plan’s ability to increase contributions mean that a plan has less capacity to recover from an underfunded position or to make up for investment returns that fall short of expectations. A decline in the number of active workers can also accelerate plan “maturity,” as measured by the ratio of nonworking to working participants. Plan maturity has implications for a plan’s investment practices and the time frame over which the plan must be funded. According to PBGC’s data for the multiemployer plans it insures, there were approximately three active participants for every nonworking participant in 1980 (3:1); by 2014, the ratio was approximately one active worker for every two nonworking participants (1:2). Figure 6 shows the change in the percentages of active and nonworking participants for the multiemployer plans that PBGC insures. CSPF saw an even more dramatic change in its active to nonworking participant ratio from 1982 through 2015. In 1982, there were more than two active workers for every nonworking participant (2:1) and by 2016 that ratio had fallen to approximately one active worker for every five nonworking participants (1:5) (see fig. 7). Because CSPF’s contributing employers were largely trucking companies, stakeholders said this made the fund especially vulnerable to industry-wide shocks. Like the industry as a whole, CSPF was unable to attract new employers to replace exiting employers, in part because of the lack of new unionized employers. CSPF officials said that changes to the trucking industry and its workforce also led to other challenges for the plan. For example, contributions to the plan declined with the shrinking number of active workers. CSPF officials told us they could not significantly increase the contribution rate paid by remaining employers because of the financial hardship it would cause, and as a result, the plan’s ability to recover from its underfunded position was limited. CSPF officials said that this increased the plan’s reliance on investment returns to try to close the gap between its assets and liabilities. Funding Challenges and Investment Practices Stakeholders we interviewed cited challenges inherent in multiemployer plans’ funding and investment practices, and described how the challenges may have contributed to the critical financial condition of some plans, including CSPF. Stakeholders said that CSPF and many other multiemployer plans have been challenged by employer withdrawals. An employer withdrawal reduces the plan’s number of active worker participants, thereby reducing its contribution base and accelerating plan maturity. A withdrawing employer generally must pay a share of any unfunded benefits. Stakeholders identified several ways in which the withdrawal liability framework could result in a withdrawing employer underpaying its share of an unfunded liability. We have previously reported on the challenges associated with withdrawal liability, including: withdrawal liability assessments are often paid over time, and payment amounts are based on prior contribution rates rather than the employer’s actual withdrawal liability assessment. withdrawal liability payments are subject to a 20-year cap, regardless of whether an employer’s share of unfunded benefits has been fully paid within this 20-year timeframe; plans often did not collect some or all of the scheduled withdrawal liability payments because employers went bankrupt before completing their scheduled payments; and fears of withdrawal liability exposure increasing over time could be an incentive for participating employers to leave a plan and a disincentive for new employers to join a plan; Stakeholders we interviewed also added that the calculation used to determine withdrawal liability may use an investment return assumption that inherently transfers risk to the plan. When exiting employers do not pay their share of unfunded benefits, any remaining and future employers participating in the plan may effectively assume the unpaid share as a part of their own potential withdrawal liability as well as responsibility for the exiting employer’s “orphaned” participants. Participating employers may negotiate a withdrawal if they perceive a risk that the value of their potential withdrawal liability might grow significantly over time. In its MPRA application, CSPF cited employer withdrawals and bankruptcies as a significant challenge for the plan. CSPF reported that after deregulation, the number of contributing employers dropped by over 70 percent. While some of the drop could be due to the consolidation of trucking companies after deregulation, CSPF officials cited several cases in which employers went bankrupt or withdrew from the plan, which reduced the plan’s contribution base and accelerated its maturity. Additionally, when employers went bankrupt, they often did not pay their full withdrawal liability. For example, CSPF said two of its major contributing employers left the plan between 2001 and 2003, and left $290 million of more than $403 million in withdrawal liability unpaid after they went bankrupt. Stakeholders identified funding timeframes as a factor that contributed to the challenges facing many multiemployer plans, including CSPF. ERISA’s minimum funding standards have historically allowed multiemployer plans to amortize, or spread out the period of time for funding certain events, such as investment shortfalls and benefit improvements. For example, CSPF began a 40-year amortization of approximately $6.1 billion in underfunding on January 1, 1981, giving the plan until the end of 2021 to fully fund that amount. Longer amortization periods increase the risk of plan underfunding due to the number and magnitude of changes in the plan’s environment that may occur, such as a general decline in participants or deregulation of an industry. The Pension Protection Act of 2006 shortened amortization periods for single- employer plans to 7 years and the amortization periods for multiemployer plans to 15 years. Shorter amortization periods provide greater benefit security to plan participants by reducing an unfunded liability more rapidly. In addition, shorter amortization periods can be better aligned with the projected timing of benefit payments for a mature plan. However, shorter periods can be a source of hardship for plans with financially troubled contributing employers because they may require higher contributions. According to CSPF officials, CSPF requested and received an additional 10-year amortization extension from the IRS in 2005 after relating that contribution requirements could force participating employers into bankruptcy. One CSPF representative said an amortization extension can also help avoid subjecting the plan’s employers to IRS excise taxes for failing to make required minimum contributions. Stakeholders we interviewed said that certain common investment practices may have played a role in the critical financial condition of CSPF and other mature and declining plans. In general, multiemployer plans invest in portfolios that are expected, on average, to produce higher returns than a low-risk portfolio, such as one composed entirely of U.S. Treasury securities. Stakeholders also stated that these investment practices may have been too risky because returns can be more volatile, and the higher expected returns might not be achieved. In addition, the Congressional Budget Office has reported that if “plans had been required to fund their benefit liabilities—at the time those liabilities were accrued—with safer investments, such as bonds, the underfunding of multiemployer plans would have been far less significant and would pose less risk to PBGC and beneficiaries.” Stakeholders also told us that for mature plans like CSPF, these investment practices can pose further challenges. Mature plans, with fewer active employees, have less ability to recoup losses through increased contributions and have less time to recoup losses through investment returns before benefits must be paid. Market corrections, such as those that occurred in 2001 through 2002 and in 2008, can be particularly challenging to mature plans and their participants, especially if a mature plan is also significantly underfunded. Mature plans could mitigate these risks by investing more conservatively, however, the resulting lower expected returns from more conservative investing necessitates higher funding targets and contribution rates, which could be a hardship for employers in an industry with struggling employers. Alternatively, a plan that invests more conservatively may provide lower promised benefits to accommodate the level of contributions it can collect. Lower investment returns from a more conservative investment policy would cost employers more in contributions and could potentially result in employers leaving the plan. Further, investing in a conservative portfolio would be relatively unique among multiemployer plans, and stakeholders said plan managers may feel they are acting in a prudent fashion by investing similarly to their peers. Underfunded plans like CSPF may not see conservative investment as an option if they cannot raise the contributions necessary to fully fund their vested benefits. Officials from CSPF told us that, because they lacked the ability to significantly increase revenue or decrease accrued benefits, the named fiduciaries sought incrementally higher investment returns to meet funding thresholds required by the amortization extension they received in 2005. On the other hand, there are challenges associated with risk bearing investments. In our prior work, we reported that multiemployer plans generally develop an assumed average rate of investment return and use that assumption to determine funding targets, required contributions, and the potential cost of benefit improvements. Experts we interviewed for that report told us that using a portfolio’s expected return to value the cost of benefits increases the risk that insufficient assets could be on hand when needed. They also told us that using the portfolio’s expected return to calculate liabilities could incentivize plans to invest in riskier assets and to negotiate higher benefit levels because the higher returns expected from riskier portfolios can result in lower reported liabilities. Plan Terms Set through Collective Bargaining Stakeholders we interviewed said that plan terms, such as contribution rates, which are set through the collective bargaining process, can create an additional challenge for multiemployer plans. Employers in multiemployer plans generally are not required to contribute beyond what they have agreed to in collective bargaining, and these required employer contributions generally do not change during the term of a collective bargaining agreement. CSPF officials said that up until the early 2000s, plan officials did not request modifications to collective bargaining agreements, such as reallocating contribution dollars, to respond to adverse investment returns. Investment Performance and Market Downturns Stakeholders highlighted the effects of market downturns on multiemployer plan assets as another contributing factor to CSPF’s critical financial condition and that of other multiemployer plans. Failure to achieve assumed returns has the effect of increasing unfunded liabilities. For the multiemployer system in aggregate, the average annual return on plan assets over the 2002 to 2014 period was about 6.1 percent, well short of typical assumed returns of 7.0 or 7.5 percent in 2002. Many multiemployer plans were especially impacted by the 2008 market downturn. PBGC estimated that from 2007 to 2009, the value of all multiemployer plan assets fell by approximately 24 percent, or $103 billion, after accounting for contributions to and payments from the plans. Although asset values recovered to some extent after 2009, some plans continued to be significantly underfunded, and stakeholders said this could be due to the contribution base not being sufficient to help recover from investment shortfalls. CSPF’s investment performance since 2000 has reflected performance similar to other multiemployer plans and the plan went from 73 percent funded in 2000 to about 38 percent funded in 2017. While the plan used an assumed rate of return of 7.5 to 8.0 percent per year between 2000 and 2014, our analysis of the plan’s regulatory filings shows that the plan’s weighted-average investment return over this period was about 4.9 percent per year. CSPF officials said the 2008 downturn significantly reduced CSPF’s assets and it was unable to sufficiently recoup those losses when the market rebounded in 2009. Plan assets declined from $26.8 billion at the beginning of 2008 to $17.4 billion at the beginning of 2009, with $7.5 billion of the decline attributable to investment losses. Despite reporting a 26 percent return on assets during 2009, CSPF had only $19.5 billion in assets at the end of 2009 because benefits and expenses exceeded the contributions it collected and because it had fewer assets generating returns for the plan. By the end of 2009, CSPF’s funding target was $35.9 billion but the fund had less than $20 billion that could be used to generate investment returns. If CSPF’s portfolio had returned 7.5 percent per year over the 2000-2014 period, instead of the approximately 4.9 percent we calculated, we estimate that the portfolio value would have exceeded $32.0 billion at the end of 2014, or 91 percent of its Actuarial Accrued Liability. Effect of UPS Withdrawal In addition to the factors mentioned that affected many multiemployer plans, stakeholders we interviewed also noted the unique effect of the UPS withdrawal on CSPF. In 2007, UPS negotiated with the International Brotherhood of Teamsters for a withdrawal from CSPF and paid a withdrawal liability payment of $6.1 billion. This payment was invested just prior to the 2008 market downturn. Moreover, the loss of UPS, CSPF’s largest contributing employer, reduced the plan’s ability to collect needed contributions if the plan became more underfunded. A UPS official said that, following the market decline of 2001-2002, the company considered whether it should withdraw from all multiemployer plans because it did not want to be the sole contributing employer in any plan. According to this official, UPS considered the large number of UPS employees in CSPF and the plan’s demographics—such as an older population and fewer employers—in its decision to withdraw. CSPF officials said they did not want UPS to withdraw because its annual contributions accounted for about one-third of all contributions to the plan. CSPF officials also told us that, prior to the UPS withdrawal, they had expected the population of active UPS workers in the plan to grow over time. UPS’ withdrawal of 30 percent of CSPF’s active workers, in combination with the significant market downturn just after UPS withdrew, reflected the loss of working members and investment challenges on a large scale. Additionally, stakeholders noted that although each of the factors that contributed to CSPF’s critical financial condition individually is important, their interrelated nature also had a cumulative effect on the plan. Industry deregulation, declines in collective bargaining, and the plan’s significantly underfunded financial condition all impaired CSPF’s ability to maintain a population of active workers sufficient to supply its need for contributions when investment shortfalls developed. Given historical rules for plan funding and industry stresses, CSPF was unable to capture adequate funding from participating employers either before or after they withdrew from the plan. The plan’s financial condition was further impaired when long-term investment performance fell short of expectations. For an underfunded, mature plan such as CSPF, the cumulative effect of these factors was described by some stakeholders as too much for CSPF to overcome. CSPF’s Investment Policy Since 1982 Generally Increased Allocation to Equities, but Shifted Toward Fixed Income in 2017, Ahead of Projected Insolvency There have been three distinct periods related to CSPF’s investment policy after the original consent decree took effect: the early period, from the consent decree’s effective date in September 1982 through October 1993, during which named fiduciaries set different investment policies and sold many of CSPF’s troubled assets—mostly real estate; a middle period from November 1993 through early 2017, during which CSPF’s investment policies were consistently weighted towards equities and its asset allocation varied, with notable equity allocation increases occurring from year-ends 1993-1995 and 2000-2002; and the current period, starting in January 2017, during which named fiduciaries and CSPF trustees are moving assets into fixed income ahead of insolvency. Appendix I has a detailed timeline that includes changes to CSPF’s investment policies since the consent decree was established in 1982. Early Period: September 1982–October 1993 The original consent decree placed exclusive responsibility for controlling and managing the plan’s assets with an independent asset manager, called a named fiduciary. Additionally, the original consent decree prohibited CSPF trustees from managing assets or making investment decisions and gave a single named fiduciary the authority to set and change the plan’s investment objectives and policies, subject to court approval (see fig. 8). During this period, two successive named fiduciaries—first Equitable Life Assurance Society of the United States (Equitable) and then Morgan Stanley—set and executed the plan’s investment objectives using similar investment philosophies, but differing investment return goals and target asset allocations (see fig. 9). Both named fiduciaries planned to sell the plan’s troubled real estate assets from the pre-consent decree era. They also limited nonpublicly traded investments to 35 percent of the plan’s assets and set broad allocation targets for new real estate, fixed income, and equity assets. In 1984, Morgan Stanley considered a dedicated bond portfolio in its capacity as the plan’s named fiduciary, but after review, Morgan Stanley decided similar results could be obtained through other investment strategies. In executing these policies, the plan’s asset allocation varied from year to year. Starting in 1987 and in subsequent years during the early period, Morgan Stanley invested a majority of the plan’s assets in fixed income assets—more than half of which were passively managed—and all equity assets were allocated to domestic equity through 1992. By 1989, CSPF officials reported that nearly all troubled real estate assets had been sold and Morgan Stanley’s responsibilities and risk of potential fiduciary liability were reduced, permitting a concomitant reduction in fees paid to the named fiduciary (see fig. 10). Middle Period: November 1993–January 2017 CSPF’s Investment Policy During the middle period, CSPF’s investment policy was broad and consistently directed that asset allocations be weighted toward equities. In 1993, Morgan Stanley revised its investment policy statement for CSPF to eliminate asset allocation targets for each asset class and instead specified that the plan invest a majority of assets in equity or equity-type securities and no more than 25 percent in nonpublicly traded assets. After 1999, CSPF’s investment policy under other, successive named fiduciaries continued to be broad and generally specified that the plan should invest a majority of assets in equity or equity-type securities. Specifically J.P. Morgan’s and Northern Trust’s consecutive investment policies for part of the plan’s assets continued to specify that a majority of the plan’s assets be invested in equity or equity-type securities and no more than 15 percent be invested in nonpublicly traded assets. Goldman Sachs’ investment policy for another part of the plan’s assets did not specify asset allocation details but indicated slightly higher tolerance for risk in conjunction with its equity portfolio. CSPF trustees said that named fiduciaries considered investing in alternative assets, but instead chose to increase the plan’s allocation to equity assets. The named fiduciaries’ investment policies did not vary significantly over this period because CSPF officials said that the plan’s overarching investment objective of achieving full funding did not change, even though there were key changes to the plan’s investment management structure during this time period. Specifically, starting in 1999, the plan temporarily shifted to a dual named fiduciary structure and increased its use of passively-managed accounts—both described in detail below— changing the named fiduciary structure that had been in place since the original consent decree (see fig. 11). More specifically, the two key changes to the plan’s investment management structure were: A temporary shift to a dual named fiduciary structure. Effective in 1999, CSPF proposed and the court approved allocating plan assets between two named fiduciaries instead of one in order to diversify CSPF’s investment approach, among other things. Both named fiduciaries were in charge of setting and executing separate policies for plan assets they managed—called “Group A” and “Group B” assets—irrespective of the other named fiduciary’s allocations. During this time, the two named fiduciaries were J.P. Morgan/Northern Trust and Goldman Sachs. Specifically, J.P. Morgan was named fiduciary between 2000 and 2005 and Northern Trust between 2005 and 2007 for “Group A” assets. Goldman Sachs was named fiduciary for “Group B” assets between 2000 and 2010. In 2010, an investment consultant found the performance of two named fiduciaries under the dual named fiduciary structure had been similar and more expensive than it would be under a proposed move back to a single named fiduciary. Accordingly, CSPF officials proposed, and the court approved, consolidation of all assets allocated to named fiduciaries in August 2010, with Northern Trust as the plan’s single named fiduciary. An increased use of passively-managed accounts. Between 2003 and 2010, the portion of assets that named fiduciaries managed declined as the plan moved 50 percent of its assets into three passively-managed accounts. Specifically, in 2003, 20 percent of CSPF’s assets were transitioned into a passively-managed domestic fixed income account to lower the plan’s investment management fees. In addition, both of the named fiduciaries reported that they had not outperformed the industry index for the domestic fixed income assets they managed after they were approved as named fiduciaries in 1999 and 2000 through February 2003. Similarly, in 2007 and 2010, CSPF officials said that two more passively-managed accounts were created to further reduce plan fees. Specifically, in 2007, 20 percent of plan assets were moved into a passively-managed domestic equity account. Then, in 2010, an additional 10 percent of the plan’s assets were allocated to passively-managed accounts—5 percent were allocated to a new passively-managed international equity account and 5 percent were added to the passively-managed domestic equity account. CSPF officials and named fiduciary representatives also said that the plan’s investment policies did not change in response to a couple of the events that contributed to CSPF’s critical financial condition. For example, when UPS withdrew from the plan in December 2007, it paid $6.1 billion in a lump sum to fulfill its withdrawal liability. Consistent with the named fiduciaries’ investment policies during this time period, the majority of this withdrawal payment was invested in equity assets. Specifically, the court approved the UPS withdrawal liability payment to be allocated: $1 billion to Northern Trust to be invested primarily in short-term fixed income assets, $0.9 billion to the passively-managed domestic fixed income account, and $4.2 billion to partially fund the newly created passively- managed domestic equity account. As a result of the 2008 market downturn, the balance of each of CSPF’s accounts—Northern Trust’s named fiduciary account, the passively-managed domestic fixed income and domestic equity accounts, and Goldman Sachs’ named fiduciary account—declined because of investment losses or withdrawals from investment assets to pay benefits and expenses. Some of the declines in each account were reversed by investment gains in 2009. Process for Setting and Executing CSPF’s Investment Policy Although the changes made to CSPF’s investment management structure did not lead to investment policy changes during the middle period, they altered the process by which the policy was set and executed. In particular, trustee responsibilities in the policy process grew after CSPF trustees became responsible for developing investment policy statements and selecting and overseeing managers of the passively-managed accounts, subject to court approval. In addition, CSPF officials said the addition of passively-managed accounts between 2003 and 2010 had the effect of creating broad bounds within which the named fiduciary could set the plan’s asset allocation. For example, when the plan moved 20 percent of total plan assets into the passively-managed domestic fixed income account in 2003, this placed an upper bound on the plan’s total equity allocation at 80 percent. Similarly, since 2010 the 30 percent of total plan assets in passively-managed equity accounts has placed a lower bound on the plan’s total equity allocation at 30 percent (see fig. 12). Nevertheless, named fiduciaries maintained the largest role in setting and executing CSPF’s investment policy throughout the middle period. From 1993 to 2003, named fiduciaries managed all of the plan’s investment assets, and from 2003 to 2009, when the plan added two of the current passively-managed accounts, named fiduciaries still held the majority of the plan’s assets. It has only been since 2010 that the assets in passively-managed accounts equaled those managed by the named fiduciary. Furthermore, Northern Trust representatives said they considered the plan’s allocations to passively-managed accounts when developing the objectives and target asset allocations for the assets they managed. Northern Trust representatives also said they discussed the plan’s overall asset allocation with trustees, but the trustees, and ultimately the court, were responsible for the decision to move 50 percent of the plan’s assets into passively-managed accounts. Asset Allocation under CSPF’s Investment Policy After the 1993 policy change that specified the plan would invest a majority of assets in equity or equity-type securities, CSPF’s asset allocation changed significantly. For example, during the middle period the plan’s allocation to equities increased from 37 percent at the end of 1993 to 69 percent at the end of 2002, and its allocation to cash plus fixed income decreased from 63 percent at the end of 1993 to 27 percent at the end of 2002. In particular, Morgan Stanley increased the plan’s allocation to equity assets from 37 percent at the end of 1993 to 63 percent at the end of 1995, with the percentage in equities almost or above 50 percent through the end of 1999. From 1993 through 1999, Morgan Stanley generally decreased the plan’s allocation to fixed income assets and increased its allocation to international equity (reaching a high of about 28 percent of the plan’s assets in 1995), an asset class in which the plan had not previously invested (see fig. 13). After 1999, the plan’s asset allocation continued to be weighted towards equities. After the market downturn in 2001, CSPF trustees told us that J.P. Morgan and Goldman Sachs explicitly increased the equity allocation in an attempt to generate higher investment returns and increase the plan’s funded ratio—the plan’s overarching investment objective. Between 2000 and mid-2010, when the plan had two named fiduciaries, equity assets increased from about 58 percent at the end of 2000 to between 66 and 70 percent at the end of 2001 and each year thereafter until the end of 2009, mostly based on the named fiduciaries’ decisions to increase the plan’s allocation to domestic equity assets. When Northern Trust became the sole named fiduciary in 2010, the proportion of equity assets declined from almost 72 percent at the end of 2010 to almost 63 percent at the end of 2016. During this time, Northern Trust generally decreased the plan’s allocation to domestic equity assets, increased the allocation to actively-managed fixed income, and started investing in global infrastructure assets. Northern Trust representatives said CSPF’s recent portfolio had been kept relatively aggressive in an attempt to achieve the returns the plan would need to become fully funded while balancing risk (see fig. 14). Current Period: January 2017 – Present CSPF’s deteriorating financial condition precipitated a recent investment policy change that will move plan assets into fixed income and cash equivalent investments ahead of projected insolvency. In early 2017, Northern Trust representatives revised the plan’s investment policy because they, in consultation with the trustees, believed the plan had no additional options to avoid insolvency (see textbox). This change to the plan’s outlook led to a significant change in the plan’s investment objective, from a goal of fully funding the plan to instead forestalling insolvency as long as possible while reducing the volatility of the plan’s funding. Northern Trust representatives and CSPF officials revised applicable plan investment policy statements and started to gradually transition the plan’s “return seeking assets”—such as equities and high yield and emerging markets debt—to high quality investment grade debt and U.S. Treasury securities with intermediate and short-term maturities. Northern Trust’s new investment policy specified the assets under its control would not be invested in nonpublicly traded securities, in order to manage risk and provide liquidity. CSPF Has Limited Options to Achieve Solvency As of March 2018, the Central States, Southeast and Southwest Areas Pension Fund (CSPF) was projected to be insolvent on January 1, 2025. As of July 2017, CSPF officials said that the following measures (in isolation) could help the plan avoid insolvency: 18 percent year-over-year investment returns (infinite horizon), or 250 percent contribution increases (with no employer attrition), or 46 percent across-the-board benefit cut. However, CSPF officials said that investment returns and contribution increases of these magnitudes were untenable, and CSPF’s application to reduce accrued benefits under the Multiemployer Pension Reform Act of 2014 (MPRA) was denied in 2016. CSPF officials and Northern Trust representatives said these asset allocation changes are intended to provide participants greater certainty regarding their benefits and reduce the plan’s exposure to market risk and volatility until it is projected to become insolvent on January 1, 2025 (see fig. 15). Northern Trust is expected to continue to manage 50 percent of the plan’s investment assets until the plan becomes insolvent. While the total amount of assets in the passively-managed accounts will continue to constitute 50 percent of the plan’s assets, the trustees plan to transfer assets from the passively-managed domestic and international equity accounts into the passively-managed domestic fixed income account, which will be gradually transitioned to shorter-term or cash-equivalent fixed-income securities sometime before the end of March 2020 (see fig. 16). CSPF officials said the changes will reduce the amount of fees and transaction costs paid by the plan. Specifically, investment management fees are expected to generally decrease as the plan moves into shorter duration fixed income assets. In addition, Northern Trust’s plan is designed to reduce transaction costs in two ways: (1) in the near term, Northern Trust plans to liquidate “return-seeking assets” so the cash it receives can be used directly to pay benefits, and (2) it plans to synchronize the fund’s benefit payments with the maturity dates of fixed income assets it purchases so cash received can be used directly to pay benefits. Both of these design features are intended to eliminate the need to reinvest assets, which might entail additional transaction costs. Available Data Show That CSPF Investment Returns and Fees Were Generally Comparable to Similar Plans Our analysis of available data from several different sources shows the returns on CSPF’s investments and the fees related to investment management and other plan administration activities appear generally in line with similar pension plans or other large institutional investors of similar size. CSPF’s Investment Return History is in Line with Other Funds and Plans CSPF Investment Returns Compared to Other Large Institutional Funds The annual returns on CSPF’s investments in recent decades have generally been in line with the annual returns of a customized peer group provided by the investment consultant Wilshire. The comparison group data is from Wilshire’s Trust Universe Comparison Service (TUCS)—a tool used by CSPF to compare its investment returns to a group of peers. Over the 22 years covered by our analysis, CSPF’s returns were above the median in 12 years and below the median the other 10. Figure 17 illustrates how CSPF’s annual investment returns compare to CSPF’s customized peer group of master trusts with over $3 billion in assets. CSPF’s annual investment returns tended to fluctuate relative to the annual median of the TUCS peer group over the 1995 through 2016 period. For example, in 14 of the 22 years analyzed, its annual return was in the highest or lowest 25 percent of returns (7 years each). Further, in 3 years, its investment returns fell either within the top 5 percent of returns (1996, 2009) or bottom 5 percent (1998). In 8 of the 22 years, CSPF’s annual return was within the middle 50 percent of its TUCS peer group. The TUCS data we analyzed also included median portfolio allocations for the group of CSPF’s peers. Table 2 compares CSPF’s asset allocations for selected asset categories to the median allocations of its TUCS comparator group. In 1996, compared to the TUCS medians, CSPF had relatively lower proportions of its assets in both equities and fixed income and a relatively higher proportion in real estate. Twenty years later (2016), CSPF had relatively higher proportions of its assets in both equities and fixed income (about 15 and 7 percentage points, respectively) than the respective medians for its peer group. However, the relatively large difference between CSPF’s 2016 equity allocation and the median allocation of its peer group may be a result of the peers moving into different asset classes. For example, there is a relatively large difference, in the other direction, in the allocation to alternative investments (see table 2). We did not identify an alternative asset category in CSPF’s asset reports for 2016, but the TUCS comparator group median asset allocation in that year is 11.8 percent of assets. CSPF Investment Returns Compared to Other Similar Multiemployer Plans Similar to our findings when comparing the returns on CSPF’s investments to a customized peer group of other large institutional funds, the annual returns on CSPF’s investments in recent decades have also generally been in line with the annual returns for a group of similar multiemployer pension plans. To create a group of comparable plans, we selected plans that had a similar degree of “maturity” to CSPF in 2000, as such plans may face similar cash flow challenges to those facing CSPF. This comparator group ultimately consisted of 15 plans in addition to CSPF. Relative to less mature plans, mature plans generally have a greater proportion of liabilities attributable to retired participants receiving benefit payments and a lower proportion attributable to active participants (i.e., workers) earning benefits. Mature plans may have limited capacity to make up for insufficient investment returns through employer contributions. Similar to the comparison against other large institutional fund returns based on TUCS data, our comparison against other mature multiemployer plan returns based on Form 5500 data shows that CSPF’s annual returns fluctuate relative to the median annual return for the mature plan comparator group (see fig. 18). For example, in 12 of the 15 years, CSPF’s annual return was in the highest or lowest 25 percent of returns (7 times high and 5 times low). In 3 of the 15 years analyzed, CSPF’s annual return fell within the middle 50 percent of the peer group. Overall, from 2000 to 2014, CSPF’s annual return was above the group median return in 9 of the 15 years—and lower than the median return in the other 6 years. Relative to its peers, CSPF’s annual returns performed worst during economic downturns and best in years coming out of such downturns. CSPF’s annual investment return was in the bottom 10 percent of returns in 2001, 2002, and 2008. Alternatively, its annual return was in the top 10 percent of returns from 2003 to 2006, in 2009, and in 2012. Additionally, the dollar-weighted average annual return for CSPF over the 2000 through 2014 period was roughly the same as the median for the mature plan comparison group. Specifically, the dollar-weighted average annual return over this period for CSPF was roughly 4.9 percent, while the median dollar-weighted average annual return over this period among the comparison plans with continuous data was 4.8 percent. Our analysis of investment returns for mature plans compares investment returns for a set of peers that includes only multiemployer defined benefit plans. However, as with the comparison against other large institutional funds, the comparisons against other mature plans are not measures of over- or under-performance relative to an index or benchmark. Similarly, as with the earlier comparison, the analysis does not account for variations in the levels of investment risk taken by the plans. Fees and Expenses Paid by CSPF Were Similar to Other Large Multiemployer Plans Our analysis of Form 5500 data shows CSPF’s investment fees and administrative expenses were in line with other large multiemployer plans. Plan investment fees and administrative expenses are often paid from plan assets so many plans seek to keep these fees and expenses low. Additionally, investment fees are likely to be related to the value of assets under management, and plans with greater asset values tend to be able to negotiate more advantageous fee rates. According to a pension consultant and a DOL publication, investment management fees are typically a large defined benefit plan’s largest category of expense, but a pension plan also incurs a number of lesser expenses related to administering the plan. Administrative expenses (other than investment fees) may include those for: audit and bookkeeping/accounting services; legal services to the plan (opinions, litigation and advice); administrative services provided by contractors; plan staff salaries and expenses; plan overhead and supplies; and other miscellaneous expenses. These administrative expenses relate to plan operations beyond the management of the assets, including the day-to-day expenses for basic administrative services such as participant services and record keeping. Furthermore, some of these expenses can vary based on the number of participants in the plan. To compare CSPF’s fees and expenses against similarly sized plans, we tallied various investment fee-related and other administrative expenses and compared CSPF to a group of multiemployer defined benefit plans that were among the 19 largest plans in terms of assets as of January 1, 2014. According to CSPF’s 2014 Form 5500, CSPF spent about $46.5 million on investment fees (or $47.6 million in 2016 dollars) and had about $17.4 billion in assets (or $17.8 billion in 2016 dollars) as of the end of the year—resulting in an investment fee expense ratio of about 27 basis points, or 0.27 percent. Over the 2000 to 2014 period, CSPF’s average annual investment fee expense ratio was 34 basis points (0.34 percent) while the median of the averages for our large plan comparison group was 37 basis points (0.37 percent). While CSPF’s average investment fee expense ratio was below the median for its comparison group over the period we examined, the relationship of CSPF’s annual ratio to the annual median changed over time. CSPF’s annual investment-fee expense ratio was consistently at or above the median from 2000 through 2006, but was below the median thereafter. In addition, CSPF’s average investment fee expenses over the period that followed 2006 were 26 percent less than the average over the period before 2007. (They averaged 39 basis points, or 0.39 percent, from 2000 through 2006 and 29 basis points, or 0.29 percent, from 2007 through 2014.) Two events may have contributed to this change. First, CSPF introduced the passively-managed accounts beginning in 2003—as noted earlier, and CSPF moved certain assets to those accounts in an effort to reduce fees. Second, the change back to a single named fiduciary, which was suggested as an expense saving move, occurred in the middle of the 2007 to 2014 period analyzed. Figure 19 illustrates how CSPF’s investment fee expense ratio compares to other large plans. Our analysis uses investment fee data reported in the Form 5500 that does not include details about the sources of the fees. Investment fees may be sensitive to a plan’s particular investment strategy and the way assets are allocated. For example, with CSPF, a named fiduciary has responsibility for executing the investment strategy and allocations. According to a representative from Northern Trust—the current named fiduciary—CSPF pays a fee of about 5 basis points for named fiduciary services, and this, combined with investment management fees, is in line with investment fees for other clients (though the overall fees depend on the types of asset allocations and investment strategies). Figure 20 shows how CSPF’s administrative (or non-investment) expenses compare to those of other large plans on a per participant basis. According to CSPF’s 2014 Form 5500, CSPF spent about $38 million on administrative expenses ($39 million in 2016 dollars)—the third most among the 20 peer plans. However, when these expenses are expressed relative to the number of plan participants, CSPF had per participant expenses of $98 in 2014, which is about 16 percent less than the median of the large comparator group, $117. Over the period studied, CSPF’s administrative expenses per participant were at or above the large comparator median in 3 years (2001, 2004, and 2005), but lower than the median in all other years of the 2000 to 2014 period. CSPF’s administrative expenses were also in line with a broader group of comparators. For example, PBGC reported on 2014 administrative expenses of a population of large multiemployer plans (plans with more than 5,000 participants). By closely replicating the methodology of that study, we found CSPF’s expenses of $98 per participant in 2014 fell below the median expense rate of $124 dollars per participant but above the lowest quartile of this group of large multiemployer plans. In comparing administrative expenses as a percentage of benefits paid, we found CSPF’s administrative expenses were among the lowest 5 percent of this group of large multiemployer plans. We performed a similar comparison against the peer group of large plans. CSPF had the lowest administrative expense rate among the large plan peer group in 2014, paying administrative expenses at a rate of 1.4 percent of benefits paid. In addition, CSPF’s annual administrative expenses as a percentage of benefits were below the median of our peer group of large plans in all years we reviewed. Our analysis of administrative expenses is highly summarized and does not account for possibly unique sources of administrative expenses. Plans may have unique organizational structures and attribute expenses differently. For example, one plan may contract a significant portion of administrative duties with a third-party, while another plan may administer the plan in-house. According to an actuary we interviewed, most multiemployer plans are administered by a third-party, but the plan’s in- house staff will still retain a number of duties. Additionally, the amount of individual administrative expenses could vary significantly by plan depending on the importance of the related administrative function in the plan’s organization. Agency Comments and Our Evaluation We provided a draft of the report to the U.S. Department of Labor, U.S. Department of the Treasury, and the Pension Benefit Guaranty Corporation for review and comment. We received technical comments from the U.S. Department of Labor and the Pension Benefit Guaranty Corporation, which we incorporated as appropriate. The U.S. Department of the Treasury provided no comments. We will send copies to the appropriate congressional committees, the Secretary of Labor, the Secretary of the Treasury, Director of the Pension Benefit Guaranty Corporation, and other interested parties. This report will be available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact Charles Jeszeck at (202) 512-7215 or jeszeckc@gao.gov or Frank Todisco at (202) 512-2700 or todiscof@gao.gov. Mr. Todisco meets the qualification standards of the American Academy of Actuaries to address the actuarial issues contained in this report. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix IV. Appendix I: Objectives, Scope, and Methodology Our objectives were to review: (1) what is known about the factors that contributed to the Central States, Southeast and Southwest Areas Pension Fund’s (CSPF) critical financial condition; (2) what has been CSPF’s investment policy, and the process for setting and executing it, since the consent decree was established; and (3) how has CSPF performed over time, particularly compared to similar pension plans. For all objectives, we reviewed relevant federal laws and regulations, literature, and documentation the U.S. Department of Labor (DOL) and CSPF officials provided, including reports prepared by the court- appointed independent special counsel. We interviewed knowledgeable industry stakeholders, participant advocates, CSPF officials, International Brotherhood of Teamsters officials and members, and federal officials— including officials from the Pension Benefit Guaranty Corporation (PBGC), DOL, and the U.S. Department of the Treasury (Treasury). To describe the major factors that led to CSPF’s critical financial condition, we conducted semi-structured interviews and reviewed CSPF documentation, relevant scholarly materials, trade and industry articles, government reports, conference papers, research publications, and working papers. We also collected actuarial, financial, and other data on current and historical measures of plan assets, liabilities, investment performance, and other factors, and performed our own analyses of these data. The data and documentation collected were generally from the plan or agencies that oversee pensions. We determined the information to be generally reliable for the purposes of our objectives. To describe CSPF’s investment policy and the process by which it was set and executed we (1) reviewed CSPF’s investment policy statements, court orders and consent decree amendments, and other documentation provided by CSPF officials; (2) interviewed CSPF officials, including pension plan staff, the board of trustees, and the investment advisor, and representatives of the named fiduciary serving the plan at the time of our review; and (3) summarized certain aspects of CSPF’s assets using year- end performance reports prepared by the named fiduciaries. To describe how CSPF has performed over time compared to similar pension plans, we analyzed investment and fee data from DOL’s Form 5500, the government’s primary source of pension information. We also examined CSPF’s investment returns in comparison to a customized Wilshire Associates’ (Wilshire) Trust Universe Comparison Service (TUCS) benchmark of trusts with $3 billion or more in assets. CSPF provided these data and the data are included in the independent special counsel reports. Wilshire provided supplemental data using the same benchmark specifications. CSPF and DOL Document Reviews We reviewed three types of documentation provided by CSPF for changes in named fiduciaries; changes in investment policy, strategy, and asset allocation; major issues that affected funding; and how these issues affected CSPF’s investment strategy and policy. Select independent special counsel reports. CSPF officials provided 4th quarter reports for each year from 1982 through 2002 and available quarterly reports from 2003 through 2007. We downloaded all available quarterly reports from 2008 through 2017 from CSPF’s website. Select board of trustee meeting minutes. We requested board of trustee meeting minutes from 1983, 1994-95, 1998-2005, 2007-2010, and 2016 so we could review trustee discussions from the first full year the plan was covered by the 1982 consent decree, the most recent full year; periods that included a recession and/or when the plan’s assets performed poorly; and periods that preceded a change or reappointment of the named fiduciary. CSPF officials selected portions of the trustee meeting minutes from those years that pertained to the following topics: named fiduciary reports concerning investment performance; discussions relating to the amortization extension the Internal Revenue Service (IRS) granted to the plan and the contribution rate increases the plan required of participating employers in an effort to comply with funding targets required as condition of the IRS-approved amortization extension; major amendments to the plan; significant reports concerning the plan’s financial condition; amendments to the consent decree; discussions relating to any inquiries or issues DOL raised; discussions of named fiduciary appointments or resignations; discussions of particularly significant contributing employer delinquencies, bankruptcies, and settlements; and discussions relating to the independent special counsel. In addition to the board of trustee meeting minutes, CSPF officials provided select documentation on similar topics a former secretary of the board of trustees retained (1995 through 2008). Select correspondence between CSPF and DOL. CSPF officials provided select correspondence with DOL from 1987 through 2017 relating to DOL’s oversight of the plan. CSPF officials said they provided all records of those communications that related to significant, substantive, and nonroutine issues. The correspondence excluded other documents, such as periodic reports concerning asset rebalancing and correspondence related to fairly noncontroversial motions to the consent decree. In addition, DOL provided documentation throughout the course of our engagement, including documentation it provided between September and October 2017 that it had not previously identified as being relevant to our review. We completed an on-site file review at DOL in September 2017, and DOL sent us additional electronic documentation in September and October 2017. Overall, we reviewed extensive documentation from DOL—spanning over 10,000 pages of paper-based and electronic files— and spent substantial time cataloging and categorizing it. However, DOL officials reported that certain documentation related to CSPF was no longer available because it had only been retained for the time specified in the records retention policy of the relevant office. Semi-structured Interviews We conducted 23 semi-structured interviews with federal agency officials and other stakeholders, including affected parties, and persons knowledgeable about unions, participants and retirees, the trucking industry, collective bargaining agreements, and multiemployer pension plans. We also interviewed three stakeholders with actuarial expertise to specifically understand actuarial standards and procedures. We selected knowledgeable stakeholders based on review of literature and prior GAO work, and recommendations from other stakeholders. We judgmentally selected stakeholders whose expertise coincided with the scope of our objectives and who would be able to provide a broad range of perspectives. In our semi-structured interviews we asked about key factors affecting the plan, the broader regulatory and financial environment in which multiemployer plans operate, and solvency options for plans like CSPF. Investment Policy Statement Review We reviewed CSPF’s investment policy statements after CSPF entered into a consent decree in 1982, most of which are documented in the consent decree or other court orders. Seven of the investment policy statements were developed by named fiduciaries in consultation with the plan’s board of trustees and four were developed by the trustees. (See fig. 21.) From each investment policy statement, we compiled relevant information on: (1) investment philosophy and plan characteristics considered in developing it, (2) investment return benchmarks, (3) asset allocation, and (4) strategies and assets. See table 3 for select asset allocation information. Asset Summary To describe how CSPF’s investment policy was executed, we compiled information from performance reports prepared by named fiduciaries. We reported CSPF’s asset allocation generally based on the aggregate asset allocation categories CSPF’s named fiduciaries assigned in those reports. CSPF provided these reports for the end of each year 1984 through 2016—except 1992 and 1995, for which it provided reports as of the end of November. Information we compiled included the plan’s: account breakdown (i.e., assets in named fiduciary and passively- asset allocation; and investment assets withdrawn to pay benefits and administrative expenses. When possible, we checked the information from year-end performance reports against that in other sources. Specifically, to ensure we captured the vast majority of the plan’s assets in our asset summary we compared the total amount of plan assets named fiduciaries reported with Net Assets reported in CSPF’s Form 5500 filings, available from 1982 through 2016. We generally found these totals to be similar for each year—in most years the difference was about or under 1 percent. Also, named fiduciary performance reports included information on withdrawals from investment assets to meet pension and administrative expense obligations as of the end of each year, except for 1995 and 1999-present. For 1995 through 2016, we compiled this information from independent special counsel reports. For years in which we had overlapping information, 1996 through 1998, we found the reported totals were similar—no more than about 0.6 percent difference in each of those years. Based on our review we believe that the differences were insignificant to our overall analysis and did not impact our findings. Form 5500 Data Analysis To determine investment returns, investment fees, and administrative expenses for CSPF and related comparator group multiemployer defined benefit plans, we analyzed electronic Form 5500 information, the primary, federal source of private pension data. We analyzed information from 2000 through 2014, the most current and complete year at the time we performed our analysis. We began our analysis with 2000 data as data on investment returns and plan fees is primarily found in the Schedule H. Schedule H information was first collected in 1999. But we begin our analysis with 2000 data as electronic data became more reliable the year after the schedule was introduced. We have previously reported on the problems associated with the electronic data of the Form 5500. To mitigate problems associated with the data, we used Form 5500 research data from PBGC. PBGC analysts routinely and systematically correct the raw 5500 data submitted by plans, and PBGC’s Form 5500 research data are thought to be the most accurate electronic versions. Although we did not independently audit the veracity of the PBGC data, we took steps to assess the reliability of the data and determined the data to be sufficiently reliable for our purposes. For example, we performed computer analyses of the data and identified inconsistencies and other indications of error and took steps to correct inconsistencies or errors. A second analyst checked all computer analyses. Funded Status Funded status is a comparison of plan assets to plan liabilities. One measure of funded status is the funded percentage, which is calculated by dividing plan assets by plan liabilities. Another measure of funded status is the dollar amount of difference between plan assets and plan liabilities; the excess of plan liabilities over plan assets is the unfunded liability (or surplus if assets exceed liabilities). In this report, we measured funded status using the Actuarial Value of Assets and the Actuarial Accrued Liability, which are the basic measures used to determine the annual required minimum contribution for multiemployer plans under ERISA. We chose these measures because of the consistent availability of data for these measures. There are other ways to measure plan assets and plan liabilities. The Actuarial Value of Assets can be a “smoothed” value that differs from the market value of plan assets. The Actuarial Accrued Liability depends on the choice of actuarial cost method and discount rate, and on whether it is determined on an ongoing plan basis or a plan close-out basis. While different measures of plan assets and liabilities will produce different measures of funded status at any particular point in time, we found that our use of the Actuarial Value of Assets and the Actuarial Accrued Liability was sufficient for our purposes, which included examining the plan’s progress relative to statutory funding standards as well as its trend over time. Comparator Group Construction We developed multiple comparison groups for our analysis. The general rationale behind these comparator groups is to identify plans with similar fundamental characteristics, such as plan size or plan maturity, for purposes of investment return and fee and expense comparisons. We created the following two comparator groups: 1. Large plans (in terms of assets). We ordered multiemployer defined benefit plans by descending 2014 plan assets (line 2a of the 2014 of the Schedule MB). Because one of our key analyses of the data involves comparing investment returns across plans, we also limited the comparable plans to those that share a common plan year to CSPF (specifically if they have the same plan year-end of December 31). We selected the 20 plans that had the largest plan asset values. This includes CSPF, which was the second largest multiemployer plan as of the beginning of 2014. Because these comparator plans are among the largest, they should have similar cost advantages. For example, for investment management services, they should have similar advantages in obtaining lower fees and thus garner greater net returns due to the more favorable fee structures. 2. Mature plans (in terms of retiree liability proportions). We ordered multiemployer defined benefit plans by their similarity to CSPF’s ratio of retiree to total liabilities as of the beginning of calendar year 2000. The ratio of retiree to total liabilities is defined as line 2(b)1(3) of the 2000 Schedule B divided by total liabilities of line 2(b)4(3) of the 2000 Schedule B. To compare retiree to total liability ratios, we created a variable for the absolute value of the difference between CSPF’s ratio and that of a given plan. We ordered the plans by ascending differences in the ratios (excluding any with missing differences). CSPF was the top plan because its difference is zero by definition. Because one of our key analyses of the data involved comparing investment returns across plans, we also limited the comparable plans to those that shared a common plan year with CSPF (specifically if they have the same plan year-end of December 31). Of the plans that had the same plan year as CSPF and assets over $300 million, we selected the 20 plans (including CSPF) that had the smallest absolute difference from CSPF in the retiree-to-total liability ratio. Plans with a high ratio of liabilities attributable to retirees will have a relatively large portion of future benefit payments attributable to those that are older and retired. By selecting plans that were similarly mature to CSPF (and had $300 million in assets as of the beginning of 2000), we identified plans that may have had a similar basis for their plan investments, similar cash flow characteristics, or similar potential deviations between time-weighted and dollar-weighted average investment returns over time (see section below entitled “Calculation of Average Investment Return over Multiple Years”). That is, these plans should have roughly similar cost advantages and similar considerations in their investment objectives such as the balance of cash flows into and out of the fund and the plans’ investment horizons. Similarity in the balance of cash flows is important because it helps to mitigate the influence of plan maturity on the weighted average investment return over multiple years. The year 2000 was used to select the group because the primary purpose of this group is comparison of investment returns for plans that are similarly situated at the beginning of the period being analyzed. Calculations of Annual Investment Returns Our calculation of investment returns is based on the investment return calculation expressed in the Form 5500 instructions for the Schedule MB. Specifically the instructions of the 2014 Schedule MB state: Enter the estimated rate of return on the current value of plan assets for the 1-year period ending on the valuation date. (The current value is the same as the fair market value—see line 1b(1) instructions.) For this purpose, the rate of return is determined by using the formula 2I/(A + B – I), where I is the dollar amount of the investment return, A is the current value of the assets 1 year ago, and B is the current value of the assets on the current valuation date. Enter rates to the nearest .1 percent. If entering a negative number, enter a minus sign (“ - “) to the left of the number. After preliminary analysis of the variable and consultation with a GAO senior actuary, we determined that Form 5500, Schedule H contains all the information necessary to derive the calculation for years prior to 2008—as far back as 1999 when the Schedule H first came into existence. Additionally, we made adjustments for the timing of cash flows, to the extent indicated by the data. For example, employer and employee contributions that were considered receivable at the end of the prior year and thus included in the Schedule MB calculation were instead included in the year when the plan received the cash for the contribution. Thus, our calculation of annual rate of return is expressed as line items of the 2014 Schedule H to be: 2 * / [[{item1(f)a} – {item 1(b)1(a)} - {item 1(b)2(a)} - {item 1j(a)}] + [{item1(f)b} – {item 1(b)1(b)} - {item 1(b)2(b)} - {item 1j(b)}] – [{item 2d} - {item 2a(3)} – {item 2c}]] Or expressed with expository names as: (2 * (TLINCOME - TOTLCON - OTHERINCOMEW)) / ((TASSTSBY - (ERCONBOY + EECONBOY + OTHER_LIAB_BOY_AMT)) + (TASSTSEY - (ERCONEOY + EECONEOY + OTHER_LIAB_EOY_AMT)) - (TLINCOME - TOTLCON - OTHERINCOMEW)) For purposes of data reliability and validation of our results, we ran permutations of the calculation to see how, if at all, certain items could influence the calculation. In two permutations, we changed the timing of net asset transfers to or from other plans. (This occurs when, for example, there is a plan merger.) A senior actuary determined whether the calculations with/without net asset transfers affected our calculation. If the timing of the net transfer caused the investment return calculation to vary by more than 0.1 percent, we excluded the data for that particular plan in that particular year. We also ran another calculation that did not include “other” income so we could estimate the impact of not adjusting for such information. Calculation of Average Investment Return over Multiple Years Historical average investment returns over multiple years can be calculated in at least two different ways. One measure is the “time- weighted” average return, calculated as a geometric average of the annual returns during the period. A time-weighted average measures average investment performance without regard to the order of the annual returns or the impact of different plan circumstances over time. Another measure is the “dollar-weighted” average return–also known as the “internal rate of return” (and also referred to as the “cash flow weighted” return in this report)—which reflects the impact of the plan’s cash flow pattern. The dollar-weighted average return is the rate that, when applied over time to the asset value at the beginning of the period and to each year’s net cash flow into or out of the plan over the period, reproduces the asset value at the end of the period. We calculated dollar-weighted average returns (along with some time- weighted returns for comparison), for both CSPF and for the multiemployer system as a whole, as discussed in the report. We used a market value of plan assets for this purpose. The dollar-weighted average captures the impact of negative cash flow on average investment return. For example, with negative cash flow, investment results in an earlier year can have a bigger impact than investment results in a later year because more money is at stake in the earlier year. Using the same beginning-of-period asset value, and subsequent annual net cash flows into or out of the plan, used in calculating the dollar- weighted average return, we also performed a hypothetical calculation of what CSPF’s end-of-period asset value would have been if the plan had earned 7.5 percent per year instead of its actual return. Calculation of Fees and Expenses Conceptually, there are multiple ways to express investment fees, but our analysis used the following two methods for calculating them: Investment fee ratio. Investment fees [line 2i(3) of the 2014 Schedule H] divided by end-of-year net assets [line 1l(b) of the 2014 Schedule H] less receivables [line 1b(1)(b); line 1b(2)(b); and line 1b(3)(b) of the 2014 Schedule H]. Investment fees per participant. Investment fees [line 2i(3) of the 2014 Schedule H] divided by total (end-of-year) participants [line 6f of the 2014 main form]. Administrative Expense Calculations We define administrative expenses as all other expenses besides investment fees. In part, we used this definition of administrative expenses as it represents the expenses that remain after excluding investment fees. In addition, according to a PBGC analyst, this is the unit of analysis that they also used in their study of administrative expenses. Administrative expense to benefits paid. This is administrative expenses (professional, contract and other) divided by benefits paid. For administrative expenses we derived the value by taking total expenses less investment fees . For benefits paid, we used the 2014 Schedule H, line 2e(1), “Benefit payment and payments to provide benefits directly to participants or beneficiaries, including direct rollovers.” However, if the benefit payment value for such payments is missing or zero, we used the 2014 Schedule H, line 2e(4) “Total Benefit Payments” since the plan may be expressing their benefit payments on another line. Administrative expense per participant. Administrative expenses (professional, contract and other) divided by total (end-of-year) participants . For administrative expenses we derived the value by taking total expenses [line 2i(5) of the 2014 Schedule H] less investment fees [line 2i(3) of the 2014 Schedule H]. PBGC Study on Administrative Expenses PBGC has reported on administrative expenses and included various breakouts of these data in past data book supplements. The calculations of administrative expenses in this report are similar to those used by PBGC. Certain differences may exist because our calculation did not include certain multiemployer plans that reported missing data. Additionally, our population of multiemployer plans included only those plans exclusively associated with defined benefit features. The table below compares our results for plans with 5,000 or more participants, which is a subset of plans analyzed in the PBGC study. Our results used a sample that includes three fewer plans than the PBGC study, but our distributional results were within one-tenth percent for the administrative expense ratio and within $5 of the administrative expenses per participant (see table 4). Comparing the administrative expenses across reports using other statistics such as the minimum, maximum and standard deviation shows similar results for the PBGC and our analysis (see table 5). The mean administrative expenses per participant differ by $2.47. This difference is 1.5 percent lower than the PBGC estimate and could be a result of the difference in sample size. We also performed additional analyses as summarized below. We compared CSPF’s annual returns against plans that have the largest assets among multiemployer defined benefit plans (with the same plan year as CSPF) and CSPF’s results against these plans were broadly similar to results for the mature plans (see fig. 22). We compared CSPF’s administrative expenses as a percentage of benefit paid against other large plans. As noted in this report, CSPF has the lowest relative administrative expenses among the comparators in 2014 with administrative expenses at 1.4 percent of benefits (see fig. 23). In addition, CSPF’s administrative expenses as a percentage of benefits were consistently below the median. Analysis of Wilshire TUCS Data For our analysis of Wilshire TUCS data, we used two sources of data. Data from 1999 through 2016 was provided by CSPF. CSPF provided reports of their TUCS custom comparison group, master trusts with greater than $3 billion in assets. These data also included the year-end return results for the total fund (also known as the combined fund) as well as returns by subcategory such as a specific named fiduciary or fund. For example, subcategories listed for year-end 2006 included the results for both named fiduciaries (Goldman Sachs and Northern Trust) as well as the passively-managed accounts (then known as the CSSS fund). The custom comparison groups for the 1999 through 2016 data were determined each year in early-February of the year following the December 31 return results for the prior year. Thus, over time more master trusts were added (or subtracted) depending on the level of assets for the master trusts in that year. For example, the return results for year- end 1999 are determined as of February 10, 2000 and the group of master trusts with more than $3 billion contains 62 observations. The number of trusts in the custom group of master trusts with more than $3 billion generally grew over time with the number peaking with the return results for year-end 2014 (determined as of February 9, 2015), which contains 124 observations. The TUCS data from 1995 through 1998 was provided by Wilshire. The comparison group for these data were not selected each year, but, instead, selected retrospectively. For example, the comparison group of master trusts with more than $3 billion from 1995 through 1998 was selected as of January 9, 2017. There were 99 reported observations in 1995 and 132 observations in 1998. In addition, the 1995 through 1998 TUCS data did not include specific returns for CSPF. We were able to find the annual year-end return in the December (i.e. year-end) management report, which for these years was provided by the named fiduciary, Morgan Stanley. Appendix II: Selected Events Affecting the Central States, Southeast and Southwest Areas Pension Fund Below is a list of selected events that have affected the Central States, Southeast and Southwest Areas Pension Fund (CSPF) as identified through a review of relevant documentation and interviews with stakeholders and agency officials. It is not intended to be an exhaustive list of the events that have impacted CSPF, nor is it intended to include comprehensive descriptions of each event. Appendix III: Key Provisions of the Central States, Southeast and Southwest Areas Pension Fund’s Consent Decree Brief History and Current Status of Consent Decree On September 22, 1982, the Department of Labor (DOL) entered into a court-enforceable consent decree with the Central States Southeast and Southwest Areas Pension Fund (CSPF) to help ensure the plan’s assets were managed for the sole benefit of the plan’s participants and beneficiaries as required by the Employee Retirement Income Security Act of 1974 (ERISA). The consent decree has been amended several times and currently remains in effect, as amended, under the jurisdiction of the Federal Court for the Northern District of Illinois, Eastern Division. Below is a description of the key parties to and their primary responsibilities under the consent decree. Key Parties and Their Primary Roles under Consent Decree The consent decree defines roles and responsibilities for its parties, including the court, the court-appointed independent special counsel, DOL, the plan and its Board of Trustees, and the independent asset manager, which is called the named fiduciary. Court The primary role of the court is to oversee and enforce the consent decree. Specifically, the court: appointed an independent special counsel to assist it in administering has approval over the appointment of named fiduciaries and trustees; has approval over the appointment of investment managers of the may, for good cause shown, remove a named fiduciary after 60 days’ notice provided to the named fiduciary and DOL; and may, upon request by the plan, dissolve the consent decree absent good cause shown by DOL why the consent decree should continue in effect. Independent Special Counsel The court-appointed independent special counsel is intended to serve the court by assisting in identifying and resolving issues that arise in connection with the plan’s compliance with the consent decree and Part 4 of Title I of ERISA, and to report on the plan to the court. Specifically, the independent special counsel: has full authority to examine the plan’s activities and oversee and report on the plan’s performance of the undertakings of the consent decree; may, with court approval, employ attorneys, accountants, investigators, and others reasonably necessary and appropriate to aid him in the exercise of his responsibilities; has full access to all documents, books, records, personnel, files, and information of whatever type or description in the possession, custody, or control of the plan; may attend meetings of the plan, including meetings of the board of trustees and any meetings at which plan-related matters are discussed or considered; can petition the court to compel the plan to cooperate with the independent special counsel in the performance of his duties and responsibilities; may consult with DOL, the Internal Revenue Service, and other agencies, as appropriate, but must provide access to DOL upon its request to any documents prepared by the independent special counsel within the exercise of his power; is required to file quarterly reports, as well as any other reports the independent special counsel deems necessary or appropriate, with the court, and provide copies to DOL and the plan; may have other powers, duties, and responsibilities that the court may later determine are appropriate; and cannot be discharged or terminated during the duration of the consent decree except for leave of court, and upon the termination, discharge, death, incapacity, or resignation of an independent special counsel, the court will appoint a successor. Department of Labor Under the consent decree, DOL has an oversight role and may object to certain proposed plan changes. Specifically, DOL: may request and review certain reports provided by the plan and any documents prepared by the independent special counsel in the exercise of his authority; may object to the appointment of proposed trustees, named fiduciaries, investment managers of the passively-managed accounts, and asset custodians; receives notice of proposed changes to the plan’s investment policy statements from the plan; and may object to the dissolution of the consent decree. CSPF (including Board of Trustees and Internal Audit Staff) The plan must operate in full compliance with the consent decree, with ERISA, and with any conditions contained in determination letters it receives from the Internal Revenue Service. Specifically, CSPF, its board of trustees, and its internal audit staff must meet certain requirements. is required to use an independent asset manager known as the named fiduciary; must rebid the named fiduciary role at least once within every 6 years, with the option to extend the appointment for 1 calendar year; may remove a named fiduciary without cause shown on 6 months’ written notice to the named fiduciary and DOL; must cooperate with the independent special counsel in the performance of his duties and responsibilities and with DOL in its continuing investigation and enforcement responsibilities under ERISA; is required to recommend to the court three replacement candidates, agreeable to DOL, to replace an outgoing independent special counsel; and is required to maintain a qualified internal audit staff to monitor its affairs. is required to appoint, subject to court approval, the investment managers of the passively-managed accounts; is prohibited from authorizing any future acquisitions, investments, or dispositions of plan assets on a direct or indirect basis unless specifically allowed by the consent decree; and is required to comply with ERISA fiduciary duties, such as monitoring the performance of the assets of the plan, under Part 4 of Title I of ERISA. is required to review benefit administration, administrative expenditures, and the allocation of plan receipts to investments and administration; and is required to prepare monthly reports setting forth any findings and recommendations, in cooperation with the executive director of the plan, and make copies available to the independent special counsel and, upon request, to DOL and the court. Named Fiduciaries The independent asset managers, known as named fiduciaries, are appointed by the plan’s trustees, subject to court approval, and have exclusive responsibility and authority to manage and control all assets of the plan allocated to them. Specifically, the named fiduciaries: may allocate plan assets among different types of investments and have exclusive authority to appoint, replace, and remove those have responsibility and authority to monitor the performance of their are required to develop, in consultation with the Board of Trustees, and implement investment policy statements for the assets they manage, giving appropriate regards to CSPF’s actuarial requirements. Appendix IV: GAO Contacts and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the individual named above David Lehrer (Assistant Director), Charles J. Ford, (Analyst-in-Charge), Laurel Beedon, Jessica Moscovitch, Layla Moughari, Joseph Silvestri, Anjali Tekchandani, Margaret J. Weber, Adam Wendel, and Miranda J. Wickham made key contributions to this report. Also contributing to this report were Susan Aschoff, Deborah K. Bland, Helen Desaulniers, Laura Hoffrey, Jennifer Gregory, Sheila McCoy, Mimi Nguyen, Jessica Orr, Monica P. Savoy, and Seyda Wentworth. Related GAO Products Central States Pension Fund: Department of Labor Activities under the Consent Decree and Federal Law. GAO-18-105. Washington, D.C.: June 4, 2018. High-Risk Series: Progress on Many High-Risk Areas, While Substantial Efforts Needed on Others. GAO-17-317. Washington, D.C.: February 15, 2017. Pension Plan Valuation: Views on Using Multiple Measures to Offer a More Complete Financial Picture. GAO-14-264. Washington, D.C.: September 30, 2014. Private Pensions: Clarity of Required Reports and Disclosures Could Be Improved. GAO-14-92. Washington, D.C.: November 21, 2013. Private Pensions: Timely Action Needed to Address Impending Multiemployer Plan Insolvencies. GAO-13-240. Washington, D.C.: March 28, 2013. Private Pensions: Multiemployer Plans and PBGC Face Urgent Challenges. GAO-13-428T. Washington, D.C.: March 5, 2013. Pension Benefit Guaranty Corporation: Redesigned Premium Structure Could Better Align Rates with Risk from Plan Sponsors. GAO-13-58. Washington, D.C.: November 7, 2012. Private Pensions: Changes Needed to Better Protect Multiemployer Pension Benefits. GAO-11-79. Washington, D.C.: October 18, 2010. Private Pensions: Long-standing Challenges Remain for Multiemployer Pension Plans. GAO-10-708T. Washington, D.C.: May 27, 2010. The Department of Labor’s Oversight of the Management of the Teamsters’ Central States Pension and Health and Welfare Funds. GAO/HRD-85-73. Washington, D.C.: July 18, 1985. Investigation to Reform Teamsters’ Central States Pension Fund Found Inadequate. HRD-82-13. Washington, D.C.: April 28, 1982.

Why GAO Did This Study Multiemployer plans are collectively bargained pension agreements often between labor unions and two or more employers. CSPF is one of the nation's largest multiemployer defined benefit pension plans, covering about 385,000 participants. Since 1982, the plan has operated under a court-enforceable consent decree which, among other things, requires that the plan's assets be managed by independent parties. Within 7 years, CSPF estimates that the plan's financial condition will require severe benefit cuts. GAO was asked to review the events and factors that led to the plan's critical financial status and how its investment outcomes compare to similar plans. GAO describes (1) what is known about the factors that contributed to CSPF's critical financial condition; (2) what has been CSPF's investment policy, and the process for setting and executing it, since the consent decree was established; and (3) how CSPF's investments have performed over time, particularly compared to similar pension plans. GAO reviewed relevant federal laws and regulations; interviewed CSPF representatives, International Brotherhood of Teamsters officials and members, federal officials, and knowledgeable industry stakeholders; reviewed CSPF documentation including investment policy statements and board of trustee meeting minutes; and analyzed investment returns and fees from required, annual pension plan filings and from consultant benchmarking reports. What GAO Found The Central States, Southeast and Southwest Areas Pension Fund (CSPF) was established in 1955 to provide pension benefits to trucking industry workers, and is one of the largest multiemployer plans. According to its regulatory filings, CSPF had less than half the estimated funds needed to cover plan liabilities in 1982 at the time it entered into a court-enforceable consent decree that provides for oversight of certain plan activities. Since then, CSPF has made some progress toward achieving its targeted level of funding; however, CSPF has never been more than 75 percent funded and its funding level has weakened since 2002, as shown in the figure below. Stakeholders GAO interviewed identified numerous factors that contributed to CSPF's financial condition. For example, stakeholders stated that changes within the trucking industry as well as a decline in union membership contributed to CSPF's inability to maintain a healthy contribution base. CSPF's active participants made up about 69 percent of all participants in 1982, but accounted for only 16 percent in 2016. The most dramatic change in active participants occurred in 2007 when the United Parcel Service, Inc. (UPS) withdrew from the plan. At that time, UPS accounted for about 30 percent of the plan's active participants (i.e. workers). In addition, the market declines of 2001 to 2002 and 2008 had a significant negative impact on the plan's long-term investment performance. Stakeholders noted that while each individual factor contributed to CSPF's critical financial condition, the interrelated nature of the factors also had a cumulative effect on the plan's financial condition. Both CSPF's investment policy and the process for setting and executing it have changed several times since the consent decree was established in 1982. The original consent decree gave an independent asset manager—called a named fiduciary—exclusive authority to set and change the plan's investment policies and manage plan assets, and prohibited CSPF trustees from managing assets or making investment decisions. Initially, the named fiduciaries sold the troubled real estate assets acquired during the pre-consent decree era. Subsequent changes include the following: In 1993, the named fiduciaries started to increase investment in equities, and their policies continued to direct that asset allocations be weighted toward equities until early 2017. Between 2003 and 2010, the court approved three plan decisions to move a total of 50 percent of CSPF's assets into passively-managed accounts (passive management typically seeks to match the performance of a specific market index and reduce investment fees). An early-2017 investment policy change precipitated by CSPF's deteriorating financial condition will continue to move plan assets into fixed income investments ahead of projected insolvency, or the date when CSPF is expected to have insufficient assets to pay promised benefits when due. As a result, assets will be gradually transitioned from “return-seeking assets”—such as equities and emerging markets debt—to high-quality investment grade debt and U.S. Treasury securities with intermediate and short-term maturities. The plan is projected to become insolvent on January 1, 2025. CSPF officials and named fiduciary representatives said these changes are intended to reduce the plan's exposure to market risk and volatility, and provide participants greater certainty prior to projected insolvency. GAO found that CSPF's investment returns and expenses were generally in line with similarly sized institutional investors and with demographically similar multiemployer pension plans. For example, GAO's analysis of returns using the peer group measure used by CSPF known as the Wilshire Associates' Trust Universe Comparison Service (TUCS), showed that CSPF's annual investment returns since 1995 were above the median about as many times as they were below. Similarly, comparing CSPF's returns to a peer group of similar multiemployer defined benefit plans using federally required annual reports found that CSPF's annual investment returns were in line with those of its peers. Specifically, CSPF's annual returns were above the median nine times and below it six times—and CSPF's overall (dollar-weighted) average annual return from 2000 through 2014 was close to that of the peer median average return of 4.8 percent. In addition, GAO found that CSPF's investment fees and other administrative expenses have also been in line with other large multiemployer plans. For example: CSPF's investment fees as a percentage of assets were about 9 percent lower than the median of large defined benefit multiemployer plans over the 2000 through 2014 period—though much of that difference is accounted for by a relative reduction in investment fees since 2007. CSPF's investment fees as a percentage of assets were, on average, about 34 basis points (or 0.34 percent). CSPF's administrative expenses related to the day-to-day operations of the plan have also been in line with other large multiemployer plans. CSPF's administrative expenses per participant were below the median for large defined benefit multiemployer plans for 12 of the 15 years over the 2000 through 2014 period. As of 2014, CSPF's administrative expense was $98 per participant, which is about 16 percent less than the median for large defined benefit multiemployer plans. What GAO Recommends GAO is not making recommendations in this report.

gao_GAO-18-630

gao_GAO-18-630_0

Background The federal government has recognized 573 Indian tribes as distinct, independent political communities with tribal sovereignty. There are different categories of tribal lands, with differing implications with respect to ownership and administration. Reservations are defined geographic areas with established boundaries recognized by the United States. Tribal lands vary in size, demographics, and location. For example, those lands smallest in size are less than one square mile, and the largest, the Navajo Nation, is more than 24,000 square miles (the size of West Virginia). Tribal land locations can range from extremely remote, rural locations to urban areas. Figure 1 shows tribal lands in the United States according to the 2010 Census. The term “broadband” commonly refers to Internet access that is high speed and provides an “always-on” connection, so users do not have to reestablish a connection each time they access the Internet. Broadband service may be “fixed”—that is, providing service to a single location, such as a customer’s home—or “mobile,” that is, providing service wherever a customer has access to a mobile wireless network, including while on the move, through a mobile device, such as a smartphone. Fixed and mobile broadband providers deploy and maintain infrastructure to connect consumers to the Internet. Providers offer fixed Internet service through a number of technologies, such as copper phone lines, fiber-optic lines, coaxial cables, wireless antennas, satellites, or a mix of technologies (see fig. 2). To install fixed or wireless infrastructure, providers must obtain permits from government entities with jurisdiction over the land or permission from public utilities to deploy infrastructure on existing utility poles. The federal government has emphasized the importance of ensuring Americans have access to broadband, and a number of agencies, including FCC, currently provide funding to subsidize broadband deployment in areas in which the return on investment has not attracted private investment. The Communications Act of 1934, as amended by the Telecommunications Act of 1996, specifies that consumers in “rural, insular, and high-cost areas” should have access to telecommunication services and rates that are “reasonably comparable” to consumers in urban areas. To achieve this goal, FCC administers the High-Cost program, which provides subsidies to providers of phone service in rural, insular, and other remote areas. In 2011, FCC launched a series of reforms to its High-Cost program, including adding support for broadband services, and created the Connect America Fund, which provides subsidies to fixed and mobile providers of telecommunications and broadband services in rural, insular, and other remote areas where the costs of providing service is high. To be eligible for Universal Service Fund support from FCC, a provider must be designated an Eligible Telecommunications Carrier by the appropriate state or by FCC and must meet certain service obligations. The Connect America Fund has distributed approximately $4.5 billion per year, and has separate funding mechanisms targeted to specific goals. For example, there are funds for fixed-phone and broadband service and funds for mobile service, including a Tribal Mobility Fund (Phase 1) that awarded nearly $50 million in 2014 for the provision of 3G and 4G service to unserved tribal areas. In addition to FCC, a number of other agencies provide funding for broadband deployment in unserved or underserved areas. For example, the United States Department of Agriculture’s Community Connect Program, which provides grants to rural communities to provide high- speed Internet service to unserved areas. The American Recovery and Reinvestment Act of 2009 (Recovery Act) mandated the development of a nationwide map of broadband availability. To implement the act, the National Telecommunications & Information Administration (NTIA)—an agency within the Department of Commerce—established a grant program to enable U.S. states and territories to collect state-level broadband mapping data. NTIA used these data to launch the National Broadband Map (www.broadbandmap.gov) in February 2011. As the funding for the NTIA’s program came to an end in 2014, NTIA stopped collecting data to update the map and, according to FCC officials, created a memorandum of understanding with FCC through which FCC agreed to maintain public access to the last version of the map. FCC issued rules in 2013 to begin collecting broadband deployment data, in addition to the broadband subscription data it had collected from providers since 2000. FCC sought, but did not receive, $3 million to update the National Broadband Map in its fiscal year 2015 and fiscal year 2016 budgets. In 2018, Congress directed FCC to develop a report by March 23, 2019, evaluating broadband coverage in certain tribal lands (to include an assessment of areas that have adequate broadband coverage, as well as an assessment of unserved areas), and to complete a proceeding to address unserved areas by September 23, 2020. Currently, FCC requires broadband providers to report on their broadband deployment by filing a form twice a year (Form 477). Fixed broadband providers submit a list of the census blocks in which their broadband service is available, and mobile providers submit “shapefiles”—a geospatial depiction of the coverage area, which FCC refers to as “polygons”—of their coverage areas. FCC uses providers’ 477 data to develop a statutorily mandated annual report on advanced telecommunications capability. In addition, in 2016, FCC began publishing its own maps of broadband deployment, using the information from providers’ Form 477 filings. In February 2018, FCC launched an updated map of fixed broadband deployment (https://broadbandmap.fcc.gov/#/). This map allows users to search for broadband deployment by address and provides summary-level statistics regarding broadband deployment in specific tribal lands (see fig. 3). According to FCC officials, this new map format will support more frequent data updates. FCC also provides national maps of mobile LTE coverage; these maps do not allow users to access data at the same level of granularity as the maps of fixed broadband (see fig. 4). FCC Collects Data on Broadband Availability but Lacks Accurate and Complete Data on Broadband Access on Tribal Lands FCC collects and uses data that capture broadband availability to measure broadband access on tribal lands, leading to overstatements of broadband access on tribal lands. Specifically, FCC’s method of collecting mobile and fixed broadband data from providers (the Form 477) does not accurately or completely capture broadband access on tribal lands because it (1) captures nationwide broadband availability data— areas where providers may have broadband infrastructure—but does so in a way that leads to overstatements of availability, and (2) does not capture information on factors that FCC and tribal stakeholders have stated can affect broadband access on tribal lands, such as affordability, service quality, and denials of service. Nonetheless, FCC uses its Form 477 broadband availability data in annual broadband deployment reports to measure the percentage of Americans living on tribal lands with or without access to broadband, and to measure progress toward FCC’s strategic goal of increasing all Americans’ access to affordable broadband. By using broadband availability data to measure broadband access on tribal lands, FCC overstates broadband access on tribal lands. FCC Collects Broadband Availability Data, but Its Collection Method Leads to Overstatement of Availability on Tribal Lands FCC’s Form 477, its primary method of collecting nationwide broadband data, collects information on broadband availability, which identifies where providers have broadband infrastructure and could potentially provide broadband services but not where consumers can actually access those services. Moreover, the Form 477’s mobile broadband data- collection methods are not standardized, and its fixed broadband data- collection methods are not sufficiently granular to provide information about broadband availability on tribal lands. Mobile Broadband Data Collection FCC’s Form 477 requires mobile broadband providers to report their coverage areas by submitting geospatial data depicting the areas in which consumers could expect to receive the minimum advertised speed. FCC has previously noted the importance of collecting nationally standardized, uniform broadband data from providers to assess broadband availability and allow for easy comparison across providers. However, the Form 477 does not require that providers use a standardized method with defined technical parameters (such as signal strength, or amount of interference) when determining their coverage area, resulting in data that cannot be meaningfully compared across providers, according to FCC. To map their coverage areas, providers may use predictive models based on different measurement methods and a variety of factors known to affect mobile broadband service such as topography, tree cover, and buildings, among other factors. Providers and tribal stakeholders have expressed concern with the accuracy of FCC’s mobile broadband data, and FCC has acknowledged concerns that the lack of a standardized method resulted in data that were unreliable for the purposes of determining mobile broadband coverage for specific geographic areas, such as tribal lands. About half of the tribal government representatives we interviewed told us that they believe FCC’s data overstate mobile LTE broadband availability on their lands. For example, a few representatives expressed concerns with the accuracy of the mobile data in areas with varied terrain, such as mountains and valleys. In comments to FCC, broadband providers have also raised concerns regarding the accuracy of the mobile coverage data generated by the Form 477 for the purposes of identifying areas eligible for funding through FCC’s Mobility Fund Phase II program, which provides federal funding to increase mobile broadband services in unserved areas. In 2017, in response to such concerns, FCC reversed its prior decision to use the Form 477 data to identify specific areas eligible for federal funding through the Mobility Fund Phase II program. Instead, FCC undertook a one-time special data collection, for which it required providers to measure their coverage based on a common set of standards, in order to better identify unserved areas that would be presumptively eligible for funding. FCC plans to allow parties, including tribal governments, to challenge the data where they believe the data overstate mobile broadband coverage through August, 2018. Additionally, in an August 2017 Notice of Proposed Rulemaking, FCC requested comment on potential changes to modernize its Form 477 data collection, including whether it should require all providers to use a standardized method when submitting mobile coverage data on the form. FCC officials told us that they do not have a timeline for the development of a final rule, and as of August 2018, FCC had not yet issued a final rule on modernizing the Form 477. Fixed Broadband Data Collection The Form 477 collects fixed broadband data that are not sufficiently granular to accurately depict broadband availability on tribal lands. Specifically, FCC directs fixed broadband providers to submit a list of census blocks where service is available on the Form 477. FCC defines “available” as whether the provider does—or could, within a typical service interval or without an extraordinary commitment of resources— provide service to at least one end-user premises in a census block. Thus, in its annual reports and maps of fixed broadband service, FCC considers an entire block to be served if a provider reports that it does, or could offer, service to at least one household in the census block. FCC does not define a typical service interval or an extraordinary commitment of resources in its Form 477 instructions. However, FCC officials stated that providers should not report service in areas in which major construction would be required to provide service. A few providers told us that the lack of clear guidance from FCC regarding how to determine where broadband is available has led different providers to interpret the Form 477 directions in different ways, which can affect the accuracy and consistency of reporting from provider to provider. For example, in a filing with FCC, one provider stated that it had misapplied the definition of “available” and, as a result, overstated the availability of its services by almost 3,000 census blocks. As shown in figure 5, FCC’s definition of availability leads to overstatements of fixed broadband availability on tribal lands by: (1) counting an entire census block as served if only one location has broadband, and (2) allowing providers to report availability in blocks where they do not have any infrastructure connecting homes to their networks if the providers determine they could offer service to at least one household. Almost all the providers and private companies, and most of the representatives of tribal governments and organizations we spoke with told us that due to these issues, FCC’s definition of availability results in data that overstate broadband availability. According to FCC officials, FCC requires providers to report fixed broadband availability where they could provide service within a “typical service interval” and without “an extraordinary commitment of resources” in order to: (1) ensure that it captures instances in which a provider has a network nearby but has not installed the last connection to the homes, and (2) identify where service is connected to homes, but homes have not subscribed. FCC officials also told us that FCC measures availability at the census block level because sub-census block data may be costly to collect. In 2013, FCC considered collecting more granular nationwide data on broadband deployment but decided against collecting these data because it determined that the burden would outweigh the benefit. However, FCC, tribal stakeholders, and providers have noted that FCC’s approach leads to overstatements of availability. For example, in its 2017 Notice of Proposed Rulemaking on modernizing the Form 477 data collection, FCC acknowledged that by requiring a provider to report where it could provide service, it is impossible to tell whether the provider would be unable or unwilling to take on additional subscribers in a census block it lists as served. According to FCC, this limits the value of the data to inform FCC policies. In addition, several providers and tribal stakeholders we interviewed said that some “digital subscriber line” (DSL) and fixed wireless providers may overstate their service areas on the Form 477 because they may not take into account technological or terrain limitations that would affect their ability to actually provide service. FCC has also recognized that by measuring availability at the census block level, not every person will have access to broadband in a block that the data show as served, and FCC has noted that in rural areas, such as tribal lands, census blocks can be large and providers may only deploy service to a portion of the census block. A few representatives for tribal governments and organizations noted that the use of census blocks may uniquely overstate broadband availability on tribal lands when census blocks contain both tribal and non-tribal areas, because availability in the non-tribal portion of the block can result in the tribal area of the census block also being counted as served. FCC is considering requiring providers to report whether they are willing and able to serve additional customers in a census block and collecting sub-census block data in its 2017 proposed rulemaking on modernizing the Form 477. About one-third of the parties that commented on FCC’s proposals were not in favor of FCC collecting these more granular data on the Form 477, stating that the data would be less accurate and more burdensome for providers to collect and report, among other reasons, and questioned whether more detailed information on nationwide broadband availability is necessary. We heard similar concerns from a few of the providers and trade associations we interviewed. However, about one- third of the parties that commented on FCC’s proposals were in favor of collecting more granular data, stating that such data would be more useful for policymakers and more accurate. Additionally, a few tribally owned and non-tribal providers we interviewed told us that providers already maintain data for business purposes that would allow them to report more granular information on broadband availability. One stakeholder we spoke with pointed out that, as the federal government and states work to ensure the last remaining unserved areas—rural, low- population density areas including tribal lands—have service, sub- census-block-level data are needed to ensure that governments are making wise and accurate investments. FCC Does Not Collect Data on Several Factors That Affect Broadband Access on Tribal Lands FCC does not collect information on several factors that FCC and tribal stakeholders have stated can affect broadband access. FCC and tribal stakeholders have noted that broadband access can be affected by factors such as the affordability and quality of the broadband services being offered, and the extent to which providers deny service to those who request it. By collecting and using data on factors that can affect broadband access, FCC would have more complete information on the extent to which Americans living on tribal lands have access to broadband Internet services. Affordability: FCC has noted that affordability of broadband services can affect broadband access but does not collect information on the cost of broadband service on tribal lands on the Form 477. For example, in the National Broadband Plan, FCC cited affordable access to robust broadband service as a long-term goal, and in its Strategic Plan 2018–2022, FCC acknowledged that affordability is an important factor affecting broadband access and a key driver of the digital divide. Moreover, most of the representatives of tribal governments and organizations we spoke to told us that the affordability of broadband services is an important factor for understanding whether or not people on tribal lands could realistically access broadband services. Tribal government officials from one tribe we spoke with told us that residents on their lands cannot access broadband because it is too costly. For example, a provider that advertises services on the tribe’s land charges $130 per month for broadband services, approximately one-and-a-half times the average rate providers charge for comparable services in urban areas, according to FCC (see fig. 6). In the 2018 Broadband Deployment Report, FCC acknowledged that affordability can influence a consumer’s decision on whether to purchase broadband, but FCC did not consider cost in its assessment of broadband access on tribal lands, stating that pricing does not go to the congressional requirement to assess deployment and availability in conducting its inquiry as required by Congress under section 706 of the Telecommunications Act and also citing a lack of reliable comprehensive data on this issue. In addition, FCC officials we interviewed acknowledged that while broadband service may be technically available, it may be prohibitively expensive for some, which may make availability alone an incomplete indicator of broadband access. Quality of Service: In the Telecommunications Act of 1996 Congress recognized the importance of service quality by defining advanced telecommunications capability as any technology that enables users to originate and receive high-quality voice, data, graphics, and video telecommunications. In keeping with this legislation, FCC has consistently set thresholds for speeds that qualify as broadband services and has stated that “latency” and consistency of service figure prominently into whether a broadband service is able to provide advanced capabilities and thus whether users can access high-quality telecommunications. Likewise, almost all of the representatives for tribal governments or organizations we interviewed told us that quality of service is a key component of access to broadband and that routine outages, slow speeds, and high latency keep people on tribal lands from consistently accessing the Internet. Most tribal stakeholders and a few providers we interviewed told us that factors such as terrain, weather, and type of technology can all affect the quality of service an end user receives and, ultimately, the subscribers’ ability to access the Internet (see fig. 7). For example, some representatives of tribal governments and organizations told us issues like oversubscription— when a provider signs up more customers than its equipment can handle—and outdated or limited infrastructure result in low-quality services that cannot support advanced and, in some cases, basic functions. Though FCC uses the Form 477 to collect some data on advertised speeds from providers, FCC does not collect data on actual speeds, service outages, and latency on the form. In its 2018 Broadband Deployment Report, FCC stated that it did not consider FCC data on actual speed, latency, or consistency of service when evaluating broadband access due to the lack of appropriate data. FCC noted that the lack of Form 477 data on actual speeds in particular constrained evaluation of mobile broadband access. Service Denials: FCC has recognized that information on denials of service is pertinent to understanding actual broadband access but does not collect data on service denials in the Form 477. Specifically, in the National Broadband Plan, FCC recommended that FCC collect data to determine whether broadband service is being denied to potential residential customers based on the income of the residents in a particular geographic area. Some representatives of the tribal governments or organizations told us that that they were aware of a provider denying service to residents of tribal lands, despite the provider reporting broadband availability on at least a portion of those lands, according to our analysis of the Form 477 data. These representatives told us that they believed service was denied because of disputes with the tribal government, low demand for service, or the high costs of extending services to the home on tribal lands. Some representatives of tribal governments or organizations we spoke with also told us that providers may have denied service because their equipment was at capacity and could not accommodate new users (see fig. 8). For example, on three of the tribal lands we visited, we observed fiber optic cable located close to government and residential structures that did not have broadband access via fiber. According to tribal government officials, despite the physical proximity of the fiber optic cable, the tribal government and residents could not access it because the provider was not offering service or was unwilling or unable to build to the structures. A few providers we interviewed stated that they may not provide services to individuals who request them because of high-costs, administrative barriers, or technical limitations. However, FCC does not collect data on service denials on the Form 477. FCC Uses Broadband Availability Data to Measure Broadband Access on Tribal Lands, Overstating Access on Tribal Lands In its Strategic Plan 2018–2022 and the National Broadband Plan, FCC identified increasing all Americans’ access to affordable broadband as a long-term, strategic goal. Congress has similarly directed FCC to develop policies and programs aimed at increasing access to affordable broadband in all regions of the United States, including tribal lands, and required FCC to report annually on its progress. According to the Government Performance and Results Act (GPRA), as enhanced by the GPRA Modernization Act of 2010 (GPRAMA), agencies should use accurate and reliable data to measure progress toward achieving their goals. Additionally, Standards for Internal Control in the Federal Government state that agencies should use quality information— information that is complete, appropriate, and reliable—to inform decision-making processes and evaluate the agency’s performance in achieving goals. According to these standards, agencies should also communicate quality information externally to achieve the agency’s goals. However, FCC has used its Form 477 data, which do not accurately or completely measure broadband access on tribal lands, as its primary source to evaluate progress toward FCC’s strategic goal of increasing broadband access and to develop maps and reports intended to depict broadband access on tribal lands. For example, in its 2018 Broadband Deployment Report, FCC found that 64.6 percent of Americans residing on tribal lands have access to fixed broadband services. By using these data, FCC has overstated the extent to which Americans living on tribal lands can actually access broadband Internet services and FCC’s progress toward increasing broadband access. As a result, the digital divide may appear less significant as a national challenge, and FCC and tribal stakeholders working to target broadband funding to unserved or underserved tribal lands will be limited in their ability to make informed decisions. This increases the risk that residents living on tribal lands will continue to lack broadband access. Some tribal officials stated that inaccurate data have affected their ability to plan their own broadband networks and obtain federal broadband funding, and most of the tribal stakeholders we interviewed identified a pressing need for accurate data on the gaps in broadband access on tribal lands in order to ensure that tribes can qualify for federal funding and to effectively target the areas that need it most. For example, representatives for one tribal government that is providing broadband services said the government will not be able to use a federal grant to build broadband infrastructure in areas of their reservation that lack access, because the Form 477 data overstate actual access on the tribe’s land. As more than three quarters of the tribal governments we spoke to are working to provide broadband services on their lands in some capacity, overstating broadband access on tribal lands could affect the ability of a number of tribes to access federal funding to increase broadband access on their lands. As previously discussed, FCC is considering proposals to modify its Form 477 data collection as part of a 2017 Notice of Proposed Rulemaking, but FCC officials told us that the Commission does not have a timeline for issuance of a final rule. While some of FCC’s proposals could help address some of the limitations identified above by, for example, collecting more granular nationwide broadband availability data, FCC has not addressed specifically the collection of more accurate and complete data on broadband access for tribal lands in this proceeding. FCC has identified the need to improve broadband data for tribal lands in particular, and as previously noted, in 2018 Congress directed FCC to develop a report evaluating broadband coverage in certain tribal lands and initiate a proceeding to address the unserved areas identified in the report. FCC officials told us that FCC has not determined how it will address this requirement, but it is currently considering its options, including potentially addressing the requirement as part of its ongoing proposed rulemaking on modernizing the Form 477 data collection. An evaluation of broadband coverage on tribal lands that relies on the current Form 477 data would be subject to the limitations described above, including the overstatement of broadband access on tribal lands. Additionally, FCC has demonstrated that it is possible in some circumstances to collect more granular data when such data collection is targeted to a specific need or area. For example, in 2017 FCC began requiring certain providers that receive funding through the Connect America Fund to report the latitude and longitude of locations where broadband is available, and FCC has noted that these more granular data are extremely useful to the Commission, especially for rural areas where census blocks can be quite large. A few large providers and trade associations similarly stated in public comments on FCC’s proposed rulemaking to modernize the Form 477 process that FCC should target its collection of more granular broadband data to areas where the data are most likely to be overstated—specifically, large, rural census blocks with low population densities, such as those on tribal lands. Additionally, as discussed above, FCC undertook a one-time special data collection for Mobility Fund II to ensure that the mobile broadband data it collected would be reliable for the intended use. By developing and implementing methods for collecting and reporting accurate and complete data on broadband access specific to tribal lands, FCC would be able to better identify tribal areas without access to broadband and to target federal broadband funding to the tribal areas most in need. FCC Does Not Have a Formal Process to Obtain Tribal Input on Its Broadband Data, and Tribal Stakeholders Reported a Lack of Provider Engagement FCC Does Not Have a Formal Process to Obtain Tribal Input on Its Broadband Data FCC uses data submitted by broadband providers via the Form 477 process to develop maps and datasets depicting broadband services nationwide, and in specific locations, such as tribal lands, but does not have a formal process to obtain input from tribes on the accuracy of the broadband data. FCC’s 2010 National Broadband Plan noted the need for the federal government to improve the quality of data regarding broadband on tribal lands and recommended that FCC work with tribes to ensure that any information collected is accurate and useful. It also noted that tribal representatives should have the opportunity to review mapping data about tribal lands and offer supplemental data or corrections. Similarly, federal internal control standards note the need for federal agencies to communicate with external entities, such as tribal governments, and to enable these entities to provide quality information to the agency that will help it achieve its objectives. FCC officials told us that they address questions and concerns regarding provider coverage claims submitted to the Office of Native Affairs and Policy, which will work with tribal governments to help them identify inaccurate broadband data for tribal lands, and share tribal questions and concerns with the appropriate FCC bureaus. However, FCC does not have a formal process for tribes (or other governmental entities) to provide input to ensure that the broadband data FCC collects through the 477 process, or the resulting maps that FCC creates to depict broadband on tribal lands, are accurate. Similarly, FCC does not use other methods to verify provider-submitted Form 477 data on tribal lands against other sources of information, such as on-site tests or data collected by other agencies. When discussing the lack of a formal process for tribal representatives or other governmental entities to provide feedback on the accuracy of the 477 broadband data, FCC officials noted that if consumers and local officials have information on individual locations that lack broadband service, such information does not indicate that the entire census block lacks broadband service. Additionally, FCC officials noted that providers attest to the accuracy of the data and that FCC staff validate the data by conducting internal checks to identify possible errors, such as unlikely changes in a providers’ coverage area, and may follow-up with a provider to discuss such changes. However, these checks do not include soliciting input from tribes. About half of the tribal stakeholders we spoke to raised concerns that FCC’s broadband deployment data rely solely on unverified information submitted by providers. Additionally, most tribal stakeholders we interviewed told us that consistent with the recommendations in the National Broadband Plan, FCC should work directly with tribes to obtain information from them to improve the accuracy of its broadband deployment data for tribal lands. These stakeholders identified several ways in which FCC could work with tribes on this issue, including: conducting on-site visits with tribal stakeholders to observe the extent to which broadband infrastructure and services are present; conducting outreach and technical assistance for tribal stakeholders to raise awareness and use of FCC’s broadband data; and providing opportunities for the tribes to collect their own data or submit feedback regarding the accuracy of FCC’s data. FCC’s National Broadband Plan notes the importance of supporting tribal efforts to build technical expertise with respect to broadband issues, and federal internal control standards state that federal agencies should obtain quality information from external entities. Officials we interviewed in FCC’s Office of Native Affairs and Policy told us that they provide some outreach and technical assistance to tribal officials at regional and national workshops, and FCC officials stated that they conducted specific outreach to tribal entities regarding the Mobility Fund Phase II challenge process, while, about half of the tribal representatives we spoke to stated that they were not aware of the Form 477 data or corresponding maps, or raised concerns about a lack of outreach from FCC to inform tribes about the data. Some tribal stakeholders stated that if FCC were to solicit tribal input as part of its verification of the broadband data and maps, technical training and assistance could help tribes use and provide feedback on the data, or improve the collection and submission of their own data. A few of the stakeholders we interviewed noted that tribes can face difficulties when they attempt to challenge FCC’s broadband availability data. For example, in 2013, prior to the auction that distributed Tribal Mobility Fund Phase 1 support, FCC allowed interested parties to challenge FCC’s preliminary determinations regarding which census blocks lacked 3G or better service and would be eligible for support in the auctions. However, all of the tribal entities that challenged the accuracy of FCC’s data were unsuccessful in increasing the number of eligible areas. According to FCC officials, the tribal entities did not provide sufficient or sufficiently verifiable information to support their challenges. A few tribal stakeholders provided varying reasons for this, one of which was the need for more technical expertise to help the tribe meet FCC’s requirements. Because FCC lacks a formal process to obtain tribal input on its broadband data, FCC is missing an important source of information regarding areas in which the data may overstate broadband service on tribal lands. Tribal stakeholders are able to provide a first-hand perspective on the extent to which service is available within their lands and the extent to which factors like affordability, service quality, and service denials affect residents’ ability to access broadband. FCC plans to award nearly $2 billion in support from the Connect America Fund to areas that it has identified as lacking broadband, including tribal lands. Any inaccuracies in its broadband data could affect FCC’s funding decisions and the ability of tribal lands to access broadband in the future. Additionally, in its 2017 report on tribal infrastructure, the National Congress of American Indians stressed the importance of including tribal governments in a leadership role with respect to collecting data on local infrastructure needs. Specifically, it stressed the need for the federal government to invest in tribal data systems and researchers to generate useful, locally specific data that can inform the development and implementation of infrastructure development projects and assess the effectiveness of those projects over time. By establishing a process to obtain input from tribal governments on the accuracy of provider- submitted broadband data that includes outreach and technical assistance, as recommended in the National Broadband Plan, FCC could help tribes develop and share locally specific information on broadband access, which would in turn improve the accuracy of FCC’s broadband data for tribal lands. The success of such an effort may rely on the tribes’ knowledge of, and technical ability to participate in, the process. Half of the Tribal Stakeholders We Interviewed Reported a Lack of Provider Engagement When discussing the need to improve data regarding broadband on tribal lands, FCC’s 2010 National Broadband Plan recommended that FCC develop a process for tribes to receive information from providers about broadband services on tribal lands. In 2011, FCC required that Eligible Telecommunications Carriers (providers receiving Universal Service Funds from FCC) serving tribal lands meaningfully engage with tribes regarding communications services (including broadband). Specifically, the providers must file an annual report documenting that this engagement included a discussion of, among other things, a needs assessment and deployment planning for communications services, including broadband. FCC’s 2012 guidance on fulfilling the engagement obligations, which FCC officials confirmed is still in effect, noted that the stated goal of the engagement requirement was to benefit tribal government leaders, providers, and consumers by fostering a dialogue between tribal governments and providers that would lead to improved services on tribal lands. The guidance further noted that the tribal engagement process “cannot be viewed as simply another ‘check the box’ requirement by either party,” and states that a provider should “demonstrate repeated good faith efforts to meaningfully engage with the tribal government.” Finally, FCC noted in its 2012 guidance that the guidance would evolve over time based on the feedback of both tribal governments and broadband providers and that FCC would develop further guidance and best practices. This approach is consistent with federal internal control standards, which call for agencies to communicate with, and obtain quality information from, external parties. About half of the tribal stakeholders we interviewed raised concerns about difficulties accessing information from providers regarding broadband deployment on their tribe’s lands, a key part of the provider engagement process, according to FCC’s guidance. For example, a representative from one tribe stated that a provider declined his requests to meet more than once a year to discuss the provider’s deployment of broadband services on the tribe’s land. A representative from another tribal government stated that some providers are very focused and transparent about their broadband plans and work with the tribe, while other providers treat tribal engagement as a “box to check” and send the tribe broadband deployment information that is not useful because it is redacted. Similarly, some tribal stakeholders stated that providers heavily redacted deployment information (which providers may consider proprietary) or required the tribe sign non-disclosure agreements to access deployment data. According to one tribal stakeholder, these non-disclosure agreements could possibly require tribes to waive tribal sovereign immunity in order to view the data. Some of the industry stakeholders we interviewed stated that they attempt to engage with tribes but the level of responsiveness from tribes varies. For example, some stakeholders stated that they send letters and do not hear back from tribes. One stakeholder stated that they make repeated attempts to contact tribes when they do not hear back after their initial contact, while another stated that a provider meets regularly with some tribes. Although FCC stated in its 2012 guidance that it would update the tribal engagement guidance and develop best practices based on feedback from tribal governments and broadband providers, it has taken limited steps to obtain such feedback from providers and tribal governments to determine whether its guidance is enabling meaningful tribal engagement. Additionally, FCC has not updated the guidance or issued best practices. Thus, FCC has limited information regarding whether its tribal engagement requirement is fulfilling its intended purpose. FCC officials we interviewed said that the Office of Native Affairs and Policy (ONAP) provided information and, in some cases, held training sessions about the tribal engagement obligation during workshops with tribal representatives, and encouraged representatives to contact ONAP with any concerns. ONAP officials also noted that they handle complaints from tribes regarding a lack of provider engagement and reach out to providers to address tribal concerns. ONAP officials stated that they have had internal discussions about whether the guidance is clear or needs revision, but this has not gone beyond internal discussions. A few of the tribal stakeholders provided examples of the benefits of providers engaging with tribes to ensure tribal representatives have access to information regarding broadband availability on their lands. For example, one representative stated that this information could help the tribes plan deployments by focusing on areas that they know the provider does not plan to serve. Another representative stated that tribal engagement could help improve the accuracy of FCC’s broadband maps. By obtaining feedback from both tribal stakeholders and providers on the effectiveness of FCC’s tribal engagement guidance to determine whether changes are needed, FCC would be better positioned to ensure that tribal governments and providers are sharing information in a manner that will lead to improved services on tribal lands. Conclusions FCC has collected data and developed maps and reports depicting broadband on tribal lands and has noted the lower levels of broadband access on tribal lands, in comparison to other areas. However, limitations in FCC’s existing process for collecting and reporting broadband data have led FCC to overstate broadband access on tribal lands. By taking steps to address these limitations and to collect data that more accurately and completely depict broadband access on tribal lands, FCC would have greater assurance that it is making progress on reducing the digital divide on tribal lands and targeting broadband funding to tribal lands most in need. Without taking these steps, FCC increases the risk that residents living on tribal lands will continue to lack broadband access. Compounding the limitations in FCC’s data collection process is FCC’s lack of a formal process to obtain tribal input on the accuracy of provider- submitted broadband data for tribal lands. By developing a process to solicit tribal input and ensuring that tribes know about the process and are equipped with the technical skills and abilities necessary to provide this information, FCC would be better able to ensure the accuracy of its broadband data for tribal lands. Moreover, FCC would be able to obtain firsthand, locally specific information on broadband access that could inform FCC’s policies and funding decisions and help FCC achieve its goal of increasing broadband access for all Americans, including those living on tribal lands. Finally, by obtaining feedback from providers and tribal stakeholders on the effectiveness of FCC’s tribal engagement guidance, FCC would be better positioned to assess whether its guidance is helping providers meet requirements and ultimately whether providers’ engagement is fulling its intended purpose of fostering a dialogue between tribal governments and providers that would lead to improved services on tribal lands. Recommendations We are making the following three recommendations to the Chairman of the Federal Communications Commission. The Chairman of the Federal Communications Commission should develop and implement methods—such as a targeted data collection—for collecting and reporting accurate and complete data on broadband access specific to tribal lands. (Recommendation 1) The Chairman of the Federal Communications Commission should develop a formal process to obtain tribal input on the accuracy of provider-submitted broadband data that includes outreach and technical assistance to help tribes participate in the process. (Recommendation 2) The Chairman of the Federal Communications Commission should obtain feedback from tribal stakeholders and providers on the effectiveness of FCC’s 2012 statement to providers on how to fulfill their tribal engagement requirements to determine whether FCC needs to clarify the agency’s tribal engagement statement. (Recommendation 3) Agency Comments We provided a draft of this report to FCC for review and comment. In written comments provided by FCC (reproduced in appendix III), FCC agreed with our findings and recommendations. In its written comments, FCC described efforts, some of which are already under way, that it felt would address each recommendation and stated its intent to build upon those efforts. For example, FCC explained that it is exploring methods to collect more granular broadband deployment data and noted the need to balance the burden on Form 477 filers. FCC also noted that it is starting work to address a statutorily-required evaluation of broadband coverage on certain tribal lands. We agree that increasing the granularity of deployment data is helpful in addressing data accuracy issues, but we also note that it is important to collect data related to factors that affect broadband access on tribal lands. FCC also described informal efforts to collect tribal feedback on providers’ broadband data and stated it would explore options for a formal process to collect feedback. Regarding our recommendation related to providers’ engagement efforts, FCC outlined its existing methods by which tribal stakeholders can provide feedback on providers’ engagement efforts and agreed that seeking additional feedback from tribal stakeholders and providers would be desirable. We agree that improving feedback in these ways could help FCC determine whether it needs to clarify its tribal engagement statement. FCC also provided technical comments, which we incorporated as appropriate. We are sending copies of this report to the appropriate congressional committees, the Chairman of the Federal Communications Commission, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-2834 or GoldsteinM@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix IV. Appendix I: List of Interviewees Appendix I: List of Interviewees Representatives from tribal governments or tribally owned broadband providers Choctaw Nation of Oklahoma (OK) Confederated Tribes of the Colville Reservation (WA) Fond du Lac Band of Lake Superior Chippewa (MN) Fort Belknap Indian Community (MT) Gila River Telecommunications, Inc. (AZ) Hopi Telecommunications, Inc. (AZ) Jamestown S’Klallam Tribe (WA) Karuk Tribe (CA) Leech Lake Band of Ojibwe (MN) Makah Tribe (WA) Navajo Tribal Utility Authority (AZ, NM, UT) Nez Perce Tribe (ID) Osage Nation (OK) Pueblo of Acoma (NM) Pueblo of Pojoaque (NM) Pueblo of San Ildefonso (NM) Taos Pueblo (NM) Red Spectrum Communications (Coeur d’Alene Tribe (ID)) Saint Regis Mohawk Tribe and Mohawk Networks, LLC (NY) San Carlos Apache Telecommunications Utility, Inc. (AZ) Southern California Tribal Chairmen’s Association - Tribal Digital Village Network (CA) Spokane Tribe of Indians and Spokane Tribe Telecom Exchange (WA) Standing Rock Telecommunications, Inc. (ND, SD) Warm Springs Telecommunications Co. (OR) Yurok Tribe and Yurok Connect (CA) Representatives from tribal associations/consortiums that include tribes Affiliated Tribes of Northwest Indians Middle Rio Grande Pueblo Consortium National Congress of American Indians Native American Finance Officers Association (NAFOA) REDINet Representatives from companies/academic groups that work with tribes AMERIND Risk Arizona State University, American Indian Policy Institute and School of Public Affairs Turtle Island Communications Representatives from providers/trade associations (non-tribally owned) AT&T Representatives from companies that collect broadband data Alexicon Connected Nation Government Agencies (non-tribal) Census Bureau U.S. Department of Agriculture’s Rural Utilities Service Department of Interior’s Bureau of Indian Affairs National Telecommunications and Information Administration Minnesota Office of Broadband Development One broadband provider we interviewed did not want to be included in this appendix. Appendix II: Objectives, Scope, and Methodology This report discusses the extent to which: (1) the Federal Communications Commission’s (FCC) approach to collecting broadband availability data accurately captures the ability of Americans living on tribal lands to access broadband Internet services and (2) FCC obtains tribal input on the accuracy of provider-submitted broadband data for tribal lands. To address both objectives, we analyzed FCC’s December 2016 fixed and mobile broadband availability data—the most recent data at the time of our review—to identify the speeds, technologies, and availability providers reported for federally recognized tribal lands. Providers currently report this information to FCC by filing a “Form 477,” twice a year. We also used 2010 U.S. Census data to identify census blocks completely or partially on tribal lands. To assess the reliability of FCC’s data and 2010 U.S. Census data, we reviewed a previous GAO reliability assessment, and for FCC’s data we conducted electronic testing and analysis of the data, reviewed FCC guidance and documentation, and interviewed FCC officials. Based on the results of our analysis, we determined the data to be reliable for our purposes, which were: (1) to inform our selection of tribal governments and providers for interviews and visits, as described below, and (2) to develop maps depicting fixed and mobile broadband availability for the nine tribal lands we selected for visits, in order to obtain tribal representatives’ feedback on the data. Specifically, we mapped; fixed broadband data according to speed and technology, and mobile data for long-term evolution (LTE) services by provider for each tribal land. We used those maps during our visits to discuss the accuracy of the data with representatives for each tribal government or tribally owned provider. Though we analyzed all up and download speeds that providers reported in the Form 477, for the purposes of this report we defined “broadband” as fixed Internet service reaching at least 25 megabits per second (Mbps) download and 3 Mbps upload speeds, in accordance with FCC’s advanced telecommunications capability benchmark in its 2018 Broadband Deployment Report. We also report on the availability of mobile broadband, which, for the purposes of this report, does not have a speed threshold and refers to long-term evolution (LTE) services. To address both objectives and obtain tribal government representatives’ feedback on the accuracy of FCC’s broadband data for their lands, we interviewed representatives from 25 tribal governments or tribally owned providers, including visits to 9 tribal lands. We considered a range of factors when we selected tribal governments and tribally owned providers for interviews, including our analysis of Form 477 data, recommendations from tribal, industry, or government stakeholders regarding tribal and non- tribal representatives familiar with broadband data issues, and demographic and geographic characteristics, among others. For example, we considered demographic characteristics such as unemployment rate from the 2011– 2015 American Community Survey data, and geographic characteristics such as rurality from the United States Department of Agriculture (USDA) Rural-Urban Commuting Area Codes data. The tribes included in our review vary with respect to location, level of broadband availability according to FCC, land mass, and population size and density. The results of our interviews are not generalizable to all tribal governments or tribally owned broadband providers. In addition to tribal governments and tribally owned providers, we interviewed six tribal organizations and four stakeholders who work with tribes on broadband issues. For reporting purposes, we developed the following series of indefinite quantifiers to describe the tribal responses from the 35 entities representing tribal stakeholders we interviewed: 3 to 7 is defined as “a few;” 8 to 15 is described as “some;” 16 to 20 is described as “about half;” 21 to 27 is described as “most;” and 28 to 34 is described as “almost all.” A full list of the tribal stakeholders we interviewed can be found in appendix I. Further, to obtain industry perspectives, we reviewed public comments submitted by providers and industry associations in FCC’s ongoing 2017 Notice of Proposed Rulemaking on Modernizing the Form 477 Data Program. We also interviewed 10 non-tribally owned fixed and mobile broadband providers and three industry associations to understand providers’ views on the Form 477 and how providers interact with tribal governments. When selecting providers for interviews, we included providers that reported serving the lands of tribal governments we interviewed and selected providers that varied in the percentage of tribal lands they reported serving. The providers we interviewed represent large, nationwide carriers as well as small, local carriers, and offer broadband via a variety of technologies, including fiber optics, digital subscriber line (DSL), fixed wireless, and mobile LTE. The results of our interviews with providers are not generalizable to all broadband providers. In addition, to address both objectives, we interviewed representatives from other government entities, as well as private companies that collect and report broadband data. A full list of the industry stakeholders we interviewed can be found in appendix I. To identify the extent to which FCC’s approach to collecting broadband availability data reflects the ability of Americans living on tribal lands to actually access broadband Internet services, we reviewed documentation of the Form 477 process, including submission guidance, and FCC’s proposals and public comments in its 2017 Notice of Proposed Rulemaking on Modernizing the Form 477 Data Program and Mobility Fund Phase II proceedings. We also interviewed FCC officials, industry stakeholders, and tribally owned broadband providers to understand FCC’s current process for collecting broadband data. To understand the purpose of the Form 477 data collection process and FCC’s strategic goals, we reviewed relevant statutes, and FCC documents, including FCC’s Strategic Plan 2018––2022, the National Broadband Plan, and FCC’s broadband deployment and progress reports. Given the importance placed on broadband access in these documents, we interviewed tribal stakeholders, as described above and reviewed FCC documents to identify factors affecting the ability of Americans living on tribal lands to access broadband Internet services. We also reviewed previous GAO work that identified barriers to broadband access on tribal lands. We compared the Form 477 process to FCC’s strategic goals and to factors affecting broadband access to determine the extent to which the Form 477 was designed to collect information on those factors and to meet FCC’s goals. We further evaluated this information against the Government Performance and Results Act, as enhanced by the GPRA Modernization Act of 2010 and Standards for Internal Control in the Federal Government. We also reviewed documentation for other FCC data collection programs, including the Measuring Broadband America program and the Urban Rate Survey, to determine the extent to which FCC collected data on factors affecting broadband access outside of the Form 477 process. To determine the extent to which FCC obtains tribal input on the accuracy of provider-submitted broadband data for tribal lands, we interviewed FCC officials and analyzed FCC documents regarding the collection procedures for the Form 477 data and FCC’s policies for working with tribal governments, as well as Connect America Fund documents regarding requirements for providers to share information with tribal governments. We also reviewed documents on past FCC Universal Service Fund processes to challenge broadband data and identified prior instances in which tribal governments or tribally owned providers challenged FCC’s broadband data and the outcomes of those challenges. Additionally, we interviewed tribal stakeholders, as described above, to understand the extent to which: (1) FCC involves tribal governments and other stakeholders in the validation of Form 477 broadband data, (2) tribal governments can access broadband data from FCC or providers, and (3) FCC’s Form 477 data accurately reflected broadband access on their lands. For the nine tribal lands we visited, we asked tribal governments or tribally owned providers to identify where the data do or do not accurately reflect broadband access on maps of FCC’s data. Further, to identify how providers complied with FCC’s tribal engagement requirement and obtain their perspectives, we interviewed providers and industry associations. We compared FCC’s data validation procedures and tribal stakeholders’ feedback on the process to FCC’s policies for working with tribal governments, FCC recommendations from the National Broadband Plan and Standards for Internal Control in the Federal Government. We also interviewed and received written comments from officials from other federal agencies that have broadband programs, including USDA Rural Utilities Service, the National Telecommunications and Information Administration (NTIA), and others, in addition to a state agency and three private companies that collect and report broadband data to understand how other entities collect and validate broadband data. We conducted this performance audit from June 2017 to September 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix III: Comments from the Federal Communications Commission Appendix IV: GAO Contact and Staff Acknowledgments GAO Contact Mark L. Goldstein, (202) 512-2834 or GoldsteinM@gao.gov. Staff Acknowledgments In addition to the contact named above, Keith Cunningham (Assistant Director); Crystal Huggins (Analyst in Charge); Katherine Blair; Lilia Chaidez; Camilo Flores; Adam Gomez; Serena Lo; Jeffery Malcolm; John Mingus; Joshua Ormond; Jay Spaan; James Sweetman, Jr.; Elaine Vaurio; and Michelle Weathers made key contributions to this report.

Why GAO Did This Study Broadband furthers economic development, educational attainment, and public health and safety; however, residents of tribal lands have lower levels of broadband access relative to the U.S. population. Congress has prioritized identifying and targeting funds to unserved areas. FCC uses data from broadband providers to develop maps and reports depicting broadband availability in the United States, with specific information on tribal lands. GAO was asked to review FCC's efforts to collect broadband data for tribal lands. This report examines the extent to which: (1) FCC's approach to collecting broadband data accurately captures broadband access on tribal lands and (2) FCC obtains tribal input on the data. GAO interviewed stakeholders from 25 tribal governments or tribally owned providers, and visited nine tribal lands. The selected tribes varied geographically and in levels of broadband availability, among other characteristics. GAO also reviewed FCC's rulemakings on broadband data and interviewed other tribal stakeholders, FCC officials, and 13 non-tribal broadband providers selected to include a diversity of technologies. Provider and tribal interviews were based on non-generalizable samples. What GAO Found The Federal Communications Commission (FCC) collects data on broadband availability from providers, but these data do not accurately or completely capture broadband access on tribal lands. Specifically, FCC collects data on broadband availability; these data capture where providers may have broadband infrastructure. However, FCC considers broadband to be “available” for an entire census block if the provider could serve at least one location in the census block. This leads to overstatements of service for specific locations like tribal lands (see figure). FCC, tribal stakeholders, and providers have noted that this approach leads to overstatements of broadband availability. Because FCC uses these data to measure broadband access, it also overstates broadband access—the ability to obtain service—on tribal lands. Additionally, FCC does not collect information on several factors—such as affordability, quality, and denials of service—that FCC and tribal stakeholders stated can affect the extent to which Americans living on tribal lands can access broadband services. FCC provides broadband funding for unserved areas based on its broadband data. Overstatements of access limit FCC's and tribal stakeholders' abilities to target broadband funding to such areas. For example, some tribal officials stated that inaccurate data have affected their ability to plan their own broadband networks and obtain funding to address broadband gaps on their lands. By developing and implementing methods for collecting and reporting accurate and complete data on broadband access specific to tribal lands, FCC would be better able to target federal broadband funding to tribal areas that need it the most and to more accurately assess FCC's progress toward its goal of increasing all Americans' access to affordable broadband. FCC does not have a formal process to obtain tribal input on the accuracy of provider-submitted broadband data. In the National Broadband Plan , FCC highlighted the need for a targeted approach to improve broadband availability data for tribal lands. As outlined in the plan, such an approach would include working with tribes to ensure that information is accurate and useful. About half of the tribal stakeholders GAO interviewed raised concerns that FCC relies solely on data from providers, and most stated FCC should work with tribes to improve the accuracy of FCC's data. Establishing a formal process to obtain input from tribal governments on the accuracy of provider-submitted broadband data could help improve the accuracy of FCC's broadband data for tribal lands. What GAO Recommends GAO is making three recommendations to FCC, including that it collect and report data that accurately measure tribal broadband access as well as develop a process to obtain tribal input on the accuracy of the data. FCC agreed with the recommendations.

gao_GAO-19-8

gao_GAO-19-8_0

Background Poverty in U.S. High Schools Poverty can adversely affect academic and other outcomes in profound ways. Specifically, living in poverty is linked with negative conditions for children at home, in schools, and in neighborhoods and communities, and can include substandard housing, homelessness, inadequate nutrition and food insecurity, inadequate home-based child care, increased health care costs, and unsafe neighborhoods. Poverty has a particularly adverse effect on the academic outcomes of children that starts in early childhood and continues through the academic pipeline. Chronic stress associated with living in poverty has been shown to adversely affect children’s concentration and memory which may impact their ability to learn. Census data from 2014 show a relationship between the rate at which students dropped out (left school without obtaining a high school credential) and family income. The dropout rate of students from high- income families was 2.8 percent, while the dropout rate for individuals from low-income families was 11.6 percent. Our prior work describes how the nation’s schools have become increasingly comprised of students in poverty. In school year 2015-16, of the 12.5 million students in public high schools (schools with grades 9- 12), over 5 million (40 percent) attended schools where at least half of the students were experiencing poverty, as indicated by eligibility for free or reduced-priced lunch. Nearly 1.8 million (over 14 percent) attended schools where at least three-quarters of the students were experiencing poverty (see table 1). Our prior work has also discussed the association between poverty and race or ethnicity. High schools with a relatively large proportion of students in poverty also tend to have a higher proportion of minority students, students with disabilities, and English learners. The link between racial and ethnic minorities and poverty is long-standing, and studies have noted concerns about this segment of the population that falls at the intersection of poverty and minority status in schools and how this affects their access to quality education. Characteristics of U.S. High Schools and the College Preparation Process Of the roughly 12.5 million students who were enrolled in public high schools during the 2015-16 school year, about 87 percent attended traditional public schools, according to Education data; the remaining students were enrolled at charters, magnets, and other types of public schools (see table 2). While not all students will decide to pursue college, those who do generally must prepare for and navigate the college admissions process while in high school. This process can involve multiple administrative and financial steps, according to information from Education and college advising organizations. (See figure 2 for more information on the college application and admissions process.) U.S. Department of Education College Readiness Initiatives The Department of Education plays a role in helping students be prepared for college through initiatives in several of its offices. For example, Education’s Office of Postsecondary Education (OPE) administers several discretionary grant programs designed to increase college readiness among students from disadvantaged backgrounds, such as the Gaining Early Awareness and Readiness for Undergraduate Programs (GEAR UP). GEAR UP aims to increase the number of low- income students who are prepared to enter and succeed in postsecondary education. In 2016, OPE awarded approximately $323 million in grants through GEAR UP. In addition, Education’s Office of Elementary and Secondary Education (OESE) provides grants and technical assistance to states and districts to encourage advanced course opportunities and college and career readiness initiatives. OESE also oversees states’ and districts’ use of Title I, Part A funds under the Elementary and Secondary Education Act, as amended. These funds provide financial assistance to school districts and schools with high numbers or high percentages of children from low-income families to help ensure that all children meet challenging state academic standards, and can be used to provide additional courses and college readiness programs in schools. Finally, Education’s Office of Federal Student Aid (FSA) is responsible for managing the student financial assistance programs authorized under Title IV of the Higher Education Act of 1965. These programs provide grants, loans, and work-study funds to students attending college or career school. FSA also publishes guidance and other resources related to federal student aid and college costs. These resources are designed for students and parents who are navigating the college application and financial aid processes. (For more information on Education’s grant programs relevant to college readiness, see appendix II.) Federal Efforts to Promote Equitable Access to Educational Resources Education and the Department of Justice (Justice) promote equitable access to education resources as part of their missions in two key ways: (1) conducting investigations of discrimination complaints; (2) issuing guidance on ways to address potential disparities; and (3) providing technical assistance. Education and Justice are responsible for enforcing a number of civil rights laws that prohibit discrimination in public schools on the basis of race, sex, disability, color, and national origin. (For examples of cases resolved by Education and Justice related to access to college preparation resources, see appendix IV.) To enforce relevant civil rights laws, Education carries out complaint- driven and agency-initiated investigations, which are called compliance reviews and which target problems that Education has determined are particularly acute. For example, in a recent review, Education’s OCR reviewed whether Black students in a Virginia school district had the same access to educational opportunities as other students. OCR found a significant disparity between the numbers of Black and White high school students who take AP, advanced courses, and dual credit programs. These discrimination cases can be resolved through several means, including voluntary resolution, dismissal, or closure due to insufficient evidence. Education may also terminate federal funds if Education determines that a recipient is in violation of civil rights laws and the agency is unable to reach agreement with the parties involved. Justice has the authority to file suit in federal court to enforce the civil rights of students in public education. Specifically, Justice investigates discrimination in school resources based on complaints filed under federal civil rights laws and monitors and enforces open federal school desegregation orders where Justice is a party to the litigation. For example, in 2015 Justice entered into a court-approved agreement with a Louisiana city school board after finding that more college preparatory courses were offered in schools that predominantly serve White students than in schools that predominately serve Black students. This agreement required, among other things, that the district ensure that all students were given the opportunity to take all courses offered in the district. In addition to enforcement actions, Education and Justice help promote equitable access to education resources by issuing guidance and providing technical assistance. For example, in 2014, OCR issued guidance addressing equitable access to educational resources, in part, to address chronic and widespread racial disparities in access to rigorous courses, academic programs, and extracurricular activities which can hinder the education of students of color. In this guidance, OCR describes proactive ways to address potential disparities in academic and extracurricular programs that are differentiated based on academic rigor (e.g., gifted and talented or college preparatory programs) or content (e.g., business, music, art, or career and technical education programs). This guidance includes the following steps that states and school districts can take to help ensure equal access to educational resources: designating an employee to review policies governing how resources are distributed to and within schools; evaluating resource access across and within schools; notifying parents, students, and community members of avenues to raise concerns about resource access; and taking proactive steps to identify disparities in access to resources. Education also offers technical assistance, through various means, such as conducting webinars, sponsoring and presenting at conferences, and disseminating resource guides to schools and school districts. High-Poverty Schools Offer Fewer of the Courses That Prepare Students for Public 4- Year College At a Glance: Student Access to College Preparation Courses and Admissions Expectations Poverty and Student Demographics Schools with the highest concentration of poor students were predominantly comprised of Black and Hispanic students. Access to more advanced math and science courses (e.g., calculus and physics) decreased as the level of school poverty increased. Larger high schools offered more advanced math and science courses than smaller schools, regardless of poverty level. Charter schools offered fewer advanced math and science courses than traditional and magnet schools, regardless of poverty level. Public 4-year colleges generally expect applicants to have completed three or four math and three or four science credits in high school, but we found that the percentage of schools offering these recommended courses decreased as poverty level increased. High-Poverty High Schools Largely Comprised of Black and Hispanic Students Our analysis of Education data for school year 2015-16 showed that high- poverty high schools were predominately comprised of Black and Hispanic students, while low -poverty schools had a higher proportion of White students. Specifically, roughly 80 percent of students attending high-poverty schools were either Black or Hispanic, but were less than 20 percent of students enrolled in low-poverty schools (see fig. 3). Access to Advanced High School Courses Varies Based on School Poverty Level, Size, and Type Poverty Level Our analysis of Education data for school year 2015-16 showed that students’ access to more advanced high school courses decreased as the level of school poverty increased. High-poverty schools represented 17 percent of all high schools in 2015-16. College Admissions Perspective Admissions officials from all four public, 4- year universities we interviewed reported that they look for students to take advanced coursework in high school in order to be more competitive applicants. Some college admissions officials and college advising organizations reported that students face academic difficulties when they get to college if they did not take advanced courses that help prepare for the rigor of college. A college admissions official we interviewed reported that over 90 percent of the university’s incoming freshmen took courses in high school that could earn college credit, such as Advanced Placement (AP), International Baccalaureate (IB), or dual enrollment courses. Across all poverty levels, almost all schools offered the basic math courses (algebra I and geometry); however, disparities in offering advanced math courses grew as school poverty level increased (see fig. 4). For calculus in particular, the percentage of schools offering the course decreased as school poverty level increased, with the gap between low- and high-poverty schools widening to nearly 35 percentage points (85 percent of low-poverty schools versus about 50 percent of high-poverty schools). Generally, a similar pattern emerged for science courses. Again, the majority of all schools, at least 90 percent across all poverty levels, offered biology; but for chemistry and physics, disparities grew as poverty increased. For example, almost 90 percent of low- poverty schools offered physics, with the percentage decreasing steadily to 62 percent for high-poverty schools. For courses that allow students to earn college credit and that can help make students more competitive applicants (see text box), our analysis showed a similar trend, with disparities that deepened as school poverty increased. For Advanced Placement (AP) courses overall, our analysis showed that the gap in courses offered was widest between the lowest and highest poverty schools—with over 80 percent of low-poverty schools offering at least one AP course compared to about 60 percent of high- poverty schools. We found a similar pattern for AP math and science courses. Among schools that offered any AP courses, nearly all low- poverty schools offered AP math compared to 75 percent of high-poverty schools, a nearly 20 percentage point gap (see fig. 5). Advanced Placement courses: Upon successful completion of the course and a standardized AP exam, a student may be qualified to receive college credit and/or placement into advanced college courses. International Baccalaureate courses: The International Baccalaureate (IB) courses are designed as an academically challenging and balanced program of education, with final examinations, that prepares students, usually aged 16 to 19, for success in college. Dual Enrollment/Credit programs: Dual enrollment/dual credit programs provide opportunities for high school students to take college-level courses offered by colleges, and earn concurrent credit toward a high school diploma and a college degree while still in high school. Across all poverty levels, larger public high schools offered more advanced math and science courses than smaller schools, according to our analysis of Education’s school year 2015-16 data. As illustrated in figure 7, this pattern held true for all math and science courses. In particular, among high-poverty schools, 90 percent of large schools offered calculus, compared to 54 percent and 11 percent of medium and small schools, respectively. Similarly, among high-poverty schools, over 90 percent of large schools offered physics compared to about two-thirds of medium and about a third of small schools. A similar pattern was evident for AP courses (see fig. 8). Among high- poverty schools, 97 percent of large schools offered AP courses compared to 68 percent of medium and 11 percent of small schools. Across all poverty levels, access to advanced courses differed by school type. We found that, in general, fewer charter schools, across all poverty levels, offered math, science, and AP courses, compared to traditional and magnet schools, according to our analysis of Education’s school year 2015-16 data (see fig. 9). Further, a higher percentage of magnet schools offered advanced courses (such as physics and AP courses), compared to traditional schools. We also analyzed alternative schools and special education schools. When analyzing Education’s data by school type, these schools had the lowest percentage of schools offering college preparatory courses. We focused our analyses in the body of the report on traditional, magnet, and charter schools, the school types with larger enrollments. Alternative and special education schools enroll fewer than 1.5 percent of high school students. See appendix V for full data tables, which include breakouts for alternative and special education schools. For AP courses, across all poverty levels, a lower percentage of charter schools offered these courses compared to traditional and magnet schools (see fig. 10). In particular, among high-poverty schools, 33 percent of charter schools offered any AP courses compared to 71 percent of traditional and 94 percent of magnet schools. We also analyzed high school course offerings based on whether schools were located in an urban, suburban, or rural location, but our regression model did not find a consistent association between school locale and course offerings. For example, a lower percentage of high-poverty schools in rural areas offered advanced math and science courses compared to high-poverty urban or suburban schools. However, a higher percentage of low-poverty rural schools offered advanced math and science courses than did low-poverty urban schools. For full results by school locale, see appendix V. High-Poverty Schools Were Less Likely to Offer Math and Science Courses Needed for College Admission Colleges often look for students to have completed multiple credits of a subject in high school, such as math or science; however, our analysis suggests that some high-poverty schools may not offer the math and science courses needed to meet basic admission expectations for public 4-year colleges. Based on our analysis of a generalizable sample of U.S. public 4-year college websites, an estimated 95 percent of colleges expected applicants to have completed three or four credits of math (see text box). Further, a majority of public 4-year colleges specifically recommended that applicants take algebra I, geometry, and algebra II. With respect to science an estimated 76 percent of colleges expected students to have completed three or four credits of science, with many specifically recommending biology, chemistry, or physics. (See fig. 11). Our analysis of Education data for school year 2015-16, however, found that the percentage of schools offering these recommended math and science courses decreased as poverty level increased. With respect to math courses, 7 percent of low-poverty schools did not offer the recommended math courses (at least algebra I, geometry, and algebra II), compared to 17 percent of high-poverty schools that did not offer these courses. Further, while 12 percent of low-poverty schools did not offer the recommended science courses (at least biology, chemistry, and physics), 41 percent of high-poverty schools did not. (See fig. 12). Odds of offering at least algebra I, geometry, and algebra II Generally, no statistically significant association. Odds of offering at least biology, chemistry, and physics Higher poverty schools were associated with lower odds of offering these courses compared to lower poverty schools. Odds of offering any AP courses Higher poverty schools were generally associated with lower odds of offering any AP courses, compared to lower poverty schools. Race Higher levels of Hispanic or Asian students were associated with lower odds of offering these courses. Higher levels of Black, Hispanic, or American Indian/Alaskan Native students were associated with lower odds of offering these courses. Generally, no statistically significant association. School Size Smaller schools were associated with lower odds of offering these courses, compared to larger schools. Smaller schools are associated with lower odds of offering these courses, compared to larger schools. Smaller schools were associated with lower odds of offering AP courses, compared to larger schools. School Type Alternative schools were associated with lower odds of offering these courses compared to traditional schools; however, the results were not statistically significant for other school types. Charter schools and alternative schools were associated with lower odds of offering these courses compared to traditional schools. Charter schools were associated with lower odds of offering any AP courses and magnet schools were associated with higher odds of offering any AP courses, compared to traditional schools. Officials from All 12 High-Poverty Schools Stated That Their Students Face Multiple, Complex Challenges to Prepare for College, and Some Had Efforts in Place to Help Students in High-Poverty Schools Confront Complex Challenges to Prepare for College Across the three selected states, officials representing the 12 high- poverty schools we visited consistently reported that students confront multiple challenges to being prepared to attend college. They cited a range of academic roadblocks to college, including that students are behind academically before they get to high school; that the schools they attend lack rigorous courses, such as AP courses; and that students struggle to attain grade point averages (GPA) high enough for admission to some 4-year colleges. Officials explained that family challenges and obligations can compound the academic challenges and make navigating the college admissions and enrollment process difficult for their students. Students have not made sufficient academic progress to be admitted to college, according to officials we interviewed at 12 high-poverty schools (see fig. 13). Officials representing most of these schools (10 of 12) reported that their students were often academically behind. For example, at one urban and predominantly Black Wisconsin high school, officials said that 80 percent of 9th graders were performing below grade-level targets for reading and math, and at a Georgia high school where nearly all of the students were eligible for free or reduced-price lunch, officials said that over 30 percent of freshman students in school year 2016-17 had to repeat the 9th grade. Insufficient academic progress can be compounded by challenges high- poverty schools face in offering advanced coursework. For example, officials at five schools said they did not offer calculus; officials at three of these schools noted this was because most students typically did not take algebra I in middle school and, therefore, did not have the time to progress to calculus. Officials at a high school with over 900 students reported they did not offer calculus or AP math courses due to low student demand and that they must weigh the cost of providing a course with the number of students who would benefit. Two high-poverty high schools we visited that did not offer calculus courses were exploring offering the courses to students through videoconference. However, an official from one school district we interviewed said the district uses videoconference as a last resort because they have found students learn better with a teacher physically present allowing for more exchange of dialogue. In addition, the challenge of finding and retaining high-quality teachers can exacerbate the difficulties high-poverty schools face in offering advanced courses, according to state educational agency officials in two of the states we visited. Offering advanced courses is important to providing challenging opportunities for students and avoiding remedial coursework once in college, according to college and high school officials we interviewed. Officials we interviewed stressed that taking advanced courses provides students with challenging academic opportunities that help to prepare students for the rigor of college courses, whether they pass their AP exams or not. A representative of a college advising organization said that while it is possible to get into college without higher-level math courses, these courses often determine if a student needs remedial math in college. Officials from two college advising organizations said that when students are required to take remedial courses in college, it can have a detrimental effect. They said remedial courses generally cost money but do not provide credits towards graduation and can delay graduation, and sometimes can contribute to students leaving college without a degree. School officials for almost all the schools we visited (11 of 12) also said that students often had low GPAs and SAT or ACT scores, which made them less competitive applicants for admission or scholarships to 4-year colleges. For example, the average GPA for 11th grade students at three Wisconsin high schools we visited was below 2.0; officials at one school told us that last year’s valedictorian had a 3.0 GPA. Further, officials at multiple schools said students feared they would not do well on the ACTs or SATs; and one counselor said this means that many students did not even try. Low GPAs and college entrance exam scores may be a particularly acute roadblock to 4-year college in areas where the state university system has grown increasingly competitive due to high demand, according to a counselor at one predominantly Hispanic California high school who said the state system is looking for students with 4.0 GPAs. Difficult Life Circumstances In addition to insufficient academic progress, a confluence of family, financial, and social-emotional challenges often confronts students in high-poverty schools, making it difficult for them to prepare for college, according to our interviews with school officials (see fig. 14). School and state education officials said that a range of stressors can compound the difficulties poor students face with learning and academic achievement. Officials at most of the schools (9 of 12) we visited and one state educational agency cited adverse conditions associated with poverty––such as hunger, homelessness, living in foster care, witnessing or experiencing violence or abuse—that made it hard for students to focus on school work. In one high school, officials reported that a school staff member handed out care packages to students every Friday to ensure students had something to eat on the weekend. Officials also reported that students demonstrated behavioral and emotional issues in their schools. Officials at one Wisconsin school said they have noticed a large increase in anxiety among students. This anxiety can be paralyzing for some students and, for others, can result in explosive and violent behavior that affects other students’ ability to learn, according to the school officials. Officials in 11 of the high-poverty schools we visited said that going to college often conflicts with a student’s need to help support their families or that the cost of college can be prohibitive. Some students provide an important source of income for their family or are the caregiver for family members, according to officials in nine schools. Family obligations can also affect students’ decisions about whether to take college preparation courses, according to one school administrator. For example, the principal of a California charter school said a high-performing student dropped an AP course because the demands from family were so great. In addition, officials in six schools said that the cost of college can deter low-income students. One of these officials reported that even with financial aid and scholarships, their students may not be able to cover even small gaps in funding. According to one high school counselor, the cost of going to college plus the practicalities of getting to and from school and figuring out how to pay for meals during breaks if dorms or the cafeteria are closed, are concerns for low-income students. Parents struggling with poverty may not expect their children to go to college, according to college advising officials and officials at most schools (10 of 12) we visited. For example, officials at one Georgia high school said that many students are aiming to be the first in the family to graduate high school (first generation high school graduates), and do not prioritize college. Similarly, at another school, officials said parents and students do not have the expectation of going to college because the parents had not been to college themselves. Students from high-poverty schools may continue to harbor low expectations upon admission to college because they feel they do not belong, according to a principal and a college advising official. In addition, first generation students usually do not have the family support and knowledge to feel confident in their abilities to navigate college life, as a college admissions official noted. School officials at one high school we visited said their students, who attend high school in a highly segregated area, have felt overwhelmed and intimidated trying to transition to a college with a predominately white student population. Barriers to Navigating College Processes A variety of factors—from the availability of high school counselors to taking college entrance exams—can make the college admissions and enrollment processes difficult for students in high-poverty schools, according to school, college, and college advising organizations in the communities we visited (see fig. 15). College admission officials in two of the states we visited noted the importance of the high school counselor in navigating the college admissions process, such as taking students to college fairs and building relationships with colleges. However, counselors often face high caseloads and competing priorities, such as getting kids to graduate and handling emotional and social issues, according to multiple school officials and local college advising organizations. In one rural school we visited, one counselor handled the needs of about 400 students and was also the bus driver and occasional substitute nurse. Taking the SAT and ACT exams can also pose challenges for students. For example, according to administrators at one school, the cost of the exams may be a deterrent. At another high school, counselors noted that students may lack transportation to the test site and, at another school, officials said weekend jobs kept students from taking the tests. Applying for financial aid can also be challenging for students from high- poverty schools, according to school and college advising organization officials. At six of the schools we visited, officials said that sometimes parents are reluctant to report their income, because they are undocumented or because the process is unfamiliar. In addition, some school officials told us that even families with legal immigration status can be reluctant to submit personal information to government websites because they distrust how the information will be used. College advising officials we interviewed in two states said that complicated family financial situations, such as when a student cannot obtain income information from a parent, can also make the financial aid process difficult. In addition, officials from two college advising organizations said that financial aid award packages can be difficult to understand. For instance, they said that these packages may not clearly explain what amount the student is responsible for paying. Further, the aid letters may not indicate the additional cost associated with room and board, books, and transportation, according to one of these officials. Finally, even after a student has been admitted to college, they still may experience obstacles before classes begin, according to our interviews. Four officials reported that lack of college advisement over the summer after high school graduation has led to “summer melt,” when students do not attend college as planned. Officials from a college advising organization said that sometimes students missed a step in the enrollment process, such as paying deposits or tuition balances before the semester begins. Some High-Poverty Schools Are Trying to Ease Roadblocks to College Officials representing selected state educational agencies, school districts, and high-poverty schools we visited reported that they try to mitigate the barriers students in high-poverty schools face in being prepared to attend a 4-year college, despite resource challenges. Free access to college courses. Providing students with free access to college courses was one way some states and schools have been able to help students prepare for college. For example, Georgia’s dual enrollment program allows high school students to earn college credit for free while working on their high school diploma. The program covers tuition, mandatory fees, and books. Administrators at a Georgia high school reported that the program has allowed some students to earn an associate’s degree upon graduation from high school, helping to ease the cost burden of college. A charter school we visited in California partners with local colleges and covers tuition, text books, and transportation for college courses. The school principal said that the school does not offer calculus, but students can take it at a local community college and receive college credit. Outside supports for college advising. In Georgia, officials from a college advising organization reported helping with the college admission process in selected schools, including registering students to take the ACT or SAT, organizing college visits, helping students research colleges, and helping students and parents apply for financial aid. They also said they used text messages as a way to reach out to students and remind them to complete certain steps in the enrollment process. In addition, officials from half of the schools we visited (6 of 12) reported their schools had, or previously had, federal grants that supported college readiness activities for disadvantaged students. For example, one Wisconsin high school where most students are eligible for free or reduced-priced lunch (90 percent) and are Black (82 percent) or Hispanic (14 percent) had a GEAR UP grant that supported students in the classes of 2017 and 2018 since middle school, according to the school administrators. Strategies to exhibit a college-going culture. To help encourage students to consider college as a possibility, officials at some high-poverty schools we visited reported using strategies to exhibit a “college-going culture” within the school. For example, based on our site visit interviews and observations, schools displayed college banners; opened college and career counseling centers; provided incentives, such as prizes, to complete financial aid applications; and posted testing and scholarship information in prominent locations (see figs. 16 and 17). At one urban high school we visited in Georgia, teachers displayed their alma maters on their classroom doors and the school held “College Fridays” so students could learn about different colleges, according to school administrators. All-hands-on deck approach. One California school reported using an “all-hands-on-deck” approach to getting students through the college admission process. Teachers, counselors, and administrators work together to track and follow up with students to ensure they take the needed coursework and do not miss a step in the admissions process. Officials reported that school staff built personal connections with the students and with the community outside of the school to encourage buy- in surrounding the college application process. At a high school in Georgia where 100 percent of students were eligible for free or reduced- priced lunch, school officials said they also used an all-hands-on-deck approach to help students persevere through personal challenges they face, such as balancing work and school or dealing with trauma. The school provides a team of administrators and counselors for each grade level to better identify when a student may be struggling and help support students’ college preparation goals, according to school administrators. Alignment of graduation requirements and college admission requirements. Wisconsin officials reported that the state made changes to better align high school requirements with college and career readiness expectations, and universities’ expectations by increasing its math and science graduation requirements from two units to three units of each, starting with the 2017 graduating class. According to a 2014 analysis by the Education Commission of the States, 18 states have complete or partial alignment between state high school graduation requirements and statewide higher education minimum admission requirements. In addition, the University of California and the California State University systems have established a uniform minimum set of courses, known as A-G requirements, required for admission as a freshman. These courses, offered in California high schools and online schools, are designed to ensure students have attained a body of general knowledge for more advanced study, according to information from the University of California. Even though it is not a state requirement, one Georgia school district reported that it requires two units of foreign language because it is a requirement of the University System of Georgia. Free college admission tests. In two of the states we visited, officials reported that students may take select college entrance exams or preparatory exams during a school day free of charge. Georgia pays for all 10th graders in public schools to take the Preliminary SAT (PSAT). Wisconsin officials reported that the state requires and provides the funding for all 11th graders in the state to take the ACT. A school district in California we visited noted that it covers the cost of the PSAT for 9th, 10th, and 11th graders in the district, as well as the SAT for 11th graders. In addition, officials at several schools said they offer students free online test preparation tools. College initiatives to improve access and retention. Officials at colleges in all three states we visited reported having initiatives that helped increase admissions or ease the transition to college for low- income or first-generation students. For example, officials at the University of Georgia said the college guarantees admission to the valedictorian of every accredited high school in the state. Admissions officials said this helped students with fewer educational opportunities to be competitive for admissions. California State University (CSU)–Los Angeles, as well as other CSU campuses, has a program to help improve access and retention of low-income and educationally disadvantaged students. Under the program, the university accepts a limited number of students who do not meet regular admission criteria and provides academic, and in some cases financial, assistance to these students. The university also offers a 6-week “summer bridge” program for first generation students since they are most in danger of dropping out between high school graduation and the first day of college classes in the fall. At the University of Wisconsin-Milwaukee, an admissions official said the university develops transfer plans for students who start at a 2-year community college, to ease the transition to a 4-year college. Agency Comments We provided a draft of this report to the Departments of Education and Justice for review and comment. These agencies provided technical comments, which we incorporated as appropriate. As agreed with your offices, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies to the appropriate congressional committee, the Secretary of Education, the Attorney General, and other interested parties. In addition, the report will be available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (617) 788-0580 or nowickij@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix VIII. Appendix I: Objectives, Scope, and Methodology Overview The objectives of this report were to (1) examine the extent to which high schools of different poverty levels offer courses to prepare students academically for college and (2) describe challenges that students in high-poverty schools face in being prepared to attend college. For our first objective, we analyzed federal data on college preparatory course offerings by school poverty level quartiles; and within these quartiles, we analyzed the demographic composition of students in those schools. We also analyzed course offerings of schools in each poverty quartile by school type, size, and locale. Further, we reviewed college admissions expectations for a generalizable random sample of public 4- year colleges and compared course offerings from schools in each poverty quartile to these expectations. Lastly, we conducted a regression analysis to explore whether and to what extent certain school-level characteristics were associated with higher rates of college preparatory course offerings. For our second objective, we visited selected high-poverty high schools in three states to provide illustrative examples of challenges students face in being prepared for college. In those states, we also interviewed officials from state educational agencies, school districts, college advising organizations, and public 4-year colleges. We focused on public 4-year colleges because these institutions offer a bachelor’s degree and are generally a more affordable 4-year option, compared to private colleges. The following sections contain detailed information about the scope and methodology for this report. Analysis of College Preparatory Courses National Data To determine the extent to which schools offer courses to prepare students academically for college, we conducted statistical analyses using the U.S. Department of Education’s (Education) Civil Rights Data Collection (CRDC) and the Common Core of Data (CCD). Specifically, the CRDC is a biennial survey that is mandatory for every public school and district in the United States. Conducted by Education’s Office for Civil Rights (OCR), the survey collects data on the nation’s public schools (pre-K through 12th grade), including course offerings, student characteristics and enrollment, and disciplinary actions. The CRDC collected data from nearly every public school in the nation (approximately 17,000 school districts, 96,000 schools, and 51 million students in school year 2015-16). The course offering variables we used in our analysis are for those courses typically associated with and reported by high schools. As a result, our analysis only includes high schools that have all grades 9, 10, 11, and 12 (a total of 14,111 high schools). We thus excluded schools that had any grades K-8. Further, we excluded juvenile justice facilities—because the provision of educational offerings may function differently in those schools—and schools with fewer than 10 students. Our analysis was conducted using the public-use data file of the CRDC for school year 2015-16, the most recent data available at the time of our analysis. We matched schools in the CRDC for school year 2015-16 to schools in the CCD for school year 2015-16 to enable us to perform certain analyses based on variables that are unique to the different datasets, and excluded schools for which there was not a match. CRDC data are self-reported by districts and schools, and consequently there is potential for misreporting of information. Although our analyses of these data showed disparities, taken alone, these disparities do not establish whether unlawful discrimination has occurred. The 2015-16 CRDC survey collected data on several math and science courses that are considered by Education to be college-preparatory courses. The college-preparatory math courses included in the CRDC are: algebra I; geometry; algebra II; advanced mathematics; and calculus. The college preparatory science courses included in the CRDC are: biology; chemistry; and physics. The CRDC also collected data on a number of variables related to Advanced Placement (AP) course offerings as well as other course offerings that potentially offer students college credit. See table 4 for full definitions of key variables. Analysis by Poverty and Student Demographics To analyze course offerings by the poverty level of the school, we pulled in data on free or reduced-price lunch (FRPL) eligibility from the 2015- 2016 CCD, and matched it to our universe of 14,111 high schools in the 2015-16 CRDC, given that the CRDC does not collect FRPL eligibility data. The CCD is administered by Education’s National Center for Education Statistics (NCES), and annually collects nonfiscal data about all public schools in the nation. A student is generally eligible for free or reduced-price lunch based on federal income eligibility guidelines that are tied to the federal poverty level and size of the family. State educational agencies supply these data for their schools and school districts. We then sorted high schools into poverty quartiles based on the percentage of students eligible for free or reduced-price lunch as follows: schools with 0 to 24.9 percent of students that are FRPL eligible, which we call low-poverty schools; schools with 25 to 49.9 percent of students that are FRPL eligible; schools with 50 to 74.9 percent of students that are FRPL eligible; and schools with 75 to 100 percent of students that are FRPL eligible, which we call high-poverty schools (see table 5). The poverty thresholds and measure of poverty discussed here and throughout this report were commonly used in the literature and also aligned with how Education analyzed its data. Further, to understand which students attend schools in the different poverty quartiles, we analyzed student demographic composition for each group of schools. Beginning in the 2014-15 school year, the National School Lunch Program included a new provision for providing free meals to all students in the school, without needing to collect individual applications from students to determine eligibility. This provision—known as the Community Eligibility Provision (CEP)—was implemented to expand access to free meals to all students and decrease household and administrative burdens for participating schools. We assessed whether the CEP variable had the potential to make sorting schools into quartiles based on the percentage of students eligible for free or reduce-price lunch unreliable. Our analysis showed that the number of schools in each poverty quartile remained roughly the same as in prior years and thus, we concluded the reported FRPL data was reliable for our purposes. Analysis by School Size To analyze course offerings by the size of public school a student attended, we sorted the 14,111 high schools in our universe into three groups, based on the number of students enrolled in the school, according to the 2015-16 CRDC data (see table 6). We excluded schools with fewer than 10 students because (1) schools of this size likely do not have the resources or infrastructure to offer advanced courses and (2) to prevent minor fluctuations in the data from having large effects on our results. We grouped schools into one of three size categories based on the number of students enrolled. The Department of Education and the CRDC do not have classifications of schools by size, so we determined reasonable size categories based on our analysis of the data. To arrive at these categories, we looked at average number of advanced course offerings by school size strata in groupings of 100 students. This analysis led to three categories based on the distribution of the data: 1 to 200 students (small schools); 201 to 1000 students (medium schools); and 1,001 or more students (large schools). Analysis by School Type To analyze course offerings by the type of public school a student attended, we sorted the 14,111 schools in our universe into mutually exclusive categories using the self-reported school type variable in the CRDC. The CRDC allowed schools to self-identify as special education, magnet, charter, and alternative schools (see table 7). The categories of public schools in the CRDC were not mutually exclusive; that is, schools could select multiple school types to describe their schools, such as a charter school that was also an alternative school. To create mutually exclusive categories for analytical purposes, we applied the following criteria: Alternative school: all schools that selected “alternative” as the school type in the CRDC, even if they selected other types as well. Special education school: schools that selected “special education” as the school type in the CRDC, except those schools that also selected the alternative school type. Charter school: schools that selected “charter” as the school type, except those schools that also selected the alternative school type or the special education school type. Magnet school: schools that selected “magnet” as the school type, except those schools that also selected the alternative school type, the special education school type, or the charter school type. Traditional school: schools that did not select any other school type in the CRDC. Table 8 provides the breakdown of students and schools captured in the 2015-16 CRDC after applying these criteria. Analysis by School Locale To analyze courses offerings by the locale of public school a student attended, we pulled in the school locale variable from the 2015-16 CCD and matched it to schools in the CRDC, which did not collect data on school locale. The locale variable in the CCD is primarily based on a school’s location relative to populous areas. The locale variable is divided into four main types: City, Suburb, Town, and Rural. For the purposes of our analyses, we combined the Town and Rural variables into one Town/Rural variable because they are defined similarly (see table 9). Table 10 provides the breakdown of students and schools captured in the 2015-16 CRDC after applying the GAO Categories above. CRDC and CCD Data Reliability We determined that the data we used from the CRDC and CCD were sufficiently reliable for the purposes of this report by reviewing technical documentation, conducting electronic testing, and interviewing officials from Education’s OCR and NCES. Past releases of the CRDC have subsequently been updated by Education to correct errors and omissions in the data. For our analysis of the 2015-16 CRDC, we used the data file that was publically available as of April 24, 2018. Regression Analysis We conducted a generalized linear regression with a logistic regression model using the 2015-16 CRDC and CCD data to explore whether and to what extent certain school-level characteristics were associated with higher rates of college preparatory course offerings, while controlling for other factors. Such a model allowed us to test the association between the offering of college preparatory courses and school characteristics, including poverty, while holding other school characteristics constant (school type, school size, school locale, student demographics). Table 11 lists the variables we included in our regression model. We conducted a separate regression for each of the course offerings or sequence of offerings listed as an outcome variable. Our regression model used the same universe of 14,111 schools as our descriptive analysis of the CRDC data. Since the regression model is based on observations across all independent variables, and some variables had a small number of missing data points, our final model had 13,278 observations. All regression models are subject to limitations and for this model the limitations included: Data we analyzed were by school rather than student. Consequently, we were not able to describe the association between our independent variables and a student’s access to college preparatory courses, while controlling for characteristics of an individual student, such as sex, race or ethnicity, disability status, or grade level. Instead, the school-level nature of the CRDC data limited our description of the associations between school characteristics and course offerings to whether there was an increase, decrease, or no effect on course offerings for schools with a given characteristic, controlling for other characteristics of the entire school’s population, such as school type. Some variables that may be related to student access to advanced courses are not available in the data. For example, in this context, it could be that parent education level or household type (single- versus multiple-headed household) could be related to course access. Results of our analyses are associational and do not imply a causal relationship. Typically, a logistic regression model, which is a generalized linear regression model, is appropriate when the model assumption of normality is not appropriate, as is the case with a binary (yes/no) outcome. A logistic regression model provides an estimated odds ratio, where a value greater than one indicates a higher or positive association, in this case, between whether a course is offered and the independent variable of interest, such as being a charter school or having a higher percentage of Black students. An estimated odds ratio less than one indicates lower odds of offering a given college preparatory course when a factor is present. Given the limitations of our model as described above, we present the results of our regression model in tables 12, 13, and 14 by describing the direction of the associations, rather than the estimated odds of outcome variables. For categorical variables in these tables, we provided the comparison school characteristic in brackets. For example, the results in these tables should be interpreted as charter schools were significantly less likely than traditional schools to offer AP courses, because the association is negative. For continuous variables (i.e., those starting with “Percent”), the results in these tables should be interpreted as the likelihood of offering courses decreased, if the association was negative, as the percentage of students in the school with a given characteristic increased. For example, as the percentage of Black students increased, we found that the likelihood of offering the sequence of at least three science courses decreased. Review of College Admission Criteria To determine which academic courses colleges expect applicants to take while in high school, we reviewed websites from a generalizable stratified random sample of 100 public 4-year colleges in the United States. The sample was selected from Education’s 2015-16 Integrated Postsecondary Education Data System (IPEDS), which contains data for colleges that participate in federal student aid programs authorized under Title IV of the Higher Education Act of 1965, as amended. Our sampling frame consisted of all public 4-year degree granting colleges that participated in Title IV federal student aid programs, predominately award baccalaureate degrees, have full-time first-time undergraduate students, and that are located in a U.S. state or the District of Columbia, yielding a universe of 555 colleges. We stratified the sample by groupings colleges based on admission rates into four strata. We computed the sample size of 100 schools to achieve a precision of at least plus or minus 10 percentage points for an estimate of a population proportion at the 95 percent confidence level. We then proportionally allocated the sample size across the defined strata. This sample allowed us to make national estimates about the admission criteria for expected high school coursework at public 4-year colleges. To review comparable information across the sampled schools, we developed a standardized web-based data collection instrument that we used to examine the admission criteria for first-time freshman applicants posted on each college’s website. Specifically, we attempted to identify the minimum required or recommended units of math, science, social studies, English, Foreign Language, and Fine Arts courses applicants are expected to take in high school to be considered for admission to the college. For math and science courses, we also attempted to identify any specified courses the colleges provide to meet the required or recommended units for those subject. We also collected information on whether or not each college required students to submit SAT or ACT exam scores to be considered for admission. We reviewed websites from September 2017 through November 2017. One analyst recorded information in the data collection instrument. The information was then checked and verified by another analyst. We collected complete information for all 100 colleges in our sample. We then analyzed the information across colleges. We did not, as part of our review of college websites, assess whether the information provided on the website accurately reflected the current admission policies of the college. Instead, this review was intended to better understand the courses that colleges expect students to take in high school. High School Site Visits To obtain information on the challenges students attending high-poverty high schools face in being prepared to attend public 4-year colleges, we selected three states—California, Georgia, and Wisconsin—and conducted site visits to four high schools in each of the states (for a total of 12 high schools). To select states for our site visits, we used the 2013- 14 CRDC data—the most recent available at the time of our selection—to sort states based on the percentage of their schools offering courses commonly associated with college readiness. We selected states that fell below the national average in percentage of schools offering Algebra II. We also considered states that were at or above the national average in percent of high-poverty schools offering two or fewer math and science courses. We also selected states providing us with a mix of state policies on college readiness and geographic diversity. Within each of the three states we used 2013-14 CRDC data to select high schools to visit that had greater than 75 percent of students eligible for free or reduced-price lunch (FRPL) and that offered a range of math and science courses. We also considered the number of AP courses offered by the school. As secondary criteria, we selected schools to achieve variation in school size, school type, and locale, to gather perspectives from officials in a diverse array of high-poverty schools. At each of the 12 schools, we interviewed the principal and other key leadership staff, and high school counselors. To supplement our site visits, we interviewed by phone state educational agency officials in each of the three states, as well as school district officials for most of the schools we visited. We interviewed officials from at least one local college advising organization in each of these states. In addition, we interviewed officials from at least one public 4-year university in each of the three states, for a total of four public 4-year universities. We selected universities that admit a high percent of in-state students, to attempt to talk to officials who were familiar with the high schools that we selected. These interviews provided us with information about what college admission officers view as challenges in admitting students from high poverty schools and the challenges students face in being successful in completing college. Because we selected the schools judgmentally, based on our criteria, the findings about the challenges these schools reported or the strategies they used to help students address those challenges cannot be generalized to all schools nationwide. Additional Interviews In addition to interviews in our site visit states, we interviewed officials from the Education Commission of the States, National Association for College Admission Counseling, and the College Board. We also held interviews and reviewed documentation from the U.S. Departments of Education and Justice to gather information on their programs supporting access to college preparation opportunities. We also reviewed relevant literature, as appropriate. We conducted this performance audit from May 2017 to October 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: U.S. Department of Education Discretionary Grant Programs to Increase College Readiness in K-12 Students Appendix III: Federal Agencies Responsible for Enforcing Civil Rights Laws in Public Schools Appendix IV: Selected Federal Civil Rights Enforcement Cases Related to Access to College Preparation Courses and Programs Department of Education According to administrative data from the U.S. Department of Education (Education), the Office for Civil Rights (OCR) received over 480 civil rights cases related to college and career readiness and resource comparability from FY 2011 through 2017. Some of these cases were initiated by external complaints and other reviews were initiated by Education. In the selected cases described below Education found underrepresentation of minority students or English learners in advanced, honors, or Advanced Placement (AP) middle and high school courses or in other types of college preparatory programs. This selection of cases is not generalizable, and was selected for illustrative purposes only. Education Case 1: Equitable Access to Advanced Courses for Black Students in an Ohio School District. In a 2016 investigation, OCR identified a number of potential Title VI compliance concerns regarding equitable access to certain resources for Black students at some schools. Specifically, OCR found that students at three schools, including two predominantly Black high schools, did not have the opportunity to take advanced courses taught live at their schools and, therefore, could not engage in-person with the course instructors. According to OCR’s investigation, students participated remotely, watching the class through a video system. When the classes first started during the 2011-12 school year, the district staffed the distance classroom with paraprofessionals to assist the students. For that year, the district reported using technology to offer greater curriculum choices to its students through distance learning, especially when a sufficient number of students did not sign up for an advanced course at a specific school. After OCR notified the district of its concerns regarding this practice, the district placed teachers in these classrooms effective the 2014-15 school year. The district reported to Education that it was also pursuing efforts that would allow students to earn college credit, increase the number of courses, and improve the courses to provide high-level course choices for students. Before OCR concluded its investigation, district school officials voluntarily entered into a resolution agreement with Education, which committed the district to take certain actions, such as implementing programs designed to ensure that equally effective and qualified teachers are equitably distributed throughout the district and ensure Advanced Placement and other higher-level college preparatory courses are taught in the district’s predominantly Black high schools, and provide students the opportunity to engage in-person with course instructors. Education Case 2: Equitable Access to College Preparatory Programs for Black, Hispanic, and English Learner (EL) Students in a New York School District. In 2013, OCR investigated whether a New York school district discriminated against Black, Hispanic, and EL students by establishing and implementing policies and procedures that resulted in their exclusion from college and career ready programs and courses, such as honors courses and AP courses. OCR reviewed information that the district provided regarding its high school honors courses and analyzed data from the district that revealed that Black, Hispanic, and EL students were underrepresented to a statistically significant degree in high school honors courses and AP courses. OCR also reviewed information concerning the district’s gifted and talented program at the elementary and middle school levels and its advanced courses at the middle school level. Data provided by the district indicated that Black, Hispanic and EL students were underrepresented to a statistically significant degree in middle school advanced courses, as well as in some of the district’s enrichment programs. OCR noted that enrollment in these programs and courses could potentially have an effect on later enrollment in high school honors and AP courses. Before the conclusion of OCR’s investigation, the district voluntarily entered into a resolution agreement with Education. The agreement committed the district to take specific actions including hiring a consultant with expertise in addressing the underrepresentation of Black, Hispanic, and EL students in advanced and enrichment courses. According to the agreement, the consultant was to study the underrepresentation and make specific recommendations, as appropriate, for improving the district’s efforts to provide all students with equal access to and an equal opportunity to participate in its advanced courses and programs. Education Case 3: Representation of Black Students in Advanced Courses and Enrichment Programs in a New Jersey School District. In 2014, OCR determined that Black students in a New Jersey school district were underrepresented in high school AP courses. Specifically, OCR found that Black students comprised 51.5 percent of high school students in the district, but only 18.7 percent of students in AP courses in school year 2012-13. In addition, OCR determined that in middle schools, Black students were underrepresented in the district’s advanced math courses, as well as in the math enrichment programs at certain schools in the district. Before the conclusion of OCR’s investigation, the district voluntarily entered into a resolution agreement with Education. The agreement committed the district to take specific actions including hiring a consultant with expertise in addressing the underrepresentation of Black students in college and career preparatory courses. According to the agreement, the consultant was to study the underrepresentation and make recommendations, as appropriate, for improving the district’s efforts to provide all students with equal access to and an equal opportunity to participate in its advanced courses and programs. Education Case 4: College Preparation Opportunities for Black Students in a Virginia School District. In 2014, OCR investigated whether a Virginia school district discriminated against Black students by failing to provide them with the same resources and educational opportunities that it provided to White students to prepare them for postsecondary education or careers. As part of this review, OCR reviewed information regarding the district’s high school higher-level learning opportunities, including advanced courses, AP courses, and dual credit programs (where students enroll in courses at a local community college). In addition, OCR collected and reviewed information about other possible barriers to college and career readiness, including student discipline. OCR found a significant disparity between the numbers of Black and White high school students who take AP, advanced courses, and dual credit programs. Preliminary information provided by the district indicated disproportion in the representation of Black students in advanced math classes, gifted programs, and accelerated reading programs in elementary schools. When speaking with students about what they considered in determining whether to enroll in these courses, many students informed OCR that they took AP or advanced courses if they took advanced courses in middle school and elementary school. OCR also reviewed student discipline, particularly exclusionary disciplinary that removes students from the school setting, because, according to OCR, such removals can serve as a potential barrier to college and career readiness. Before OCR concluded its investigation, the district voluntarily entered into a resolution agreement with Education to resolve the case. The agreement committed the district to retain the services of a consultant with expertise in addressing the underrepresentation of Black students in gifted programs, elementary and middle school advanced courses, and high school AP and dual credit courses. The consultant’s role was to examine the root causes for underrepresentation and to make recommendations about what measures, if any, the district should take as part of its on-going efforts to provide all students with equal access to advanced courses and programs. According to the agreement, the consultant was to study the underrepresentation and make recommendations, as appropriate, for improving the district’s efforts to provide all students with equal access to and an equal opportunity to participate in its advanced courses and programs. Department of Justice Justice also investigates allegations of discrimination related to school resources in response to complaints filed under federal civil rights statutes and monitors and enforces open federal school desegregation orders where Justice is a party to the litigation. Justice sometimes partners with OCR on these cases. In September 2017, Justice officials stated that there were 172 open cases to which the agency was a party. The selected cases described below summarize Justice’s findings and the agreed upon remedies. This selection of cases is not generalizable, and was selected for illustrative purposes only. Justice Case 1: Equal Educational Opportunities in an Alabama School District. As part of an ongoing civil rights lawsuit against an Alabama school district, in 2015, the U.S. District Court for the Northern District of Alabama approved a consent order filed by Justice and the district to reconfigure school attendance zones, improve access to quality course offerings, and address racial discrimination in student discipline, among other areas. The proposed consent order required the district to provide equal educational opportunities to Black students by revising attendance zones and growing and strengthening magnet programs to improve diversity at many of its schools. It also required the district to expand access for Black students by taking a number of steps, including expanding access for Black students to college counseling and advance course offerings such as AP and International Baccalaureate (IB). It also required the district to expand access for Black students to pre- kindergarten, gifted programs, and academic afterschool programs. The district agreed to implement measures to promote faculty and administrator diversity and to ensure that all students are aware of and can equally participate in extracurricular activities. Justice Case 2: Equitable Access to Course Offerings in a Louisiana School District. As part of an ongoing civil rights lawsuit against a Louisiana School Board, in 2015, the U.S. District Court for the Western District of Louisiana approved a consent decree between Justice and the school board. This consent decree addressed district’s fulfillment of its desegregation obligations, terminating long-standing judicial supervision of the district in this matter. Prior to this consent decree, in 2010, the court directed the district to offer the same courses at every high school. However, 5 years later, the court found that a high school in the district, which predominantly served White students, offered 32 more courses, including college preparatory courses, than another high school, which predominantly served Black students. Similarly, across all schools in the district (elementary, middle, and high), the schools that were racially identifiable as White had far more gifted and talented course offerings than other schools. In the consent decree, the district agreed, among other things, to strive to have all courses listed in its course catalog taught at each high school. Further, if a course is ultimately not taught at a given school, students at that school would be given the opportunity to take the course at another school in the district. The district also agreed to provide free transportation, at the student’s request, and to adjust the student’s schedule and the scheduling and location of the course, as necessary, to facilitate the student’s attendance at the course. Justice Case 3: Access to College and Career Readiness Programs and Courses for American Indian Students in a New Mexico School District. In 2017, Justice and OCR resolved a compliance review of a New Mexico school district. The purpose of the review was to determine whether the district discriminated against American Indians by excluding them from college and career readiness programs and courses, such as gifted and talented, AP, and honors courses. Justice and OCR also evaluated whether the district discriminated against American Indian parents by not providing them with information surrounding the aforementioned programs and courses in a language they understand. District staff surveyed during this review recommended ways to address American Indian student underrepresentation in college and career readiness programs and courses. On February 14, 2017, the district entered into a resolution agreement with OCR and Justice, committing to take specific actions to ensure that it is providing an equal opportunity and equal access for all students to its advanced and higher level learning opportunities. The district agreed to several actions including reaching out to an equity assistance center or consultant for technical assistance in addressing the underrepresentation of American Indian students in the college and career readiness programs and courses and improving outreach to the American Indian community. Appendix V: Additional Data Tables This appendix contains several tables that show the underlying data used throughout this report, as well as additional analyses we conducted using the Department of Education’s Civil Rights Data Collection (CRDC) and Common Core of Data (CCD) for school year 2015-16. The following tables and information are included in this appendix: Table 17: High schools offering math and science courses, by school poverty level. Table 18: High schools offering math and science sequences, by school poverty level. Table 19: High schools offering Advanced Placement courses, International Baccalaureate program, and Dual Enrollment options, by school poverty level. Table 20: High schools offering different numbers of Advanced Placement courses, by school poverty level. Table 21: High schools offering math courses, by school size and poverty level. Table 22: High schools offering science courses, by school size and poverty level. Table 23: High schools offering math and science sequences, by school size and poverty level. Table 24: High schools offering Advanced Placement courses, International Baccalaureate program, and Dual Enrollment options, by school size and poverty level. Table 25: High schools offering math courses, by school type and poverty level. Table 26: High schools offering science courses, by school type and poverty level. Table 27: High schools offering math and science sequences, by school type and poverty level. Table 28: High schools offering Advanced Placement courses, International Baccalaureate program, and Dual Enrollment options, by school type and poverty level. Table 29: High schools offering math courses, by school locale and poverty level. Table 30: High schools offering science courses, by school locale and poverty level. Table 31: High schools offering math and science sequences, by school locale and poverty level. Table 32: High schools offering Advanced Placement courses, International Baccalaureate program, and Dual Enrollment options, by school locale and poverty level. Appendix VI: Additional Figures Appendix VII: College Admission Website Review As described in Appendix I, we reviewed websites from a nationally- representative sample of 100 public 4-year colleges in the United States to determine which academic courses colleges expect applicants to take while in high school. Our sampling frame consisted of all public 4-year degree granting colleges that participated in Title IV federal student aid programs, predominately award baccalaureate degrees, have full-time first-time undergraduate students, and that are located in a U.S. state or the District of Columbia, yielding a universe of 555 colleges. Based on our review, an estimated 88 percent of public 4-year colleges posted recommended or required high school coursework as admission criteria for applicants. Of the colleges that had coursework criteria posted on their websites, the results are shown in table 33 below. Appendix VIII: GAO Contacts and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Sherri Doughty (Assistant Director), Cady Panetta (Analyst-in-Charge), James Ashley, James Bennett, David Dornisch, Holly Dye, Alison Grantham, Connor Kincaid, Grant Mallie, Benjamin Sinoff, Walter Vance, and Sonya Vartivarian made key contributions to this report. Also contributing were Deborah Bland, Aaron Karty, Sheila R. McCoy, and Margie Shields.

Why GAO Did This Study Poverty can have a profound effect on academic outcomes and college readiness and students from low-income families are less likely to go to college. The low rates of degree attainment for low-income students raises questions about whether the students who wish to pursue higher education have access to courses that support their readiness for college. GAO was asked to review college preparatory course offerings in U.S. high schools. This report (1) examines the extent to which high schools of different poverty levels offer courses to prepare students academically for college, and (2) describes the challenges students in high-poverty schools face being prepared to attend college. GAO analyzed 2015-16 Education data on course offerings by school poverty level, type, and size, and developed a generalized linear regression model to explore whether certain school-level characteristics may be associated with course offerings; reviewed a generalizable sample of public 4-year college websites for course requirements for admission; and interviewed officials from Education and the Department of Justice. GAO also conducted site visits to 12 high-poverty high schools in 3 states selected to provide variation in course offerings, among other things. In this review, GAO focused on public 4-year colleges because they offer a bachelor's degree and are generally a more affordable 4-year option. What GAO Found Students in relatively poor and small schools had less access to high school courses that help prepare them for college, according to GAO's analysis of Department of Education (Education) data for school year 2015-16 (the most recent available). While most public high schools, regardless of poverty level, offered courses like algebra and biology, disparities in access were associated with school poverty level for more advanced courses like calculus, physics, and those that may allow students to earn college credit, like Advanced Placement (AP) courses (see figure). High-poverty schools were less likely to offer the math and science courses that most public 4-year colleges expect students to take in high school, according to GAO's analysis of college websites. GAO's regression analysis also showed that smaller schools and certain types of schools, like charter schools, are less likely to offer the college preparatory math or science courses that many colleges look for during the admissions process. Officials GAO interviewed in selected high-poverty high schools said their students can face a number of complex challenges in preparing for college. For instance, officials said that many students are academically behind when they enter high school and are unable to progress to more advanced courses. Further, high-poverty schools may not offer rigorous courses, such as AP courses, due to lack of resources or teaching staff. Students in high-poverty schools also face other stressors that can make going to college challenging. Officials at 9 of the 12 schools GAO visited cited the effects of poverty on their students, such as homelessness, hunger, and trauma, that make preparing for college difficult. School officials also said the steps involved in applying to and enrolling in college can be difficult to navigate for many students in high-poverty schools. Officials in selected schools reported efforts to address these challenges, such as offering free college courses and obtaining outside supports to assist with college advising.

gao_GAO-18-322

gao_GAO-18-322_0

Background HHS-OIG Exclusion Authority HHS-OIG has the authority to exclude providers and other entities that have committed certain acts from participation in federal health care programs. According to HHS-OIG guidance, exclusion is a remedial measure designed to protect federal health care programs from any entity whose participation constitutes a risk to the programs or to program beneficiaries. Federal health care programs will not pay for any items or services furnished, ordered, or prescribed by excluded entities. Exclusions are mandatory under certain circumstances and permissive in others. In particular, mandatory exclusion applies to offenses that result in convictions relating to patient abuse or neglect and other crimes related to federal health care programs. When these offenses occur, but there is no criminal conviction, HHS-OIG may exercise its permissive exclusion authority. In certain circumstances where HHS-OIG can exercise its permissive exclusion authority, it evaluates each situation and decides what action to take based on its assessment of the future risk the entity poses to federal health care programs. Actions that HHS-OIG can consider taking include the following: Exclusion. HHS-OIG will exclude the highest-risk entities from participation in federal health care programs. Require the entity to enter into an agreement. HHS-OIG can require an entity to enter into a CIA or IA in exchange for a release of HHS-OIG’s exclusion authority. According to HHS-OIG guidance, the goals of these agreements are to strengthen an entity’s compliance program and promote compliance so that any issues in the future can be prevented or identified, reported and corrected. Heightened scrutiny. According to HHS-OIG officials, heightened scrutiny is reserved for situations in which the agency determined that an agreement was warranted but the entity was uncooperative. In such situations, HHS-OIG considers what other unilateral monitoring steps it can take to impose greater scrutiny. For example, according to HHS-OIG guidance, the agency has audited, evaluated, or investigated entities after fraud settlements when the entity would not enter into an agreement with HHS-OIG and it has made referrals to the Centers for Medicare & Medicaid Services for claims reviews. Reserve exclusion authority. For certain entities, HHS-OIG may reserve its exclusion authority and take “no further action,” meaning that HHS-OIG will not exclude the entity at that time and will not require the entity to enter into an agreement. Release of exclusion authority. In certain circumstances, HHS-OIG will release its exclusion authority without imposing additional requirements. Specifically, HHS-OIG may do this in situations in which the entity has self-disclosed the fraudulent conduct to HHS-OIG or has agreed to integrity obligations with a state or the DOJ that HHS- OIG has determined are sufficient. Agreement Negotiation and Monitoring In situations in which HHS-OIG is evaluating whether to exercise its permissive exclusion authority, DOJ is often separately negotiating a settlement of the civil and/or criminal case against the entity on behalf of the federal government. Typically, such settlements resolve allegations that the entity is liable under the False Claims Act for submitting false claims to federal health care programs. According to both HHS-OIG and DOJ officials, if there is a related DOJ civil or criminal case and HHS-OIG officials are also negotiating an agreement with the entity in lieu of exclusion, the DOJ and HHS-OIG negotiations often occur at the same time or “on a parallel track.” However, according to these officials, while HHS-OIG and DOJ officials share information as needed, each engage in separate negotiations with the entity. According to HHS-OIG officials, there are also situations in which HHS-OIG enters into an agreement when there is not a related DOJ legal settlement. The Office of Counsel to the Inspector General within HHS-OIG is responsible for negotiating agreements and for monitoring them once they take effect. All agreements include provisions that identify the enforcement actions HHS-OIG can take when it finds that an entity has not complied with the terms of its agreement. These enforcement provisions outline the monetary penalties, referred to in the agreements as stipulated penalties, which HHS-OIG will demand if it identifies that the entity has failed to comply with certain agreement terms. The enforcement provisions also outline what constitutes a material breach of the agreement and indicate that exclusion can result if the entity is found to have materially breached its agreement. Examples of a material breach of the agreement include repeated violations of any of the agreement’s obligations and the failure to respond to a demand letter from HHS-OIG concerning the payment of stipulated penalties. From 2005 through 2017, HHS-OIG Entered into Dozens of New Agreements Each Year, the Majority of Which Applied to Three Types of Entities From July 2005 through July 2017, HHS-OIG entered into 652 new agreements—an average of about 50 agreements per year—ranging from a high of 83 to a low of 37. The agreements were almost exclusively CIAs, which apply to larger corporations, and IAs, which apply to individual practitioners and entities such as small physician groups. HHS- OIG has used CIAs and IAs exclusively since 2010. From 2010 to July 2017, 74 percent of agreements have been CIAs and 26 percent of agreements have been IAs. See figure 1 for more information on the number and types of agreements since July 2005. HHS-OIG officials said that the agency transitioned away from other agreement types because of certain limitations that made them less useful than CIAs and IAs. For example, one historical agreement type— Certification of Compliance Agreements—did not provide sufficient opportunities for oversight, yet it required significant resources to create, officials said. Another discontinued agreement type—Settlement Agreement with Integrity Provisions—was negotiated as part of the DOJ settlement, such that HHS-OIG needed to work through DOJ if there was a need to take action for noncompliance. Although HHS-OIG and DOJ negotiate their agreements and settlements separately now, the majority of CIAs and IAs, are still associated with a DOJ legal settlement. Of the 652 agreements from July 2005 through July 2017, 619 were paired with a DOJ settlement, while 33 were the result of HHS-OIG independently exercising its exclusion authority. The total number of agreements in effect each year for the period we reviewed, which includes new agreements and ongoing agreements from past years, has decreased. Between 2006 and 2016 (the earliest and latest full years included in HHS-OIG’s database), the number of agreements in effect for any part of the calendar year decreased by 44 percent (see fig. 2). According to HHS-OIG officials, this is because, over time, the agency has increasingly focused its resources on entities that present the highest risk of potential fraud. Specifically, HHS-OIG officials said that in 2006 they first imposed a monetary threshold for damages caused to federal health care programs, above which the agency would pursue an agreement. HHS-OIG officials told us that they initially set this threshold at $100,000, but that in 2014 the agency increased it to $500,000 for smaller entities (i.e., those eligible for IAs) and $1 million for larger entities (i.e., those eligible for CIAs). HHS-OIG officials added that the monetary threshold is one factor that triggers pursuit of an agreement, and that risk of beneficiary harm may also cause the agency to seek an agreement, even when damages are low. HHS-OIG, in using these criteria, said that it is foregoing pursuing agreements with low-damage, lower-risk entities, instead taking no further action but reserving its exclusion authority. HHS-OIG entered into agreements with a wide range of entities, but most were concentrated among a few types of entities. Specifically, HHS-OIG entered into agreements with 30 different types of entities from July 2005 through July 2017, though slightly more than half of the agreements were with 3 types—individual/small group practices, hospitals, and skilled nursing facilities. Another quarter of the agreements were with medical group practices, pharmaceutical manufacturers, clinics, medical device manufacturers, and ambulance companies. (See fig. 3.) HHS-OIG officials stated that it is rare for the agency to enter into multiple agreements with the same entity, adding that the few entities that have had multiple agreements were generally large corporations with multiple divisions or sites, and that the agreements applied to different areas of the firms’ business. Our analysis of HHS-OIG data showed that 15 entities had more than one agreement from July 2005 through July 2017. In other situations, HHS-OIG extended an ongoing agreement, rather than entering a new agreement with the same entity, in light of new allegations that arose during the time the agreement was in effect. From July 2005 through July 2017, the time periods for five agreements were extended beyond the standard five years to reflect new settlements with DOJ. Almost all of the agreements we reviewed were negotiated by HHS-OIG at the same time DOJ was negotiating a legal settlement with the entity to resolve related allegations under the False Claims Act. Many of these allegations resulted from cases filed by a whistleblower under the False Claims Act’s qui tam provisions—commonly referred to as qui tam cases. Slightly more than half of HHS-OIG agreements are with entities who settled qui tam cases. From July 2005 through July 2017, agreements imposed by HHS-OIG as a result of claims alleged by a whistleblower in a qui tam case increased in prevalence compared to agreements that were not associated with a qui tam case. (See fig. 4.) The DOJ-negotiated settlement amounts associated with qui tam cases, among those entities that also entered into an agreement with HHS-OIG, greatly exceeded the settlement amounts negotiated for non-qui tam cases and make up most of the total settlement amounts. From July 2005 through July 2017, total settlement amounts, among those entities that also entered into an agreement with HHS-OIG, were $16.1 billion for qui tam cases and $3.1 billion for all others. A spike in settlement amounts in 2012 reflects two settlements, one of $2 billion and another of $800 million, with two pharmaceutical manufacturers. (See fig. 5.) Although pharmaceutical manufacturers accounted for about 6 percent of entities subject to an agreement with HHS-OIG from July 2005 through July 2017, they represent a large share of the settlement amounts DOJ negotiated with those entities: $11.8 billion out of $19.2 billion (62 percent). The next largest shares of settlement amounts DOJ negotiated were with hospitals at $2.5 billion and medical device manufacturers at almost $900 million. Most of the pharmaceutical settlements associated with HHS-OIG agreements were qui tam cases (31 of 37 agreements), and a third of all qui tam settlement amounts were associated with just 4 pharmaceutical qui tam cases. HHS-OIG Considers Risk Factors, Such As an Entity’s Conduct, When Evaluating Whether to Exercise its Exclusion Authority and in Negotiating Agreements HHS-OIG Has Criteria to Determine Whether to Pursue Exclusion or Enter into an Agreement; Certain Initial Allegations Were Common among Entities That Entered Into Agreements HHS-OIG guidance includes the criteria that agency officials said they follow to determine whether to exercise the agency’s permissive exclusion authority, or take an alternate action, such as entering into an agreement with an entity. According to HHS-OIG officials and agency guidance, each situation is evaluated on a risk continuum and the course of action chosen is based on the agency’s assessment of the future risk the entity poses to federal health care programs. HHS-OIG has four broad categories of criteria that it applies in deciding where an entity falls on the risk continuum and which action to take. These four categories are (1) the nature and circumstances of the conduct; (2) conduct during the government’s investigation; 3) whether the entity has made efforts to improve its conduct; and 4) the entity’s history of compliance. According to HHS-OIG officials, the agency will exclude the highest-risk entities, and since fiscal year 2011, under its permissive exclusion authority, HHS-OIG has excluded 65 entities that were the subject of a related DOJ legal settlement. However, HHS-OIG guidance states that HHS-OIG often concludes that exclusion is not necessary, so long as the entity will enter into an agreement with the agency. For new agreements from July 2005 through July 2017, our review of HHS-OIG data showed that there were four main types of initial allegations that resulted in the entity entering into an agreement with HHS-OIG. This included: billing for services not rendered - 194 agreements (about 30 percent); provision of medically unnecessary services - 136 agreements (about acts prohibited under the Anti-Kickback statute - 135 agreements (about 21 percent); and misrepresentation of services and/or products – 131 agreements (about 20 percent). The majority of agreements (about 63 percent) were associated with one initial allegation. However, some agreements were associated with more than one initial allegation: about 23 percent of agreements from July 2005 through July 2017 were associated with two initial allegations and about 15 percent were associated with three or more initial allegations. Agreements Included Standard Provisions with Some Variation to Address Risks Specific to the Entities We compared the provisions required in selected agreements to those outlined in HHS-OIG’s current agreement templates and found that the provisions were generally similar. All of HHS-OIG’s templates and the agreements we reviewed were organized into the same broad sections. For example, all of the templates and agreements contained sections detailing the information entities were required to submit to HHS-OIG in an initial implementation report and in annual reports, and all agreements had a section that outlined the enforcement provisions for the agreement. In addition, there generally was a standard term for agreements of either three or five years depending on the type of agreement. All 23 of the CIAs we reviewed had a term of five years, and of the nine IAs we reviewed, five had a 5-year term and four had a 3-year term. The IAs with a longer 5-year term generally were older agreements from 2010 or 2011. According to HHS-OIG officials, the current practice is to negotiate 3-year terms for IAs and 5-year terms for CIAs. HHS-OIG has developed agreement templates that include standard provisions to address the risks an entity’s noncompliance could pose to federal health care programs. Additionally, in some templates, provisions are included to address the specific types of conduct that gave rise to the agreement. HHS-OIG has four templates for use in negotiating CIAs and two for negotiating IAs, and HHS-OIG officials said that they will use one of the six templates as a starting point when drafting an agreement. HHS-OIG officials told us that the terms included in agreements are similar across CIAs and IAs because certain provisions are non- negotiable. For example, officials said that they always include provisions requiring an entity to hire a compliance officer, submit annual reports, and provide HHS-OIG with access to the entity when requested. Across the various types of templates, there are similar standard provisions, and our review of selected agreements found many of the same provisions. For example, among the 32 agreements we reviewed: All 32 agreements required the entity to engage an independent review organization to perform the agreement’s required reviews, including claims reviews. Entities have retained a variety of individuals and businesses as their independent review organization, ranging from small regional consulting firms to large national consulting or accounting firms. For agreements HHS-OIG has entered into from July 2005 through July 2017, our review of the agency’s data found that there were 173 unique associated independent review organizations. All 32 agreements had training and education requirements, although the specifics of the required training, such as the number of hours or the specific topics, varied across agreements. 28 of the 32 agreements reviewed required the entity to have a compliance officer. The four agreements that did not require this were two IAs for small group practices, one for a medical group practice, and one for a clinic that named an individual practitioner as a party to the agreement. Although agreements shared many standard provisions, some provisions were unique to either CIAs or IAs. Many of the CIAs that we reviewed included provisions detailing specific responsibilities for the entity’s board of directors (18 of 23 CIAs) and requirements for certain high-level employees to annually certify that they were in compliance with federal health care program requirements and the provisions of the agreement (12 of 23 CIAs). None of the nine IA’s we reviewed included these provisions. On the other hand, all nine IAs we reviewed (and one CIA) had provisions regarding third-party billing. If the provider subject to the agreement contracted with a third-party billing company to submit claims on the provider’s behalf, these agreements required the provider to certify that they did not have an ownership or controlling interest in the third- party billing company. In addition to agreement type, provisions also varied due to the nature of the conduct that led to the agreement or the type of entity entering into the agreement. For example, some agreements included provisions intended to ensure compliance with the Anti-Kickback statute and Stark law (8 of 32). HHS-OIG officials told us that specific provisions related to the Anti-Kickback statute and Stark law would only be present in agreements when the conduct that had led to the agreement involved acts prohibited under those statutes, such as prohibited kickbacks or improper referral arrangements. Other agreements include provisions specific to monitoring quality of care issues. For example, one of the agreements we reviewed was a quality of care CIA that required the entity to retain an independent monitor to examine, among other things, the entity’s internal quality control systems and its response to quality of care issues. In addition, 2 of the 32 agreements we reviewed were with pharmaceutical manufacturers and contained provisions not in other agreements because they would only be relevant to a pharmaceutical manufacturer. For example, both agreements we reviewed had a requirement that the manufacturers, within 30 days, provide HHS-OIG with a copy of any written communication with the Food and Drug Administration that materially discussed the actual or potential unlawful or improper promotion of the manufacturer’s product. According to HHS-OIG data, most of the 652 agreements entered into from July 2005 through July 2017 (about 95 percent) required the entity to perform at least one review as part of the agreement. The most common types of required reviews captured in HHS-OIG’s database during this time were reviews of health care claims, unallowable costs, and arrangements. Slightly more than half of the agreements (19 of 32) we reviewed required the entity to perform a claims review. Fifteen of these were annual claims reviews and four were quarterly claims reviews. In addition, slightly more than a quarter of agreements we reviewed (9 of 32) required an unallowable costs review. Finally, a quarter of the agreements (8 of 32) required the entity to perform an arrangements review. The eight agreements requiring an arrangements review were the same agreements that included a section with provisions related to compliance with the Anti-Kickback statute and Stark law. A few agreements had required reviews that were not common across the agreements we reviewed and usually related to the types of services that the entity provided. For example, three agreements we reviewed required the entity to conduct a cardiac catheterization procedures review, described as an evaluation and analysis of the medical necessity and appropriateness of interventions performed either in the entity’s cardiac catheterization lab or by the provider. HHS-OIG Uses Multiple Strategies to Ensure Compliance with Terms of Agreements and Imposes Certain Penalties When Noncompliance Is Identified HHS-OIG Reviews Required Reports, Communicates with Entities, and Conducts Site Visits to Monitor Entities’ Compliance with Agreements According to HHS-OIG officials, the agency assigns a monitor to each agreement—an HHS-OIG staff attorney or program analyst—who, for the duration of the agreement, oversees the entity’s compliance with the terms of its agreement. Per officials and what is outlined in internal agency documents that describe how to monitor agreements, the monitors’ responsibilities include: Reviewing the information that entities provide in their initial implementation report, annual reports, and any other reports required under the agreement within the time frames established by internal HHS-OIG guidance. Communicating with entities to provide assistance to those who need help in understanding the requirements or to request additional information when a required report has missing or incomplete information. Reviewing and responding to periodic correspondence received from entities, including notifications required by the agreement, reportable event disclosures, and other communications from the entity. Drafting any letters that are sent to the entity, if noncompliance is identified, including letters demanding the payment of penalties— referred to as stipulated penalty demand letters. Conducting site visits to verify that the entities are complying with the agreements properly. According to internal HHS-OIG guidance, monitors are to select sites primarily based on concerns that they may have with specific entities, as well as other factors, such as the type of provider, the size or complexity of the entity, length of the agreement, and the severity or complexity of the offenses that resulted in the agreement. According to internal HHS-OIG guidance regarding site visit protocol and agency officials we spoke to, during site visits, HHS-OIG officials may conduct document reviews of training records, policies and procedures, or other documents; hold meetings with the compliance officer or board members; or tour the facility, among other activities. Officials said that two HHS-OIG officials typically conduct the site visit—the agreement’s monitor and one other official—and the site visits typically last about one day to a day-and-a-half. For agreements entered into from July 2005 through July 2017, we found that HHS-OIG officials conducted 211 site visits that were associated with 155 agreements. Thirty of these agreements were associated with more than one site visit ranging from 2 to 10 visits. The majority of the 211 site visits were for CIAs (about 87 percent). During the full calendar years from 2006 through 2016, HHS- OIG completed an average of 18 site visits each year. HHS-OIG Works with Entities to Ensure Compliance but May Impose Monetary Penalties or Exclude Entities for Noncompliance in Some Cases Although most entities comply with the provisions of their agreements, according to HHS-OIG officials, when noncompliance occurs, the most common issue is the late submission of required reports or reviews. According to HHS-OIG officials, other types of noncompliance range from falsely certifying the accuracy of reported information to submitting reports that do not include the required elements. According to officials and as outlined in agreements, HHS-OIG addresses noncompliance through a series of escalating steps, which, in rare instances, may result in the HHS-OIG imposing penalties on an entity as laid out in the agreement (stipulated penalties) or even exclusion of an entity from federal health care programs. Steps HHS-OIG takes to address noncompliance may include: Working with the entity before taking official action. For example, officials told us that monitors typically request additional documentation or information from providers when they identify potential issues rather than imposing stipulated penalties immediately. Demanding that the entity pay stipulated penalties. HHS-OIG will send a stipulated penalties demand letter to an entity in accordance with the breach and default terms of the agreement. The stipulated penalty amounts for noncompliance with the different provisions are specified in the agreement. According to officials, the stipulated penalty amounts in agreements are non-negotiable and the amounts associated with noncompliance with specific provisions do not change across agreements. The stipulated penalties in agreements range from $1,000 to $50,000 per violation. For example, for each day an entity fails to submit a complete annual report to HHS-OIG by the submission deadline, the stipulated penalty is $2,500 for CIAs and $1,500 for IAs. In addition, for all agreements, each false certification submitted by or on behalf of the entity results in a stipulated penalty of $50,000. For agreements entered into from July 2005 through July 2017, our review of HHS-OIG data found that HHS-OIG issued 41 letters demanding stipulated penalties (between 0 and 7 letters per year) for actions such as the failure to submit annual reports and employing individuals excluded from participation in federal health care programs. In total, HHS-OIG collected about $5.4 million in stipulated penalties during this time. Penalty amounts demanded in each letter ranged from $1,000 to over $3 million, with a median of $18,000. According to HHS-OIG, the stipulated penalty of over $3 million was a record penalty for failure to comply with an agreement. This penalty, according to HHS-OIG, resulted from the entity’s failure to correct improper billing processes and poor claims submission practices that had led to error rates and overpayments to the company by Medicare for hospice services. These issues were uncovered through the claims reviews required under the agreement and HHS-OIG’s site visits to the company’s facilities. Determining that the entity is in material breach of the agreement. As defined in agreements, this determination may result from repeated or flagrant violations of agreement obligations; failure to notify HHS-OIG of certain reportable events; failure to take corrective actions or make appropriate refunds; failure to respond to a stipulated penalties demand letter; or failure to engage an independent review organization. From July 2005 through July 2017, HHS-OIG issued 10 material breach letters to entities informing them that HHS-OIG intended to exclude them. However, the 10 material breach letters were associated with only 6 agreements, and 5 of the 10 material breach letters issued were to the same entity. These five letters were issued to the entity between March 2012 and January 2014 for a series of alleged material breaches of its agreement including, among other things, a failure to report serious quality of care reportable events or to perform training required under the terms of its agreement. This entity, which was a national chain of clinics that primarily provided dental services to children on Medicaid, was ultimately excluded in September 2014 from participation in federal health care programs, including Medicaid, for a period of five years. Excluding an entity from participation in federal health care programs. For agreements entered into from July 2005 through July 2017, we found that HHS-OIG has issued five exclusion letters to entities for failing to adhere to their agreements. These exclusion letters were associated with four agreements—2 CIAs and 2 IAs. According to HHS-OIG’s data, these exclusions occurred in 2007 (1), 2014 (1), and 2015 (2). The four entities that were excluded included a durable medical equipment provider, a national chain of clinics, a practice management company, and a medical group practice. An agreement affords the entity the opportunity to respond to a material breach letter prior to the issuance of a notice of exclusion. However, an HHS-OIG official said that, given the multiple steps involved in the breach and default process, it is unlikely that a breach would be addressed satisfactorily by the entity at this stage in the process. Of the four entities that HHS-OIG excluded, three had also previously received a notice of material breach from HHS-OIG. Agency Comments We provided a draft of this report to HHS and DOJ for review and comment. These departments provided technical comments, which we incorporated as appropriate. As agreed with your offices, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies to the Secretary of HHS, the Attorney General, and the Inspector General of HHS. In addition, the report will be available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-7114 or kingk@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix I. Appendix I: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Karen Doran (Assistant Director), Alison Goetsch (Analyst-in-Charge), and Perry Parsons made key contributions to this report. Also contributing were Sam Amrhein, Muriel Brown, Dan Ries, Jennifer Rudisill, and Merrile Sing.

Why GAO Did This Study HHS-OIG has the authority to exclude providers and other entities that have committed certain acts, such as submitting false or fraudulent claims, from participation in federal health care programs. However, HHS-OIG can enter into agreements—CIAs and IAs—with providers and other entities as an alternative to exclusion. HHS-OIG is responsible for negotiating such agreements—which it typically does at the same time the Department of Justice (DOJ) is negotiating a legal settlement to resolve related allegations—and then monitoring the entities' compliance with them. GAO was asked to review HHS-OIG's use of these agreements. This report describes (1) the number of agreements and their general characteristics; (2) the circumstances that may lead to an agreement and the standard provisions of agreements; and (3) monitoring efforts and actions taken, if any, in response to noncompliance with the agreements. GAO examined agreements entered into from July 2005 (when HHS-OIG created its database) through July 2017 (most current at the time of GAO's analyses) and used HHS-OIG data to describe agreements' characteristics and actions to address noncompliance. GAO reviewed HHS-OIG documentation, including agreement templates and a selection of agreements to identify standard provisions. GAO also interviewed HHS-OIG and DOJ officials. GAO provided a draft of this report to HHS and DOJ. The agencies provided technical comments, which were incorporated as appropriate. What GAO Found To help improve adherence to federal health care program requirements by entities that have allegedly engaged in certain acts, such as submitting false or fraudulent claims, the Department of Health and Human Services' Office of Inspector General (HHS-OIG) entered into 652 agreements with those entities from July 2005 to July 2017. Since 2010, two types of agreements have been used: Corporate Integrity Agreements (CIA) and Integrity Agreements (IA). The more commonly used CIAs apply to larger entities, compared to IAs, which apply to individual practitioners or small businesses. From July 2005 through July 2017, about half of all agreements were with 3 types of entities—individual or small group practices, hospitals, and skilled nursing facilities. For new agreements since July 2005, the most common initial allegations that led to an entity entering into an agreement included billing for services not provided and providing medically unnecessary services. When negotiating agreements, HHS-OIG uses one of six templates that address the different types of entities or conduct involved. Across agreements the provisions are generally similar—for example, requirements to provide training on specified topics or to hire a compliance officer. HHS-OIG uses multiple strategies to oversee agreements, such as requiring periodic reports from the entities that demonstrate compliance and assigning a monitor to review these reports and conduct site visits. HHS-OIG can also take certain actions to address noncompliance. For example, for new agreements from July 2005 through July 2017, HHS-OIG imposed monetary penalties 41 times, ranging from $1,000 to more than $3 million (median of $18,000), and excluded 4 entities from participation in federal health care programs.

gao_GAO-18-373T

gao_GAO-18-373T_0

Background Part of the Mariana Islands Archipelago, the CNMI is a chain of 14 islands in the western Pacific Ocean, just north of Guam and about 3,200 miles west of Hawaii. The CNMI has a total population of 53,890, according to the CNMI’s 2016 Household, Income, and Expenditures Survey. Almost 90 percent of the population (48,200) resided on the island of Saipan, with an additional 6 percent (3,056) on the island of Tinian and 5 percent (2,635) on the island of Rota. Application of Federal Immigration Laws to the CNMI The Consolidated Natural Resources Act of 2008 amended the U.S.– CNMI covenant to apply federal immigration law to the CNMI after a transition period. To provide for an orderly transition from the CNMI immigration system to the U.S. federal immigration system under the immigration laws of the United States, DHS established the CW program in 2011. Under the program, foreign workers are able to obtain, through their employers, nonimmigrant CW-1 status that allows them to work in the CNMI for up to a year. The Consolidated Natural Resources Act of 2008 requires DHS to annually reduce the number of CW-1 permits until the number reaches zero by the end of the transition period. The act was amended in December 2014 to extend the transition period through December 31, 2019. The act was further amended in August 2017 to, among other things, (1) add 350 CW-1 permits to the fiscal year 2017 cap; (2) restrict future permits for workers in construction and extraction occupations; and (3) increase the CNMI education funding fee that employers must pay for each permit from $150 to $200. DHS determines the annual cap on CW-1 permits and the terms and conditions of the CW program. In November 2017, DHS set the cap for CW-1 permits for fiscal year 2018 through the end of the program (see table 1). Proposed Legislative Changes Affecting the CW Program The proposed bill, the Northern Mariana Islands U.S. Workforce Act (S. 2325), includes the following provisions, among others, that would affect the CW program: (1) the number of permits to be allocated each year, (2) the distribution of the permits, and (3) a new CW-3 worker designation. Number of Permits Allocated Under the terms of S. 2325, the number of permits issued may not exceed 13,000 during fiscal year 2019. Starting in fiscal year 2020, the number of permits issued may not exceed a number that is 500 fewer than the number issued during the immediately preceding fiscal year. Figure 1 shows the past and future numerical limits on CNMI-Only Transitional Worker permits established by DHS and the proposed numerical limits for permits under S. 2325. The limits shown for S. 2325 in figure 1 assume that employers would petition for, and DHS would issue, the maximum number of available permits for fiscal year 2019 and for each subsequent year. Distribution of Permits Under S. 2325, as under the current law, a permit for construction and extraction occupations would be issued only to extend a permit that was first issued before October 1, 2015. Also, S. 2325 would require the Secretary of Homeland Security to consider, in good faith, any comments or advice submitted by the CNMI governor, including any recommendation to reserve a number of permits each year for occupational categories necessary to maintain public health or safety in the commonwealth. CW-3 Permits S. 2325 proposes a new CW-3 worker designation. Foreign workers who are otherwise admissible would be eligible for CW-3 permits if they were admitted to the CNMI as CW-1 workers during fiscal year 2014 and every subsequent fiscal year beginning before the date of the enactment of S. 2325. These workers would receive a permit to remain in the CNMI for a 3-year period beginning on the date of S. 2325’s enactment. CW-3 permits could be renewed in 3-year increments during the transition period for workers who remain outside the United States for a continuous period of not less than 30 days during the 180-day period immediately preceding each such renewal. CW-3 permits would count against the numerical caps specified in S. 2325. Recent Trends in the the CNMI Economy The CNMI’s inflation-adjusted gross domestic product (GDP) has grown each year since 2012. The U.S. Department of Commerce’s Bureau of Economic Analysis (BEA) estimates that the CNMI’s GDP increased by almost 29 percent in 2016 after increasing by about 4 percent in 2015. BEA attributes this economic growth to a significant increase in visitor spending, particularly for casino gambling, and investment in the construction of a casino resort in Saipan and other hotel construction. Figure 2 shows the casino’s development site in Saipan before and during construction. The new casino opened for business on July 6, 2017. CNMI Tourism Trends Since 1990, the CNMI’s tourism market has experienced considerable fluctuation, as evidenced by total annual visitor arrivals (see fig. 3). Visitor arrivals to the CNMI declined from a peak of 726,690 in fiscal year 1997 to a low of 338,106 in fiscal year 2011, or by 53 percent. However, since 2011, visitor arrivals have nearly doubled, reaching 653,150 in fiscal year 2017, and increased by 30 percent from 2016 to 2017. Data from the Marianas Visitors Authority show that the downward trend in Japanese arrivals from 2013 to 2017 was offset by the growth in arrivals from China and South Korea. While eligible Japanese and South Korean visitors enter the CNMI under the U.S. visa waiver program, Chinese visitors are not eligible for the program and are allowed to be temporarily present in the CNMI under DHS’s discretionary parole authority, according to DHS officials. DHS exercises parole authority to allow, on a case-by-case basis, eligible nationals of China to enter the CNMI temporarily as tourists when there is significant public benefit, according to DHS. CNMI Labor Market Following consecutive annual decreases in the total number of employed workers from 2005 through 2013, employment has increased annually since 2014, according to CNMI tax data. Figure 4 shows the numbers of employed workers, both foreign and domestic, in the CNMI from 2001 through 2016. From 2013 to 2016, the number of employed workers increased by approximately 25 percent, from 23,344 to 29,215. As figure 4 shows, while the number and percentage of foreign workers fell between 2001 and 2016, foreign workers still constitute the majority of the CNMI workforce. Of the nearly 30,000 employed workers in the CNMI in 2016, more than half were foreign workers, according to CNMI tax data. The number of foreign workers fell from a peak of over 38,000 in 2002—roughly 75 percent of employed workers—to fewer than 16,000 in 2016. In contrast, since 2002, the number of domestic workers has fluctuated from year to year, ranging from about 10,500 to almost 13,700, but increased by 28 percent from 2013 to 2016. Preliminary Observations on CW- 1 Permits for Fiscal Years 2012-2018 Numbers of CW-1 Permits The CNMI economy continues to experience growing demand for workers. In fiscal years 2012 through 2016, the number of CW-1 permits almost doubled, and since fiscal year 2016, the number of permits has approached the numerical limits on permits for those years. Our preliminary analysis indicates that the number of approved CW-1 permits grew from 7,127 in fiscal year 2012 to more than 13,000 in fiscal year 2016. On October 14, 2016—2 weeks into fiscal year 2017—USCIS announced that it had received enough petitions to reach the CW-1 cap for fiscal year 2017 and would not accept requests for new permits for that year during the remaining 11 months. In May 2017, USCIS announced that it had received a sufficient number of petitions to reach the CW-1 cap for fiscal year 2018. Table 2 shows the CW-1 permit caps and numbers of permits approved for fiscal years 2012 through 2018. According to USCIS officials, as of January 26, 2018, fiscal year 2018 petitions were still being adjudicated. Characteristics of Foreign Workers with CW-1 Permits Our preliminary analysis of USCIS CW-1 permit data for fiscal years 2012 through 2018 identified trends in CW-1 workers’ country of birth, occupation, and duration of employment. Country of Birth USCIS data showed a decline in the numbers of CW-1 permits for fiscal years 2017 through 2018 for workers born in each of the five countries listed most frequently on the petitions—the Philippines, China, South Korea, Bangladesh, and Japan (see table 3). As of January 17, 2018, the number of permits approved for workers born in the Philippines, who received the most permits for all 7 years of the CW program, had declined by 13 percent from fiscal year 2017 to fiscal year 2018 and by 26 percent from fiscal year 2015 to fiscal year 2018. Concurrent with construction of the casino and other tourism infrastructure, the number of CW-1 permits for workers born in China increased by almost 3,800 from fiscal year 2015 to fiscal year 2016 and declined by about 3,500 from fiscal year 2017 to fiscal year 2018. Our preliminary analysis indicates that as of January 17, 2018, USCIS had approved 750 CW-1 permits for construction workers for fiscal year 2018. This number represents a 75 percent decline from the nearly 3,000 permits approved for fiscal year 2017 (see table 4). This decline reflects new restrictions on future permits for workers in construction occupations. Of the 8,228 foreign workers who had been granted fiscal year 2018 permits as of January 17, 2018, 2,352 had maintained continuous employment in the CNMI since fiscal year 2014 (see table 5). Of the 2,352 workers with continuous employment in fiscal years 2014 through 2018, 1,905 workers (81 percent) were born in the Philippines. Chairman Murkowski, Ranking Member Cantwell, and Members of the Committee, this concludes my prepared statement. I would be pleased to respond to any questions you may have at this time. GAO Contacts and Staff Acknowledgments For further information regarding this statement, please contact David Gootnick, Director, International Affairs and Trade at (202) 512-3149 or gootnickd@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this statement. Individuals making key contributions to this testimony included Emil Friberg (Assistant Director), Julia Ann Roberts (Analyst-in-Charge), Sada Aksartova, Andrew Kurtzman, Reid Lowe, Moon Parks, and John Yee. Technical support was provided by Chris Keblitis, Mary Moutsos, and Alexander Welsh. This is a work of the U.S. government and is not subject to copyright protection in the United States. The published product may be reproduced and distributed in its entirety without further permission from GAO. However, because this work may contain copyrighted images or other material, permission from the copyright holder may be necessary if you wish to reproduce this material separately.

Why GAO Did This Study Pub. L. No. 110-229, enacted in 2008, amended the U.S.-CNMI covenant to apply federal immigration law to the CNMI after a transition period. The law required the Department of Homeland Security (DHS) to establish a temporary work permit program for foreign workers. DHS is required to decrease the number of permits issued annually, reducing them to zero by the end of the transition period, scheduled for December 31, 2019. To implement the law, DHS established a new work permit program in 2011. Under the program, foreign workers can obtain, through their employers, nonimmigrant CW-1 status that allows them to work in the CNMI. The law was amended in August 2017 to, among other things, restrict future permits for workers in construction and extraction occupations. Proposed legislation—Senate bill S. 2325—would, among other things, extend the transition period through December 31, 2029; increase the number of available permits from the 2018 level; and set required decreases in the annual numerical limit for the permits. (See figure for past numerical limits established by DHS and future limits proposed by S. 2325.) This testimony discusses (1) recent trends in the CNMI economy and (2) preliminary observations about the number of approved CW-1 permits and characteristics of permit holders, drawn from GAO's ongoing work. GAO updated information about the CNMI's economy that it reported in May 2017 (see GAO-17-437 ). GAO also analyzed data and documents from U.S. agencies and the CNMI government. What GAO Found The Commonwealth of the Northern Mariana Islands' (CNMI) inflation-adjusted gross domestic product (GDP) has grown each year since 2012, according to the Bureau of Economic Analysis. In 2016, the CNMI's GDP rose by 29 percent, partly as a result of construction investment. While tourism has fluctuated in recent years, visitor arrivals in the CNMI rose by nearly a third from 2016 to 2017. After nearly a decade of annual decline, the total number of workers employed in the CNMI increased from 2013 through 2016, according to the most recent available CNMI tax data. Foreign workers made up 53 percent of those employed in 2016, compared with roughly 75 percent in 2002. GAO's preliminary analysis indicates that the number of approved CNMI-Only Transitional Worker (CW-1) permits for foreign workers in the CNMI grew from over 7,100 for fiscal year 2012 to nearly 13,000 for fiscal year 2017. In addition, GAO identified trends in the country of birth, occupation, and employment duration of foreign workers with CW-1 permits approved for fiscal years 2012 through 2018. Workers born in the Philippines received the highest number of CW-1 permits each year. As of January 2018, 750 CW-1 permits had been granted to construction workers for fiscal year 2018—a 75 percent decline from the prior fiscal year. GAO estimated that approximately 2,350 foreign workers with approved CW-1 permits maintained continuous employment in the CNMI from fiscal year 2014 through January 2018. About 80 percent of these workers were born in the Philippines.

gao_GAO-18-71

gao_GAO-18-71_0

Background Spectrum is a natural resource used to provide a variety of communication services to businesses and consumers, as well as federal, state, and local governments. Businesses and consumers use spectrum for a variety of wireless services including mobile voice and data, WiFi- and Bluetooth-enabled devices, broadcast television, radio, and satellite services. Federal, state, and local governments’ uses of spectrum include national defense, law enforcement communication, air- traffic control, weather services, military radar, and first responder communications. IoT applications that rely on spectrum are highly diverse and include connected vehicles, devices in the home, and personal mobile devices. IoT devices communicate using wireless networks, including wide area networks that use cellular networks to cover large areas (e.g., cellular transmission), local area networks that cover about 100 meters (e.g., Wi-Fi within a house), and personal networks covering about 10 meters (e.g., Bluetooth inside a room) (see fig. 1). Each of these wireless devices, like other wireless IoT devices, communicates using spectrum, and the number of connected devices is expected to increase. In 2013, the number of devices connected to the internet globally was estimated to be over 9 billion. In 2015, the Organisation for Economic Cooperation and Development (OECD) estimated that a family of four had an average of 10 devices connected to the Internet in their household, and that this average will increase to 50 devices by 2022. As companies bring new IoT technologies and services to market and government users develop new mission needs, the demand for spectrum will increase. The frequencies, or frequency bands, of spectrum have different characteristics that make them more or less suitable for specific purposes, depending on the specific band (see fig. 2). These bands have different levels of ability to penetrate physical obstacles and cover distances, known as “propagation,” and different limits to the amount of information that they can carry, known as data capacity, and are used for different communication purposes. Low frequency bands are characterized by strong propagation, and are used by numerous IoT devices, some of which may only transmit small amounts of information such as temperature, location, or activity status. The strong propagation of low bands means they can transmit over long distances. Mid-band frequencies have higher data capacity than low bands (because, in part, frequency allocations in higher bands are larger, allowing wider channels), as well as, stronger propagation qualities than higher bands. The bands above 30 GHz have high data capacity but relatively poor propagation, to the point that bands at the highest frequencies can be easily obstructed. This spectrum is currently used by a variety of services, including satellite, fixed microwave, and radio astronomy, and is expected to be important for the next generation wireless technology (5G). FCC is the federal agency responsible for allocating spectrum for various consumer and commercial purposes, assigning spectrum licenses, and making spectrum available for use by unlicensed devices. Licensing assigns frequencies of spectrum, in a specific area, to a specific entity, such as a telecommunications company that operates a network using licensed spectrum. We refer to these bands as licensed spectrum. In some frequency bands, FCC authorizes unlicensed use of spectrum bands—generally referred to as unlicensed spectrum—that is, users do not need to obtain a license to use spectrum. Rather, users of unlicensed devices can share frequencies on a non-interference basis, such as with home wireless networks, cordless phones, and garage door openers. In addition, FCC supports federal emergency-communications activities. NTIA is responsible for establishing policy on regulating federal spectrum use, assigning spectrum bands to government agencies, and maintaining spectrum use databases. Additionally, like FCC, NTIA participates in federal emergency communications activities. NTIA also determines what spectrum bands reserved for the federal government can be made available for commercial use. In managing spectrum, one factor that FCC and NTIA consider is the potential for interference. Harmful interference occurs when two communication signals are either at the same frequencies or close to the same frequencies in the same vicinity, a situation that may lead to degradation of a device’s operation or service. Co-channel interference occurs when two communications systems operate on the same frequency assignment in the same vicinity. Adjacent band interference occurs between two communication systems operating on different, but adjacent frequencies in the same geographic area. Another source of interference can be signals on adjacent spectrum bands leaking into another band. FCC and NTIA work to make more efficient use of spectrum that has been assigned. One means of more efficiently using spectrum is to share it, between and among both federal users and commercial users. In 2017, FCC and NTIA continued oversight of the development of a new- spectrum sharing mechanism called the Spectrum Access System (SAS) in the 3.5 GHz band. Among other things, the SAS allows multiple users access to the same band at different times or places. Within this spectrum band the SAS establishes a three-tiered system of access priority, with federal and non-federal incumbent users having first priority, new non- federal users who have paid for licensed access as second priority, and other users as third priority. This system relies on the SAS to assign frequencies by determining if a frequency is in use by a higher priority user before assigning it to a lower priority user. Selected Stakeholders Identified Spectrum Availability and Managing Interference as Challenges Affecting IoT Devices Stakeholders representing IoT network providers, device manufacturers, users, and federal regulators consistently identified two spectrum-related challenges to the continued growth and development of IoT 1) ensuring the availability of sufficient spectrum and 2) managing the harmful interference from the increasing number of IoT devices. Spectrum Availability While not currently a crisis, the stakeholders we spoke to agreed that ensuring the availability of sufficient amounts and the right kinds of spectrum is a key challenge for supporting the growth of IoT. Specifically, stakeholders cited three dimensions of the spectrum availability challenge: the amount, the balance between licensed and unlicensed, and the variety of spectrum bands available. According to some reports, incorrectly anticipating industry needs in any of these areas could weaken IoT growth and development in the United States. Amount of spectrum: The amount of spectrum needed for IoT devices is expected to increase with their growth. According to a majority of stakeholders we interviewed, FCC will need to continue to make additional spectrum commercially available in order to meet the demand from expected rapid growth in wireless devices, including IoT devices. FCC officials told us the current amount of available spectrum will be sufficient for the growth of IoT unless devices that use a disproportionally large amount of spectrum become more prevalent. Such devices, like those that stream video, could lead to a spectrum shortage that negatively impacts IoT growth. According to several stakeholders spectrum availability will become an issue as use of these devices increases. FCC officials said that cellular providers experienced similar issues when they introduced smart phones, spurring rapid, exponential growth in consumer demand to send and receive wireless data. Despite the potential for a shortage of spectrum for IoT devices, most of the stakeholders agreed that there should not be specific spectrum set aside for IoT devices; rather, some noted spectrum policies should remain flexible, allowing licensees to determine the best use. Licensed and unlicensed spectrum: A majority of stakeholders said that the spectrum availability challenge includes making both licensed and unlicensed spectrum available. According to FCC staff, FCC is responsible for ensuring sufficient spectrum exists for commercial purposes and will continue to identify new spectrum that can be used for a variety of uses, including by IoT and other wireless devices. This identification of new spectrum includes making spectrum available on both a licensed and unlicensed basis to meet the needs of IoT and other wireless devices. For example, some devices may need to send a signal over a long distance and with a high quality of service to ensure a signal will go through, such as a fire alarm, something licensed spectrum can provide. However, for other devices, cost is a more important consideration. Licensed spectrum has costs that can come from purchasing the license or accessing the spectrum. For example, an official from a supply-chain automation company that develops radio- frequency identification tags told us the lack of inexpensive, low power networks that provide broad coverage is a challenge for their business. With such a network, the company’s tags could send out small amounts of data at intervals to help manufacturers track their goods. However, the cost of such a service is important if these tags are to attach to all size packages because paying for GPS or a wireless connections for each would make it unfeasible. According to several stakeholders, the correct balance between licensed and unlicensed spectrum is difficult to know. Spectrum bands: Several stakeholders indicated that the need to make various spectrum bands available for IoT devices contributes to the spectrum management challenge. As previously described, each band of spectrum has different characteristics, such as the ability to carry data long distances and penetrate obstacles. IoT devices have diverse spectrum needs, such as needing to send a signal over a distance or send a constant stream of information. For example, in the package delivery industry there could be IoT devices, sending signals over a distance, that read the location of the vehicle and direct the driver on a different route based on traffic and deliveries. In addition, there are IoT devices that can monitor containers being delivered including their location, temperature within the container, and other characteristics. In both these examples, the devices can send signals over long distances to systems that can monitor the information. Spectrum Interference Some stakeholders and FCC staff also agreed that managing interference caused by the increasing number of IoT devices will challenge the continued growth of IoT. As previously stated, interference occurs when signals in the same vicinity attempt to access the same spectrum bands or bands close to each other, causing the signals to degrade. This can lead to intermittent access, poor reception, or no reception. As the number of wireless IoT devices grows, the chances of harmful interference increases. The number of IoT devices is predicted to grow so fast the instances of harmful interference could be difficult to track. Furthermore, according to one stakeholder, with devices being made by more manufacturers, not all devices are created of equal quality, potentially further increasing the chance that such devices will cause interference. A recently issued GAO report found that according to FCC staff, the expansion in wireless services and devices, not just IoT, has contributed to interference becoming more of a challenge for FCC. FCC staff agreed that managing interference is becoming more challenging as the number of wireless IoT devices grows. However, according to FCC staff, relatively few complaints pertaining to licensed services involve devices that are compliant with FCC regulations and operating properly. Managing interference may be particularly difficult in homes where many devices rely on unlicensed spectrum. The FCC Technical Advisory Council’s (TAC) report from 2014, expressed concerns that the rapid growth of IoT could exacerbate interference issues in the home. Particularly, the growing reliance on unlicensed spectrum for many consumer IoT devices has contributed to this concern. For example, many IoT devices using unlicensed spectrum, such as digital assistants or wireless speaker systems, use Wi-Fi, Bluetooth or similar technology to transmit a short distance to a smart phone or Wi-Fi router. Not all agree however, that this use is an issue. One spectrum expert we interviewed for a recently-issued report said that interference among consumer devices is less likely to be an issue because they only transmit for short durations and over short distances. If the devices only transmit a short distance then many devices can transmit on the same spectrum. Similarly, if devices only transmit for short durations then they can take turns transmitting over the same spectrum. The Federal Government Has Plans to Meet Spectrum Needs but Does Not Track IoT Devices That Could Cause Congestion FCC Spectrum Planning To plan for spectrum needs, FCC has repurposed spectrum by making additional spectrum available for commercial purposes and, according to FCC officials, the agency is continuing to look for additional opportunities to do so. For example, in 2016, FCC issued a final order that opened up high-band spectrum (above 24 GHz) for use with 5G networks and applications. This particular rulemaking from FCC opened up a total of 10.85 GHz of spectrum, 3.85 GHz for licensed mobile use and 7 GHz for unlicensed use. According to FCC, this order follows a technology neutral approach to planning by allowing spectrum users to develop technologies for the spectrum and not have FCC dictate its specific use. Advances in technology that now allow use of spectrum above 24 GHz for high-speed mobile services led the FCC to initiate the proceeding resulting in this order. Previously, this spectrum was best suited for various satellite or fixed microwave applications. As shown in table 1, in recent years FCC has freed up spectrum for licensed use, unlicensed use, and sharing between the two. In 2016, FCC issued a proposed rule to allow mobile uses in an additional 17.7 GHz of spectrum. In 2017, the FCC issued a Notice of Inquiry seeking input on potential opportunities for additional flexibility, particularly for wireless broadband services, in spectrum bands between 3.7 and 24 GHz. However, according to FCC staff, the process of identifying and freeing up new spectrum can take a significant amount of time as FCC must complete a rulemaking and either relocate existing users or define sharing arrangements between the existing users and new users. FCC has also proposed sharing mechanisms it hopes will allow some bands to be used by existing users as well as additional uses in the future. Other efforts to make additional spectrum commercially available have included examining the potential for sharing the 5.9 GHz band that FCC designated for transportation safety. This band was allocated over 15 years ago and designated exclusively for safety communication between vehicles and between vehicles and infrastructure. In recent years, FCC has worked with the automobile industry and Department of Transportation to assess whether all or a portion of that spectrum could be shared. FCC is also monitoring development of specifications to support 5G—the next generation of wireless networks. According to FCC, the 5G technologies that providers develop are projected to bring wireless networks lower latency, better coverage, faster Internet connections, and allow for more connections than the existing cellular network, all of which may enable more IoT devices to be connected. However, 5G technology is still being developed, and while specifications are not fully defined, according to the plans from the standards-making bodies there will be particular standards designed to support IoT communications. NTIA’s Data Gathering and Research In 2016, NTIA issued a report on the potential roles of the federal government in support of the growth of IoT. It addressed specific questions regarding the spectrum needs and potential interference related to IoT devices and reaffirmed the government’s role in supporting technology growth. Furthermore, the report identified ongoing initiatives that support IoT as well as proposed future steps the Department of Commerce can take to further support IoT development. For example, NTIA’s report proposed that it continue to analyze the usage and growth of IoT devices through its survey used to collect its Digital Nation data. Recent Digital Nation surveys have asked about wearable devices, use of smart televisions, and use of Internet-enabled mobile phones, all uses that include IoT applications. The most recent survey, in 2015, also asked Internet users whether they interact with household equipment or appliances via the Internet. NTIA officials recently told us that they will continue to monitor these connected items to track trends in their use but do not intend to expand the survey to include questions about additional IoT devices. Specifically, in January 2017, NTIA sought out public comment on its November 2017 Digital Nation survey including comment on a proposed questionnaire. NTIA subsequently submitted its proposed questionnaire to Office of Management and Budget for final approval. NTIA also has ongoing spectrum studies through its Institute for Telecommunications Sciences and the findings may apply to IoT’s use of spectrum. As shown in table 2, these studies touch on a number of areas related to IoT including interference issues and spectrum use. NTIA also co-chairs the Wireless Spectrum Research and Development Interagency Working Group that coordinates spectrum-related research and development activities across the federal government, academia, and the private sector. Among other activities, this working group has developed the Wireless Spectrum Research and Development Inventory that, in its 2016 iteration, provides information on completed projects or those scheduled to be completed between January 1, 2015 and December 31, 2018. Tracking IoT’s Growth FCC has a strategic goal of promoting economic growth, and one way FCC pursues that goal is by ensuring that there is sufficient spectrum to support commercial demand. Most stakeholders agree that the growth in mobile IoT devices will eventually require additional spectrum to operate effectively. According to some stakeholders we interviewed and reports we reviewed, rapid, unexpected growth in two areas could lead to congestion and interference that could slow the growth of IoT in the United States: (1) high-bandwidth devices and (2) devices that operate in unlicensed bands. Federal standards for internal control instruct agencies to address risks such as these by estimating the significance of the risk, analyzing the likelihood of it occurring, and assessing its nature. Such assessments can be used to determine how to respond to the potential risks that could prevent agencies from meeting their goals. Rapid growth in high-bandwidth and unlicensed spectrum devices represent risks to FCC achieving its goal of promoting economic growth by ensuring that sufficient spectrum is available. FCC officials said that the agency tracks industry-produced trends and projections related to spectrum demand and use but does not focus on specific devices. Rather, it relies on network providers to manage and track the spectrum related to specific device types. When more spectrum is needed, FCC officials said that FCC identifies additional spectrum and makes it available to the commercial sector. However, this reactive approach may not adequately address the risks caused by high- bandwidth and unlicensed-spectrum devices. High-bandwidth devices: Some stakeholders we interviewed and FCC officials said that rapid increases in high-bandwidth IoT devices could overwhelm current wireless networks. Such IoT devices could include video-streaming devices or unmanned drones, which have much higher data needs and will require a lot of bandwidth. FCC officials said that the supply of spectrum has not always kept pace with demand caused by rapid increases in high-bandwidth devices. For example, the officials said that wireless networks were overwhelmed when providers introduced smart phones. Until then, ringtones represented the bulk of demand for wireless data, but mobile Internet browsing caused the demand for wireless data to increase several fold. In 2014, the FCC TAC warned that new IoT applications could overwhelm networks the same way smartphones and other new technologies have in the past. The TAC recommended that FCC monitor IoT wireless networks with a specific focus on high-bandwidth devices. Unlicensed spectrum use: Some stakeholders also said that unlicensed bands are particularly vulnerable to congestion and potential interference because of expected growth in IoT devices. For example, all the commercial, industrial, and personal devices that connect using WiFi and Bluetooth networks use unlicensed spectrum. In 2014, the TAC indicated that the majority of wireless IoT devices will rely on unlicensed spectrum and recommended FCC make sufficient unlicensed spectrum available for devices operating on local and personal area networks, like WiFi and Bluetooth. However, FCC may not have enough information to determine when the amount of unlicensed spectrum is sufficient. While network providers can manage the number of devices on their own licensed networks, this approach does not work for devices that use unlicensed spectrum, and FCC does not track unlicensed spectrum utilization. It does not track use of unlicensed spectrum because congestion of unlicensed spectrum is geographically and technically challenging to track. Specifically, it is geographically challenging because network congestion and demand can vary over very short distances and technically challenging because there are so many bands of spectrum that would have to be tracked at one time and unlicensed spectrum typically propagates over relatively short distances. However, there may be ways to track unlicensed use that does not require monitoring. For example, NTIA’s Digital Nation survey provides information on select IoT devices using unlicensed spectrum that could help track unlicensed spectrum use. While FCC makes additional spectrum available when needed, it lacks an early warning system for high-risk sectors, like high-bandwidth and unlicensed-spectrum devices. The process of identifying and reallocating spectrum is a lengthy process that can take years, including the need to identify new bands, address the needs of existing users on the bands, establish service rules, and license or assign the spectrum for commercial uses. Without tracking the high-bandwidth and unlicensed-spectrum devices, FCC is not assessing a key risk associated with its goal of promoting economic growth. Rapid, unexpected growth in these IoT sectors could lead to spectrum congestion and interference that could slow or halt the economic growth associated with IoT until FCC can make additional spectrum available. Selected Leading Countries Vary in Spectrum-Planning Approaches for IoT Approaches to Spectrum Planning in Selected Leading Countries Like the United States, France, Germany, the Netherlands, and South Korea are among the world leaders in the development of IoT. We contacted public and private officials in these countries to identify their approaches to spectrum planning to address the growth of IoT. Those officials described approaches to planning for future spectrum needs that are similar to the United States in one area but different in others (see table 3). Specifically, we found that all four countries practice technology neutral spectrum planning, an approach that was broadly supported by the stakeholders we interviewed, including wireless carriers, a technology manufacturer, academics, and a nonprofit group. Some of these stakeholders indicated that this approach to spectrum planning encourages innovation as it allows developers to choose the most appropriate spectrum bands for new technology without having to take the extra step of getting regulators’ permission for each new device or application. Two of the selected leading countries, Germany and South Korea, have developed national IoT plans focused on developing IoT for industry; however, only South Korea has a plan that specifically addresses spectrum issues. South Korea’s national IoT plan seeks to increase collaboration among IoT stakeholders, promote innovation, and develop services for the global market in order to promote productivity and efficiency in Korean business. South Korea also developed a mid- to long-term spectrum plan to respond to the expected growth in demand for spectrum as IoT expands and 5G cellular networks are deployed. Released in 2016, the plan intends to makes more spectrum available to support new services such as smart homes, smart factories, smart cities, remote medical treatment, and unmanned vehicles. Specifically, the South Korean spectrum plan that includes IoT and establishes the following goals: almost doubling the amount of available spectrum available, expanding from 44 GHz of available spectrum to 84 GHz by 2026, and increasing the efficiency of spectrum use, promoting spectrum sharing, and advancing international coordination in spectrum planning. Officials from France and the Netherlands told us that making more unlicensed spectrum available is a high priority in their spectrum planning. These officials told us that unlicensed spectrum promotes greater innovation by lowering barriers to access, and many IoT devices are expected to be designed to operate on unlicensed bands. German and Dutch officials told us that numerous smart city IoT applications have been developed in their respective countries, most of which operate on unlicensed spectrum. For example, German and Dutch networks use unlicensed spectrum for purposes that include managing street lighting, preventing the theft of property such as bicycles, monitoring parking spaces, and managing agricultural resources. To provide service options for low power IoT devices, private companies in France, the Netherlands, and South Korea developed nationwide low- power wide-area networks (LPWAN) which use unlicensed spectrum to transmit data. These LPWANs use the 800 and 900 MHz bands to transmit data from wireless IoT devices such as sensors and location trackers. Signals in these bands can be transmitted over long distances and can penetrate obstacles. According to one LPWAN provider, the distance served by a LPWAN site is greater than a single cellular network site. However, according to the same LPWAN operator, the bands used for LPWAN networks have limited data capacity compared to those used by cellular networks. According to officials and telecommunications industry stakeholders in these countries, LPWANs offer several potential benefits including low barriers to entry, low costs, and broad coverage. According to a Dutch telecommunications industry stakeholder most devices that use LPWANs transmit only small amounts of data. A telecommunications industry stakeholder in France told us that the long range and strong propagation of these LPWANs make them useful for utility metering data and South Korean official told us that LPWANs are used to transmit location or temperature data. For example, in the Netherlands, LPWANs are used to monitor water depth and quality, manage street lighting, and to track the location of business inventory and personal property. In France, LPWANs are used for similar tracking as well as smoke detectors. Other uses for the LPWANs are currently in development. For example, a representative of a Dutch telecommunications company told us that in the Netherlands, IoT devices operating on the nationwide LPWAN are being tested at an airport for use in logistical processes such as baggage handling. Additionally, a Dutch railway station is experimenting with IoT technology that monitors rail switches using the LPWAN, and depth sounders at the port of Rotterdam have been fitted with devices to connect them to the network. South Korean officials said that the LPWAN in their country also provides specialized location-tracking services. Selected Leading Countries Face International Coordination and Potential Spectrum Congestion Challenges Selected leading countries take many similar approaches to each other and the United States to managing spectrum in order to address related challenges (see table 4). Like the United States, spectrum-planning officials in France, Germany, and the Netherlands told us that it was necessary to coordinate spectrum planning with other countries on their borders. Officials in each of these countries told us that European spectrum planning is complicated by the number of countries that share borders. Germany, for example, borders nine other countries. As each country is responsible for its own spectrum planning, if their plans are not closely coordinated, there is a potential for cross-border interference. This coordination is complicated by the fact that European countries have legacy spectrum allocations, and these must be accommodated in spectrum planning. The United States, by contrast, shares its border with only Mexico and Canada. According to FCC officials, both of these countries generally align their spectrum plans to those of the United States, reducing interference issues. In order to facilitate international coordination of spectrum planning, each of the four selected leading countries, like the United States, belongs to a regional spectrum-planning association that works to harmonize spectrum planning among member states. Officials of regional groups we spoke with told us that harmonizing can reduce interference issues across borders and facilitate interoperability of devices across different countries. Officials from the manufacturing and telecommunications industries told us that this interoperability creates a larger potential market for IoT devices, thereby improving the economies of scale for the manufacture of IoT devices and reducing production costs. Regional planning associations are also taking steps to prepare their member countries for the spectrum needs of IoT. For example, an official of one association, the Inter-American Telecommunication Commission, told us that in 2016 it held a workshop on “machine-to-machine” technologies that brought together spectrum planners and stakeholders from IoT-related industries. Regional-planning associations also represent their member countries at World Radiocommunications Conferences (WRC). An official from one association told us that due to the diverse nature of IoT devices and applications it is unnecessary for IoT to be explicitly addressed as an agenda item at WRCs. However, the official further stated that the spectrum needs of specific IoT applications— including low power sensors, robotics, and connected vehicles—are included on the agenda. For example, the next WRC is scheduled for 2019 and includes an agenda item addressing connected vehicles, which are closely linked to IoT. Spectrum-planning officials in each of the selected leading countries told us they are concerned about the potential for spectrum congestion, due to growth in the number of IoT devices. However, like FCC in the United States, these officials do not currently believe such congestion presents an immediate problem. Representatives of the four countries we spoke with told us that one way that they address the potential challenge of spectrum congestion is through the use of spectrum-sharing arrangements. Representatives from Germany specifically stressed the importance of finding additional sharing arrangements in response to the expected spectrum needs for IoT. In 2016, both France and the Netherlands initiated pilot programs for spectrum sharing in which multiple users’ access the same bands while prioritizing use by the licensee. These pilot programs are similar to the dynamic-sharing model that FCC adopted in 2015, as described previously. However, whereas the model adopted by FCC has three tiers of users, the model used by France and the Netherlands has only two, and lacks the third tier of general access users. Unlike the United States, officials from Germany and France told us that they directly monitor spectrum congestion. For example, German officials told us that there are spectrum-monitoring services at six locations around the country, and that they perform mobile measurements of spectrum congestion. FCC officials told us that their primary means of tracking congestion is to communicate with spectrum licensees. According to officials from the Netherlands, the Dutch spectrum management agency takes a similar approach and has struck an agreement with a group of telecommunications companies to share information concerning IoT’s interference and congestion issues. Officials also told us that it is easier to monitor spectrum congestion in smaller countries, as there is simply less geographical space to monitor. Nevertheless, officials in France, Germany, and the Netherlands told us that monitoring spectrum is a challenging task, as it is difficult to determine how many wireless devices are active at any given time. Conclusions FCC has a strategic goal to promote economic growth and effective spectrum management represents a key way that FCC can support meeting that goal. To that end, FCC officials said that the agency continuously seeks to make additional spectrum available and broadly tracks spectrum demand. However, stakeholders and FCC’s own technical advisors have identified rapid, unexpected growth in both high- bandwidth devices and unlicensed spectrum as risks to effective spectrum management. By overwhelming existing networks before FCC can make more spectrum available, rapid growth in spectrum demand could slow or halt IoT’s potential to facilitate economic growth. Absent additional efforts to assess the risks to effective spectrum management by focusing on high-bandwidth and unlicensed-spectrum devices, spectrum congestion and interference could slow IoT growth. Recommendations We are making the following two recommendations to the Chairman of FCC. The Chairman of FCC should track the growth in high bandwidth IoT devices, such as video-streaming devices and optical sensors. (Recommendation 1) The Chairman of FCC should track the growth in IoT devices relying on unlicensed spectrum. (Recommendation 2) Agency Comments and Our Evaluation We provided a draft of this report to FCC and the Department of Commerce for their review and comment. FCC provided comments in a letter, which is reprinted in appendix IV. FCC and the Department of Commerce provided technical comments that we incorporated as appropriate. In its written comments, FCC did not concur with our recommendation that it track growth in high-bandwidth devices. FCC noted that it continues to believe that the best approach to track growth of devices is by monitoring overall traffic statistics and forecasts and how these devices affect aggregate spectrum requirements for all applications and services. However, FCC noted that it would task the Technological Advisory Council (TAC) to periodically review the state of the IoT ecosystem to ensure that the planned communications infrastructure is sufficient to support the needs of the growing sector and advise on any actions the FCC should take. We continue to believe that tracking the growth of high- bandwidth devices is necessary to avoid the potential spectrum shortage and that the TAC may be able to help FCC accomplish that. FCC did not concur with our recommendation to track IoT devices that rely on unlicensed spectrum. FCC noted that it may not be practical to determine which devices qualify as IoT or quantify their effect on spectrum utilization. As a result, FCC said that the best way to monitor growth in unlicensed IoT devices is to continue to monitor published papers and conferences and work with industry. However, since most of the projected IoT growth is expected to occur in unlicensed bands that are not protected from interference, we continue to believe that FCC should place a greater focus on tracking IoT devices in these bands. For example, the TAC may also be well positioned to help FCC track unlicensed IoT devices. As agreed with your offices, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies to the Secretaries of Homeland Security and Commerce, the Chairman of FCC, and appropriate congressional committees. In addition, the report will be available at no charge on the GAO website at http://www.gao.gov. If you or members of your staff have any questions about this report, please contact me at (202) 512-2834 or goldsteinm@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. Major contributors to this report are listed in appendix IV. Appendix I: Objectives, Scope, and Methodology We were asked to examine the challenges facing federal spectrum managers and the steps they are taking to address those challenges. In this report we discuss: (1) the spectrum-related challenges stakeholders identified due to the anticipated growth of IoT, (2) steps FCC and NTIA are taking to plan for the anticipated growth in the demand for spectrum as a result of IoT, and (3) efforts that selected leading countries are undertaking to plan for IoT’s spectrum needs and ways that these efforts compare with those of the United States. To identify the spectrum-related challenges stemming from the expected growth of IoT, we reviewed documents from the Federal Communications Commission (FCC) and the National Telecommunications and Information Administration (NTIA), the two federal agencies that have direct authority over spectrum planning. Further, in order to identify relevant literature for review, we (1) conducted a key word search of data bases; (2) searched IoT and spectrum related websites, such as those of cellular carriers, telecommunications industry groups, and nonprofit organizations; (3) reviewed prior GAO reports on IoT and spectrum issues; and (4) asked FCC and NTIA officials, researchers, and non-profit organizations to identify relevant documents. Through our literature search, we identified a number of documents, including academic reports, government reports, congressional committee hearings, and trade journals addressing the projected growth of IoT to understand the number of devices that would be relying on the spectrum in the coming years. We also reviewed literature concerning the growth of other wireless devices, such as smart phones, and the burden they place on the spectrum, to assess if there are any lessons learned from the demand these devices placed on the spectrum that could be applied to the expected growth of IoT. In addition, we interviewed FCC and NTIA officials, and conducted 24 telephone and in-person interviews with officials from industry associations, industrial and commercial users of IoT, nonprofit groups, subject matter experts, manufacturers, and telecommunications companies to obtain their perspectives on the challenges presented by the expected growth of IoT. The experiences of the stakeholders are not generalizable to those of all IoT stakeholders in the United States; however, we believe that the information we gathered from selected stakeholders provides a balanced and informed perspective on the topics discussed. We identified relevant stakeholders by reviewing comments submitted to NTIA in response to its request for comment on the government’s role in planning for IoT growth, reviewing congressional hearings, and conducting a literature review encompassing academic articles, government reports, and trade journals. We interviewed officials from businesses that manufacture Internet-connected devices or equipment that would be considered part of IoT, including agriculture, telecommunications, and manufacturing. We spoke with these officials to gather information about the spectrum challenges they face as businesses working with and developing IoT devices. We then analyzed the results of these interviews and related documents to identify the main themes and develop summary findings. To characterize the views captured during the interviews, we defined the terms to quantify the views as follows: “most” users represents 18 to 24 users, “a majority of” users represents 11 to 17 users, “several” users represents 6 to 10 users, and “some” users represents 3 to 5 users.” To identify the steps FCC and NTIA are taking to plan for the anticipated growth in the demand for spectrum as a result of IoT, we interviewed FCC and NTIA officials and reviewed agency reports and documents. We interviewed officials to understand any agency plans to address spectrum needs for IoT devices and how these plans aligned with the spectrum planning for other wireless devices. We reviewed agency reports and documents on spectrum planning, IoT planning, and the role of the federal government in planning for IoT. Specifically, we reviewed comments submitted in response to NTIA’s request for comment and the final report developed in response to the comments received on the role of the federal government. To identify other relevant reports and literature from FCC and NTIA, we asked officials at the meetings and conducted a literature search. We also compared those planning efforts against FCC’s and NTIA’s strategic goals and the federal internal control standards related to risk management. Specifically, we compared FCC’s planning against its strategic goal to promote economic growth and national leadership in telecommunications, and NTIA’s efforts against its mission to expand the use of spectrum by all users and to ensure that the Internet remains an engine for continued innovation and economic growth. We also assessed the efforts of both agencies against leading practices that we previously developed for identifying, analyzing, and responding to risks related to achieving agency objectives. To identify the efforts that selected foreign governments are taking to plan for the expected spectrum needs of IoT and ways their efforts compare with those of the United States, we surveyed trade journals, industry publications, and foreign governments’ websites and publications. Through this survey, we identified seven countries of potential interest, all of which have conducted spectrum planning in support of IoT: China, France, Germany, Netherlands, Japan, Singapore, and South Korea. We selected four of these countries—France, Germany, the Netherlands, and South Korea—as being like the United States and leaders in IoT development based on additional criteria including the level of their economic development, the maturity of their telecommunications infrastructures, the comparability of their governments to the United States, and the accessibility of their spectrum-planning information. We categorized a country’s economy as fully developed if the United Nations Statistics Division categorized it in 2016 as existing in a developed economic region. When determining the maturity of a country’s telecommunications infrastructure, we followed the United Nation’s International Telecommunication Union (ITU) in categorizing a country’s telecom infrastructure as mature if it was included in the top quartile of the 175 countries ranked in ITU’s 2016 Information and Communications Technology Development Index. We considered a country to have a government structure comparable to that of the United States if Freedom House’s 2016 Freedom in the World report rated it as “free” and the Polity Project categorized it as a “democracy” in 2015. Finally, we considered the extent to which information could be efficiently procured from each country under consideration. We reviewed documents and conducted telephone and written interviews with officials from the spectrum management agencies in each of these four countries. We also conducted eight telephone and written interviews with officials from foreign telecommunications companies, IoT manufactures, and international spectrum-planning groups to gather information about IoT development, challenges, and responses to these challenges in the leading countries that we contacted. We conducted this performance audit from August 2016 to November 2017 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Agencies, Organizations, and Individuals GAO Interviewed Barcoding Case IH Deere & Co. O-I Consumer Technology Association CTIA National Association of Manufacturers Telecommunications Industry Association U.S. Chamber of Commerce Wi-Fi Alliance World Shipping Council New America Foundation Public Knowledge Technology and Innovation Foundation Jeffrey Reed, Ph.D. (Virginia Polytechnic Institute and State University) Douglas Sicker, Ph.D. (Carnegie Mellon University) AT&T Sigfox Verizon Agence Nationale des Fréquences (France) Agentschap Telecom (Netherlands) Bundesnetzagentur (Germany) Ministry of Science, ICT, and Future Planning (South Korea) Appendix III: Regional Spectrum- Management Associations and Their Member States Appendix IV: Comments from the Federal Communications Commission Appendix V: GAO Contact and Staff Acknowledgment GAO Contact Staff Acknowledgments In addition to the individual named above, Keith Cunningham (Assistant Director); Eric Hudson (Analyst-in-Charge); Camilo Flores; Adam Gomez; Josh Ormond; Andrew Stavisky; Hai Tran; and Michelle Weathers made key contributions to this report.

Why GAO Did This Study IoT generally refers to devices (or “things”), such as vehicles and appliances, that use a network to communicate and share data with each other. The increasing popularity of wireless IoT devices that use spectrum has created questions about spectrum needs. GAO was asked to examine issues related to spectrum and IoT. This report discusses, among other things, (1) spectrum challenges related to IoT, (2) how the federal government plans for IoT’s spectrum needs, and (3) how selected leading countries prepare for IoT’s spectrum needs. GAO reviewed documents and interviewed officials from FCC and the National Telecommunications and Information Administration as well as 24 officials from a variety of sectors, including government, commercial, and manufacturing. Stakeholders were selected based on a literature review, among other factors. GAO interviewed government and commercial representatives from four leading countries regarding IoT planning and development and reviewed associated documents. These countries were selected based on criteria that included level of economic development among other criteria. What GAO Found The stakeholders GAO spoke with identified two primary spectrum-related challenges for the internet of things (IoT)—the availability of spectrum and managing interference. Although not considered an immediate concern, Federal Communications Commission (FCC) staff and some stakeholders noted that rapid increases in IoT devices that use large amounts of spectrum—called high-bandwidth devices—could quickly overwhelm networks, as happened with smart phones. Stakeholders and FCC staff also indicated that managing interference is becoming more challenging as the number of IoT and other wireless devices grows, particularly in bands that do not require a spectrum license. The figure below illustrates the uses of radio frequency spectrum, including unlicensed use. FCC plans for IoT’s spectrum needs by broadly tracking spectrum demand and making additional spectrum available as needed. Ensuring sufficient spectrum to support commercial demand is one way FCC pursues its strategic goal of promoting economic growth. FCC has made additional spectrum publicly available at least four times since 2015 by repurposing over 11 gigahertz of spectrum. However, FCC does not track the growth of IoT devices in two areas that pose the greatest risk to IoT’s growth—high bandwidth and unlicensed-spectrum devices. In 2014, FCC’s Technical Advisory Council (TAC) recommended that FCC monitor high-bandwidth IoT devices and make sufficient unlicensed spectrum available. FCC officials said that FCC monitors spectrum use broadly and makes spectrum available as needed. However, since the process of reallocating spectrum is lengthy, FCC may not have adequate time to take actions to avoid a shortage, possibly hindering IoT’s growth and associated economic growth. Spectrum planners in four leading countries—France, Germany, the Netherlands, and South Korea—have taken steps similar to those taken by the United States in preparation for IoT’s expansion, including taking a technology-neutral approach that stakeholders believe encourages innovation. Unlike the United States, officials from two leading countries said they are concerned about spectrum congestion from the growth of IoT devices, but only one is actively monitoring congestion. In addition, three leading countries have developed nationwide low power wide-area networks that use unlicensed spectrum with potential benefits including low costs and low barriers to entry. What GAO Recommends FCC should track the growth in (1) high-bandwidth IoT devices and (2) IoT devices that rely on unlicensed spectrum. FCC did not believe these actions are necessary but noted that it would ask its TAC to periodically review and report on IoT’s growth. GAO continues to believe the recommendations are valid.

gao_GAO-18-187T

gao_GAO-18-187T_0

FirstNet’s Progress Establishing and Financing the Network and Consulting Stakeholders In our June 2017 report, we found that FirstNet has conducted key efforts to establish the network, namely releasing the request for proposal for the network in January 2016 and awarding the network contract to AT&T in March 2017. As the contractor, AT&T will be responsible for the overall design, development, production, operation, and evolution of the network, as well as the marketing, product management, sales, distribution, and customer care. Further, we found that FirstNet has established a framework to meet the financial requirements established in the 2012 Act, as depicted in figure 1. This framework focuses on leveraging FirstNet’s spectrum through the use of payments and fees with the aim of ensuring that the network is financially sustainable over the life of the contract and that FirstNet sustains self-funding operations. By establishing a single, dedicated network for public safety use, FirstNet’s network is expected to foster greater interoperability and meet public safety officials’ reliability and other needs. However, the actual use (or “adoption”) of the network by public safety users will be voluntary. Thus, even with the establishment of this framework, substantial unknowns remain regarding how many public safety users will adopt the network, the extent to which AT&T will be successful in monetizing the spectrum to retain revenue from commercial users, and the extent to which this revenue will be sufficient or appropriate in relation to the capital needed to build, operate, and maintain the network. Therefore, we noted that, at the time of our report, we could not assess the viability of this framework and whether FirstNet’s structures for overseeing the contractor’s use of the spectrum for commercial users will be appropriate. We also found that FirstNet has made progress consulting with state and local, federal, and tribal stakeholders through a variety of mechanisms. State officials we contacted were generally satisfied with FirstNet’s efforts to engage them. However, tribal stakeholders we contacted expressed concern with FirstNet’s efforts to consult with tribes per the 2012 Act’s requirements. In particular, four of the five tribal organizations we contacted said that FirstNet has not fully engaged in effective communication or has relied on state points of contact too much as opposed to engaging directly with tribes; the other tribal organization was not aware of FirstNet or its mission at all. Further, tribes noted that individuals with first-hand knowledge of tribes’ experiences are not able to represent tribal views directly among FirstNet’s key decision makers. FirstNet has stated that, indeed, the 2012 Act requires that it consult with tribes through state points of contact. Nevertheless, several federal agencies have identified seeking a full understanding of tribal concerns— and reaching consensus where possible—as a key principle of effective tribal communication, noting that agencies should adapt to changing circumstances, contemplate creative problem solving, identify options for addressing concerns, and exhaust alternatives to achieve mutually agreeable solutions. We concluded that, by fully exploring and proposing actions to address tribal stakeholders’ concerns, FirstNet could help improve its relations with tribes and better meet stakeholders’ needs. As such, we recommended in our report that FirstNet fully explore tribal concerns and propose actions, as needed, to address those concerns. FirstNet agreed with this recommendation and, in September 2017, described to us the actions it has taken to implement it. For example, according to FirstNet, in September 2017 it began a process to formally explore the tribal outreach concerns raised in our report and expects to propose improvements by the end of this year. FirstNet has also said that it adopted an organization-wide tribal consultation policy which it expects to take effect towards the end of this year. If implemented as planned, these actions should address the intent of the recommendation. FirstNet’s Network Reliability, Security, and Interoperability Challenges and Efforts to Address Them In our report, we found that—according to stakeholders we contacted— FirstNet faces various challenges to ensure the network’s reliability, security, and interoperability. For example, stakeholders raised concerns related to: providing network coverage to rural areas, in buildings, or ensuring the network’s overall resiliency and cybersecurity; and managing frameworks for user identity, credentialing of users, access management, and prioritization of users on the network. However, we also found that both FirstNet and the PSCR have begun research and other efforts to help ensure the reliability, security, and interoperability of the network and address the challenges raised by stakeholders. For example, in November 2016, FirstNet opened an Innovation and Test Lab at its technical headquarters in Boulder, Colorado. According to FirstNet documentation, FirstNet plans to use— and allow AT&T to use—the lab to test public safety devices and applications before deploying them on the network. Additionally, the PSCR has conducted research on behalf of FirstNet and, using $300 million in funds provided to NIST by the 2012 Act, is also planning for and implementing other research activities to support FirstNet. For instance, in January 2016, PSCR launched its Public Safety Innovation Accelerator Program to support these research activities, and in December 2016, NIST issued a funding announcement to fund research in several areas. At the time of our report, we found that PSCR’s research process generally aligned with key phases of sound research programs identified by leading national organizations, including the American Evaluation Association and the National Academy of Sciences. For example, PSCR has established a structured process for developing research priorities that includes both internal and external stakeholders, and has identified criteria it uses to help it select the research areas to fund and procedures to help it guide and monitor its research. Similarly, FirstNet has determined its research priorities to date based on its network- planning needs and in consultation with internal and external stakeholders, and worked with the PSCR to define criteria to help it select research areas. Further, we found that the majority of stakeholders we contacted were satisfied with the planning efforts to ensure the reliability, security, and interoperability of the network. However, many stakeholders also said that there is much remaining uncertainty about how this will be implemented in practice. Additionally, one public safety official we contacted told us that FirstNet and its contractor will have to balance the costs associated with implementing features that make the network reliable and secure with the need to establish compelling and competitively priced service packages and fees that will encourage user adoption of the network. Indeed, numerous stakeholders we contacted cited the cost of subscribing to the network as a key factor affecting user adoption, noting that the pricing must be comparable to what they pay for commercial service now, that budgets are constrained in the public safety community, or that local governments do not want costs to increase. Further, commercial carriers could choose to compete with FirstNet. FirstNet has stated that it expects AT&T to provide services at a competitive price and deliver affordable, high-quality services that will encourage public safety users to adopt the network. Ultimately—because the network must be self-funding and FirstNet has stated that revenue from network users will be critical to this funding—the success of the network depends on whether FirstNet and AT&T generate enough revenue to operate it over the long term and whether public safety users adopt it, no matter how reliable and secure it is. FirstNet’s Contract Oversight Mechanisms FirstNet must manage and oversee the implementation of the network contract to build, operate, and maintain the network. Federal internal- control standards also state that an entity’s management retains responsibility for the performance of processes assigned to service organizations (such as contractors) and that management should hold these organizations accountable for their performance. In our report, we found that FirstNet has taken a number of steps to establish contract oversight mechanisms, but has not fully assessed the staffing needs of its oversight workforce. FirstNet’s oversight mechanisms include developing policies and procedures to guide contract administration and establishing offices to oversee its network contractor. In particular, FirstNet established the Network Program Office to oversee the contractor’s performance and facilitate quality assurance of contract deliverables, among other things. FirstNet is also receiving assistance from the Department of the Interior, which has experience with contract administration, although FirstNet plans to assume full responsibility for contract administration in the future. We also found that FirstNet’s efforts to develop contract oversight mechanisms aligned with several key actions that we identified as contributing to effective contract oversight. However, although FirstNet’s Network Program Office will perform essential contract administration functions, FirstNet had not conducted long-term projections of staffing needs for the office as of April 2017. Planning for and assigning adequate resources, including people, and performing an assessment of the resources needed to oversee projects is one of the key actions we identified for planning and executing effective contract oversight. We concluded that FirstNet lacks reasonable assurance that it will have sufficient resources to handle increases in its responsibilities over time and that, by performing a long-term staffing assessment for the Network Program Office, FirstNet would be in a better position to fully understand its staffing needs and respond to staffing changes and risks as it assumes full responsibility of contract administration in the future. As such, we recommended in our report that FirstNet assess the long-term staffing needs in the Network Program Office prior to assuming full responsibility for administering the network contract. FirstNet agreed with this recommendation and, in September 2017, described the actions it has taken to implement it. According to FirstNet, in August 2017 the Network Program Office adopted a strategic workforce plan for fiscal years 2018 to 2022, which it expects to update annually. According to FirstNet, this plan provides a comprehensive view of current and future human capital needs required to support the implementation of the network and identifies strategies the office will employ to fill gaps between current and future needs, among other things. If implemented as planned, this action should address the intent of the recommendation. Chairman Donovan, Ranking Member Payne, and Members of the Subcommittee, this concludes my prepared statement. I would be pleased to respond to any questions that you may have at this time. GAO Contact and Staff Acknowledgments If you or your staff have any questions about this testimony, please contact Mark L. Goldstein, Director, Physical Infrastructure Issues at (202) 512-2834 or goldsteinm@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this statement. GAO staff who made key contributions to this testimony are Sally Moino and Nalylee Padilla. Other staff who made contributions to the report cited in this testimony are identified in the source product. This is a work of the U.S. government and is not subject to copyright protection in the United States. The published product may be reproduced and distributed in its entirety without further permission from GAO. However, because this work may contain copyrighted images or other material, permission from the copyright holder may be necessary if you wish to reproduce this material separately.

Why GAO Did This Study FirstNet is charged with establishing a nationwide public-safety broadband network that is reliable, secure, and interoperable. To inform this work, FirstNet is consulting with a variety of stakeholders. In March 2017, FirstNet awarded a 25-year contract to AT&T to build, operate, and maintain the network. FirstNet's oversight of AT&T's performance is important given the scope of the network and the duration of the contract. This testimony provides information on (1) FirstNet's efforts to establish the network; (2) stakeholder views on network reliability, security, and interoperability challenges FirstNet faces and its efforts to address them; and (3) FirstNet's plans to oversee its network contractor. This statement is based on GAO's June 2017 report ( GAO-17-569 ). For this report, GAO reviewed FirstNet documentation, key contract oversight practices identified in federal regulations and other sources, tribal communication practices identified by federal agencies, and assessed FirstNet's efforts and plans against these practices. GAO also interviewed FirstNet officials and a nongeneralizable selection of public safety, tribal, and other stakeholders selected to obtain a variety of viewpoints. What GAO Found In June 2017, GAO reported that the First Responder Network Authority (FirstNet) had conducted key efforts to establish the network, namely releasing the request for proposal (RFP) for the network and awarding the network contract to AT&T. As the contractor, AT&T will be responsible for the overall design, development, production, operation, and evolution of the network. Additionally, FirstNet consulted with state and local, federal, and tribal stakeholders. State officials GAO contacted were generally satisfied with FirstNet's efforts to engage them. However, tribal stakeholders GAO contacted expressed concern that FirstNet has not fully engaged in effective communication with tribes. FirstNet engaged tribes through a variety of mechanisms, such as through state points of contact and a working group, but tribes noted that individuals with first-hand knowledge of tribes' experiences are unable to represent tribal views directly among FirstNet's key decision makers. Although FirstNet is required to consult with tribes through state points of contact, a key principle of effective tribal communication is to seek full understanding of tribal concerns and reach consensus where possible. By fully exploring and proposing actions to address tribal stakeholders' concerns, FirstNet could help improve its relations with tribes and better meet stakeholders' needs. According to stakeholders GAO contacted, FirstNet faces various challenges to ensure the network's reliability, security, and interoperability. For example, stakeholders raised concerns related to: providing coverage to rural areas, in buildings, or underground; ensuring the network's overall resiliency and cybersecurity; and managing frameworks for user identity, credentialing of users, access management, and prioritization of users on the network. FirstNet has taken action to address these challenges, such as by opening a test lab to test public safety devices and applications before deploying them on the network. The majority of stakeholders GAO contacted were satisfied with FirstNet's efforts but many noted that much uncertainty remains about how the network will be implemented. FirstNet established offices to oversee its network contractor, developed policies and procedures to guide contract administration—including management and oversight—and is receiving assistance from another federal agency with contract administration experience, although FirstNet plans to assume full responsibility in the future. For example, FirstNet established the Network Program Office to oversee the contractor's performance and facilitate quality assurance of contract deliverables, among other things. Although this office will perform essential contract-administration functions, FirstNet had not conducted long-term projections of staffing needs for the office as of April 2017. As a result, FirstNet lacks reasonable assurance that it will have sufficient resources to handle increases in its responsibilities over time. Planning for and assigning adequate resources, including people, and assessing resource needs is a key practice for planning and executing effective contract oversight. By performing a long-term staffing assessment for the Network Program Office, FirstNet would be in a better position to fully understand its staffing needs and respond to staffing changes and risks as it assumes full responsibility of contract administration in the future. What GAO Recommends In June 2017, GAO recommended that FirstNet fully explore tribal stakeholders' concerns and assess its long-term staffing needs. FirstNet agreed with GAO's recommendations and described actions to address them.

gao_GAO-18-671

gao_GAO-18-671_0

Background The Veterans Access, Choice, and Accountability Act of 2014 provided up to $10 billion in funding for veterans to obtain health care services from community providers through the Choice Program when veterans faced long wait times, lengthy travel distances, or other challenges accessing care at VA medical facilities. The temporary authority and funding for the Choice Program was separate from other previously existing programs through which VA has the option to purchase care from community providers. Legislation enacted in August and December of 2017 and June 2018 provided an additional $9.4 billion for the Veterans Choice Fund. Authority of the Choice Program will sunset on June 6, 2019. Responsibilities of the Choice Program TPAs In October 2014, VA modified its existing contracts with two TPAs that were administering another VA community care program—the Patient- Centered Community Care program—to add certain administrative responsibilities associated with the Choice Program. For the Choice Program, each of the two TPAs—Health Net and TriWest—are responsible for managing networks of community providers who deliver care in a specific multi-state region. (See fig. 1.) Specifically, the TPAs are responsible for establishing networks of community providers, scheduling appointments with community providers for eligible veterans, and paying community providers for their services. Health Net’s contract for administering the Choice Program will end on September 30, 2018, whereas TriWest will continue to administer the Choice Program until the program ends, which is expected to occur in fiscal year 2019. Choice Program Claim Processing and Payment VA’s TPAs process claims they receive from community providers for the care they deliver to veterans and pay providers for approved claims. Figure 2 provides an overview of the steps the TPAs follow for processing claims and paying community providers. VA’s contracts with the TPAs do not include a payment timeliness requirement applicable to the payments TPAs make to community providers. Instead, a contract modification effective in March 2016 established a non-enforceable “goal” of processing—approving, rejecting or denying—and, if approved, paying clean claims within 30 days of receipt. To be reimbursed for its payments to providers, the TPAs in turn submit electronic invoices—or requests for payment—to VA. TPAs generate an invoice for every claim they receive from community providers and pay. VA reviews the TPAs’ invoices and either approves or rejects them. Invoices may be rejected, for example, if care provided was not authorized. Approved invoices are paid, whereas rejected invoices are returned to the TPAs. The federal Prompt Payment Act requires VA to pay its TPAs within 30 days of receipt of invoices that it approves. VA’s Planned Consolidated Community Care Program The VA MISSION Act of 2018, among other things, requires VA to consolidate its community care programs once the Choice Program sunsets 1 year after the passage of the Act, authorizes VA to utilize a TPA for claims processing, and requires VA to reimburse community providers in a timely manner. Specifically, the act requires VA (or its TPAs) to pay community providers within 30 days of receipt for clean claims submitted electronically and within 45 days of receipt for clean claims submitted on paper. In December 2016, prior to enactment of the VA MISSION Act of 2018, VA issued an RFP for contractors to help administer the Veterans Community Care Program. The Veterans Community Care Program will be similar to the current Choice Program in certain respects. For example, VA is planning to award community care network contracts to TPAs, which would establish regional networks of community providers and process and pay those providers’ claims. However, unlike under the Choice Program, under the Veterans Community Care Program, VA is planning to have medical facilities—not the TPAs—generally be responsible for scheduling veterans’ appointments with community providers. The Length of Time Taken by TPAs to Pay Claims from Community Providers Has Varied Widely, and VA’s Monitoring of This Timeliness Has Been Limited The Time TPAs Have Taken to Pay Claims Has Varied Widely, and Available Data Do Not Account for Payment Delays Due To Rejected or Denied Claims From November 2014 through June 2018, VA’s TPAs paid a total of about 16 million clean claims—which are claims that contain all required data elements—under the Choice Program, of which TriWest paid about 9.6 million claims and Health Net paid about 6.4 million. Data on the median number of days VA’s TPAs have taken to pay clean claims each month show wide variation over the course of the Choice Program—from 7 days to 68 days. As discussed previously, in March 2016, VA established a non-enforceable goal for its TPAs to process and, if approved, pay clean claims within 30 days of receipt each month. Most recently, from January through June 2018, the median number of days taken to pay clean claims ranged from 26 to 28 days for TriWest, while it ranged from 28 to 44 days for Health Net. (See fig. 3.) In addition to the 16 million clean claims the TPAs paid from November 2014 through June 2018, during this time period they also paid approximately 650,000 claims (or 4 percent of all paid claims) that were classified as non-clean claims when first received after obtaining the required information. Non-clean claims are claims that are missing required information, which the TPA must obtain before the claim is paid. From November 2014 through June 2018, TriWest paid around 641,000 non-clean claims (or 6 percent of all paid claims) while Health Net paid about 9,600 non-clean claims (or less than 1 percent of all paid claims). Data on the median number of days VA’s TPAs have taken to pay non- clean claims each month also show wide variation over the course of the Choice Program—from 9 days to 73 days. (See fig. 4.) The data on the time TPAs have taken to pay approved clean and non- clean claims do not fully account for the length of time taken to pay providers whose claims are initially rejected or denied, as, according to the TPAs, providers are generally required to submit a new claim when the original claim is rejected or denied. Thus, providers that submit claims that are rejected or denied may experience a longer wait for payment for those claims or may not be paid at all. In some cases, providers’ claims may be rejected or denied multiple times after resubmission. Stakeholders Identified Three Key Factors Affecting the Timeliness of Claim Payments to Community Providers under the Choice Program VA and its TPAs identified three key factors affecting the timeliness of claim payments to community providers under the Choice Program: (1) VA’s untimely payments of TPA invoices; (2) Choice Program contractual requirements related to provider reimbursement; and (3) inadequate provider education on filing Choice Program claims, as discussed below. VA’s untimely payments of TPA invoices. According to VA and TPA officials, VA made untimely invoice payments to its TPAs—that is, payments made more than 30 days from the date VA received the TPAs’ invoices—which resulted in the TPAs at times having insufficient funds available to pay community providers under the Choice Program. A VA Office of Inspector General (OIG) report estimated that from November 2014 through September 2016, 50 percent of VA’s payments to its TPAs during this time frame were untimely. VA officials stated that VA’s untimely payments to the TPAs resulted from limitations in its fee-basis claims system, which VA used at the beginning of the Choice Program to process all TPA invoices. In addition, the VA OIG found that VA underestimated the number of staff necessary to process Choice Program invoices in a timely manner. Choice Program reimbursement requirements. According to VA and TPA officials, three Choice Program requirements, some of which were more stringent than similar requirements in other federal health care programs, led to claim denials, which, in turn, contributed to the length of time TPAs have taken to pay community providers when the providers did not meet these requirements: 1. Medical documentation requirement. Prior to a March 2016 contract modification, VA required providers to submit relevant medical documentation with their claims as a condition of payment from the TPAs. According to TriWest officials, those Choice Program claims that did not include medical documentation were classified by TriWest as non-clean claims and placed in pending status until the documentation was received. When community providers did not provide the supporting medical documentation after a certain period of time, TriWest typically denied their claims. According to Health Net officials, Choice Program claims that did not include medical documentation were denied by Health Net. 2. Timely filing requirement. VA requires providers to file Choice Program claims within 180 business days from the end of an episode of care. TPAs deny claims that are not filed within the required time frame. 3. Authorization requirement. VA requires authorizations for community providers to serve veterans under the Choice Program and receive reimbursement for their services; however, if community providers deliver care after an authorization period or include services that are not authorized, the TPAs typically deny their claims. According to TPA data, denials related to authorizations are among the most common reasons the TPAs deny community provider claims. Inadequate provider education on filing Choice Program claims. According to VA and TPA officials as well as providers we interviewed, issues related to inadequate provider education may have contributed to the length of time it has taken the TPAs to pay community providers under the Choice Program. These issues have included providers submitting claims with errors, submitting claims to the wrong payer, or otherwise failing to meet Choice Program requirements. For example, some VA community care programs require the claims to be sent to one of VA’s claims processing locations, while the Choice Program requires claims to be sent to TriWest or Health Net. Claims sent to the wrong entity are rejected or denied and have to be resubmitted to the correct payer. Ten of the 15 providers we interviewed stated that that they lacked education and/or training on the claims filing process when they first began participating in the Choice Program, including knowing where to file claims and the documentation needed to file claims that would be processed successfully. Four of these 10 providers stated that they learned how to submit claims through trial and error. VA’s Monitoring of Claim Payment Timeliness Has Been Limited under the Choice Program, but VA Plans to Strengthen Monitoring under the Veterans Community Care Program At the infancy of the Choice Program, November 2014 through March 2016, VA was unable to monitor the timeliness of its TPAs’ payments to community providers because it did not require the TPAs to provide data on the length of time taken to pay these claims. Effective in March 2016, VA modified its TPA contracts and subsequently began monitoring TPA payment timeliness, requiring TPAs to report information on claims processing and payment timeliness as well as information on claim rejections and denials. However, because VA had not established a payment timeliness requirement, VA officials said that VA had limited ability to penalize TPAs or compel them to take corrective actions to address untimely claim payments to community providers. Instead, the March 2016 contract modification established a non-enforceable goal for the TPAs to process and pay clean claims within 30 days of receipt. As of July 2018, according to VA officials, VA did not have a contractual requirement it could use to help ensure that community providers received timely payments in the Choice Program. Officials from VA’s Office of Community Care told us that VA’s experience with payment timeliness in the Choice Program informed VA’s RFP for new contracts for the Veterans Community Care Program, which includes provisions that strengthen VA’s ability to monitor its future TPAs. For example, in addition to requiring future TPAs to submit weekly reports on claim payment timeliness as well as claim rejections and denials, VA’s RFP includes claim payment timeliness standards that are similar to those in the Department of Defense’s TRICARE program. Specifically, according to the RFP, TPAs in the Veterans Community Care Program will be required to process and pay, if approved, 98 percent of clean claims within 30 return claims, other than clean claims, to the provider with a clear explanation of deficiencies within 30 days of original receipt, and process resubmitted claims within 30 days of resubmission receipt. The RFP also identifies monitoring techniques that VA may employ to assess compliance with these requirements, including periodic inspections and audits. VA officials told us that VA will develop a plan for monitoring the TPAs’ performance on these requirements once the contracts are awarded. VA Has Addressed Some but Not All of the Key Factors Affecting the Timeliness of Claim Payments to Community Providers under the Choice Program We found that VA has made system and process changes that improved its ability to pay TPA invoices in a timely manner. However, while VA has modified two Choice Program requirements that contributed to provider claim payment delays, it has not fully addressed delays associated with authorizations for care. Furthermore, while VA and its TPAs have taken steps to educate community providers in order to help prevent claims processing issues, 9 of the 15 providers we interviewed reported poor customer service when attempting to resolve these issues. VA Has Changed Its System and Processes for Paying TPA Invoices in order to Improve Its Ability to Pay TPAs in a Timely Manner VA has taken steps to reduce untimely payments to its TPAs, which contributed to delayed TPA payments to providers, by implementing a new system and updating its processes for paying TPA invoices so that it can pay these invoices more quickly. Specifically, VA has made the following changes: In March 2016, VA negotiated a contract modification with both TPAs that facilitated the processing of certain TPA invoices outside of the fee basis claims system from March 2016 through July 2016. According to VA officials, due to the increasing volume of invoices that the TPAs were expecting to submit to VA during this time period, without this process change, VA would have experienced a high volume of TPA invoices entering its fee basis claims system, which could have exacerbated payment timeliness issues. In February through April 2017, VA transitioned all TPA invoice payments from its fee basis claims system to an expedited payment process under a new system called Plexis Claims Manager. VA officials told us that instead of re-adjudicating community provider claims as part of its review of TPA invoices, Plexis Claims Manager performed up front checks in order to pay invoices more quickly, and any differences in billed and paid amounts were addressed after payments were issued to the TPAs. In January 2018, VA transitioned to a newer version of the Plexis Claims Manager that enabled VA to once again re-adjudicate community provider claims as part of processing TPA invoices, but in a timelier manner compared with the fee basis claims system. According to VA officials, this is due to the automation of claims processing under Plexis Claims Manager, which significantly reduced the need for manual claims processing by VA staff that occurred under the fee basis claims system. Based on VA data, as of July 2018, VA is paying 92 percent of TriWest’s submitted invoices within 7 days, with payments being made in an average of 4 days, and 90 percent of Health Net’s invoices within 7 days, with payments being made in an average of 4 days under the newer version of Plexis Claims Manager. In addition to steps taken to address untimely payments to the TPAs under the current Choice Program contracts, VA has taken steps to help assure payment timeliness in the forthcoming Veterans Community Care Program. Specifically, the RFP includes a requirement for VA to reimburse TPAs within 14 days of receiving an invoice. VA officials stated that to achieve this metric, they are implementing a new payment system that will replace Plexis Claims Manager and will no longer re-adjudicate TPA invoices prior to payment. VA Has Modified Two Choice Program Requirements That Contributed to Provider Payment Delays, but Has Not Fully Addressed Delays Associated with Authorizations for Care VA has issued a contract modification and waivers for two Choice Program contract requirements that contributed to provider payment delays—(1) the medical documentation requirement and (2) the timely filing requirement. However, while VA issued a contract modification to amend the requirements for obtaining authorizations for Choice Program care, provider payment delays associated with requesting these authorizations may persist, because VA is not ensuring that VA medical centers review and approve these requests within required time frames. Elimination of medical documentation requirement. Effective beginning March 2016, VA issued a contract modification that eliminated the requirement that community providers must submit medical documentation as a condition of receiving payment for their claims. Data from one TPA showed a reduction in non-clean claims following the implementation of this contract modification. For example, starting in April 2016, after this modification was executed, almost 100 percent of claims submitted to TriWest were classified as clean claims, as opposed to 49 percent of claims submitted in March 2016. However, when the modification first went into effect in March 2016, TriWest and Health Net officials stated that they processed a large amount of claims from community providers that had previously been pended or denied because they lacked medical documentation and, in turn, submitted a large number of invoices to VA for reimbursement. As previously discussed, to help address the increased number of TPA invoices, VA issued lump-sum payments to the TPAs during this time period. Modification of timely filing requirement. In February and May 2018, VA issued waivers that gave TPAs the authority to allow providers to resubmit rejected or denied claims more than 180 days after the end of the episode of care if the original claims were submitted timely—that is, within 180 days of the end of the episode of care. VA officials stated that the waivers were intended to reduce the number of rejected and denied claims by giving community providers the ability to resubmit previously rejected or denied claims for which the date of service occurred more than 180 days ago. VA’s waivers were implemented as follows: In February 2018, VA issued a waiver that allowed community providers to resubmit certain claims rejected or denied for specific reasons when the provider or TPA could verify that the provider made an effort to submit the claim prior to the claims submission deadline. In May 2018, VA issued a second waiver that removed the 180 day timeliness requirement for all Choice Program claims. The waiver also provided instructions to the TPAs on informing providers that they may resubmit claims rejected or denied for specific reasons and how the TPAs are to process the resubmitted claims. In regards to the first waiver, TPA officials stated that the processing of those resubmitted claims adversely affected the timeliness of the TPAs’ payments to community providers because the waiver resulted in a large influx of older claims. As the second waiver was in the process of being implemented by the two TPAs at the time we conducted our work, we were unable to determine if the second waiver affected the TPAs’ provider payment timeliness. Changes to authorization of care requirement. VA issued a contract modification in January 2017 to expand the time period for which authorizations for community providers to provide care to veterans under the Choice Program are valid. In addition, in May 2017, VA expanded the scope of the services covered by authorizations, allowing them to encompass an overall course of treatment, rather than a specific service or set of services. According to VA officials, the changes VA made related to the authorization of care requirement were also intended to reduce the need for secondary authorization requests (SAR). Community providers request SARs when veterans need health care services that exceed the period or scope of the original authorizations. Community providers are required to submit SARs to their TPA, which, in turn, submits the SARs to the authorizing VA medical facility for review and approval. Both Health Net and TriWest officials told us that since VA changed the time frame and scope of authorizations, the number of SARs has decreased. Despite efforts to decrease the number of SARs, payment delays or claim denials are likely to continue if SARs are needed. We found that VA is not ensuring that VA medical facilities are reviewing and approving SARs within required time frames. VA policy states that VA medical facilities are to review and make SAR approval decisions within 5 business days of receipt. However, officials from one of the TPAs and 7 of the 15 providers we interviewed stated that VA medical facilities are not reviewing and approving SARs in a timely manner. According to TriWest officials, as of May 2018, VA medical facilities in their regions were taking an average of 11 days to review and make approval decisions on SARs, with four facilities taking over 30 days for this process. According to an official from VA’s Office of Community Care, VA does not currently collect reliable national data to track the extent of nonadherence to the VA policy to review and make SAR approval decisions within 5 business days. The official told us that instead, VA relies on employees assigned to each Veterans Integrated Service Network to monitor data on VA medical facilities’ timeliness in making these SAR approval decisions. If a VA medical facility is found not to be in adherence with the SAR policy, the official told us that staff assigned to the Veterans Integrated Service Network attempt to identify the reasons for nonadherence, and perform certain corrective actions, including providing education to the facility. Despite these actions, the official told us that there are still VA medical facilities not in adherence with VA’s SAR approval policy. According to a VA official, VA is in the process of piloting software for managing authorizations that will allow VA to better track SAR approval time frames across VA medical facilities in the future. However, even after this planned software is implemented, if VA does not use the data to monitor and assess SAR approval decision time frames VA will be unable to ensure that all VA medical facilities are adhering to the policy. Standards for internal control in the Federal Government state that management should establish and operate monitoring activities to evaluate whether a specific function or process is operating effectively and take corrective actions as necessary. Furthermore, monitoring such data will allow VA to identify and take actions as needed to address any identified challenges VA medical facilities are encountering in meeting the required approval decision time frames. Without monitoring data to ensure that all VA medical facilities are adhering to the SAR approval time frames as outlined in VA policy, community providers may delay care until the SARs are approved or provide care without SAR approval. This in turn increases the likelihood that the community providers’ claims will be denied. Further, continued nonadherence to VA’s SAR policy raises concerns about VA’s ability to ensure timely approval of SARs when VA medical facilities assume more responsibilities for ensuring veterans’ access to care under the forthcoming Veterans Community Care Program. TPAs Have Taken Steps to Improve Provider Education to Help Providers Resolve Claims Processing Issues, but Many Providers Still Report Poor Customer Service We found that VA and its TPAs have taken steps to educate community providers in order to help prevent claims processing issues that have contributed to the length of time TPAs have taken to pay these providers. Despite these efforts, 9 of the 15 providers we interviewed reported poor customer service when attempting to resolve claims payment issues. While VA’s contracts with the TPAs do not include requirements for educating and training providers on the Choice Program, both TPAs have taken steps to educate community providers on how to successfully submit claims under the Choice Program. Specifically, TriWest and Health Net officials told us that they have taken various steps to educate community providers on submitting claims correctly, including sending monthly newsletters, emails, and faxes to communicate changes to the Choice Program; updating their websites with claims processing information; and holding meetings with some providers monthly or quarterly to resolve claims processing issues. Officials from both TPAs also told us that they provided one-on-one training to some providers on the claims submission process to help reduce errors when submitting claims. In addition, VA’s RFP for the Veterans Community Care Program contracts includes requirements to provide an annual training program curriculum and an initial on-boarding and ongoing outreach and education program for community providers, which includes training on the claims submission and payment processes and TPA points of contact. VA and the TPAs have also made efforts to help providers resolve claims processing issues and outstanding payments. For example, VA launched its “top 20 provider initiative” in January 2018 to work directly with community providers with high dollar amounts of unpaid claims and resolve ongoing claims payment issues. This initiative included creating rapid response teams to work with community providers to settle unpaid claim balances within 90 days and working with both TPAs to increase the number of clean claims paid in less than 30 days. In addition, VA has developed webinars on VA’s community care programs and—in conjunction with trade organizations and health care systems—has delivered provider education on filing claims properly. TriWest officials stated that it has educated the customer service staff at its claims processing sub-contractor, who field community provider calls regarding claims processing issues, to help ensure that the staff are familiar with Choice Program changes and can effectively assist community providers and resolve claims processing issues. Internal TriWest data show that providers’ average wait time to speak to a customer service representative about claims processing issues decreased from as high as 18 minutes in 2016 to as low as 2.5 minutes in 2018. Health Net officials were unable to provide data, but stated that since the fourth quarter of 2017, Health Net has decreased the time it takes for a community provider to speak with a customer service representative by adding additional staff and extending the hours in which providers can call with questions. In addition, Health Net officials stated that they have required customer service staff to undergo additional training related to resolving claims processing issues. Despite these efforts, 7 of the 10 providers that participate in the Health Net network and 2 of the 7 providers that participate in the TriWest network we interviewed between April and June 2018 told us that when they contact the TPAs’ customer service staff to address claim processing questions, such as how to resolve claim rejections or denials, they experience lengthy hold times, sometimes exceeding one hour. In addition, 7 of the 15 providers we spoke with told us they typically reach employees who are unable to answer their questions. According to these providers, this experience frustrated them, as they often did not understand why a claim had been denied or rejected, and they required assistance correcting the claim so it could be resubmitted. One community provider stated that their common practice to resolve questions or concerns was to call customer service enough times until they received the same answer twice from a TPA representative. In addition, 5 of the 10 Health Net providers we interviewed stated that they have significant outstanding claim balances owed to them. One of these providers—who reported over $3 million in outstanding claims—stressed the importance of being able to effectively resolve claims issues with TPA customer service staff, as the administrative burden of following up on outstanding claim balances takes time away from caring for patients. The issues concerning customer service wait times and TPA staff inability to resolve some claims processing issues reported by community providers appear to be inconsistent with VA contractual requirements. VA’s current Choice Program contracts require the TPAs to establish a customer call center to respond to calls from veterans and non-VA providers. The contract requires specified levels of service for telephone inquiries at the call center. For example, VA requires TPA representatives to answer customer service calls within an average speed of 30 seconds or less and requires 85 percent of all inquiries to be fully and completely answered during the initial telephone call. However, VA officials explained that VA does not enforce the contractual requirement for responding to calls from community providers. Furthermore, according to these officials, VA allows the TPAs to prioritize calls from veterans. Officials from VA’s Office of General Counsel, Procurement Law Group, confirmed that this requirement does apply to the TPAs’ handling of calls from community providers. Because VA does not enforce the customer service requirement for providers, VA has not collected data on or monitored the TPAs’ compliance with these requirements for providers’ calls. As previously stated, standards for internal control in the Federal Government state that management should establish and operate monitoring activities to evaluate whether a specific function or process is operating effectively and take corrective actions as necessary. Without collecting data and monitoring customer service requirements for provider calls, VA does not have information on the extent to which community providers face challenges when contacting the TPAs about claims payment issues that could contribute to the amount of time it takes to successfully file claims and receive reimbursement for services under the Choice Program. This, in turn, poses a risk to the Choice Program to the extent that community providers who face these challenges decide not to serve veterans under the Choice Program. Looking forward, VA has included customer service requirements in its RFP for the Veterans Community Care Program contracts, and VA officials have told us that these requirements are applicable to provider calls. For example, the RFP includes a requirement for its future TPAs to establish and maintain call centers to address inquiries from community providers and has established customer service performance metrics to monitor call center performance. Monitoring data on provider calls under the contracts will be important as Veterans Community Care Program TPAs will continue to be responsible for building provider networks, processing claims, and resolving claims processing issues. Conclusions The Choice Program relies on community providers to deliver care to eligible veterans when VA is unable to provide timely and accessible care at its own facilities. Although VA has taken steps to improve the timeliness of TPA claim payments to providers, VA is not collecting data or monitoring compliance with two Choice Program requirements, and this could adversely affect the timeliness with which community providers are paid under the Choice Program. First, VA does not have complete data allowing it to effectively monitor adherence with its policy for VA medical facilities to review SARs within 5 days of receipt, which impacts its ability to meet the requirement. To the extent that VA medical facilities delay these reviews and approvals, community providers may have to delay care or deliver care that is not authorized, which in turn increases the likelihood that the providers’ claims will be denied and the providers will not be paid. Second, VA requires the TPAs to establish a customer call center to respond to calls from veterans and non-VA providers. However, VA does not enforce the contractual requirement for responding to calls from community providers and allows the TPAs to prioritize calls from veterans. Consequently, VA is not collecting data, monitoring, or enforcing compliance with its contractual requirements for the TPAs to provide timely customer service to providers. As a result, VA does not have information on the extent to which community providers face challenges when contacting the TPAs about claims payment issues, which could contribute to the amount of time it takes to receive reimbursement for services. To the extent that these issues make community providers less willing to continue participating in the Choice Program and the forthcoming Veterans Community Care Program, they pose a risk to VA’s ability to successfully implement these programs and ensure veterans’ timely access to care. Recommendations for Executive Action We are making the following two recommendations to VA: Once VA’s new software for managing authorizations has been fully implemented, the Undersecretary for Health should monitor data on SAR approval decision time frames to ensure VA medical facilities are in adherence with VA policy, assess the reasons for nonadherence with the policy, and take corrective actions as necessary. (Recommendation 1) The Undersecretary for Health should collect data and monitor compliance with the Choice Program contractual requirements pertaining to customer service for community providers, and take corrective actions as necessary. (Recommendation 2) Agency Comments We provided a draft of this report to VA for review and comment. In its written comments, reproduced in appendix I, VA concurred with our two recommendations and said it is taking steps to address them. For example, VA plans to implement software in spring 2019 that will automate the SAR process and allow for streamlined reporting and monitoring of SAR timeliness to ensure ongoing compliance. Additionally, VA has included provider customer service performance requirements and metrics in its Veterans Community Care Program RFP, and will require future contractors to provide a monthly report to VA on their call center operations and will implement quarterly provider satisfaction surveys. We are sending copies of this report to the Secretary of Veterans Affairs, the Under Secretary for Health, appropriate congressional committees, and other interested parties. This report is also available at no charge on the GAO Web site at http://www.gao.gov. If you or your staff have any questions about this report, please contact Sharon M. Silas at (202) 512-7114 or silass@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs are on the last page of this report. GAO staff who made major contributions to this report are listed in appendix II. Appendix I: Comments from the Department of Veterans Affairs Appendix II: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Marcia A. Mann (Assistant Director), Michael Zose (Analyst-in-Charge), and Kate Tussey made major contributions to this report. Also contributing were Krister Friday, Jacquelyn Hamilton, and Vikki Porter.

Why GAO Did This Study Questions have been raised about the lack of timeliness of TPAs' payments to community providers under the Choice Program and how this may affect the willingness of providers to participate in the program as well as in the forthcoming Veterans Community Care Program. You asked GAO to review issues related to the timeliness of TPAs' payments to community providers under the Choice Program. This report examines, among other things, (1) the length of time TPAs have taken to pay community providers' claims and factors affecting timeliness of payments, and (2) actions taken by VA and the TPAs to reduce the length of time TPAs take to pay community providers for Choice Program claims. GAO reviewed TPA data on the length of time taken to pay community provider claims from November 2014 through June 2018, the most recent data available at the time of GAO's review. GAO also reviewed documentation, such as the contracts between VA and its TPAs, and interviewed VA and TPA officials. In addition, GAO interviewed a non-generalizable sample of 15 community providers, selected based on their large Choice Program claims volume, to learn about their experiences with payment timeliness. What GAO Found The Department of Veterans Affairs' (VA) Veterans Choice Program (Choice Program) was created in 2014 to address problems with veterans' timely access to care at VA medical facilities. The Choice Program allows eligible veterans to obtain health care services from providers not directly employed by VA (community providers), who are then reimbursed for their services through one of the program's two third-party administrators (TPA). GAO's analysis of TPA data available for November 2014 through June 2018 shows that the length of time the TPAs took to pay community providers' clean claims each month varied widely—from 7 days to 68 days. VA and its TPAs identified several key factors affecting timeliness of payments to community providers under the Choice Program, including VA's untimely payments to TPAs, which in turn extended the length of time TPAs took to pay community providers' claims; and inadequate provider education on filing claims. VA has taken actions to address key factors that have contributed to the length of time TPAs have taken to pay community providers. For example, VA updated its payment system and related processes to pay TPAs more quickly. According to VA data, as of July 2018, VA was paying at least 90 percent of the TPAs' invoices within 7 days. In addition, VA and the TPAs have taken steps to improve provider education to help providers resolve claims processing issues. However, 9 of the 15 providers GAO interviewed said they continue to experience lengthy telephone hold times. According to VA and TPA officials, steps have been taken to improve the customer service offered to community providers. However, VA officials do not collect data on or monitor TPA compliance with customer service requirements—such as calls being answered within 30 seconds or less—for provider calls because they said they are not enforcing the requirements and are allowing TPAs to prioritize calls from veterans. Without collecting data and monitoring compliance, VA does not have information on challenges providers may face when contacting TPAs to resolve payment issues. What GAO Recommends GAO is making two recommendations, including that VA should collect data on and monitor compliance with its requirements pertaining to customer service for community providers. VA concurred with GAO's recommendations and described steps it will take to implement them.

gao_GAO-18-39

gao_GAO-18-39_0

Background We have consistently stressed the importance of IRS conducting tax compliance research. Likewise, analyzing compliance data can help IRS identify ways to enhance taxpayer compliance. Without such data, it is more difficult for IRS to decide whether its efforts to address specific areas of noncompliance should focus on non-enforcement activities, such as improved forms or publications, or enforcement activities. Analyzing the data can also help identify changes that could be made to tax laws and regulations that may improve compliance. In April 2016, IRS released its most recent tax gap estimate. IRS estimated that taxpayers should have paid an average of about $2.5 trillion dollars per year in federal taxes for tax years 2008 to 2010. IRS also estimated that taxpayers paid approximately $2.04 trillion voluntarily and on time for those years, on average, leaving $458 billion in unpaid taxes per year. However, IRS estimates that through late payments and enforcement actions it eventually will collect an additional $52 billion on average for those years, leaving the average net tax gap at $406 billion for 2008 to 2010, as shown in figure 1. The tax gap estimate is an aggregate estimate of the five types of taxes that IRS administers—individual income, corporation income, employment, estate, and excise taxes. For each tax type, IRS attempts to estimate the tax gap based on three types of noncompliance: (1) underreporting of tax liabilities on timely filed tax returns; (2) underpayment of taxes due from timely filed returns; and (3) nonfiling, when a taxpayer fails to file a required tax return altogether or on time. IRS has not developed specific estimates for the tax gap related to income earned from illegal activities (e.g., prostitution) or certain forms of fraud. However, if IRS discovers these types of income over the course of an audit, it could be included in the tax gap estimates. In general, refund fraud related to identity theft would not be included in the tax gap estimate because it does not involve evading a tax liability. The tax gap includes unintentional errors as well as intentional evasion, such as intentionally underreporting income, intentionally overreporting expenses, and engaging in abusive tax shelters or frivolous tax schemes. There is no single approach to estimating all of the components of the tax gap. IRS uses NRP examinations of a stratified, random sample of tax returns along with statistical modeling to produce estimates of noncompliance for the population of individual income tax return filers. Other areas of the tax gap are estimated using payment data or other statistical models. Each approach IRS uses is subject to non-sampling error and the areas of the estimate that are samples are subject to additional sampling errors. The uncertainty of the estimates is not readily captured by standard errors that typically accompany estimates based on sample data. For that reason, IRS does not report standard errors, confidence intervals, and statistical comparisons across years. Prior to the 2008–2010 estimate, IRS had released two other tax gap estimates that used data from NRP examinations: (1) a tax year 2001 tax gap estimate that was released in February 2007 and (2) a tax year 2006 tax gap estimate that was released in December 2011. As shown in figure 2, the gross tax gap estimate has increased, in nominal terms, from $345 billion for 2001 to $458 billion, on average, for 2008–2010, an increase of $113 billion or about 33 percent. However, when we adjusted for inflation (using fiscal year 2016 dollars), the gross tax gap estimates amount to $460 billion in 2001, $530 billion in 2006, and $509 billion in 2008–2010. The inflation-adjusted gross tax gap increased by about 11 percent from 2001 to 2008–2010; however, the 2008–2010 estimate is slightly lower, by about 4 percent, than the 2006 estimate. Over the three estimates, the voluntary compliance rate (VCR)—the percentage of owed tax for a given year that is paid voluntarily and timely—has decreased slightly from 83.1 percent for 2006 to 81.7, on average, for 2008–2010. IRS also estimated the VCR and distribution of tax liability for each component of the tax gap. Generally, employment taxes have the highest estimated compliance rates while individual income and estate taxes have the lowest estimated compliance rates. There was a decrease in the VCR for individual income tax from 79 percent for tax year 2001 to 74 percent in the recent 2008–2010 estimate. According to IRS, this decline, along with the individual income tax’s increase in the share of liability, contributes to the slight decline in the overall VCR. The amount of the tax gap that IRS estimated it would collect through enforcement efforts and late payments varied across the three estimates. In 2001, IRS estimated it would collect $55 billion and that figure would increase by $10 billion, to $65 billion in 2006. However, for the 2008– 2010 tax gap estimate, the enforced collections or late payments decreased by $13 billion per year, on average, from the 2006 estimate, to $52 billion. According to IRS, the methodology used to calculate prior estimates may have been too optimistic, resulting in an overstatement of as much as 25 percent. Additionally, IRS stated that the economic downturn in 2008 reduced the total tax liability from which IRS could collect revenue over this period. However, differences in the tax gap estimates across years may not all be attributed to changes in taxpayer behavior (voluntary compliance) or IRS enforcement activities. According to IRS, the tax gap estimates have increased in part because IRS included some new tax gap components and updated some methods, which it believes increased the comprehensiveness and accuracy of the estimates. IRS reported that changes in economic activity and changes in tax law and administration also contribute to differences in tax gap estimates over time. Enforcement of Tax Laws and GAO’s High-Risk List In our 2017 High-Risk Report we continued to include Enforcement of Tax Laws as a high-risk area. Key components of this high-risk area include both addressing the tax gap and improving tax compliance. IRS enforcement of the tax laws helps fund the U.S. government by collecting revenue from noncompliant taxpayers and, perhaps more importantly, promoting voluntary compliance by giving taxpayers confidence that others are paying their fair share. However, IRS still faces challenges to its capacity for implementing new initiatives and carrying out ongoing enforcement and taxpayer service programs under an uncertain budgetary environment. Given the estimated size of the tax gap, even modest reductions would yield significant financial benefits for the country. We have made numerous recommendations over time to help IRS, the Department of the Treasury (Treasury), and Congress address tax noncompliance. Nonetheless, as we have reported in the past, closing the entire gap is not feasible since it could entail more burdensome record keeping or reporting than the public is willing to accept or more resources than IRS is able to commit. For example, third-party information reporting has shown to improve accuracy of income reporting by individual taxpayers; however, it requires increased record keeping and reporting by the third party and requires IRS resources to properly match the third-party information to individual tax returns. Underreporting by Individual Taxpayers Accounts for the Largest Portion of the Tax Gap Underreporting of tax liabilities accounted for most of the tax gap estimate for tax years 2008–2010, making up 84 percent of the entire estimated gross tax gap, as shown in figure 3. Individual income taxes made up the largest portion of underreporting, followed by employment taxes and corporation income taxes. Individual Income Tax Underreporting Underreporting of business income accounted for nearly half of the individual income tax underreporting gap. This includes income from sole proprietors, which accounted for the largest share of individual income tax underreporting, as shown in table 1. Most business related income tax return items also had high net misreporting percentages, which is the sum of the net misreported amount divided by the sum of the absolute values of the amounts that should have been reported, as a percentage. To show additional detail on aspects of the tax gap, we conducted some additional analysis on selected line items of the individual NRP data, which is presented in appendix II. As we have previously reported, the extent to which individual income tax taxpayers accurately report their income is closely aligned to the amount of income that is reported to them and to IRS by third parties. For example, according to 2008–2010 IRS data, taxpayers misreported over half of the types of income for which there is little or no third-party information reporting, such as business income. In contrast, when employers withhold taxes and report the wages and salaries to employees and IRS through Form W-2, Wage and Tax Statement (W-2), there is better compliance. As shown in figure 4, 1 percent of these types of income were misreported while nearly 99 percent were accurately reported on individual income tax returns. Similarly, taxpayers misreported less than 10 percent of income for which banks and other financial institutions provided information returns (Forms 1099) to account holders and IRS that show taxpayers’ annual income from some types of investments. Generally, new requirements on third parties to submit information returns would require statutory changes. We have previously identified additional information reporting opportunities as well as improvements that IRS could make on its own to existing forms and how it uses them. For example, we suggested in August 2008 that Congress may wish to make all taxpayers with rental real estate activity subject to the same information reporting requirements as other taxpayers operating a trade or business; however, no legislative action has been taken on this suggestion. We recommended in August 2010 that IRS require mortgage-secured property addresses to be reported on other forms (Forms 982 and 1099-C) to help IRS detect taxpayers who fail to pay taxes on certain forgiven mortgage debts. Legislative and executive actions have been taken consistent with our recommendation. However, IRS has not revised two forms to collect specific information from taxpayers and lenders concerning the amount of forgiven debt attributable to a principle residence and the locations of a taxpayer’s principle residence. For items subject to substantial third-party information reporting, IRS is able to use automated processes to address noncompliance. The automated underreporter program, through which IRS matches amounts reported on returns with amounts reported on information returns submitted by third parties, is one such process. This computer matching program allows IRS to identify discrepancies between tax returns and information returns and propose automatic changes to taxpayers. For items with little to no third-party information reporting, IRS has to rely on more resource-intensive methods, such as correspondence or face-to- face examinations, to address noncompliance. While these examinations may be started by reviewing specific line items, they may also be expanded to cover other areas of the tax returns if there are indications of misreporting in areas of the return not previously identified. However, it is harder for IRS to detect noncompliance in areas with little third-party information reporting. Employment Tax Underreporting The second largest part of the underreporting tax gap is made up of employment taxes, which are comprised of three main components: self- employment tax, Federal Insurance Contribution Act (FICA) Social Security and Medicare withholding, and Federal Unemployment Tax Act (FUTA) taxes, as shown in table 2. The self-employment component is estimated from IRS’s NRP individual income tax data. However, IRS lacks NRP data for other components of employment tax. Therefore, it estimates both FICA and FUTA by applying the estimated compliance rates from a 1993 employment tax gap report, which used tax year 1984 employment tax return data, to the current reported taxes. IRS recently completed an NRP study of employment tax returns that reviewed federal income tax withholding and FICA, the first such study IRS had conducted in over 30 years. We reported that although the examinations for the study were completed, IRS had not developed formal plans to analyze the results to (1) identify areas of noncompliance, (2) address such noncompliance, or (3) update its employment tax gap estimate. According to IRS officials, they had not developed formal plans due to competing research priorities and limited resources and because the NRP results had not yet been finalized. We recommended that IRS develop plans to analyze the NRP results in 2017 to address areas of noncompliance identified and update its employment tax gap estimates. IRS agreed with our recommendations and stated that it will be determining how to use the data from this new study to update the employment tax gap estimate. Corporation Income Tax Underreporting As shown in table 3, IRS developed corporation income tax underreporting estimates for two types of corporations: small corporations (those without a balance sheet or with assets less than $10 million) and large corporations (those with assets of $10 million or more). IRS estimated the voluntary compliance rate for all corporations to be 83 percent for tax years 2008–2010. The estimates are based on adjusted data from operational examinations, which focus on the tax returns most likely to have substantive noncompliance rather than examinations of a statistically representative sample of corporation tax returns. IRS does not have a program comparable to NRP for corporation income tax because of the difficulty of constructing a representative sample for a small group of highly diverse large corporations, among other reasons. The limited scope and selection criteria for non-NRP examinations introduce statistical bias, meaning that the examination issues and results from examinations of corporation tax returns are not necessarily representative of the overall corporation population. However, IRS has developed some methods to project the results of the examinations to the larger population of corporations, and, despite these limitations, IRS considers the corporation estimates to provide a rough gauge of corporation income tax noncompliance. IRS’s divisions responsible for large and small corporation examinations each have management systems in place to track issues identified from corporation examinations. While this information is not derived from the tax gap estimates, IRS has identified several common examination issues for both large and small corporations, as shown in table 4. Net Tax Gap For the 2008–2010 tax gap estimate, IRS for the first time estimated the net tax gap by each type of tax, as shown in table 5. Unlike most of the tax gap, IRS can tabulate late payments. Since enforcement and other late payments often happen many years after a given tax year has ended, IRS must project into the future to estimate how much tax it will eventually collect for that tax year. IRS expects to recoup the smallest percentage of taxes from the gross individual income tax estimate. IRS Has Confidence in Most Aspects of the Tax Gap Estimate and Is Taking Steps to Improve It IRS officials stated IRS believes the tax gap estimates are sufficiently reliable for the intended purpose of providing a snapshot of tax compliance as a whole. The tax gap estimate is actually many estimates used together to develop one overall picture of tax compliance. IRS has more certainty in some areas of the estimate than others. Generally, IRS officials consider those components of the tax gap estimate that are based on the most current data (2008–2010 data) to be more robust. However, as shown in table 6, IRS recognizes that some component estimates of the tax gap estimate are more uncertain than others, in part because some component estimates rely on older data and it is inherently difficult to estimate some types of noncompliance. IRS has no estimates for some areas of the tax gap. According to IRS officials, IRS has higher amounts of confidence in all of the underpayment components, because they are based on data from systems that can distinguish enforcement and late payments from other payments (IRS has the most confidence in these components); the individual income nonfiling component, because it is based on a new methodology combining two methods that incorporate improvements to the methods used in prior estimates; the individual income underreporting component, because it is primarily based on adjusted NPR examination data, which is a statistically representative sample of individual tax returns; and the corporation underreporting component, because it is based on operational examination data, adjusted for selection bias. IRS has a lesser amount of confidence in the estate nonfiling and underreporting data, because they are forecasts updating estimates based on assumptions made in studies completed in 2000; and the withholding taxes (FICA and FUCA) part of the employment underreporting data, because the data are partially a forecast based on data from an older compliance study. The methodologies used to develop the component estimates differ by component, resulting in a mix of statistical sample and operational-based data being used, as well as forecasts from earlier estimates. IRS is therefore unable to calculate confidence intervals for any of the tax gap estimates. IRS officials stated that that they continue to try to identify a value for those components without estimates, such as corporation income, excise tax nonfiling, and excise tax underreporting, but have not yet found a sufficiently reliable data source nor method upon which to base estimates. To increase its confidence in the estimates of underreported individual income, IRS uses an econometric technique called detection controlled estimation (DCE). This regression-based model controls for variables that could affect the amount of underreporting IRS examiners detected. IRS uses this adjustment because it knows its examiners do not detect all underreported income during examinations, and therefore it adjusts the NRP data to account for such undetected income when estimating the tax gap. The statistical technique estimates the noncompliance detected by a hypothetical “best practices” examiner—an ideal that is unattainable—and then statistically estimates the noncompliance detected by the hypothetical examiner to adjust upward the findings from research examinations conducted by actual examiners. The DCE adjustment accounts for more than half (about $150 billion) of the total individual income tax underreporting estimate. IRS also used the DCE adjustment for the self-employment tax estimates used in the employment tax underreporting estimate. Appendix III provides more information on the extent to which DCE adjustments contributed to the 2008–2010 tax gap estimate. IRS has taken steps to improve the tax gap estimate. For example, IRS used an updated methodology to calculate the estimated nonfiling amount for the 2008–2010 estimate that combined two prior methodologies. Specifically, IRS expanded how it matches information between the U.S. Census Bureau’s annual Current Population Survey and IRS data to estimate the amount of taxes that were not filed. IRS believes this updated methodology allows it to create a better matched dataset and identify nonfilers more accurately. See appendix IV for details on changes to IRS’s tax gap methodology. IRS plans to release its next tax gap estimate in 2019 to cover tax years 2011 to 2013. IRS is also undertaking several additional studies that may offer data IRS can use to improve the tax gap estimate, including these examples: Taxpayers’ tipping behavior: IRS is surveying taxpayers to help estimate total tip income and tipping rates by industry/occupation and by major method of payment (e.g., credit card, debit card, and cash). Limited studies on C corporations and other midsize corporations: IRS studied compliance of C corporations with assets less than $250,000 and with a balance sheet, and corporations with assets of $10 to $50 million for tax year 2010. These studies plan to identify potential areas of noncompliance. Partnership misreporting pilot: In 2016, IRS initiated this study to measure reporting compliance for certain partnerships, as well as to estimate tax misreported at the taxable partner level as a result of partnership misreporting. This study was initiated in response to a recommendation from a prior report. NRP employment tax estimates: As previously mentioned, IRS is determining how it will use the NRP employment tax study it concluded in 2017 to improve the tax gap estimates. IRS Has Broad Compliance Goals but Lacks Specific Quantitative Goals for Increasing Voluntary Compliance IRS’s current strategic plan (2014–2017) discusses general approaches to make voluntary compliance easier for taxpayers and to ensure taxes owed are paid. However, in some areas, the plan does not include specific tactics for improving compliance strategies. Rather, it addresses the elements of voluntary compliance and enforcement actions through two of its goals: Delivering high-quality and timely service to reduce taxpayer burden and encourage voluntary compliance. Effectively enforcing the law to ensure compliance with tax responsibilities and combat fraud. IRS officials stated the strategic plan goals are a component of IRS’s Future State—which is a vision for agency-wide operations that aims to improve services across different taxpayer interactions such as individual account assistance, exams, and collections—and are directly reflected in three of its six themes: facilitate voluntary compliance by empowering taxpayers with secure innovative services, tools, and support; understand noncompliant taxpayer behavior, and develop approaches to deter and change it; and select highest value work using data analytics and a robust feedback loop. According to IRS officials, the remaining themes support these goals indirectly by seeking to improve IRS’s effectiveness. IRS officials noted the IRS Future State vision has outlined two measures that will support the overall goal of increased compliance. The first will be the percentage of compliance issues resolved within 1 year of filing. The second is the percentage of taxpayers with recurring compliance issues. However, IRS has not yet determined the target levels for these goals. According to officials, the levels for these goals will be published with the next IRS strategic plan (2018–2022), which IRS is scheduled to release in mid- 2018. IRS Lacks Specific Quantitative Goals to Improve Voluntary Compliance IRS previously set or acknowledged quantitative goals to improve voluntary compliance. However, IRS has since moved away from that approach. In 2005, we recommended that IRS establish a long-term quantitative voluntary compliance goal for individual income tax underreporting and for tax underpayment, as well as for other areas of noncompliance. IRS agreed with the concept of our recommendation and, in response, established a voluntary compliance rate goal of 85 percent by 2009, which was published in IRS’s fiscal year 2007 budget request. In 2006, Treasury issued its Comprehensive Strategy for Reducing the Tax Gap, with a seven-component strategy for reducing the tax gap: (1) reduce opportunities for evasion, (2) make a multiyear commitment to research, (3) continue improvements in information technology, (4) improve compliance activities, (5) enhance taxpayer service, (6) reform and simplify the tax law, and (7) coordinate with partners and stakeholders. In 2007, IRS developed a more detailed report that emphasized the same seven components outlined in the Treasury report and also outlined projects, initiatives, legislative proposals, and other actions designed to combat the sources of noncompliance. In the 2007 report, IRS acknowledged the goal set by the then Chairman of the Senate Finance Committee for IRS to meet a 90 percent voluntary compliance rate by tax year 2017 and the goal set by the IRS Oversight Board of 86 percent by tax year 2009. In 2009, IRS published another report that followed up on the efforts discussed in the 2006 and 2007 Treasury and IRS reports. However, IRS has not published any reports since that time that focus on goals for reducing the tax gap. In 2012, Treasury, along with IRS, set an agency priority goal to increase voluntary tax compliance from 83.1 to 86 percent by September 30, 2013. However, Treasury and IRS decided not to renew the agency priority goal because they said the measure did not satisfy the criteria of having indicators and quarterly milestones against which to track process or being able to determine whether the goal has been achieved by the end of a 2-year period, as established by the Office of Management and Budget. Since the tax gap estimates are only updated every few years, Treasury and IRS officials said there was no way for Treasury or IRS to show improvements or declines in meeting the goal on a quarterly basis or over the 2-year goal term. More recently, IRS officials told us that while they want to achieve a high level of voluntary compliance, neither IRS nor Treasury has set a recent high level department- or agency-wide quantitative goal aimed at reducing the tax gap or increasing voluntary compliance. Establishing clear compliance goals and measuring progress toward them benefit both IRS and external stakeholders and are consistent with the results-oriented performance management principles set forth in the Government Performance and Results Act of 1993 and the GPRA Modernization Act of 2010. As we have previously reported, setting long-term strategic goals is essential for results-oriented management, because such goals explain in greater specificity the results an agency is intending to achieve. The goals form a basis for an organization to identify potential strategies for fulfilling its mission and for improving its operations to support achievement of that mission. Directly aligning strategic goals and strategies for achieving those goals is important for assessing an agency’s ability to achieve those goals. Further, when program results could be influenced by external factors, agencies can use intermediate goals and measure to identify the program’s discrete contribution to a specific result. IRS has moved away from specific numeric goals to improve compliance because it now believes there are limited benefits to them. According to IRS officials, IRS actions alone do not determine the level of taxpayer compliance and there are also several challenges associated with establishing meaningful and useful compliance goals. IRS officials reported that many of these challenges are due to the tax gap only being estimated every few years. As a result, fluctuations in the estimate over time may not be generally attributed to changes in compliance behavior but the fluctuations might instead result from the imprecision of the estimates or updated methodologies. According to IRS, changes in the economy may also have an effect on tax compliance rates. For example, a downturn in the economy would likely result in less tax needing to be paid, but it might also cause some taxpayers to comply less than they otherwise would. Further, separating those two effects would be difficult, particularly given the unobserved nature of most noncompliance. Finally, IRS officials stated other factors, such as IRS services, enforcement efforts, evolving social norms, or changes in legislation, may affect the overall compliance rate. Although IRS may not have full control over all of the factors that affect voluntary compliance, it does have an impact on taxpayer’s compliance through its service and enforcement programs. Furthermore, IRS is not alone in not having full control over the results it seeks to achieve. A number of methods can be used to map or model the causal relationships among the inputs, processes, and outputs produced by various strategies and the forces that influence achievement of outcomes, such as results mapping and logic modeling. Recognizing that outside influences may present risks or challenges to achieving outcomes, OMB Circular Number A-11 states that while agencies cannot mitigate all risks related to achieving strategic objectives and performance goals, they should identify, measure, and assess challenges related to mission delivery, to the extent possible. We have previously reported that setting long-term quantitative goals for IRS offers several benefits. First, compliance goals coupled with periodic measurements of compliance levels would provide IRS with a better basis for determining to what extent its various service and enforcement efforts contribute to compliance. Second, long-term, quantitative goals would help IRS consider new strategies to improve compliance, especially since these strategies could take several years to implement. Third, focusing on intended results can promote strategic and disciplined management decisions that are more likely to be effective because managers who use fact-based performance analysis are better able to target areas most in need of improvement and to select appropriate interventions. Fourth, agency accountability can be enhanced when both agency management and external stakeholders—such as Congress—can assess an agency’s progress toward meeting its goals. Likewise, a survey of the Organization for Economic Cooperation and Development countries and other advanced economies found that some governments are paying increased attention to estimating tax gaps for their major types of taxes. Several countries shared their quantitative goals for reducing the tax gap or increasing their tax revenue with the survey. For example, Denmark set a target to ensure that the tax gap does not exceed 2 percent of estimated total tax liability. Without long-term, quantitative voluntary compliance goals and related performance measures, it will be more difficult for IRS to determine the success of its strategies, adjust its approach when necessary, and remain focused on results, especially since factors that affect compliance change over time. Having compliance goals as IRS has had in the past, coupled with data, would provide a solid base upon which IRS could develop a more strategic, results-oriented approach to improving compliance. IRS Uses National Research Program Data to Guide Compliance Efforts but Has Not Documented a Long- Term Strategy IRS officials told us tax gap data are used as a high-level overview of tax compliance. IRS officials also stated they use the underlying tax gap data (i.e., NRP data and other data) in several ways to update compliance efforts. However, IRS has not documented a comprehensive strategy that shows, for example, how it intends to analyze and use the tax gap data, particularly from the NRP, to develop or improve compliance programs. IRS officials told us they use the NRP data to study specific compliance behaviors. For example, IRS has studied taxpayer behavior in claiming the Earned Income Tax Credit (EITC), the child tax credit, and the additional child tax credit. Sometimes these studies are used to develop legislative proposals that are included as part of the annual budget process and outlined in Treasury’s annual revenue proposals. Officials stated the legislative proposals are based on the knowledge of compliance derived from analysis of NRP data and they are organized into themes, such as reducing the tax gap, improving voluntary compliance, or improving tax administration. However, these revenue proposals are requests for changes to the tax laws and, ultimately, it is at Congress’s discretion whether to enact them. IRS officials reported that other times NRP data are used to compute the annual improper payment rate for the EITC. According to officials, IRS also uses these data to annually categorize the root causes of EITC noncompliance. IRS uses the NRP data to update compliance plans by updating the Discriminate Function (DIF) formulas. DIF formulas are designed to score returns for the likelihood that the tax reported on the return is significantly underreported. DIF scores help IRS ensure that noncompliant taxpayers are more likely to be selected for examination and compliant taxpayers are less likely to be unnecessarily examined. IRS determines DIF scores for individual income, small corporation income, partnership, and S corporation returns. IRS officials described high-level concepts of how the various NRP and other studies contribute to compliance and enforcement strategies. Officials from IRS’s Office of Research, Applied Analytics, and Statistics (RAAS) said they think the various uses of the NRP are widely known from general documentation about the NRP and its study design. RAAS officials also noted the fiscal year 2009 budget justification that led to special funding for the NRP still reflects IRS’s current strategy for undertaking various NRP studies. Further, IRS officials pointed us to documentation from the 2000s that discussed a need for a multiyear research commitment and IRS’s goal to move NRP studies beyond the individual income tax to include other taxes. We previously recognized IRS’s commitment to multiyear research and have noted the gains it has made in regularly estimating compliance. Our analysis of the decade-old documents found evidence of a commitment to research that generally seemed sufficient for that period. However, IRS officials did not provide us more recent documents that describe its current efforts to study compliance or show how it plans to use NRP data to update compliance strategies. IRS officials also provided us with business plans for some of their other business units and divisions. The plans we reviewed noted, at a high level, that data and analysis will be used to improve workload selection but did not discuss how specific research efforts or the results of those efforts would be integrated into the missions. The Internal Revenue Manual section on the NRP states that IRS needs to measure taxpayer compliance with federal income tax laws along with contributing factors so that customer-focused programs and services can be enhanced or developed and so that compliance information and tools can be improved. A 2001 NRP prospectus states the NRP will help to increase public confidence in the fairness of our tax system by helping IRS identify where compliance problems occur and focus its resources accordingly. Further, it states that for strategic planning and budget purposes, IRS requires regular estimates of compliance. The NRP research efforts support this critical need. According to IRS, the NRP will also improve IRS’s ability to detect noncompliance; reduce the burden of unnecessary IRS contacts on compliant taxpayers; and support the strategic goals, program development, and resource allocation of IRS operating divisions. Using quality information, such as NRP data, to achieve the agency’s objectives is one of the 17 principles for internal controls. Further, the standards for internal controls also recognize documentation is a necessary part of an effective system, but the level and nature of documentation vary based on the size of an entity and the complexity of the operational process. Documentation is required to demonstrate the design, implementation, and operating effectiveness of a system. Considering the size and relative importance of the tax gap, documenting a strategy for how IRS plans to use NRP data to reduce the tax gap would be consistent with internal controls. However, without a strategy that provides an overall picture of how NRP data are used, it may be difficult for Congress and other decision makers to understand the merits of what they are being asked to fund. Further, without developing and documenting a strategy for incorporating the results of NRP data, IRS risks not fully leveraging the compliance data it collects or not allocating enforcement resources in the most cost-effective manner. We have a long-standing history of reporting on the need for IRS to develop a comprehensive compliance strategy: In 1994, we concluded that until IRS produces a comprehensive compliance strategy, existing data could be used as part of an interim compliance strategy that directs resources at the most noncompliant taxpayers. We found that using such a starting point, IRS could focus more of its efforts on highly noncompliant areas, such as small corporation income and sole proprietorship income that made up almost a third of the tax gap. In 2005, we concluded that reducing the tax gap will be a challenging task given persistent levels of noncompliance and will not likely be achieved through a single solution. Rather, the tax gap must be attacked on multiple fronts and with multiple strategies over a sustained period, thus building a foundation to help taxpayers voluntarily comply. Between 2005 and 2007, we testified six times on the need for IRS to develop a strategy to attack the tax gap on multiple fronts with multiple approaches. In 2007, we reported on the need for IRS to develop a strategy to address noncompliant sole proprietor income, which accounts for a significant share of the tax gap. Further, documenting a strategy for using NRP data to guide compliance efforts would be consistent with two key criteria for removal from the High-Risk List: Action plan: A corrective action plan exists that defines the root cause and solutions and that provides for substantially completing corrective measures, including steps necessary to implement solutions we recommended. Demonstrated progress: Ability to demonstrate progress in implementing corrective measures and resolving the high-risk area. Conclusions The nation’s long-term fiscal projections show that the federal government is on an unsustainable fiscal path. One way to help improve the nation’s fiscal position would be to reduce the tax gap. Reducing the tax gap will be a challenging task given persistent levels of taxpayer noncompliance. However, even modest reductions would yield significant financial benefits and help improve the government’s fiscal position. IRS has shown a continued commitment to study sources of noncompliance and has made strides in improving NRP and other tax gap data. However, additional efforts could further assist IRS in addressing the tax gap. A long-term, quantitative goal for improving voluntary compliance may provide IRS with a concrete target the agency can use in fulfilling its mission. Without a quantitative goal, it will be more difficult for IRS to determine the success of its strategies, adjust its approach when necessary, and remain focused on results, especially since factors that affect compliance change over time. Likewise, a strategy that outlines how IRS plans to use NRP data to update compliance strategies would help IRS determine resource tradeoffs and more fully leverage the investment it makes in compliance research, while providing Congress with a better understanding of the merits of the research it is being asked to fund. Recommendations We are making the following two recommendations to IRS: The Commissioner of Internal Revenue should re-establish long-term, quantitative goals for improving voluntary compliance. (Recommendation 1) The Commissioner of Internal Revenue should instruct the appropriate officials to develop and document a strategy that outlines how IRS will use National Research Program data to update compliance strategies that could help address the tax gap. (Recommendation 2) Agency Comments and Our Evaluation We provided a draft of this report to the Commissioner of Internal Revenue. IRS provided written comments, which are summarized below and reprinted in appendix V. IRS also provided technical comments, which we incorporated where appropriate. IRS disagreed with our recommendation that it re-establish long-term, quantitative goals for improving voluntary compliance. In its letter, IRS stated that the voluntary compliance rate is ill-suited as a strategic or performance metric for IRS for various reasons. For example, IRS stated that improving voluntary compliance, determining the success of its strategies, and adjusting its approach could be accomplished in the absence of a quantitative goal. However, as we note in the report, setting long-term strategic goals is essential for results-oriented management, because such goals explain in greater specificity the results an agency is intending to achieve. Further, focusing on intended results can promote strategic and disciplined management decisions that are more likely to be effective because managers who use fact-based performance analysis are better able to target areas most in need of improvement and to select appropriate interventions. IRS also stated that its actions alone do not determine the level of voluntary compliance, which is determined by the interaction of many factors, such as taxpayer behavior, tax law complexity, and IRS resources. We agree that IRS may not control all factors that affect voluntary compliance. However, IRS does influence taxpayer compliance through its service and enforcement programs. Furthermore, as we point out in the report, while agencies cannot mitigate all outside influences that may present risks or challenges to achieving outcomes, they should identify, measure, and assess such challenges to the extent possible. Given the benefits of setting long-term quantitative goals—as discussed in this report—we continue to believe it is prudent for IRS to establish such goals. IRS agreed with our recommendation that it develop and document a strategy that outlines how IRS will use National Research Program data to update compliance strategies that could help address the tax gap. As agreed with your office, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies to the Chairmen and Ranking Members of other Senate and House committees and subcommittees that have appropriation, authorization, and oversight responsibilities for IRS. We will also send copies of the report to the Commissioner of Internal Revenue and other interested parties. In addition, this report will be available at no charge on the GAO website at http://www.gao.gov. If you or your staff has any questions about this report, please contact me at (202) 512-9110 or mctiguej@gao.gov. Contact points for our offices of Congressional Relations and Public Affairs are on the last page of this report. GAO staff members who made major contributions to this report are listed in appendix VI. Appendix I: Objectives, Scope, and Methodology Our objectives were to determine (1) the main drivers of the tax gap; (2) the Internal Revenue Service’s (IRS) confidence in the tax gap estimates; (3) IRS goals, if any, for reducing the tax gap; and (4) the extent to which IRS uses tax gap estimates and underlying data to develop strategies and actions to reduce the tax gap. To determine the main drivers of the tax gap and IRS’s confidence in the estimates, we reviewed IRS’s tax gap estimates and underlying data; IRS technical papers and reports; third-party reviews of the data; and past GAO and Treasury Inspector General for Tax Administration reports on the 2001, 2006, and 2008–2010 tax gap estimates. We also interviewed officials from IRS’s Office of Research, Applied Analytics, and Statistics (RAAS) who are responsible for estimating the tax gap. We determined that IRS’s tax gap and compliance estimates were sufficiently reliable for the purposes of this report, particularly since IRS already has publicly released its tax gap estimates and disclosed their limitations. These purposes include discussing the major tax gap components, the orders of magnitude for various components, and IRS’s opinions about the certainty of its estimates. To determine IRS’s goals for increasing voluntary compliance, we reviewed IRS’s and the Department of the Treasury’s (Treasury) strategic plans and Treasury’s General Explanations of the Administrations Fiscal Year Revenue Proposals (commonly referred to as the Green Book) from 2011 to 2017. We also reviewed other IRS and Treasury documentation, such as the strategies for improving voluntary compliance that were developed in the mid-2000s. Additionally, we reviewed Treasury’s agency priority goals, including the goal it set in 2012 to increase voluntary compliance. We also reviewed the statutory requirements for agency performance goals under GPRAMA. We interviewed Treasury officials in the Office of Tax Analysis and the Office of Strategic Planning and Performance Improvement about prior goals that were set and the goal- setting process. We interviewed IRS officials responsible for developing strategies and establishing goals to reduce the tax gap, specifically officials in the Deputy Commissioner’s Office for Service and Enforcement and the Small Business/Self-Employed, Large Business and International, and Wage and Investment divisions. To determine the extent to which IRS uses the tax gap and other underlying data to update compliance efforts, we requested documentation on any plans or strategies that show how the various tax gap and other compliance studies work together toward a larger compliance strategy. We found that the agency had not developed these documents. We interviewed staff from the Deputy Commissioner’s Office for Service and Enforcement and RAAS about the agency’s plans to use tax gap data and other compliance studies when developing compliance strategies. To show additional detail on aspects of compliance using the same data upon which the individual income tax underreporting tax gap estimates are based, we examined IRS’s tax gap estimates for tax years 2008– 2010 and the underlying data from its National Research Program (NRP) study of individual income tax returns. This information is presented in appendix II. Unlike other IRS examinations, NRP examinations can be used to estimate taxpayer reporting compliance because they are drawn from a stratified, statistically representative sample of the population of individual income tax returns. We interviewed IRS officials from RAAS about their research and analysis of the NRP data, and we gathered related documentation where available. IRS officials described the quality review and data reliability processes they used to collect data from the NRP examinations. Because the NRP sample was based on random selections, the sample was only one of a large number of samples that IRS could have drawn. Since each sample could have provided different estimates, we express our confidence in the precision of our estimates based on the sample as a 95 percent confidence interval plus or minus a margin of error. This is the interval that would contain the actual population value for 95 percent of the samples that could have been drawn. The estimates presented in appendix II have margins of error of less than 10 percent or 10 percentage points. In analyzing the NRP data, we conducted several reliability tests to ensure the data we used were sufficiently complete. For example, we electronically tested the data for obvious errors. We concluded that the data were sufficiently reliable for the purposes of this report based on these steps and on our previous reviews of tax gap estimates and NRP data. We conducted this performance audit from July 2016 to October 2017 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Additional Analyses of National Research Program Individual Data We analyzed the raw National Research Program (NRP) data for individual income tax returns, including itemized deductions, for tax years 2008–2010. These data do not account for the undetected income that the Internal Revenue Service (IRS) adjusts for when developing tax gap estimates. Therefore these data are not comparable to the tax gap data presented elsewhere in this report, but they allow additional analysis of areas of misreporting. Table 7 provides greater details on certain line items from the individual income tax return that align with various third-party information reporting levels determined by IRS. Line items with substantial or some information reporting tended to have lower overall misreporting percentages. Line items that had the largest percentage of taxpayers with misreported income were those for sole proprietor and farming income, both of which are subject to little to no information reporting. Based on the analysis of these data, we also estimate that line items with the largest mean amounts of underreporting are business sale and supplemental income or loss; the business sale line item also has the largest mean amount of overreporting, averaging $21,000 per return; and sole proprietor income and farm income have very similar misreporting data; over 75 percent of the income is misreported and, of that, over 80 percent is under reported while under 20 percent is overreported. We also analyzed selected credits and deductions to determine the average noncompliance rates of these line items, as shown in table 8. Because credits and deductions offset taxpayers’ income or tax owed, taxpayers who underreported a credit or deduction overreport their tax liability or tax owed; taxpayers who overreport a credit or deduction underreport their tax liability or tax owed. However, we did not determine the extent to which misreporting was because of issues specific to how taxpayers claimed the credits and deductions or because taxpayers misreported their income, which could affect eligibility for certain credits and deductions. For example, the 2016 income limit to claim the Earned Income Tax Credit (EITC) is $39,296 for single, surviving spouse, or head of household taxpayers with one qualifying child. If IRS determined that such taxpayers who claimed the EITC and reported less than $39,296 in income on their tax returns had underreported their income in order to stay under the threshold, those taxpayers would no longer be eligible for the credit and would have overreported the credit. Based on the analysis of these data, we estimate that of the selected credits, EITC is the most claimed credit and about half of the time it is misreported; thirty-five percent of the time the child tax credit was misreported with about two- thirds of that being underreported, which is inconsistent with filing patterns for most of the other credits and deductions; of the selected deductions, the deduction for real estate taxes is most often claimed and about a quarter of the time it is misreported; the medical expenses line item has the highest percentage of misreporting; 80 percent of the time the deduction is overreported; and the mortgage interest deduction has the highest average overreported amount, over $4,000 per return. Appendix III: Detection Correction Factor’s Effect on Tax Gap Estimation In estimating the individual income tax gap, the Internal Revenue Service (IRS) applies an econometric technique called detection controlled estimation (DCE), which is a regression-based model that controls for variables that could affect the amount of underreporting detected. The statistical technique first produces a hypothetical ‘‘best practices’’ examiner—an ideal which is unattainable—based on who conducted the examinations and the observed examination results. It then statistically estimates the noncompliance detected by the hypothetical examiner to adjust upward the findings from research examinations conducted by actual examiners. The technique estimates total undetected underreporting by imputing the average underreporting undetected by IRS’s National Research Program (NRP) examination to the detected underreporting, controlling for certain return line item characteristics. According to IRS officials, this approach is reasonable and the best currently available to attempt to estimate the full amount of underreported individual income. As shown in table 9, more than half of the underreporting component of the tax gap consists of income that IRS did not detect during examinations. DCE is not applicable to the underpayment and nonfiling components of the tax gap. Appendix IV: Additional Information on IRS’s Tax Gap Methodology Changes In efforts to improve the tax gap estimates, the Internal Revenue Service (IRS) updated two areas of the 2008–2010 tax gap estimate by developing updated methodologies. Nonfiling individual income tax: IRS summed together the total of taxpayers who do not file tax returns and those who filed late. The 2008–2010 estimate of taxpayers not filing a return was made by combining the two methods used to estimate the 2006 and 2001 tax gap estimates. In 2006, IRS used a sample of individuals not appearing on filed tax returns. In 2001, IRS conducted an “exact match” between the Census Bureau’s annual Current Population Survey and IRS data. The 2008–2010 estimate of late filers was based on the total balance due from late filed tax returns, adjusted for income reported to IRS on information returns. IRS believes its current methodology is an improvement over the 2006 estimate as it uses the population data rather than a sample, avoiding disadvantages resulting from sampling. IRS reported improvements in the Census and IRS information associated with nonfilers allowed them to create a better matched dataset and identify nonfilers more accurately. Nonfiling self-employment data: The methodology IRS used to calculate self-employment tax nonfiling is the same that it used for individual income tax. However, for the 2008–2010 estimate, IRS changed where it reports self-employment tax nonfiling within the tax gap estimate. For the 2008–2010 estimate, this tax is now reported in the employment tax category, whereas for the 2001 and 2006 estimates it was reported in the individual income tax category. IRS officials stated that for the 2006-2010 estimate IRS decided to break self-employment out separately and report it with the employment tax because they believe it allows a more comprehensive view of employment taxes. IRS also updated how the underlying data supporting the tax gap are organized in two ways: Changes in net tax gap estimates: As previously mentioned, IRS published net tax gap estimates by each tax type (individual income, corporation income, employment, and estate tax) for the first time in the 2008–2010 estimate. IRS officials stated that they made progress by providing this data. However, IRS is unable to break out the net tax gap further by tax component because during examinations, adjustments are not categorized by component (i.e., underreporting, underpayment, and nonfiling). Changes in individual income underreporting: Starting in 2008– 2010, IRS modified the categories it uses to break down the individual income underreporting component. These changes affect how IRS calculates the net misreporting percentage (NMP) for individual income tax, but not how it calculates the tax gap. IRS uses the NMP to show the relationship between third-party information reporting and individual income tax reporting compliance. IRS reported the changes reflect an improvement in methodology. IRS cautions that any comparison of the 2008–2010 NMP to the 2006 NMP estimates should consider those improvements. The prior calculation method involved adding offsets to income, such as deductions, exemptions, and adjustments, which distorted the comparison across categories. IRS determined a better approach was to combine income items into categories and to report offsets to income as a separate category. (See figure 5.) Appendix VI: GAO Contact and Staff Acknowledgments GAO Contact: Staff Acknowledgments: In addition to the contact named above, Jeff Arkin (Assistant Director), James Ashley, Jehan Chase, Charles Fox, John Hussey, Donna Miller, John Mingus, Edward Nannenhorn, Cynthia Saunders, Robyn Trotter, and Elwood White made significant contributions to this report.

Why GAO Did This Study The tax gap—the difference between tax amounts that taxpayers should have paid and what they actually paid—has been a persistent problem for decades. The tax gap estimate is an aggregate estimate of the five types of taxes that IRS administers—individual income, corporation income, employment, estate, and excise taxes. For each tax type, IRS attempts to estimate the tax gap based on three types of noncompliance: (1) underreporting of tax liabilities on timely filed tax returns; (2) underpayment of taxes due from timely filed returns; and (3) nonfiling, when a taxpayer fails to file a required tax return altogether or on time. GAO was asked to review IRS's tax gap estimate for tax years 2008 to 2010. This report provides information on (1) the main drivers of the tax gap; (2) IRS's confidence in the tax gap estimates; (3) IRS's goals, if any, for reducing the tax gap; and (4) the extent to which IRS uses tax gap estimates and underlying data to develop strategies to reduce the tax gap. GAO reviewed IRS tax gap data and reports and interviewed IRS officials. What GAO Found The Internal Revenue Service's (IRS) latest tax gap estimate found that taxpayers voluntarily and timely paid about 81.7 percent of the taxes they should have paid for tax years 2008-2010. As with past estimates, underreporting of individual income taxes accounted for the largest portion of the 2008-2010 tax gap. IRS believes the tax gap estimates are sufficiently reliable to provide a snapshot of tax compliance as a whole because much of the estimates are based on the most current data available, such as from IRS's National Research Program (NRP). IRS previously set or acknowledged goals to improve voluntary compliance. However, IRS has since moved away from that approach. IRS officials now believe there are limited benefits to establishing goals because IRS cannot control all aspects of compliance and updated methodologies may cause fluctuations in the estimates. IRS does, however, have an impact on taxpayers' compliance through its service and enforcement programs. Without long-term, quantitative goals for improving voluntary compliance, it will be difficult for IRS to determine the success of its compliance efforts and adjust its approaches. The Internal Revenue Manual states IRS needs to measure taxpayer compliance and other factors so compliance information and tools can be improved. IRS uses tax gap data to study compliance behaviors and update computer formulas designed to identify tax returns with a high likelihood of noncompliance. Yet IRS has not documented a comprehensive strategy that shows how it intends to use NRP data to update compliance strategies. Officials said the uses of NRP are widely known from general documentation about NRP. Without developing and documenting a strategy for using the NRP data to update compliance strategies, IRS may not fully leverage the compliance data or allocate enforcement resources in the most cost-effective manner, and it may be difficult for Congress and others to understand the merits of what they are being asked to fund. What GAO Recommends GAO recommends that IRS re-establish goals for improving voluntary compliance and develop and document a strategy that outlines how it will use its data to update compliance strategies to address the tax gap. IRS disagreed with GAO's recommendation on establishing goals and agreed with the recommendation on compliance strategies. GAO continues to believe that goals are essential for results-oriented management.

gao_GAO-18-164

gao_GAO-18-164_0

Few Grantees Certified They Were Exempt from Statutory Restrictions on Religious-Based Hiring Nine DOJ Grantees Certified They Were Exempt from Statutory Restrictions on Religious- Based Hiring From 2007 to 2015, HHS, DOL, and DOJ awarded funding to at least 2,586 grantees through at least 53 grant programs that were subject to statutory restrictions on religious-based hiring. Specifically: HHS identified one grant program subject to statutory restrictions on religious-based hiring for which nonprofits were eligible to be primary recipients—the Projects for Assistance in Transition from Homelessness (PATH) program, which is administered by the Substance Abuse and Mental Health Services Administration (SAMHSA). Generally, only states are eligible to be primary recipients for PATH grant awards. However, HHS may award PATH grants directly to public or nonprofit entities if a state does not submit an application or does not meet program requirements. From this program, no grants were awarded to nonprofit organizations and therefore no FBOs were awarded grants. DOL identified 18 grant programs subject to statutory restrictions on religious-based hiring for which nonprofits were eligible to be primary recipients. All 18 of these grant programs were in DOL’s Employment and Training Administration (ETA). From these 18 programs, 931 grantees were awarded grants, including 19 we identified as potential FBOs. DOJ identified at least 34 relevant grant programs administered by OVW, COPS, and six different program offices within OJP that were subject to statutory restrictions on religious-based hiring. The 34 relevant grant programs represent the minimum number of grant programs that were subject to nondiscrimination provisions and for which nonprofit organizations were eligible from fiscal years 2007 through 2015. The number of relevant grant programs could be higher. As discussed below, OJP was unable to identify the total number of relevant grant programs and total number grantees awarded grants under these programs, including potential FBOs. More specifically within DOJ: OVW identified 20 grant programs subject to statutory restrictions on religious-based hiring. From these 20 programs, 604 grantees were awarded grants, including 25 that OVW identified as potential FBOs. OJP identified at least 10 grant programs subject to statutory restrictions on religious-based hiring. According to officials, OJP was not able to readily identify grant solicitations that were available to nonprofit organizations from fiscal years 2007 to 2015 and subject to statutory restrictions on religious-based hiring. This effort, according to OJP, would have required a manual search of each grant solicitation. However, OJP was able to identify at least 10 grant programs subject to statutory restrictions on religious- based hiring. From these 10 programs, at least 1,113 grantees were awarded grants, including 74 we identified as potential FBOs. COPS identified four grant programs subject to statutory restrictions on religious-based hiring. From these four programs, 57 grantees were awarded grants, none of which were potential FBOs. Of the 117 potential FBOs we identified across the three agencies, nine faith-based grantees, all of which were awarded DOJ grants, certified that they were exempt from statutory restrictions on religious-based hiring (see figure 1). These 9 grantees were, therefore, allowed to consider a prospective employee’s religious faith when making employment decisions in connection with the grant. DOL and HHS reported that none of their grantees have sought exemptions from religious-based hiring provisions. As shown in Table 1, 8 of the 9 faith-based grantees that certified that they were exempt were awarded funding through DOJ grant programs from fiscal years 2008 to 2010. The remaining exempted grantee received a funding award in 2015. The total funding awarded to the 9 grantees was approximately $3.2 million, which is less than 1 percent of the $804 million in grants that DOJ awarded that are subject to statutory restrictions from fiscal years 2007 to 2015. DOJ reported that 8 of these grantees received the awards on a noncompetitive basis because they were identified for funding in a DOJ appropriation or accompanying committee report. Exempted Faith-Based Grantees Stated that Hiring Staff to Assist with Grant Activities on the Basis of Religion Was Critical to Their Mission We interviewed 6 of the 9 grantees that certified that they were exempt from religious-based hiring restrictions. Each of the 6 grantees that we interviewed stated that: hiring individuals who share their religious beliefs to assist with grant activities was critical to their mission and organizational success; they include a “statement of faith” on their organization’s job application form and ask the applicant to attest to the statement of faith, or hired individuals of the same faith already employed within their organization; and had the RFRA exemption not been available to them, they likely would not have sought the grant or they would have had to seek executive- level approval within their organization to apply for the grant. At least 3 of the 6 grantees stated that they were a recipient of other federal grant funding, but those grants were not subject to statutory restrictions on religious-based hiring, and therefore did not require an exemption to make hiring decisions based on religion. Based on grant award documentation, 6 of the 9 grantees used the funding to provide assistance to at-risk youth. However, other services that the remaining grantees provided included first responder training and programs to reduce homelessness, among others, and support and response efforts for victims of sexual assault. As discussed earlier, we also selected 35 potential faith-based grantees that received funding in fiscal years 2014 and 2015 and that agencies reported had not filed a self-certification to be exempted from religious- based hiring restrictions. We interviewed 5 of these 35 grantees to discuss, among other things, whether the grantees were familiar with the exemption options. The five faith-based grantees said they did not recall seeing information about the exemption option in the grant application or grant award documentation, or were not looking for information about the exemption because they were not considering religion in their hiring decisions. Two of the faith-based grantees that did not certify as exempt told us that, while they ask that the applicant have an understanding of the traditions, culture, or languages of their religion, they do not require applicants to share the same faith. Federal Agencies Primarily Use Grant Documentation to Notify Grantees of Restrictions on Religious-Based Hiring and Requirements for Demonstrating Eligibility for Exemptions Agencies Notify Grantees of Statutory Restrictions on Religious-Based Hiring through Grant Materials that Identify Relevant Regulations DOJ, DOL, and HHS inform grant applicants and recipients of statutory restrictions on religious-based hiring and processes for obtaining an exemption from such restrictions through grant announcements. The agencies also use additional methods that varied across all three agencies for providing this information to grantees. DOJ specifically made this information available on agency web pages as well as in the documentation that is provided to grant recipients. DOJ’s Center for Faith-Based and Neighborhood Partnerships has a web page specifically for FBOs that have applied for or received grant funding. This web page includes a list of Frequently Asked Questions, including one that addresses hiring employees with federal grant funds. The Office for Civil Rights within OJP also provides information on its web page regarding how FBOs may certify that they are exempt from statutory restrictions on religious-based hiring. Additionally, it includes a link to a copy of DOJ’s exemption certification form. We interviewed representatives from four potential faith-based grantees that received a DOJ grant in fiscal years 2014 or 2015 and did not certify for an exemption. All four grantees said they could not recall seeing information in the grant application or award documentation about the exemption option or were not looking for it because they were not considering religion in their hiring decisions. Similarly, DOL has a web page devoted specifically to explaining statutory restrictions on religious-based hiring to faith-based grant applicants and recipients, which also covers the process for seeking exemptions from the restrictions. The web page makes reference to DOL’s regulations related to religious-based hiring by FBOs and also has a link to the June 2007 OLC opinion. Additionally, DOL has prepared a guidance document—available from its grants program overview web page—that explains in detail the process for seeking exemptions and how they are reviewed and approved. A representative from the one potential FBO we interviewed that received a DOL grant in fiscal years 2014 or 2015 but did not certify that they were exempt could not recall seeing information about the exemption option. Lastly, in addition to providing information in grant announcements, HHS provides all SAMHSA grant applicants seeking funds for substance abuse prevention and treatment services with a form that cites laws and regulations governing religious organizations that receive SAMHSA funding, including the regulation that outlines the exemption process. HHS requires the applicants to sign the form, and in doing so, the applicants are certifying that they are aware of and will comply with applicable laws that allow FBOs to provide SAMHSA-funded services without impairing their religious character and without diminishing the religious freedom of those who receive their services. Agencies Rely on Grantees to Self-Certify that They Meet Eligibility Requirements for Exemptions from Statutory Restrictions on Religious- Based Hiring DOJ, DOL, and HHS all require grantees that seek to make employment decisions based on religion to self-certify that they meet requirements to be eligible for an exemption from statutory restrictions on religious-based hiring, but vary in how they review and approve requests for exemptions. Department of Justice DOJ faith-based grantees that wish to demonstrate they are eligible for an exemption from statutory restrictions on religious-based hiring must complete and sign a “Certificate of Exemption for Hiring Practices on the Basis of Religion.” If an applicant is awarded a grant, it must submit a copy of the signed version of this form through DOJ’s Grants Management System. By signing the form, the grantee is certifying that: federally-funded services will be offered to all qualified beneficiaries without regard for the religious or nonreligious beliefs of those individuals; activities that contain inherently religious content will be kept separate from grant-related activities or offered to clients voluntarily; and the organization believes that the services provided are an expression of its religious beliefs, employing persons of a particular religion is important to its mission, and not being able to hire such persons would be a substantial burden to the organization. DOJ does not review these self-certification submissions to approve or deny the requests. It only reviews them for any indication that the applicant may not be an FBO, in which case DOJ officials said they would follow up with the grantee to get clarification. Agency officials also said DOJ would review any self-certifications as part of grantee compliance reviews and in response to complaints from other parties. The self- certification form covers the entire grant award period, and can cover multiple DOJ grants as long as all of the grant programs are subject to the same statutory restrictions on religious-based hiring. There is no deadline for submitting the self-certification and DOJ officials told us that while it is understood that self-certifications should be submitted before grant funds are dispersed, grantees do not need to do so. Department of Labor DOL faith-based grantees that wish to demonstrate they are eligible for an exemption also self-certify, but are required to submit their request to DOL for review and approval by the Assistant Secretary responsible for issuing or administering the grant. In its request, the grantee must certify that: providing the services to be funded by the grant is an exercise of its without the grant, its ability to provide the services funded by the grant would be substantially diminished, and providing those services is demonstrably tied to the recipient’s religious beliefs; employing individuals of a particular religious belief is important to its religious identity, autonomy, or communal religious exercise; conditioning the grant award on compliance with the nondiscrimination provision creates substantial pressure on it, in providing the services being funded, to abandon its belief that hiring based on religion is important to its religious exercise; and it will comply with the requirements of 29 C.F.R. part 2, subpart D, Equal Treatment in Department of Labor Programs for Religious Organizations; Protection of Religious Liberty of Department of Labor Social Service Providers and Beneficiaries. The Assistant Secretary’s office then reviews exemption requests and approves them or provides a reason for denial. DOL has instituted a 30- day deadline to reply back to the grant applicant with its decision. DOL implemented this process in response to the 2007 OLC opinion. However, agency officials said they have never used this process because, as explained earlier in this report, DOL has not received any exemption requests. They also told us exemptions are only valid for the grant award period and new requests must be re-submitted if the grant is renewed. However, an exemption can cover multiple grants to the same grantee as long as those grants are received from the same DOL component. Lastly, the officials said that grant funds can be disbursed before the grantee has submitted an exemption request. Department of Health and Human Services HHS faith-based grantees seeking to demonstrate that they are eligible for an exemption from statutory restrictions on religious-based hiring must self-certify that they meet several requirements outlined in HHS regulations. To demonstrate its eligibility for an exemption, a grantee must certify that: it sincerely believes employing individuals of a particular religion is important to the definition and maintenance of its religious identity, autonomy, and/or communal religious exercise; it makes employment decisions on a religious basis in analogous programs; it believes the grant would materially affect its ability to provide the type of services in question; and providing the services in question is expressive of its values or mission. Grantees must then submit their self-certification to HHS requesting an exemption, and maintain supporting justification documentation on file if needed for future review. However, as explained earlier in this report, there is currently only one HHS grant program that is subject to a statutory restriction on religious-based hiring and for which FBOs are eligible to be primary recipients—the PATH program. We did not identify any faith-based recipients of grants from this program from fiscal years 2007 through 2015, and HHS officials confirmed that no nonprofit entities received any grants from the program during this time. Agency Comments We provided a draft of this report to the Departments of Labor, Justice, and Health and Human Services. Although the agencies did not provide formal comments, the Departments of Justice and Health and Human Services did provide technical comments that we incorporated, as appropriate. As agreed with your office, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time we will send copies of this report to the Secretaries of Health and Human Services and Labor; the Attorney General; and appropriate congressional committees. In addition, the report will be available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact either Diana Maurer at (202) 512-8777 or maurerd@gao.gov; or Cindy Brown Barnes at (202) 512-7215 or brownbarnesc@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. Key contributors to this report are listed in Appendix I. Appendix I: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Mary Crenshaw, Adam Hoffman, and Kristy Love (Assistant Directors); David Ballard; Dominick Dale; Michele Fejfar; Melissa Hargy; Joel Marus; Heidi Nielson; Kelly Rolfes- Haase; and Katrina Taylor made key contributions to this report.

Why GAO Did This Study The federal government provides billions of dollars in grant funding to organizations offering social services, including FBOs. In carrying out their mission, some FBOs prefer to hire individuals who share their religious beliefs. Although the 1964 Civil Rights Act prohibits employment discrimination based on religion, section 702(a) of the Act exempts FBOs from this prohibition, thereby allowing them to hire based on religion. However, some federal grant programs contain statutory restrictions prohibiting this practice. Since a 2007 DOJ legal opinion, federal agencies allow faith-based grantees to use RFRA as a basis for seeking an exemption to allow religious-based hiring. GAO was asked to review the extent to which faith-based grantees have sought RFRA exemptions from statutory restrictions on religious-based hiring. This report describes (1) what is known about faith-based grantees that have certified exemption from statutory restrictions on religious-based hiring, per RFRA, since 2007; and (2) how agencies inform grantees of statutory restrictions on religious-based hiring and requirements for demonstrating their eligibility for an exemption. GAO reviewed information from DOJ, HHS, and DOL grantees from fiscal years 2007 to 2015 that were subject to statutory restrictions on religious-based hiring. GAO interviewed faith-based grantees that certified as exempt and a selection of those that did not. GAO also reviewed agency grant documentation and guidance provided to grantees and interviewed cognizant officials to understand the processes FBOs must follow to certify as exempt. What GAO Found From fiscal years 2007 through 2015, few faith-based grantees sought an exemption based on the Religious Freedom Restoration Act of 1993 (RFRA) from nondiscrimination laws related to religious-based hiring. Specifically, GAO found that the Department of Justice (DOJ), Department of Health and Human Services (HHS), and Department of Labor (DOL) awarded funding to at least 2,586 grantees through at least 53 grant programs containing nondiscrimination hiring restrictions during this time. The number of relevant grant programs could be higher, because GAO could not identify all such programs due to data limitations. Across the 3 agencies, GAO identified 117 grantees that were potential Faith-Based Organizations (FBOs). Of the 117 potential FBOs, 9 DOJ grantees were FBOs that certified as being exempt from statutory restrictions on religious-based hiring. GAO interviewed 6 of these FBOs, all of which stated that hiring individuals who share their religious beliefs was critical to their mission, and that had the RFRA exemption not been available to them, they likely would not have sought the grant. DOJ, DOL, and HHS inform grant applicants and recipients of statutory restrictions on religious-based hiring and processes for obtaining an exemption from such restrictions generally through grant materials. DOJ and DOL also provide relevant information on their web sites. All three agencies require grantees that seek to make employment decisions based on religion to self-certify that they meet requirements to be eligible for an exemption, but vary in how they review and approve requests for exemptions. For example, DOJ, DOL, and HHS have policies requiring grantees to submit their exemption self-certification, but only DOL reviews exemption requests and either approves them or provides a reason for denial.

gao_GAO-19-123

gao_GAO-19-123_0

Background This section presents information on the Superfund program and the stages of the cleanup process, the relationship between federally recognized tribes and the federal government, the laws and policies that govern EPA’s consultation with federally recognized tribes regarding Superfund cleanup actions, and EPA’s administration of the Superfund program. The Superfund Program and Remedial Cleanup Process CERCLA established the Superfund program to clean up contaminated sites to protect human health and the environment from the effects of hazardous substances. Under CERCLA, potentially responsible parties are liable for conducting or paying for the cleanup of hazardous substances at contaminated sites. Under the Superfund program, EPA and potentially responsible parties can undertake two types of cleanup actions: removal actions and remedial actions. Removal actions are usually short-term cleanups for sites that pose immediate threats to human health or the environment. Remedial actions are generally long- term cleanups—consisting of one or more remedial action projects—that aim to permanently and significantly reduce contamination; these actions can take a considerable amount of time and money, depending on the nature of the contamination and other site-specific factors. The Superfund process begins with the discovery of a potentially hazardous site or notifications to EPA regarding the possible release of hazardous substances that may threaten human health or the environment. EPA delineates the Superfund remedial cleanup process in nine phases: 1. Preliminary Assessment and Site Investigation. EPA’s regional offices may discover sites with releases of hazardous substances or potential for releases of hazardous substances, or such sites may come to EPA’s attention through notifications—either reports from state agencies or citizens. As part of this first phase of the process, EPA’s regional offices use a screening system called the Hazard Ranking System to guide decision making and, as needed, to numerically assess the site’s relative potential to pose a threat to human health or the environment. 2. NPL Site Listing Process. EPA may propose sites that score at or above an established level for listing on the NPL. EPA regions submit sites to EPA headquarters for possible listing on the NPL based on a variety of factors, including the availability of alternative state or federal programs that may be used to clean up the site. Sites that EPA proposes to list on the NPL are published in the Federal Register. After a period of public comment, EPA reviews the comments and makes final decisions on whether to list the sites on the NPL. 3. Remedial Investigation and Feasibility Study. EPA or a potentially responsible party will generally begin the remedial cleanup process for an NPL site by conducting a two-part study of the site: (1) a remedial investigation to characterize site conditions and assess the risks to human health and the environment, among other actions and (2) a feasibility study to evaluate various options to address the problems identified through the remedial investigation. 4. Record of Decision. At the culmination of the remedial investigation and feasibility study, EPA issues a record of decision that identifies EPA’s selected remedy for addressing the contamination. A record of decision typically lays out the planned cleanup activities for each operable unit of the site. 5. Remedial Design and Remedial Action. EPA or a potentially responsible party plans the implementation of the selected remedy during the remedial design phase, and then, in the remedial action phase, EPA or a potentially responsible party carries out one or more remedial action projects. 6. Construction Completion. EPA generally considers the construction to be complete for a site when all physical construction at a site is complete, including actions to address all immediate threats and to bring all long-term threats under control. 7. Post-Construction Completion. The potentially responsible party or the state generally conducts operation and maintenance to maintain the remedy, such as operating a groundwater extraction and treatment system. EPA generally performs reviews of the remedy at least every five years to evaluate whether it continues to protect human health and the environment. 8. NPL Deletion. EPA may delete a site, or part of a site, from the NPL when the agency and the relevant state authority determine that no further site response is needed. 9. Site Reuse and Redevelopment. EPA works with communities to ensure that site cleanups are consistent with the site’s future use and to make sure sites or portions of sites are used safely. Relationship between Federally Recognized Tribes and the Federal Government The federal government recognizes Indian tribes as distinct, independent political communities that possess certain powers of self-government and sovereignty. As of January 9, 2019, there were 573 federally recognized Indian tribes. The federal government has a government-to-government relationship with Indian tribes, so EPA works directly with tribes. The federal government also has a trust responsibility to Indian tribes and their members based on treaties, federal laws, and court decisions. In addition, treaties between tribes and the federal government may reserve rights to a tribe that could be affected by a proposed EPA action. For example, an NPL site may contaminate fish or wildlife that a tribe has a treaty right to fish or hunt. EPA guidance notes that certain types of EPA actions, namely those that are focused on a specific geographic area, are more likely than others to have potential implications for treaty-protected natural resources. Laws and Policies Governing EPA Consultation with Tribes Regarding Superfund Cleanup Actions CERCLA includes a requirement for EPA to consult with Indian tribes in certain circumstances regarding cleanup actions at Superfund sites. Specifically, under CERCLA, EPA is required to treat tribes substantially the same as states with regard to consultation on remedial actions on lands for which an Indian tribe has jurisdiction, among other things. In addition to this CERCLA requirement, the following government-wide and agency policies apply when EPA consults with tribes regarding cleanup actions at Superfund sites: Executive Order 13175 (2000). Directs agencies to have an accountable process to ensure meaningful and timely input by tribal officials in the development of regulatory policies that have tribal implications. EPA policies and guidance EPA Policy for the Administration of Environmental Programs on Indian Reservations (1984). Sets forth principles to guide EPA in dealing with tribal governments and responding to the problems of environmental management on reservations in order to protect human health and the environment. EPA Policy on Consultation and Coordination with Indian Tribes (2011). Provides a general, agency-wide policy for consultation and coordination with tribes in cases in which EPA actions and decisions may affect tribal interests. EPA developed this policy in response to Executive Order 13175 and a 2009 presidential memorandum on tribal consultation. The policy notes that EPA submits annual progress reports to the Office of Management and Budget (OMB) on the status of its consultation actions pursuant to this 2009 presidential memorandum. This policy provides guiding principles for consultation, outlines a four- phase process for conducting consultation, and establishes the roles and responsibilities for specific EPA officials. Some EPA regional offices have their own specific guidance for consulting with tribes that include the elements of EPA’s agency-wide consultation policy, but may include more specific guidelines. For example, Region 2’s consultation guidance includes a list of specific subjects to include in notification letters to tribes. EPA Policy on Environmental Justice for Working with Federally Recognized Tribes and Indigenous Peoples (2014). Affirms EPA’s commitment to provide federally recognized tribes and indigenous peoples in the United States fair treatment and meaningful involvement in EPA decisions that may affect their health or environment. EPA Guidance for Discussing Tribal Treaty Rights (2016). The guidance states that it is intended to enhance EPA’s consultations in situations where tribal treaty rights may be affected by a proposed EPA action. EPA Memorandum on Considering Traditional Ecological Knowledge During the Cleanup Process (2017). Provides direction to improve the Superfund decision-making process to ensure EPA considers a tribe’s traditional ecological knowledge when tribes willingly provide such information. EPA Memorandum on Consideration of Tribal Treaty Rights and Traditional Ecological Knowledge in the Superfund Remedial Program (2017). Provides recommendations for regional Superfund Remedial Program staff to consider when (1) evaluating tribal treaty rights and treaty-protected resources in program implementation and (2) considering traditional ecological knowledge during the cleanup process when the information is freely provided by the tribe or tribes with interests at the site. EPA’s Administration of the Superfund Program EPA’s 10 regional offices are responsible for carrying out many of the implementation and management responsibilities for NPL sites, and are guided by the Superfund Program Implementation Manual, as well as CERCLA, CERCLA’s implementing regulations, supplementary guidance, and agency policy. The Superfund Program Implementation Manual states that its purpose is to provide overarching program management priorities, procedures, and practices for EPA’s Superfund remedial and removal programs, providing a link between EPA’s strategic plan and Superfund program internal processes, among other things. Further, the manual includes definitions for Superfund program accomplishments and outlines processes for planning and tracking accomplishments through milestones, including site-wide milestones specific to how the agency manages the release of hazardous substances (e.g., human exposure under control). Using its SEMS and TCOTS data systems, EPA tracks NPL sites that are on tribal property or that affect federally recognized Indian tribes, as well as the agency’s efforts to consult with Indian tribes regarding cleanup decisions at NPL sites. SEMS is EPA’s primary database to track Superfund program accomplishments and milestones and to answer Superfund-related questions from Congress, federal and state agencies, and the public. SEMS is EPA’s primary system for Superfund data collection, reporting, and tracking and serves as the Superfund program’s data management system for accomplishment planning and tracking. According to the Superfund Program Implementation Manual, EPA regional staff are to input data into SEMS regarding planned or actual accomplishments, and EPA headquarters staff are to use SEMS data as the basis for tracking, managing, and reporting on the performance of the Superfund program. SEMS is the system of record for NPL site data, including information on tribes that have an interest in the site. We looked at three of the variables SEMS uses for tracking sites that are located on tribal property or that affect tribes: On tribal property. This variable indicates whether the release of hazardous materials is on Indian country and any other land owned by an Indian tribe or an Alaska Native entity. NAI. This variable identifies sites that may be of interest to one or more Native American entities whose members or land would be directly affected by the release of hazardous materials. Associated tribe. This variable identifies the specific Indian entity or entities associated with a site with NAI. TCOTS tracks information about potential future tribal consultation opportunities and serves as a repository for consultation-related documents for active consultations for all EPA programs, including Superfund. EPA uses TCOTS to (1) track current and forecasted consultation, (2) publicize current EPA consultation opportunities for tribal governments, and (3) provide reports to OMB, as called for in the 2009 presidential memorandum on tribal consultation. EPA Does Not Have Reliable Data Identifying NPL Sites Located on Tribal Property or That Affect Tribes EPA Data Identifying NPL Sites Located on Tribal Property Are Not Accurate EPA data identifying NPL sites that are located on tribal property or that affect tribes are not reliable. Specifically, EPA data identifying sites that are on tribal property, sites that have NAI, and the tribes that have interest in NAI sites are not accurate or complete based on our reviews of agency data and interviews with EPA officials. EPA data identifying NPL sites that are on tribal property are not accurate. EPA headquarters officials told us that the SEMS data variable for identifying sites “on tribal property” may not always accurately identify whether NPL sites are located on tribal property. Because EPA officials told us that the agency’s data regarding NPL sites on tribal property may not be accurate and provided explanations for why these data are unreliable, we did not evaluate these data to determine the total number of inaccuracies. EPA officials we interviewed provided a number of reasons why the agency’s data regarding NPL sites located on tribal property may not be accurate: First, EPA officials told us that some site location information was inaccurately transposed during maintenance of the former database of record used prior to adopting SEMS, and that these errors, in some cases, carried over to SEMS. According to these officials, the transposed information resulted in some sites appearing in the incorrect geographic hemisphere (i.e., sites located in the western hemisphere appeared to be located in the eastern hemisphere in the incorrectly transposed data). These officials told us that they have worked over the past year to correct these errors and to verify the accuracy of site coordinates. Second, EPA officials told us that accurately documenting which sites are on tribal property can be complicated due to difficulties identifying tribal property boundaries and evolving site boundaries. For example, tribal property boundaries may be difficult to establish without reviewing land titles and other documents. Further, EPA officials told us they use the best available data to identify tribal property but there are limitations in that data. In addition, EPA officials we interviewed told us that site boundaries can be difficult to define or change over time. For example, an agency official told us NPL sites may not have clearly delineated boundaries until after the remedial investigation is complete and the full extent of contamination has been determined. Further, the official said that site boundaries may change during the cleanup process or during post-cleanup reviews if EPA discovers new or more widespread contamination. According to EPA headquarters officials, EPA regional officials are responsible for tracking changes to site boundaries in their respective regions, but specific information on the location of site boundaries is not documented in SEMS. Additionally, for one site—the Tar Creek site in Oklahoma (Region 6)—EPA’s publicly-available information states that there are no clear site boundaries. One EPA regional official we interviewed told us that he was not aware of guidance for regions regarding changing tribal property information in circumstances in which site boundaries change to include land that is tribal property. Additionally, EPA officials told us that regional offices may be inconsistent in how they determine site boundaries. EPA released recommended practices for collecting geospatial data for Superfund sites in 2017 that included guidance for determining and documenting NPL site boundaries. Further, in May 2018, EPA provided national standards intended to provide a uniform method for collecting, documenting, and managing geospatial information for Superfund sites, including information identifying site boundaries. Third, EPA headquarters officials stated that EPA checks the accuracy of these data infrequently. Headquarters officials told us there are several standardized automated reports that officials at the headquarters and regional levels can use to review SEMS data and identify quality issues, including quality issues in the variables for NAI and the associated tribes. However, these reports do not contain the on tribal property variable, and SEMS currently does not have the ability to run automated checks of site proximity to tribal property based on location data. Officials told us that they review the on tribal property data periodically outside of these reports; however, EPA currently lacks a regular review process for these data. Under federal standards for internal control, management should use quality information to achieve the entity’s objectives. Quality information is appropriate, current, complete, accurate, accessible, and provided on a timely basis. In addition, under federal standards for internal control, management should design control activities to achieve objectives and respond to risks, such as by conducting reviews at the functional or activity level. According to EPA officials, data identifying NPL sites that are on tribal property may not be accurate for a number of reasons. Because SEMS automated reports do not contain the on tribal property variable, EPA regions cannot regularly conduct quality reviews of information in SEMS on tribal property using those reports. Without a regular review process to ensure the quality of SEMS data identifying sites on tribal property and the ability to use automated reports to check the accuracy of on tribal property data in SEMS, EPA does not have reasonable assurance that regional officials have accurately identified sites on tribal property. EPA Data Identifying Sites as Having NAI Are Not Accurate or Complete EPA data identifying which sites have NAI are inaccurate and incomplete, based on our reviews of the data. We found three types of errors in these data. First, we found that SEMS did not include some sites with known tribal interest as having NAI. Second, we found some sites that EPA identified in SEMS as having NAI when there was no tribal interest. Third, we found that EPA regional officials inconsistently used the NAI variable in SEMS when there was no longer tribal interest in a site. SEMS does not include some NPL sites with known tribal interests as having NAI. We found nine sites with tribal interest that EPA did not identify as having NAI in SEMS. For six of these sites, EPA regional officials told us that they knew the sites were of interest to one or more tribes, even though they were not identified as having NAI in SEMS. For example, we found that EPA Region 10 had invited the Cow Creek Band of Umpqua Tribe of Indians to consult regarding the Black Butte Mine site, but the site was not identified as having NAI in SEMS. For two additional sites, following our request to review the SEMS data, officials from Region 4 contacted tribal officials in their region to inquire about their potential interest in NPL sites and found that the Eastern Band of Cherokee Indians had interest in two sites in North Carolina not previously identified as having NAI: Barber Orchard and Benefield Industries. EPA designated both sites as ready for their intended use—meaning that construction of the remedy had been completed—in 2011 and 2014, respectively. For the remaining site, EPA officials in Region 5 stated that they learned of tribal interest in the Petoskey Manufacturing Company Groundwater site when the Little Traverse Bay Bands of Odawa Indians contacted them in December 2017, after coverage of the site’s contamination hazards on the local news. SEMS incorrectly includes some sites as having NAI when no tribal interest exists. When responding to our request to verify the accuracy of data in SEMS, EPA regional officials identified 10 sites that were incorrectly included in SEMS as having NAI when there was no actual tribal interest. For example, officials from Region 4 stated that they removed the NAI designation from three sites because the sites are situated more than 100 miles from the nearest federally recognized tribe’s property and the officials were not aware of any tribal interest in the sites. Similarly, EPA regional officials determined that two other sites—Eielson Air Force Base in Region 10 and Seneca Army Depot in Region 2—were incorrectly identified as having NAI. These officials told us that these sites should not have been designated as NAI because no tribes had expressed interest in either site. EPA inconsistently identified sites with prior NAI in SEMS. We found that EPA regional officials inconsistently used the NAI variable in SEMS when tribes were no longer interested in a site. For example, Region 2 officials stated that they maintained the NAI designation for the Hooker Hyde Park site in order to preserve the historical record after EPA identified that the Seneca Nation of Indians no longer had an interest in the site. Conversely, Region 8 officials indicated that they removed the NAI designation for the Arsenic Trioxide site when it was determined that the relevant tribe no longer had interest in the site. Based on our review of EPA guidance and data provided by EPA officials, we identified several possible reasons that the agency’s data for identifying tribal interests are not accurate or complete. One possible reason that NAI data in EPA’s SEMS may be inaccurate and incomplete is because EPA’s guidance for making NAI determinations is unclear, resulting in EPA regional officials inconsistently determining and documenting sites with NAI. EPA’s Superfund Program Implementation Manual, which provides guidance to EPA regional officials for identifying sites as having NAI, contains one sentence regarding how EPA regional officials are to determine when to designate a site as having NAI. The manual states that EPA regional officials should designate NAI in SEMS when a site “may be of interest to tribes whose members or land are directly affected” by the release of hazardous materials from the site, but the manual does not specify criteria EPA regional officials should consider for determining what constitutes NAI. For example, the manual does not specify whether ancestral lands, areas where tribes have treaty rights, or areas otherwise of interest to a tribe but that are not tribal property should be considered in making this determination. It also does not specify what types of tribal interests to consider. However, officials from tribes we interviewed for our case studies told us that tribal interests in NPL sites may be related to a variety of factors, including contamination potentially affecting tribal members living in or around the contaminated area or land where the tribe has treaty hunting or fishing rights. Furthermore, EPA’s Superfund Program Implementation Manual does not specify whether officials should remove the NAI designation if officials determine tribes no longer have interest in a site. In the case of the Petoskey Manufacturing Company Groundwater site in Michigan, EPA Region 5 officials we interviewed told us that they were uncertain as to whether they should identify the site as having NAI, because they were unsure if the level of the tribe’s interest was significant enough. EPA officials we interviewed provided additional reasons for the lack of accuracy and completeness in the agency’s data regarding sites with NAI. EPA headquarters officials told us they periodically, but infrequently, review SEMS data on Superfund sites identified as having NAI. In addition, EPA officials told us that, in some cases, they did not identify sites as having NAI where there was tribal interest or incorrectly identified sites as having NAI when no tribal interests were involved due to errors. Additionally, some regional officials expressed that identifying NAI can be complicated by the fact that tribes may have interest in sites not located near their current property due to historical interest or treaty rights. Under federal standards for internal control, management should design control activities to achieve objectives and respond to risks, such as by clearly documenting internal control in management directives, administrative policies, or operating manuals. Although EPA has documented guidance, it is not clear about how EPA officials should make determinations about designating sites as having NAI. Without clear guidance to regional offices on how to determine whether sites have NAI—including criteria to assist regions in determining when a site should be designated as having NAI in the SEMS database and how, if at all, to adjust the NAI data for sites that no longer have tribal interest—EPA does not have reasonable assurance that its data on tribes that may be affected by hazardous releases at NPL sites are accurate or complete. EPA Data on Tribes with Interest in Sites That Have NAI Are Not Accurate or Complete EPA data do not accurately or completely identify the tribes that have interest in the sites that EPA identified as having NAI. Specifically, through reviewing EPA’s data with officials in each region, we found examples of sites that EPA indicated as having NAI but that (1) did not identify any tribes with an interest in the sites, (2) did not identify all tribes with an interest in the sites, and (3) incorrectly identified tribes associated with a site. SEMS does not include tribes for all sites. We found eight sites with NAI for which EPA did not identify an interested tribe in SEMS. For these eight sites, EPA officials added the tribes’ names prior to sending us the data. SEMS does not include all tribes that have an interest in some sites. We identified eight sites for which EPA did not identify in SEMS all the tribes that had interest in the site. For example, for the Smurfit Stone Mill Frenchtown site in Missoula, Montana, EPA data listed the Confederated Salish and Kootenai Tribes of the Flathead Reservation as having an interest in the site. However, after speaking with EPA Region 8 officials, we learned that the Kalispel Indian Community of the Kalispel Reservation also has an interest in the site but could not be included in SEMS because the tribe resides in the state of Washington, and the site is located in Montana. In providing technical comments on a draft of this report, EPA identified a ninth site, the St. Louis River site, for which an additional tribe should be added to the data in SEMS. SEMS incorrectly identified an interested tribe associated with one site determined to have NAI. For the Velsicol Chemical Corporation site in Michigan, EPA identified in SEMS the interested tribe as the Sault Ste. Marie Tribe of Chippewa Indians, when the actual interested tribe was the Saginaw Chippewa Indian Tribe of Michigan. Additionally, in providing technical comments on a draft of our report, EPA also made corrections to the tribes originally listed for the Tar Lake site and clarified the tribe originally listed for the St. Louis River site. EPA officials we interviewed told us that a possible reason for the inaccuracies in the data regarding the tribe or tribes interested in NPL sites that have NAI is that, until recently, regional officials could not enter the names of additional tribes to a SEMS site record that was created in the agency’s previous database of record. In addition, officials from two EPA regions told us that they could not record tribes as having an interest in a site when the tribe is headquartered in a state other than the state address on file for the site. EPA headquarters officials told us they submitted a request in August of 2017 to have the issue resolved and that, as of April 2018, the issue had been corrected and that regions can now add additional tribes, or tribes from other states outside of the state where the site is headquartered. Officials told us that prior to the correction in SEMS, officials at the headquarters level could manually enter data to record the names of additional tribes with NAI in a site or identify tribes interested in a site that reside in states other than the state in which the site is located. EPA Does Not Have Reliable Data about the Agency’s Consultation with Tribes Regarding NPL Sites EPA does not have reliable data on the agency’s consultation with tribes regarding NPL sites. Additionally, based on our analysis of EPA data and related documentation, as well as discussions with officials from EPA and Indian tribes, we found that EPA officials more frequently coordinated informally with tribes than conducted consultation. EPA Does Not Have Reliable Data on Consultation with Tribes Regarding NPL Sites EPA does not have reliable data on the NPL sites at which it has conducted tribal consultation. According to data in TCOTS, consultation had occurred or was projected to occur at 18 sites since EPA’s consultation and coordination policy went into effect in 2011. However, TCOTS data are incomplete and did not include records for 7 NPL sites where, based on our interviews with EPA regional officials and a review of agency documents, we determined that consultation had occurred since 2011. One possible reason that EPA data on consultation with tribes are incomplete is that the agency’s guidance regarding what constitutes consultation, and therefore is to be recorded in TCOTS, is unclear. EPA officials told us they consider consultation a specific, formal interaction that involves government-to-government interaction between tribal governments and senior EPA officials, such as Regional Administrators, and generally happens at major decision points or at the request of a tribe. Several EPA officials we interviewed clarified that the majority of day-to-day interaction with tribes do not require consultation and are less formal coordination efforts. EPA’s 2011 consultation policy provides a broad definition of consultation and makes specified program and regional officials responsible for determining when consultation may be appropriate, but the policy does not provide specific criteria for regions to use to determine if consultation with a tribe should be considered. The policy initially states that it is EPA’s policy to “consult on a government-to- government basis with federally recognized tribal governments when EPA actions or decisions may affect tribal interests.” According to the policy, the broad scope of consultation contemplated by the policy creates “a large number of actions that may be appropriate for consultation.” To provide “a general framework from which to begin the determination of whether any particular action or decision is appropriate for consultation,” the policy provides a list of general EPA activity categories, including Superfund response actions. However, the policy does not provide any further guidance on the circumstances under which consultation should be considered. For example, it does not specify any particular points in the Superfund process at which consultation should be considered or any further detail on what tribal interests should be considered when determining if tribal interests are affected. Under federal standards for internal control, management should design control activities to achieve objectives and respond to risks, such as by clearly documenting internal control in management directives, administrative policies, or operating manuals. Although EPA has documented guidance about consulting with tribes, it does not provide clear direction to regions about the circumstances under which the agency should consider consulting with tribes during the Superfund process. Without clarifying guidance on tribal consultation to clearly identify the circumstances under which the agency should consider consulting with tribes, EPA does not have reasonable assurance that regions are applying the consultation policy consistently and uniformly. In addition, EPA regional officials do not consistently document invitations to consult with tribes in TCOTS, which could result in incomplete or inaccurate data on EPA consultation with tribes. EPA headquarters officials told us that invitations to consult should be entered in TCOTS, because the database has a specific field for this information. Officials we interviewed from EPA Regions 6 and 10, the two regional offices that combined manage nearly half of Superfund sites that EPA identified as having NAI, told us that they do not document all invitations to consult in TCOTS. Specifically, an official we interviewed from Region 6 told us that consultation invitations that were not made in writing are generally not entered into TCOTS, and an official from Region 10 told us that officials in the region would not document invitations to consult that did not lead to actual consultation. In providing technical comments on our draft report, EPA noted that Region 10 now documents all invitations to consult with tribes in the TCOTS database. Although EPA headquarters officials told us that invitations to consult should be entered in TCOTS, agency guidance does not direct officials to do so. EPA has developed guidance on key points in the Superfund process at which regional officials should document consultation if it occurs, but this guidance does not direct regional officials to document invitations to consult in TCOTS. Moreover, officials we interviewed from 6 of EPA’s 10 regional offices were unaware of this guidance. An EPA headquarters official we interviewed told us that EPA regional officials may be unaware of this guidance because EPA has not conducted annual training regarding documenting tribal consultation and has decided to offer the training on an as-needed basis. This guidance identifies five decision points in the Superfund process at which EPA regional officials should, at a minimum, document any associated consultation with tribes in TCOTS, outlined in figure 1 below. Under federal standards for internal control, management should design control activities to achieve objectives and respond to risks, such as by clearly documenting internal control in management directives, administrative policies, or operating manuals. By developing or revising guidance to clearly direct regional officials to document all invitations to consult with tribes in the TCOTS database and providing the guidance to those officials, EPA would have greater assurance that its regional offices are accurately and consistently documenting invitations to consult and that the data that EPA provides to OMB regarding agency consultations with tribes are accurate and complete. Consultation Is Relatively Infrequent Compared to Coordination Based on our analysis of EPA data and documentation, as well as interviews with EPA and tribal officials, we found that EPA more frequently coordinated informally with tribes regarding cleanup decisions at NPL sites than conducted consultation with tribes. Consultation between EPA and tribes, as defined in EPA’s 2011 tribal consultation policy, is relatively infrequent compared to less-formal coordination efforts. For example, officials from the Kalispel Indian Community told us that consultation is reserved for instances in which regular communication and coordination is not working. Additionally, EPA officials in Region 8 told us that most of their day-to-day interactions with tribes are considered coordination, and that consultation only occurs at key decision points in the Superfund process. Most EPA regional officials we interviewed as part of our case studies stated that consultation was relatively infrequent. At the same time, these officials stated that they frequently coordinate with tribes during the Superfund cleanup process. Additionally, EPA’s policy says that tribal officials may request consultation with the agency. Tribal officials we interviewed as part of our case studies expressed varying levels of satisfaction with EPA’s coordination and consultation efforts, as well as with EPA’s cleanup decisions overall. In the case of the General Motors Central Foundry site in Massena, New York, officials we interviewed from the Saint Regis Mohawk Tribe told us that they were dissatisfied with the consultation and the remedy at the General Motors Central Foundry site. Specifically, tribal officials stated that they were dissatisfied with EPA’s decision to install a permanent cap over an industrial landfill at the site, rather than removing all of the waste, to address the contamination at the site. Officials from the tribe told us that they felt EPA was disregarding the tribe’s health and safety concerns at the site. EPA acknowledged in its amended record of decision for the site that the tribe only partially agreed with the remedy; however, EPA notes that they took some steps to revise the remedy to address the tribe’s concerns. For example, the amended record of decision was created in part, due to tribal opposition, and includes a contingency remedy that expands the scope of the amended decision to include removal of contaminated soil located on the tribe’s property rather than on-site treatment. In other cases, officials of some tribes told us that the working relationship with their local EPA region was good and that coordination had been effective. For example, officials from the Pueblo of Laguna reported that communication and coordination with EPA region 6 regarding the cleanup of the Jackpile-Paguate Superfund site in Laguna Pueblo, New Mexico, was effective, and that the EPA remedial project manager for the site had been responsive to the tribe’s needs. EPA Has Taken Various Actions to Address Unique Tribal Needs When Making NPL Site Cleanup Decisions EPA has taken various actions to address the unique needs of tribes when making cleanup decisions at NPL sites. These actions include efforts to minimize tribal members’ exposure to contaminants and limit potential damage to tribal archeological sites. For example: EPA Regions 1 and 10 took steps to protect tribal cultural resources at NPL sites. EPA officials we interviewed from Region 1 told us that at one site, regional officials rerouted and improved roads used to remove contaminated materials to minimize the impact of cleanup activities’ on historically significant cultural resources. In addition, EPA officials we interviewed from Region 10 told us that they coordinated with tribal cultural resource program officials to ensure that tribal officials were present during excavation work at the Midnite Mine site in Wellpinit, Washington, to observe and ensure that EPA was taking appropriate measures to protect sites that are culturally important to the tribe. EPA Region 2 officials revised risk assessments at an NPL site. Because of concerns about the potential health impacts to the Saint Regis Mohawk Tribe, EPA Region 2 officials revised the risk assessment for a site with polychlorinated biphenyl contamination to more accurately reflect the typical exposure of tribal members. EPA’s revised hazard exposure assessment for the General Motors Central Foundry site assumed a higher rate of exposure to contaminants for tribal members, given that they, on average, live on the reservation longer than an adult non-tribal member may live in the same place for most of his or her life. Specifically, EPA’s exposure estimate was based on an exposure duration of 64 years for an adult tribal member and an exposure duration of 30 years for adult non-tribal member. EPA Region 9 incorporated tribal information into risk assessments for some NPL sites. EPA officials we interviewed from EPA’s Region 9 office told us about several sites where they had considered tribal members’ heightened exposure to contamination. For example, at one site, officials told us they worked closely with tribal officials to gather data on tribal members’ uses of vegetation and tribal game consumption. These EPA officials stated that they used these data to develop risk assessment plans that were sensitive to unique tribal needs. EPA officials we interviewed also provided examples of the use of traditional ecological knowledge at some NPL sites. Traditional ecological knowledge sometimes represents unique tribal needs. For example, EPA officials we interviewed described instances in which a tribe provided EPA with selected information about their traditional hunting sites and their traditional use of plants, and EPA was able to use this information when developing risk assessments and standards for safe consumption of fish and wildlife. For example, officials in EPA Region 9 told us that a tribe shared information with them about how tribal members hold reeds in their mouths as part of traditional basket making practices. These officials reported that after learning of the tribe’s use of such reeds, the agency considered this information when determining how to evaluate contamination in the area where the reeds grow. EPA and tribal officials told us that, for confidentiality reasons, some tribes may be reluctant to share some traditional ecological knowledge; however, headquarters and EPA regional officials told us that this was relatively infrequent and that, in these situations, EPA was able to work with the tribe to find ways to use more general information to inform decisions regarding Superfund cleanups. Conclusions EPA has policies and procedures for consulting with tribes when its actions and decisions at NPL Superfund sites may affect tribal interests. To carry out these policies and procedures, EPA must be able to identify when its actions and decisions may affect a tribe. The agency has developed two systems—SEMS and TCOTS—that it uses to identify and track sites that are on tribal property or that affect tribes, and the agency’s efforts to consult with affected tribes, respectively. However, based on our analysis of some of the data in these systems, these data are not reliable. Data on sites that are on tribal property are not accurate, and there is no regular, standardized review process officials can use to review the quality of these data. Without developing such a review process, EPA will not have reasonable assurance that regional officials have accurately identified the sites that are on tribal property. Additionally, data on sites that have NAI are not accurate or complete due, in part, to unclear guidance for how regions should determine whether a site has NAI. Clarifying guidance to regional offices on how to determine whether sites have NAI can help provide EPA reasonable assurance that its data on tribes that are directly affected by hazardous releases at NPL sites are accurate and complete. Moreover, we found that the data tracking consultation with tribes at NPL sites were unreliable, and may not contain all invitations to consult. Clarifying guidance to clearly identify the circumstances under which the agency should consider consulting with tribes could improve the quality of EPA’s data on consultation, and could help ensure EPA regions are applying the consultation policy consistently and uniformly. In addition, explicitly directing regional officials to document all invitations to consult with tribes, regardless of whether further consultation results after the invitation, would provide EPA greater assurance that its regional offices are accurately and consistently documenting invitations to consult, and that the data that EPA provides to OMB regarding tribal consultations are accurate and complete. Recommendations for Executive Action We are making the following four recommendations to EPA: The Director of EPA’s Office of Superfund Remediation and Technology Innovation should develop a regular review process to ensure the quality of SEMS data identifying NPL sites on tribal property and revise automated reports used to check the accuracy of SEMS data to include on tribal property data. (Recommendation 1) The Assistant Administrator of EPA’s Office of Land and Emergency Management should clarify guidance to regional offices on how to determine whether sites have NAI, including by adding criteria for when a site should be designated as having NAI in the SEMS database and how, if at all, to adjust SEMS data if a tribe is no longer interested in a site. (Recommendation 2) The Director of EPA’s Office of Superfund Remediation and Technology Innovation should clarify agency guidance regarding tribal consultation for the Superfund program to clearly identify the circumstances under which the agency should consider consulting with tribes. (Recommendation 3) The Assistant Administrator of EPA’s Office of International and Tribal Affairs should develop or revise existing guidance to clearly direct regional officials to document all invitations to consult with tribes in the TCOTS database and provide the guidance to those officials. (Recommendation 4) Agency Comments and Third-Party Views We provided a copy of this report to EPA, the Confederated Salish and Kootenai Tribes of the Flathead Reservation, the Kalispel Indian Community of the Kalispel Reservation, the Little Traverse Bay Bands of Odawa Indians, the Mashpee Wampanoag Tribe, the Pueblo of Laguna, the Saint Regis Mohawk Tribe, the Spokane Tribe of the Spokane Reservation, and the Wampanoag Tribe of Gay Head (Aquinnah) for review and comment. EPA generally agreed with our recommendations, and their comments are reproduced in appendix IV. EPA also provided technical comments, which we incorporated as appropriate. The Confederated Salish and Kootenai Tribes of the Flathead Reservation and the Pueblo of Laguna also provided written comments (reproduced in appendixes V and VI) and technical comments, which we incorporated as appropriate. The Kalispel Indian Community of the Kalispel Reservation, the Little Traverse Bay Bands of Odawa Indians, the Mashpee Wampanoag Tribe, the Saint Regis Mohawk Tribe, the Spokane Tribe of the Spokane Reservation, and the Wampanoag Tribe of Gay Head (Aquinnah) did not comment on our report. EPA concurred with our recommendation to develop a regular review process to ensure the quality of SEMS data identifying NPL sites on tribal property and revise automated reports used to check the accuracy of these data. EPA stated that during the course of our work on this report, SEMS tribal data was reviewed for quality control and corrections were made to the existing data. In addition, EPA’s Office of Superfund Remediation and Technology Innovation plans to create a schedule to review tribal data in SEMS and disseminate tribal data to Superfund regional coordinators annually for their quality assurance review starting in March 2019. EPA generally agreed with our recommendation to clarify guidance to regional offices on how to determine whether sites have NAI, including by adding criteria for when a site should be designated as having NAI in SEMS and how, if at all, to adjust SEMS data if a tribe is no longer interested in a site. EPA noted that there are a variety of circumstances under which a tribe may have interest in a site, and the agency plans to identify relevant criteria in the Superfund Program Implementation Manual that may be used to support the decision of whether or not to apply the NAI indicator. Additionally, the agency plans to create a headquarters and regional workgroup to review and update tribal data collected in SEMS. The workgroup will provide guidance to clarify the NAI determination, including identifying criteria for designating a site NAI, and identifying a process to update SEMS when a tribe is no longer interested in a site, as needed. EPA plans to complete this no later than October 2019. EPA concurred with our recommendation to clarify agency guidance regarding tribal consultation on Superfund sites to clearly identify the circumstances under which the agency should consider consulting tribes. In its letter, EPA pointed out that our original recommendation did not specify that the recommendation was about guidance regarding tribal consultation on Superfund sites, so we adjusted the language of the recommendation accordingly. EPA plans to issue a memo to the regions that clarifies circumstances under which regions may consider tribal consultation for the Superfund program no later than March 2020. EPA concurred with our recommendation that it should develop or revise existing guidance to clearly direct regional officials to document all invitations to consult with tribes in the TCOTS database and provide the guidance to those officials. EPA is planning four actions to respond to this recommendation: (1) issuing a memorandum from the Office of International and Tribal Affairs to EPA Regional Administrators on the importance of following EPA’s Tribal Consultation and Coordination Policy and documenting consultation actions into TCOTS, estimated to occur in January 2019; (2) issuing a monthly TCOTS report to Deputy Assistant Administrators and Regional Assistant Administrators on the status of consultations recorded in TCOTS, starting in January 2019; (3) initiating trainings specifically targeted to EPA's Regional Superfund staff on when and how to document consultation actions in TCOTS, estimated to begin in February or March 2019; and (4) conducting training on tribal consultation topics, with a specific emphasis on entering consultation information into TCOTS, beginning in March or April 2019. In their comments on our report, the Confederated Salish and Kootenai Tribes of the Flathead Reservation noted that our report is thorough and provides valuable insight into EPA’s policies and procedures for tribal consultation at NPL sites. The tribe provided some additional detail on the Smurfit Stone Mill Frenchtown case study which we incorporated as appropriate. The tribe also noted that they had interest in a site not identified by EPA as having NAI, the Anaconda Aluminum Co. Columbia Falls Reduction Plant site. In response, we added this site to our list of NPL sites known to be on or affecting tribal land, shown in appendix I. The Pueblo of Laguna commented that while the scope of the report was limited, the Pueblo appreciated GAO’s efforts to study EPA’s tribal consultation practices. The Pueblo emphasized their belief that EPA’s duty to consult with tribes should be an active one, not a passive one, and presented three associated comments. First, the Pueblo believes EPA should affirmatively consider offering consultation at each stage of the Superfund process beginning with preliminary investigation and site assessment. Second, the Pueblo believes EPA should continue to contact potentially interested tribes throughout the life of an NPL site, even if the tribe had not expressed interest at a previous stage of the process to ensure that newly interested tribes are identified. Finally, the Pueblo believes EPA should document all offers to consult, including ones made orally. The Pueblo provided comments and edits on the Jackpile-Paguate Mine case study in their letter, which we incorporated. The Pueblo also provided technical comments on the report, which we incorporated as appropriate. As agreed with your offices, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies of this report to the appropriate congressional committees, the Administrator of the Environmental Protection Agency, the Chairman of the Confederated Salish and Kootenai Tribes of the Flathead Reservation, the Chairman of the Kalispel Indian Community of the Kalispel Reservation, the Chairman of the Little Traverse Bay Bands of Odawa Indians, the Chairman of the Mashpee Wampanoag Tribe, the Governor of the Pueblo of Laguna, the Chiefs of the Saint Regis Mohawk Tribe, the Chairwoman of the Spokane Tribe of the Spokane Reservation, the Chairwoman of the Wampanoag Tribe of Gay Head (Aquinnah), and other interested parties. In addition, the report will be available at no charge on the GAO website at http://www.gao.gov. If you or your staff members have any questions about this report, please contact me at (202) 512-3841 or gomezj@gao.gov. Contact points for our Office of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to the report are listed in appendix VII. Appendix I: Site-wide Cleanup Status of National Priorities List Sites with Known Native American Interest This appendix provides information on the site-wide cleanup status of National Priorities List (NPL) sites with known Native American Interest (NAI), as of December 2017. We worked with the Environmental Protection Agency (EPA) to correct inaccuracies in the Superfund Enterprise Management System (SEMS) data identifying sites as having NAI, and we identified 87 NPL sites—74 sites on the NPL, 8 deleted from the NPL, and 5 proposed for addition—known to have NAI. In addition, in providing technical comments on the draft of this report, the Confederated Salish and Kootenai Tribes of the Flathead Reservation identified one additional site, bringing the total to 88 NPL sites known to have NAI. Of these 88 sites known to have NAI out of the total 1,785 NPL sites that were proposed, final, or deleted as of December 2017, many have reached site-wide milestones that EPA uses to track the cleanup status of NPL sites. EPA measures four site-wide milestones, including one that measures the progress in the Superfund process and three that describe the management of the release, such as human exposure under control: 1. Construction completion. Indicates that the physical construction of the remedy EPA has selected to address the contamination is complete. 2. Human exposure under control. Measures the incremental progress EPA achieved in controlling unacceptable exposures to people at a site. A site may achieve this measure by reducing the level of contamination, preventing people from contacting the contaminants in-place, or controlling activities near the site (e.g., by reducing the potential frequency or duration of exposure of people to contaminants). 3. Groundwater migration under control. Assesses whether groundwater contamination is below protective, risk-based levels or, if not, whether the migration of contaminated groundwater is stabilized and there is not unacceptable discharge to surface water and monitoring will be conducted to confirm that affected groundwater remains in the original area of contamination. EPA only uses this in sites with known past or present groundwater contamination. 4. Site-Wide Ready for Anticipated Use. All cleanup goals that may affect current and reasonably anticipated future land uses of the site have been achieved, so that there are no unacceptable risks and all institutional or other controls have been put in place. Table 1 below shows the site-wide cleanup status, according to EPA, of the 83 sites on or deleted from the NPL with known NAI. This table provides data on site-wide milestones obtained from EPA’s SEMS database, as well as a brief overview of each site using information from publicly available EPA documents, the EPA website, and additional information provided by EPA officials. Table 2 below lists the 5 sites with known NAI that EPA has proposed for the NPL. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The Salt Chuck Mine site is an inactive former gold, silver, copper, and palladium mine on Prince of Wales Island in southeast Alaska. Operations at the site were suspended in 1941. The site includes abandoned mine workings and mine mill equipment. Contaminants include polychlorinated biphenyls (PCBs), copper, lead, and arsenic. In 2011, EPA started a remedial investigation of the upland and adjacent marine areas to evaluate potential risk to human health and the environment. The investigation was completed in March 2018, and EPA determined that there are currently no unacceptable human health risks identified for the site and that ecological risks are limited to copper in marine sediment in areas used for tailings disposal. The Tucson International Airport Area site comprises a 10-square-mile area in and next to Tucson, Arizona. The site includes the Tucson International Airport, portions of the Tohono O'Odham Indian Reservation, residential areas of Tucson and South Tucson, and the Air Force Plant #44 Raytheon Missile Systems Company. Former aircraft and electronics manufacturing activities, fire drill training activities, and unlined landfills have contaminated groundwater and soil with volatile organic compounds, metals and PCBs. Remedial activities include: groundwater pumping and treatment, soil removal, and soil vapor extraction. Groundwater cleanup actions, operation and maintenance activities, and site monitoring are ongoing. As of July 2018, EPA reports that water treatment systems have significantly reduced the groundwater plume size and chemical concentrations in groundwater. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The 4,400-acre Iron Mountain Mine site near Redding, California produced iron, silver, gold, copper, zinc and pyrite. Though mining operations were discontinued, underground mine workings, waste rock dumps, piles of mine tailings, and an open mine pit remain at the site. Much of the acidic mine drainage is channeled into the Spring Creek Reservoir. About 70,000 people use surface water within 3 miles of the mine as their source of drinking water. The installation and operation of a full- scale neutralization system, capping of areas of the mine, and the construction and operation of a retention reservoir to collect contaminated runoff for treatment have significantly reduced acid and metal contamination in surface water at the site. Site investigations and cleanup are ongoing. The 3.2-acre Celtor Chemical Works site, located on the Hoopa Valley Indian Reservation, is the location of a former ore concentrating facility that processed sulfide ore. Wastes from the operations and processed ore generated acidic runoff and elevated metal concentrations in the soils throughout the site. The Trinity River flows along the site boundary and is the only local fish source for the Hoopa Indians. Cleanup included off- site disposal of contaminated materials; backfilling and contouring land; and revegetation and diversion of springs away from contaminated areas. After cleanup, EPA took the site off the NPL in 2003. According to EPA officials, in 2016, additional waste was discovered at the site, resulting in additional remedial investigation to determine the nature and extent of contamination. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The Leviathan Mine is an abandoned open-pit mine near Markleeville, California, on the eastern slope of the Sierra Nevada Mountains at an elevation of 7,000 feet. The site is drained by Leviathan and Aspen Creeks, which are tributaries to the East Fork of the Carson River, a major western Nevada water supply source. The mine operated intermittently between 1863 and 1962. In the early days of mining, copper sulfate was mined from the property and utilized for processing silver ore at the Comstock Mines in Virginia City, Nevada. According to EPA officials, mine operations were originally underground, but surface mining of sulfur ore began in the 1950s. These officials told us that, mining operations disturbed and exposed existing mineral-rich rock and soil, which produced residual mine waste rock. Surface runoff from snowmelt and precipitation become contaminated by contact with the mineral-rich rock and associated waste rock. Officials told us that water capture and treatment plants at the site have improved the quality of downstream surface water and watershed health. These officials also noted that site assessment and cleanup is ongoing. The 150-acre Sulphur Bank Mercury Mine site near Clearlake Oaks, California, is an abandoned open pit mercury mine located on the shoreline of Clear Lake. This mine operated intermittently between 1865 and 1957 and mined sulphur and mercury. Former mining activities at the site contaminated soils, sediment, and surface water with mercury and arsenic. Approximately 2 million cubic yards of mine wastes and tailings remain on the mine site. Mercury contaminates lake sediment and is bio-concentrated in the food web of Clear Lake. The levels of mercury in fish from the lake led the State to issue an advisory to limit consumption of local fish. Clear Lake is also a drinking water source for 4,700 people. Cleanup has included erosion control, soil removal from residential yards, and surface water diversion. After immediate actions to protect human health and the environment, site investigations and long- term cleanup planning are ongoing. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The Bonita Peak Mining District site consists of 48 historic mines or mining-related sources of contamination in unincorporated parts of Colorado. Historic mining operations have contaminated soil, groundwater, and surface water with heavy metals. Additionally, ongoing releases of metal-contaminated water and sediment are occurring within the Mineral Creek, Cement Creek, and Upper Animas River drainages in San Juan County, Colorado. EPA and other stakeholders conducted a remedial investigation and feasibility study in 2017. Ongoing cleanup activity includes an interim water treatment plant to treat acid mine drainage and management of non-hazardous sludge. EPA plans to use the remedial investigation to determine further cleanup options at the site. The 890-square-mile Idaho National Engineering Laboratory site is located near Idaho Falls, Idaho. The site consists of a number of major facilities that contribute contaminants to and draw water from the Snake River Plain Aquifer. One of these facilities is a National Reactor Testing Station built by the Atomic Energy Commission in 1949 to build, test, and operate various nuclear reactors, fuel processing plants, and support facilities. Site activities also led to the discharge of liquid wastes to several unlined ponds and an earthen ditch. The site includes contaminated soil, sludge, and groundwater that contain hazardous chemicals, heavy metals, and radioactive constituents. The site is divided into several cleanup areas to better address site cleanup. Remedy construction has been completed in several of these areas, and remedial design and construction are underway at the remaining areas. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) Also known as the Coeur d’Alene Basin Cleanup, the Bunker Hill Mining and Metallurgical Complex site is located in northern Idaho and eastern Washington, in one of the largest historical mining districts in the world. The site spans 1,500 square miles and includes 166 miles of rivers. Mining operations began in the area in 1883 and continue today. Historical mining and milling methods led to disposal of tailings in rivers and streams, which resulted in the spread of contaminants throughout the floodplain of the South Fork Coeur d’Alene River. Smelter operations also resulted in emissions and piles of waste rock. Soil, sediment, groundwater, and surface water are contaminated with heavy metals such as lead, which pose serious risks to people and the environment. Since 1983, EPA and its partners have made progress in cleaning up contamination, including cleaning some mine and mill sites, and establishing waste repositories to securely contain contaminated soil to reduce impacts to people and the environment. Site remediation is ongoing. The 2,530-acre Eastern Michaud Flats Contamination site near Pocatello, Idaho, consists of two phosphate ore processing facilities that began operations in the 1940s. One facility continues to produce solid and liquid fertilizers using phosphate ore, sulfur, air, and natural gas. The other produced elemental phosphorus for use in a variety of products from cleaning compounds to foods. Cleanup at this facility is largely located within Fort Hall Indian Reservation boundaries. Operations at both plants contaminated groundwater and soil with metals including arsenic, lead, and cadmium. Cleanup includes capping contaminated soils, extraction and containment of contaminated groundwater, and groundwater monitoring. Site cleanup began in 2010 and is ongoing. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The Cherokee County Superfund site is a former mining area in southeast Kansas covering about 115 square miles. It is part of a larger regional mining area known as the Tri-State Mining District, where more than 100 years of mining for lead and zinc created piles of mine tailings covering more than 4,000 acres. The mine tailings contaminated groundwater with lead, zinc, and cadmium. Millions of cubic yards of mine tailings are present at the surface, in addition to impacted soils, surface water, sediment, and groundwater. Several cleanup activities have been completed and others are underway. Site- wide, nearly 3 million cubic yards of mining wastes have been remediated on nearly 2,000 acres, more than 700 residential yards have been remediated, and more than 500 homes have been supplied with a clean, permanent source of drinking water. Otis Air National Guard Base and Camp Edwards together form Joint Base Cape Cod, a 22,000-acre property used for military training activities since 1911. It is the sole source aquifer for 200,000 year-round and 500,000 seasonal residents of Cape Cod. Parts of the aquifer have been contaminated by fuel spills, training activities, waste disposal, and other past activities at the base. Cleanup of a portion of the site is managed by the U.S. Air Force, which is addressing the sources of and groundwater contamination primarily on Otis Air National Guard under the authority of Superfund. Contaminated areas were the result of chemical and fuel spills, fire training activities, landfills, and drainage structures. Since 1984, when contaminants were first detected in monitoring wells, numerous investigations and cleanups have been undertaken and completed. Currently, nine groundwater plumes are undergoing extraction and treatment. The Air Force’s land use control program ensures that groundwater remedies are protective until cleanup levels are met. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The Creese and Cook Tannery site is located in Danvers, Massachusetts. Leather tanning operations took place on-site from about 1903 through the 1980s. Solid tanning wastes were disposed of in two landfills at the site. Liquid waste was discharged to the Crane River until 1975 and later to sewers, while sludge waste was deposited in an on-site lagoon system. Operations led to contamination of surface and subsurface soils with tannery wastes, and contaminants, particularly arsenic, exceed state health-based standards in multiple locations. In 2012 EPA conducted a removal of contaminated surface soil and disposed of this soil off- site. EPA issued a proposed cleanup plan for the site in October 2018. The New Bedford harbor is an 18,000-acre urban estuary with sediment highly contaminated with PCBs and heavy metals. From the 1940s until EPA banned the production of PCBs in the 1970s, two manufacturing facilities improperly disposed of industrial wastes containing PCBs, contaminating the harbor bottom for about 6 miles from the Acushnet River into Buzzards Bay. After extensive testing of water quality, harbor sediment, air quality, and locally caught fish and shellfish, EPA concluded that the PCBs in the sediment posed a serious risk to human health and the environment. EPA has placed restrictions on fishing, shellfishing and lobstering in and around the harbor. EPA has addressed approximately 450,000 cubic yards of contaminated sediment in the upper harbor as of April 2017 and plans to dredge and dispose of over 200,000 cubic yards of contamination from the lower harbor. According to EPA, the site cleanup will require an additional 5 to 7 years and significant funding to finish. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The Loring Air Force Base site is located in Limestone, Maine. Loring Air Force Base was one of the first to be designed and built to accommodate high-speed aircraft, and construction ended in 1953. Activities at the site, including maintenance of jet engines, generated waste oils, recoverable fuels, spent solvents and cleaners. These wastes contaminated soil, groundwater, surface water, and sediment at a number of areas across the former base. Cleanup activities include relocation of contaminated soil, bioremediation of groundwater, and capping of disposal areas. The Air Force is leading the site cleanup until goals have been achieved. The Air Force is conducting operation and maintenance and long-term monitoring activities. The 25-acre Eastland Woolen Mill Superfund site is located in the Town of Corinna, Maine. Prior to closing in 1996, the mill manufactured dyed wool and blended woven fabric. The dyeing operation utilized various chemicals, including dyes and dye-aids that reportedly contained biphenyl and chlorinated benzene compounds. Liquid wastes were discharged to the ground beneath mill buildings until 1977. As a result, soil and bedrock underlying the mill were contaminated with chlorinated benzene compounds. Long-term cleanup and environmental monitoring are ongoing. In 2012, EPA completed a partial deletion action to remove 80% of the land area from NPL designation and facilitate reuse. EPA completed the second Five-Year Review in 2015. The Eastern Surplus site is a 5 acre area in Meddybemps, Maine. From 1946 through the early 1980s, the Eastern Surplus Company, a retailer of army surplus and salvage items, operated on the site. Facility operations contaminated soil and groundwater with hazardous chemicals, including volatile organic compounds and calcium carbide. After immediate actions to protect human health and the environment, remediation activities included excavating soils, extracting and treating contaminated groundwater, and disposing of gas cylinders. Operation and maintenance activities and monitoring are ongoing. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) Velsicol Chemical Corporation produced various chemical compounds and products at its 54-acre plant in St. Louis, Michigan, from 1936 through 1978. Products included the fire retardant polybrominated biphenyl and the pesticide DDT. To address contamination on-site, Velsicol agreed to construct a slurry wall around the former plant and put a clay cap over it. The Pine River, which borders the former main plant site on three sides, was significantly contaminated. In response, the state of Michigan issued a no-consumption advisory for all fish species. Over 670,000 cubic yards of DDT-contaminated sediment were removed and disposed of off-site in an approved landfill. DDT levels in fish have been reduced by more than 98 percent. In the early 2000s, studies showed the slurry wall and clay cap at the main plant site were failing to keep contamination out of the river. In response, EPA and Michigan's Department of Environmental Quality (MDEQ) launched a remedial investigation and feasibility study at the main plant site and concluded that soil and groundwater were contaminated. In June 2006, EPA selected a remedy that included a comprehensive cleanup of the main plant site and a residential soil cleanup. During the residential cleanup, EPA excavated and disposed of 50,000 tons of contaminated soil at an off-site landfill. Currently, EPA and MDEQ are completing a remedial investigation in the Pine River downstream of the former chemical plant property. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The Allied Paper, Incorporated/Portage Creek/Kalamazoo River site affects Kalamazoo, Michigan, 80 miles of the Kalamazoo River (from Morrow Dam to Lake Michigan), and 3-mile stretch of Portage Creek. Paper mill properties, riverbanks and floodplains have been contaminated with PCBs. EPA has removed contaminated materials from the site, cleaned and restored 7 miles of the Kalamazoo River and banks and capped 82 acres worth of contaminated materials. In the portions of the site where cleanup has concluded, EPA conducts maintenance activities and monitors groundwater. For two areas contaminating the river that have not yet been cleaned up, EPA has decided on cleanup plans and has taken actions to prevent migration of contamination to the Kalamazoo River or Portage Creek. EPA has decided on cleanup plans for approximately a portion of the 80 mile stretch of the Kalamazoo River and Portage Creek that require remediation. The Petoskey Manufacturing Company, or PMC, contained a die casting plant from the 1940s and a painting operation from the mid- to late-1960s. Disposal of spent solvents and paint sludge onto the ground outside the PMC building contaminated soil and groundwater at the site with volatile organic compounds. Contaminated groundwater reached a nearby municipal well that provided drinking water to city residents. The city replaced the contaminated well with a new groundwater source. Currently, EPA and Michigan Department of Environmental Quality are evaluating the site for potential vapor intrusion issues into condominiums built on top of the former PMC source area. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) Grand Traverse Overall Supply was a commercial laundering and dry cleaning facility opened in 1953. Activities at the site between 1955 and 1968 included construction of a dry well and seepage lagoons to collect waste. In 1977 the facility began discharging waste to the sewer. A year later, the Michigan Department of Environmental Quality discovered groundwater contaminated with volatile organic compounds such as trichloroethylene and perchloroethlyene that impacted at least 10 wells, including one that supplied water to an adjacent elementary school. Contaminated wells were abandoned and new wells drilled. Waste lagoons were drained and filled with gravel, and the contaminated soils around the dry well and on-site barrels of waste sludge were removed in the 1970s. In providing technical comments on a draft of this report, EPA officials told us that remedial actions at the site began with soil removal activities around 2009, and that a groundwater pump and treat system was installed in 2012 and improved in 2015. These officials told us the site is expected to reach cleanup goals within approximately 5 years. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) Northwestern Leather Company operated a tannery on the 75-acre Cannelton Industries Incorporated site in Sault Sainte Marie, Michigan from 1900 to 1958. A portion of the site is located within the 100-year floodplain of the St. Mary's River. Waste disposal operations contaminated soils, sediment and the river with heavy metals, including chromium, lead, cadmium, arsenic and mercury. EPA’s initial long-term remedy for the site included the excavation and consolidation of contaminated waste material, soils, and river sediment into an on-site landfill, collection and treatment of groundwater, groundwater monitoring, and land use restrictions for the landfilled area. In commenting on a draft of our report, EPA officials told us the remedy was amended to include excavation and removal of contaminated soil and tannery waste and other waste materials from portions of the site, Construction of these remedies took place in 1999. In 2006 and 2007, additional dredging operations removed 40,000 cubic yards of contaminated sediment, about 500,000 pounds of chromium and 25 pounds of mercury from Tannery Bay and nearby wetlands. Subsequent sampling in 2014 showed mercury or chromium in Tannery Bay and an adjacent wetland. In providing technical comments on a draft of this report, officials noted that 2016 sampling also showed mercury in Tannery Bay surface water and adjacent wetland. EPA is reviewing the current monitoring requirements and protocols, as well as the cleanup goals. The monitoring portion of the operations and maintenance plan will be revised based on EPA's findings. EPA officials told us that the agency has initiated a partial deletion of the site from the NPL to enable reuse of some remediated site areas. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The 200-acre Tar Lake site in Mancelona Township, Michigan was an iron works facility from 1882 through 1945. Disposal of tar waste contaminated soil and groundwater with hazardous chemicals, including tar waste and creosote. Cleanup activities included excavation and disposal of tar and contaminated soils, and groundwater extraction and treatment. After initial cleanup, operation and maintenance activities are ongoing. EPA has conducted several 5-year reviews of the site’s remedy. EPA did additional sampling at the site in 2011 and 2012 and identified the need for additional soil excavation and expansion of the groundwater treatment system. In providing technical comments on a draft of this report, EPA officials told us that additional cleanup will begin in 2020 and last several years. EPA has deleted part of the site from the NPL. The Torch Lake site is located on the Keweenaw Peninsula in Michigan. The site includes several areas ranging in size from about 10 acres to more than 200 acres. Copper mining activities in the area from the 1890s through 1969 produced mill tailings that contaminated lake sediment and the shoreline. Cleanup included covering 800 acres of slag piles and tailings with soil and vegetation, and long-term monitoring of Torch Lake. After cleanup, operation and maintenance activities are ongoing. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) Minnesota Chippewa Tribe, Minnesota (Grand Portage Band and Fond du Lac Band); Lac du Flambeau Band of Lake Superior Chippewa Indians; Sokaogon Chippewa Community, Wisconsin. The St. Louis River site is located at the west end of Duluth, Minnesota, and includes several areas of land next to the St. Louis River, several boat slips, and a wide section of the river known as Spirit Lake. The site overall has been divided into two smaller sites, both managed by the state of Minnesota. The first area, known as the St. Louis River/Interlake/Duluth Tar (SLRIDT) site includes 255 acres of land, boat launch ramps and bays of the St. Louis River. From the 1890s through 1962, a variety of industrial plants operated at the site, including a coking plant, and tar and chemical plants. The second site, U.S. Steel comprises 500 acres of land and 200 acres of the St. Louis River. The area was contaminated by a steel mill that operated on-site between 1916 and 1981. Operations at both sites contaminated soil and underwater sediment with hazardous chemicals, including solid wastes, PCB liquids and drums. The sites are currently in different phases of cleanup. Cleanup of the land portion of the SLRIDT was substantially completed by 2001, and cleanup of the contaminated sediment by 2010. However, in its most recent 5-year review, the Minnesota Pollution Control Agency noted several smaller areas of contaminated materials that will require additional cleanup. U.S. Steel conducted multiple cleanups at their site since the 1990s and many of the actions required by EPA’s record of decision have been completed. However, in its most recent 5-year review, the Minnesota Pollution Control Agency concluded that while some cleaned-up areas continue to be protective of human health and the environment, some areas of the site are not protective. EPA officials also told us that the U.S. Steel site has also contaminated a part of the St. Louis River known as Spirit Lake. According to these officials, the cleanup of Spirit Lake, including associated tribal consultation, is planned through a partnership led by EPA’s Great Lakes National Program Office. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) Minnesota Chippewa Tribe, Minnesota (Leech Lake Band) The 125-acre St. Regis Paper Company site is located within the external boundaries of the Leech Lake Band of Ojibwe Indian Reservation in Cass Lake, Minnesota. The wood-treatment facility operated from the 1950s through the 1980s using creosote and pentachlorophenol (PCP). The facility’s operations contaminated soil and groundwater with hazardous chemicals, including PCP, dioxin and polycyclic aromatic hydrocarbons (PAH). Remedies put in place include water treatment and soil containment. Subsequent assessment demonstrated unacceptable potential risks from groundwater and surface soil contamination. EPA proposed a cleanup plan in March 2016 to address soil contamination in residential areas. EPA has determined there are no current unacceptable human risks. The 300-square-mile Anaconda Company Smelter site is near Anaconda, Montana. Anaconda operated a large copper concentrating and smelting operation on the north side of Warm Springs Creek until about 1901. Around 1902, ore processing and smelting operations began at a separate facility that is included in the site. Operations at the Anaconda Smelter ceased in 1980 and the smelter facilities were dismantled soon thereafter. More than a century of milling and smelting operations resulted in high concentrations of arsenic, lead, copper, cadmium, and zinc in groundwater and surface water. Cleanup included testing and remediation of domestic wells, removal of waste from the nearby community, construction of nearly 1,000 acres of wetland, and 30,000 feet of stream restoration. Operation and maintenance activities are ongoing in areas where cleanup is complete. In other areas, cleanup is still in progress. EPA has determined that remedies that have been completed are protective of human health and the environment. Where remedies are not complete, access is controlled to prevent human exposure to waste. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The Anaconda Aluminum Co. Columbia Falls Reduction Plant site is located two miles northeast of Columbia Falls in Flathead County, Montana. The site includes approximately 960 acres north of the Flathead River, a fishery that includes the federally designated, threatened bull trout and the federally sensitive westslope cutthroat trout. From 1955 through 2009, an aluminum smelting plant operated at the site, and produced significant quantities of hazardous wastes as a byproduct of the aluminum smelting process. The types of hazardous wastes produced at the site are known to contain cyanide compounds that can leach into groundwater. In 1988, EPA requested a site investigation that revealed that there were high concentrations of polycyclic aromatic hydrocarbons at the site, primarily in soils and sediments, and that there had been a release of cyanide to groundwater and surface water; both of these findings were attributed to activities at the former smelting plant. The remedial investigation and feasibility study of the site is in progress, and the results of the investigation will determine cleanup needs and identify potential cleanup options at the site. The Silver Bow Creek and Butte Area site is in Butte, Montana, and includes 26 miles of stream and streamside habitat. Since the late 1800s, mining wastes have been dumped into streams and wetlands near mining operations. These activities contaminated soil, groundwater, and surface water with heavy metals. From 1988 to 2005, EPA completed several removal actions to clean up areas around former smelter sites, mine waste dumps, railroad beds, stream banks and channels, and residential yards to address immediate human health and environmental risks. Operation and maintenance, sampling, and monitoring actions are ongoing. EPA agreed to future cleanup work at the site in January 2018, including removal of contaminated soils, removal of sediment and floodplain waste, and construction of stormwater basins and sedimentation bays. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The Milltown Reservoir Sediments site near Missoula, Montana includes about 540 acres in the Clark Fork River and Blackfoot River floodplain and 120 miles of the Clark Fork River upstream of the Milltown Dam and Reservoir, which are located at the confluence of the Clark Fork and Blackfoot Rivers. From the 1860s until well into the 20th century, mineral- and arsenic-laden waste from mining activities in the region flowed into the Clark Fork River. As contaminated sediment and mine- mill waste moved downstream, about 6.6 million cubic yards of sediment accumulated behind the Milltown Dam. Mining activities and the downstream transport of mining- related wastes contaminated sediment, surface water, and groundwater with heavy metals. Remedy construction began in 2006, much of the site has been cleaned up, and remedy construction is underway to address remaining contamination. The site’s long-term remedy includes construction of a bypass channel at the reservoir; removal of contaminated reservoir sediment; off-site disposal and use of contaminated sediment as vegetative cap material; removal of the Milltown Dam; continuation of a replacement water supply program and implementation of temporary groundwater controls until the Milltown aquifer recovers; and long-term monitoring of surface and groundwater. Remedy construction is ongoing. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The 438-acre Barber Orchard site in Haywood County, North Carolina, includes the area where Barber Apple Orchard operated from 1908 through 1988. Facility operations resulted in contaminated groundwater and soil. Contaminants include arsenic, lead, and pesticides such as DDT, aldrin, and dieldrin that can be found in groundwater or soils on residential properties built on the former orchard. EPA removed soil in contaminated areas and, in a 2011 proposed cleanup plan proposed long- term monitoring of contaminated groundwater with the expectation that soil remediation will positively affect groundwater contamination. EPA has determined that the contaminated groundwater does not currently threaten people living and working near or on the site. EPA officials told us that in 2004, the town of Waynesville extended its municipal water system throughout the Orchard, and since the completion of the soil cleanup in 2011, new homes have been constructed within the boundaries of the Orchard. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The 3.5-acre Benfield Industries site in Waynesville, North Carolina, includes the area where Benfield Industries mixed and packaged materials bought in bulk for resale in smaller amounts from 1971 through 1983. The facility handled and stored paint thinners, solvents, sealants, cleaners, de-icing solutions and wood preservatives. Between 1990 and 1992, EPA conducted the remedial investigation and feasibility study using federal funding. The cleanup included excavating and washing contaminated soil, biotreating contaminated slurries, and placing the cleaned soil and slurry in excavated areas. Following soil treatment, EPA graded and planted seed. According to EPA officials, a groundwater extraction system was installed and was operated between 2001 and 2007. However, a 2007 report concluded that it was no longer an effective groundwater remedy, and that monitored natural attenuation may be a more effective remedy. Consequently, EPA shut down the system in June 2007. Agency officials told us the agency recently completed a pilot scale treatability study in which chemicals were injected into the subsurface to destroy residual wood preservatives that were adversely impacting groundwater quality. According to EPA, the agency will be using the information gained from this treatability study in the forthcoming remedial design. The Homestake Mining Company site in Cibola County, New Mexico includes a former uranium mill demolished from 1993 through 1995 and the impacted portions of the underlying groundwater aquifers. Uranium milling operations began at the site in 1958 under a license issued by the Atomic Energy Commission. Site operations and seepage from two tailings impoundments contaminated soil and groundwater with hazardous chemicals including uranium, selenium, radium-226, radium-228, thorium-230 and nitrate. Nearly 4.5 billion gallons of contaminated water have been removed and 540 million gallons of treated water have been injected into the aquifer. An average of 2 feet of contaminated soil was removed from the mill area and placed in the tailings impoundments. Cleanup is ongoing. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The 125-acre United Nuclear Corporation site near Gallup, New Mexico, includes a former uranium ore tailings disposal area and processing mill that operated from 1977 through 1982. The facility processed uranium ore using a combination of crushing, grinding and acid- leach solvent extraction methods. Milling produced acidic slurry of ground rock and fluid tailings. Disposal of about 3.5 million tons of tailings took place in on-site impoundments. Facility operations contaminated soil and groundwater. Surface reclamation stabilized the mill tailings and protected the Rio Puerco from contamination spills. However, EPA notes that groundwater treatment has been difficult due to low groundwater recharge rates and extraction wells proved to accelerate movement of contaminated water rather than contain it. Consequently, EPA installed additional extraction wells in 2010. Cleanup activities and monitoring are ongoing. The 70-acre Prewitt Abandoned Refinery site is located near Prewitt, New Mexico. The refinery operated between 1938 and 1957. Refinery operations contaminated soil and groundwater with hazardous chemicals including asbestos and lead. Potentially responsible parties removed the refinery and other site structures; however, scattered demolished structures, foundations and exposed fill remained on-site. The remedy for surface soil is complete. The remedy for subsurface soil and water continues to be protective in the short term; however, EPA could not determine if the remedy is protective of human health and the environment in the long term, and the agency recommends new evaluations to characterize the quantity, composition and extent of various contaminants and exposure pathways at the site. EPA further recommends the evaluation of an alternative cleanup plan to enhance protectiveness at the site. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The 58-acre North Railroad Avenue Plume site is a contaminated groundwater plume in Española, New Mexico. The Norge Town laundromat and dry cleaning operation contaminated groundwater with tetrachloroethylene, trichloroethylene, cis-1,2- dichloroethylene and trans-1,2- dichloroethylene. The contaminated groundwater aquifer is the sole-source drinking water aquifer for the residents of City of Espanola and, the Pueblo of Santa Clara, as well as individual water supply wells near the site. The remedy consists of enhanced on-site bioremediation. The areas targeted for cleanup are the source area, soils with high contaminant levels, and contaminated shallow groundwater. EPA indicated that the remedy has reduced contamination in shallow groundwater but has not been effective in the deep aquifer; consequently, EPA initiated additional analysis in 2015. The Jackpile-Paguate Uranium Mine site is located on the Pueblo of Laguna, New Mexico, reservation and consists of three former leases. The former leaseholder, Anaconda Minerals Company, mined and operated a uranium mine at the site from 1952 through 1982. Out of a total of 7,868 leased acres, 2,656 acres were disturbed by mining. This disturbance originally included three open pits, 32 waste dumps and 23 sub-grade ore stockpiles, 4 topsoil stockpiles, and 66 acres of buildings and roads. Mining operations detrimentally affected surface water with hazardous chemicals in quantities sufficient to support listing onto the EPA National Priorities List for Superfund cleanup. Atlantic Richfield is currently undertaking the remedial investigation and feasibility study at the site. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) EPA officials told us that the Carson River Mercury site extends over more than a 130-mile length of the Carson River, beginning near Carson City, Nevada, and extending downstream to the Lahontan Valley. Contamination at the site is a legacy of the Comstock mining era of the late 1800s, when mercury was imported to the area for processing of gold and silver ore. The site includes mercury-contaminated soils at former mill sites; mercury contamination in fish and wildlife; and mercury contamination in waterways adjacent to the mill sites, including the water, sediment, and adjacent floodplain of the Carson River, Lahontan Reservoir, Carson Lake, Stillwater Wildlife Refuge, and Indian Lakes. Following excavation and removal of mercury- contaminated tailings and soils from the site to protect human health and the environment, site investigations and cleanup planning are ongoing. The Hooker (Hyde Park) site is located in Niagara Falls, New York. The 15-acre area was used for the disposal of about 80,000 tons of waste, some of it hazardous material, from 1953 through 1975, resulting in sediment and groundwater contamination with hazardous chemicals, including Aroclor 1248, chloroform, phenol, benzoic acid and chlorendic acid. Cleanup included establishment of a drain system around the landfill; treatment of liquids leaching from the landfill; capping of the landfill; and removal of contaminated soils and sediment. Site construction finished in 2003. EPA has determined that, since cleanup, the site no longer poses a threat to nearby residents or the environment. Long- term groundwater treatment and monitoring are ongoing. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The General Motors (Central Foundry Division) site is located near Massena, New York. General Motors operated an aluminum diecasting plant on the site beginning in1959 and used PCBs in the manufacturing process through 1980. Contamination resulted from General Motors’ waste disposal practices. Completed cleanup actions include the installation of a cap on an industrial landfill to prevent the surface flow of contaminants and reduce potential air exposure from contaminants; dredging of the St. Lawrence River and placement of a cap on remaining sediment; remediation of two inactive lagoons; and creation of a 150-foot landfill setback along the border with the Saint Regis Mohawk reservation. The final significant cleanup is a 10-million- gallon industrial lagoon. EPA has conducted three 5-year reviews at the site and the owner is actively marketing the property for re-use or redevelopment. The Peter Cooper site in Gowanda, New York, was the location of an animal glue and industrial adhesive manufacturing factory. Contamination was caused by the improper disposal of wastes derived from chrome-tanned hides. The waste material has been shown to contain elevated levels of chromium, arsenic, zinc, and several organic compounds. Remedial activity for the landfill contained more than 8 million tons of waste and included capping the landfill, putting in a gas venting system, and controlling leachate. A retaining wall prevents contaminants from reaching Cattaraugus Creek. Site investigations and cleanup are complete, and monitoring is ongoing. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The Onondaga Lake site includes a 4.6-square-mile lake bordering the City of Syracuse, New York, and four nearby towns and villages. The site also includes seven major and minor tributaries and upland sources of contamination from a 285-square-mile drainage basin. Onondaga Lake has been the recipient of industrial and municipal sewage discharges from the site for more than 100 years. Contaminants include chlorinated benzenes, mercury, and PCBs. Between 1998 and 2018 EPA selected cleanup remedies for several areas within the site. Cleanup activities include removing chlorobenzene from existing wells, cleaning storm drainage systems, construction of a lakeshore barrier wall, and groundwater collection and treatment systems. Site investigations and cleanup activities are ongoing in several areas of the site, including the Lower Ley Creek and Willis Avenue areas. The Cayuga Groundwater Contamination site covers about 4.8 square miles extending from Auburn to Union Springs, New York. The site is the former location of a facility where General Electric Company and its partners manufactured semiconductors. The site includes residential properties mixed with farmland, woodlands, and commercial areas. Contaminated groundwater at the site contains volatile organic compounds that are potentially harmful contaminants that easily evaporate in the air. EPA conducted a remedial investigation and feasibility study to determine the sources, nature, and extent of site contamination and to evaluate remedial alternatives. Remediation will depend on the characteristics identified, but will include bioremediation for the most contaminated area as well as natural processes to reduce the level of contamination to meet groundwater standards. EPA is requiring periodic collection and analyses of groundwater samples to verify that the level and extent of contaminants is declining. EPA is deferring a decision on how to clean up the groundwater in Area 3, and intends to further investigate that area prior to issuing a final cleanup decision. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The Eighteen Mile Creek site consists of contaminated sediment, soil, and groundwater along approximately 15 miles of creek in Niagara County, New York. The site has a long history of industrial use dating to the 19th century. Contamination, including PCBs and heavy metals, spans two areas: Eighteen Mile Creek corridor and the creek sediment to Lake Ontario. Possible sources of the contamination include releases from hazardous waste sites, industrial or municipal wastewater discharges, and disposal practices of manufacturers around the creek. EPA has demolished five contaminated residential properties and relocated the residents, completed the remedial investigation and issued a record of decision for the creek corridor in 2017, and is currently conducting the remedial investigation in the length of the river to Lake Ontario. The approximately 145-acre Wilcox Oil Company site in Bristow, Oklahoma includes the inactive and abandoned Lorraine and Wilcox Oil Refineries, which operated from approximately 1915 through 1963. The main components of the refinery included a skimming plant, cracking unit, and redistillation battery with a vapor recovery system and continuous treating equipment. Refinery operations contaminated soil and sediment and left behind refinery waste material such as oil waste and sediment skimmed from crude oil, and potentially lead. Planning and implementation of the site’s remedial investigation and feasibility study is ongoing. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The 200-acre Hudson Refinery site housed an oil refinery from 1922 until 1982. The site included aboveground storage tanks, wastewater treatment impoundments, separators, stained soils, a land treatment unit, and loose and friable asbestos-containing material. Refinery operations contaminated soil, groundwater, surface water, and sediment. The site’s long-term remedy, selected in 2007 and amended in 2010, included removal of asbestos-containing materials, coke tar, and scrap metal; soil and waste excavation with off-site disposal; excavation, stabilization, and off-site disposal of sediment from waste ponds and sumps; treatment of surface water from ponds with contaminated sediment; groundwater monitoring; and institutional controls, among others. Cleanup construction started in early 2010 and finished in October 2010. Operation and maintenance activities and monitoring are ongoing. The 160-acre Oklahoma Refining Company site in Cyril, Oklahoma contained an oil refinery operated by several different owners until 1984. Site operations contaminated soil, sediment, surface water, and groundwater with PAHs, volatile organic compounds, and metals. Long- term remedies included bioremediation; stabilization; neutralization, containment, and treatment of surface water and groundwater; and on-site disposal of excavated materials in a hazardous waste landfill. Remediation was completed in 2001 on the southern part of the site. Removal of hazardous waste was completed in 2006. EPA is currently evaluating long-term cleanup activities on the northern portion of the site. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The Tar Creek site is located in Ottawa County, Oklahoma. According to EPA, the site itself has no clearly defined boundaries, but consists of areas within Ottawa County impacted by historical mining wastes. The site is part of the larger Tri-State Mining District that consists of historical lead and zinc mining areas in northeast Oklahoma, southeast Kansas, and southwest Missouri. The site first came to the attention of the State of Oklahoma and EPA in 1979, when water began flowing to the surface near Commerce, Oklahoma from underground mine areas, through abandoned boreholes. This surface discharge flowed into Tar Creek, and soon other discharge locations were observed near Tar Creek and the abandoned mining town of Douthat and Quapaw. As a result, Tar Creek and Beaver Creek were significantly impacted. EPA has defined five areas to focus on: surface water and groundwater; waste in residential areas that causes high blood lead levels in children; chemicals found in an office and laboratory complex; piles of mine and milling waste and smelter waste; and sediment and surface waters in seven watersheds within three states and nine tribal areas. Remedial efforts include plugging abandoned wells to prevent contamination of aquifers, cleanup of public areas and residences, removal of mining chemicals, and relocating mining waste on the surface. The Quapaw Tribe has led remedial efforts on portions of tribally owned properties located within Tar Creek. Cleanup is ongoing. The 61-acre Tulsa Fuel And Manufacturing site in Collinsville, Oklahoma, is the location of a former zinc smelter and lead roaster that operated from 1914 through 1925. Historical operations contaminated soil, sediment, and surface water with hazardous materials including zinc and lead. EPA selected a cleanup plan for the site that included on-site consolidation and capping of soil, sediment and waste material. Construction of the remedy began in August 2014 and is now completed. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The McCormick and Baxter Creosoting Company site is a former creosote wood treating facility located on the east bank of the Willamette River in Portland, Oregon. The company was founded in 1944 and continued operations until October 1991.This site is located within the Portland Harbor Superfund site, but was not included in the January 2017 Portland Harbor record of decision. The site encompasses approximately 41 acres of land and an additional 23 acres of contaminated river sediment. Site investigations confirm releases of wood- treating chemical compounds to soils, groundwater, and sediment. Remedial investigations identified three plumes of contaminated groundwater migrating toward surface waters. Completed cleanup activities include demolition of the McCormick and Baxter plant; soil excavation, treatment, and disposal; upland soil capping; installation of a subsurface barrier wall; contaminant recovery; construction of a multi-layer sediment cap in the Willamette River; monitoring and engineering; and institutional controls. Construction of site remedies finished in September 2005. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) Taylor Lumber and Treating operated a wood-treating plant at the site near Sheridan, Oregon, from about 1946 until 2001. EPA found that wood-treating chemical spills, including creosote and pentachlorophenol, contaminated soil, roadside ditches, and groundwater at the site. In response, EPA constructed an underground slurry wall as part of the remedy beneath the wood-treating area to contain and extract the most contaminated groundwater to maintain hydraulic control within the barrier wall. The final cleanup included excavation of contaminated soils from 5 upland acres and from adjacent ditches flowing to the South Yamhill River; replacement of an existing asphalt cap in the wood-treating area with a new low permeability asphalt cap overlaying the underground slurry wall; disposal of material from stockpiled soil storage cells off-site; and upgrades to the storm water conveyance systems. EPA completed final cleanup in 2008. The property is now owned and operated by a private company, which has ongoing obligations related to property use restrictions, operations, and maintenance on the property. EPA conducted its second 5-year review in 2017. The 4.2-acre Harbor Oil Incorporated site is located in Portland, Oregon, in an industrial area adjacent to Force Lake. A waste oil recycling facility currently operates on the site. Past site operations included a tank truck cleaning business, which was destroyed by a fire in 1979 that ruptured five 20,000-gallon aboveground used oil tanks. Site activities, the fire, and a large oil spill in 1974 contaminated soil, sediment and groundwater with metals, oil, pesticides, and PCBs. EPA ordered a previous operator to empty, clean, and dismantle a tank containing petroleum wastes. Remedial investigations determined that contamination does not pose an unacceptable risk to human health or the environment; therefore, no further cleanup is required. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The 10-acre Gould, Incorporated site in Portland, Oregon housed a lead smelter and lead oxide production facility from 1949 until 1981. Site activities included on-site disposal of about 87,000 tons of battery casings and discharge of about 6 million gallons of acid into a nearby lake, which resulted in contaminated soils and lake sediment. EPA transferred the contaminated soils and sediment into a lined containment area at the site as part of the cleanup. EPA monitored groundwater at the site to determine if historic wastes adversely impacted shallow groundwater at the site. Based on this data, in 2000, EPA determined that no further groundwater cleanup actions were necessary. Groundwater monitoring near the containment area continues to ensure that the containment area has no adverse impact. The North Ridge Estates site is a residential subdivision 3 miles north of Klamath Falls, Oregon that is contaminated with asbestos as a result of the improper demolition of approximately 80 1940s-era military barracks buildings. Asbestos-containing materials and soil are being removed from the old military barracks site during three seasons of cleanup from 2016 through 2018. Additional contamination at the nearby Kingsley Firing Range, also part of the site, will be investigated and completed at a later time. According to EPA, cleanup and restoration will be completed by the end of 2018. The 76-acre Formosa Mine site is located on Silver Butte in Douglas County, Oregon. The site was originally mined for copper and silver from about 1910 through1937. The abandoned mine discharges millions of gallons of acid rock drainage and toxic metals into the upper reaches of Middle Creek and South Fork Middle Creek every year. These discharges have contaminated surface water, groundwater, soil, and sediment with heavy metals. EPA is currently designing the remedy for all mine-impacted material on the surface and will address risks to surface and groundwater separately. The remedy for surface contamination consists of excavating, contouring, or capping various areas to prevent leaching during precipitation events. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The Portland Harbor site includes portions in the Willamette River and about 12 river miles upstream of the Willamette River in and around Portland, Oregon, that have been contaminated from decades of industrial use. Areas of the site housed manufactured gas plants, a pesticide manufacturing facility, and boat maintenance facilities, among other industrial uses. Water and sediment at the site are contaminated with many hazardous substances, including PCBs, PAHs, dioxins/furans, pesticides, and heavy metals. The harbor is an international portal for commerce, and dozens of industries within the site provide economic sustainability to the community. The Lower Willamette is also a popular area for recreation, including fishing and boating. The river provides a critical migratory corridor and rearing habitat for salmon and steelhead, including endangered runs of steelhead and chinook. The area also holds great importance to several tribes as a natural and cultural resource. EPA issued its record of decision in January 2017 and finished its baseline sampling plan in December 2017. The record of decision specifies the remedy selected, which is designed to reduce risks to human health and the environment to acceptable levels and actively remediate (using dredging, capping, enhanced natural recovery, and monitored natural recovery) on 394 acres of contaminated sediment and 23,305 lineal feet of river bank. This final remedy is estimated to cost approximately $1.05 billion and take about 13 years to complete. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The Black Butte Mine site is located near Cottage Grove, Oregon. Mercury mining from the late 1880s through the late 1960s included extracting ore from the mine, crushing it on-site, roasting it in kilns to volatilize the mercury, and bottling and shipping the mercury. Mining operations, tailings piles left at the site, and erosion from Furnace Creek contaminated soil, sediment, surface water, and groundwater with mercury and other toxic metals. EPA and its contractors are working in the Furnace Creek area of the site to excavate mine tailings and contaminated soils/sediment for safe disposal in an off-site repository. Removing the mine tailings will reduce mercury leaking into Furnace Creek and reduce the potential for mercury leaching into groundwater. Site investigations for the long-term cleanup are under way. The Newport Naval Education/Training Center site was used by the U.S. Navy as a refueling depot from 1900 through the mid-1970s. The site encompasses 1,063 acres on the west coast of Aquidneck Island in Portsmouth, Middletown, and Newport, Rhode Island. The site includes multiple areas of contamination, including a landfill, a fire training area, a former shipyard, and five tank farms. The areas contain varying degrees of groundwater contamination. The Navy is the lead agency for site investigation and cleanup. Site cleanup has included installation of a soil cover, use of a groundwater pump and treat system, and removal of contaminated debris. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The Centredale Manor Restoration Project site is located in North Providence, Rhode Island, where the main “source area” consists of about 9 acres down the Woonasquatucket River, south to the Lyman Mill Dam, and includes the restored Allendale Dam. The site was a chemical production and drum reconditioning facility from the 1940s to the 1970s that resulted in the release of dioxin and other contamination. Past site operations led to chemicals released directly to the ground, buried and emptied directly into the river. This resulted in contamination of soil, groundwater, surface water and sediment in the adjacent river and downstream ponds. A major fire in 1972 destroyed most structures at the site. Residential apartments were constructed at the site in the late 1970s and early 1980s and still occupy the site. To address immediate risks, EPA conducted several activities including fencing the site, capping contaminated soil, and reconstructing Allendale Dam. EPA developed the cleanup plan, with amendments, in 2012. EPA, the state of Rhode Island, and potentially responsible parties agreed in July 2018 on a plan to clean up contamination at the site. The Whitewood Creek site covers an 18-mile stretch of Whitewood Creek in Lawrence, Meade, and Butte counties in South Dakota. Since the 1870s, gold mining operations in the area included the discharge of millions of tons of mine tailings into the creek. These mine tailings settled along the Whitewood Creek floodplain, contaminating soil, groundwater, and surface water with heavy metals. EPA excavated 4,500 cubic yards of contaminated soil from residential yards, disposed of contaminated soil, and established institutional controls and surface water monitoring. EPA took the site off the Superfund program’s National Priorities List in 1996 when cleanup finished and affected counties restricted future development in impacted areas. Surface water monitoring is ongoing. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The 360-acre Gilt Edge Mine site is located about 6.5 miles east of Lead, South Dakota. The primary mine disturbance area encompasses a former open pit and a cyanide heap-leach gold mine, as well as prior mine exploration activities from various companies. Mining and mineral processing at the site began in 1876 and early gold miners developed extensive underground workings that wind through the central portion of the site. There was also some surface mining. Historical operations at the site contaminated surface water and groundwater with acidic heavy-metal-laden water. In 1986, mine owners commenced development of a large-scale open pit, cyanide heap leach gold mine operation. In the late 1990s, site owners abandoned the site and their responsibilities to address acidic heavy-metal-laden water generated from the exposed highwalls of the three open mine pits and from the millions of cubic yards of acid-generating spent ore and waste rock. Investigation and cleanup activities at the site are ongoing. Interim remedies are currently in place for the water treatment, Lower Strawberry Creek, and Ruby Gulch Waste Rock Dump; and remedial action construction is in progress for the primary mine disturbance area. The Lower Duwamish Waterway site is a 5-mile segment of the Duwamish, Seattle, Washington’s only river. The river flows between residential areas as well as through the industrial core of Seattle into Elliott Bay. The waterway has served as Seattle’s major industrial corridor since the early 1900s, resulting in sediment contaminated with toxic chemicals from industrial practices, stormwater runoff, and wastewater. EPA has also found contamination in fish and shellfish, including PCBs, arsenic, polycyclic PAHs, dioxins, and furans. As a result, consumption of resident fish and shellfish, and contact with contaminated sediment pose a risk to human health. EPA signed the record of decision in 2014 that includes plans to clean up about 177 acres in the waterway, including dredging, capping, and natural sedimentation. By the end of 2015, 50 percent of PCB contamination in the river bottom was removed through these early action cleanups. Cleanup and monitoring activities are ongoing. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The 340-acre Naval Undersea Warfare Engineering Station site is located on a peninsula 15 miles west of Seattle. Site activities included torpedo maintenance, fuel storage, welding, painting, carpentry, plating, and sheet metal work. Site activities and waste disposal practices contaminated soil, sediment and groundwater with hazardous chemicals, including 1,4-Dioxane, chromium, and vinyl chloride. The site’s long-term cleanup remedy included demolition of the plating shop building; removal and disposal of contaminated soil and sediment; removal of underground storage tanks; long-term monitoring of groundwater, sediment and shellfish; institutional controls; and phytoremediation to treat contaminated landfill soil. Remedy construction took place between 1995 and 2000. Site operation and maintenance activities, and site monitoring, are ongoing. Four sites on the NPL are part of the 586-square-mile Hanford Nuclear Reservation near Richland, Washington, where waste was created as a by-product of producing plutonium from 1943 through1987. The 25- square-mile Hanford 100-Area site, also referred to as the River Corridor, is focused on cleanup of contamination that originated from nine nuclear reactors. Cooling water contaminated with radioactive and hazardous chemicals was discharged into both the adjacent Columbia River and on-site infiltration cribs and trenches. Site operations also included burying contaminated solid wastes on-site. These activities contaminated soil and groundwater with radioactive constituents, heavy metals, and other hazardous chemicals. Contaminants have been addressed by demolishing buildings, removing contaminated soil, and employing pump and treat systems for contaminated groundwater, among others. EPA has selected eight interim remedies for the 100-Area and remedial investigations are under way to support selection of final cleanup remedies. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) Four sites on the NPL are part of the 586-square-mile Hanford Nuclear Reservation near Richland, Washington where waste was created as a by-product of producing plutonium and other nuclear materials for nuclear weapons from 1943 through 1987. The 79-square-mile 200-Area site is located 17 miles north-northwest of Richland, Washington. The 200-Area site is located in the center portion of the Hanford site, known as the Central Plateau, and contains former chemical processing plants and waste management facilities. During processing activities, massive quantities of carbon tetrachloride were discharged into the ground. Site activities also included processing, finishing and managing nuclear materials, including plutonium. About 1 billion cubic yards of solid and diluted liquid wastes (radioactive, mixed, and hazardous substances) were disposed in trenches, ditches, and in an on-site landfill. About 1,000 facilities and structures were built to support processing activities which contaminated soil, groundwater and surface water with hazardous chemicals and radioactive constituents. Thousands of containers and drums holding radioactive waste were placed in burial grounds. Remedial investigations, removal actions, and remedy design and construction are under way for more than 800 waste areas at the site. Cleanup actions included decontamination and demolition of contaminated structures; treatment of contaminated soil; excavation and off-site disposal of drummed wastes; institutional controls; and natural attenuation of groundwater contaminants. According to EPA, a remedy for one of the large canyon-type buildings is about halfway complete and is awaiting investigation and remediation of surrounding waste sites before it can be completed. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) Four sites on the NPL are part of the 586-square-mile Hanford Nuclear Reservation near Richland, Washington where waste was created as a by-product of producing plutonium and other nuclear materials for nuclear weapons from 1943 through 1987. The 56 square mile Hanford 300 Area site was home to fuel manufacturing operations at Hanford as well as experimental and laboratory facilities. The 300-Area site includes an unlined liquid disposal area north of the on-site industrial complex area, landfills, and miscellaneous disposal sites associated with operations at the industrial complex. The 300-Area site contains about 27 million cubic yards of solid and diluted liquid wastes mixed with radioactive and hazardous wastes in ponds, trenches, and landfills. The areas used for liquid discharges had no outlets; therefore, liquids percolated through the soil into the groundwater and the Columbia River. Cleanup actions completed to date include decontamination and demolition of contaminated structures; natural attenuation of groundwater contaminants; and disposal of building rubble, contaminated soil, and debris. Remedy construction has been completed in several areas of the site and remedial investigations, removal actions, and remedy design and construction are under way at the remaining areas. Four sites on the NPL are part of the 586-square-mile Hanford Nuclear Reservation near Richland, Washington where waste was created as a by-product of producing plutonium and other nuclear materials for nuclear weapons from 1943 through 1987. Waste areas in the 120-square-mile Hanford 1100-Area site include a landfill, drains, underground tanks and a sand pit where as many as 15,000 gallons of waste battery fluids may have been disposed. Past site activities and waste disposal practices contaminated soil and groundwater with heavy metals and hazardous chemicals such as PCBs and trichloroethene. Remedial activities include off-site disposal of PCB-contaminated soils, capping of the landfill, and establishing continuing institutional controls to prevent future exposure and contamination from buried asbestos.Following cleanup, EPA deleted the site from the NPL in 1996. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The 300-acre Jackson Park Housing Complex site is located in eastern Kitsap County, about 2 miles northwest of Bremerton, Washington. From 1904 through 1959, the facility operated as a Navy ammunition depot and included ordnance, manufacturing, processing, and disassembly. Residual ordnance powders were disposed of by open burning. Hazardous dust deposited on floors during ordnance handling was washed into floor drains that led into Ostrich Bay. The site also included incinerators; paint, battery, and machine shops; and a boiler plant. Site activities contaminated surface water and soil with hazardous chemicals and heavy metals. The site’s long-term remedy included installation of a soil and vegetation cover over contaminated soil, shoreline stabilization, implementation of a shellfish sampling program, and signs along the shoreline to notify local residents of any harvest restrictions. Site cleanup also included the removal and off-site disposal of wooden pilings from abandoned Navy structures, excavation and disposal of contaminated soil, establishment of an environmental monitoring program, and subsurface placement of oxygen-releasing chemicals. Remedy construction began in 2000 and is ongoing. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The 53-acre Old Navy Dump/Manchester Laboratory site is located north of Manchester, Washington, along the western shore of Clam Bay in Puget Sound. Federal ownership of this site started in 1898 with the U.S. Army. In 1924, the entire site was transferred to the U.S. Navy. From the 1940s through the 1960s, the Navy used the site primarily for construction, repair, maintenance, and storage of submarine nets and boats, but also used the site for firefighter training and as a dump for wastes generated at the site. Former firefighter training activities contaminated soil with dioxins and petroleum hydrocarbons. The Navy also dumped demolition debris and industrial waste, including asbestos, into a former tidal lagoon, contaminating soil, sediment, seep water, and shellfish in Clam Bay with PCBs and metals. Clam Bay has been used primarily for recreational shellfishing and is a known habitat for the bald eagle and chinook salmon, a threatened species under the Endangered Species Act. In the early 1970s, EPA and the National Oceanic and Atmospheric Administration (NOAA) acquired portions of the property. The site is currently occupied by an EPA analytical laboratory and a NOAA fisheries research laboratory. The Army Corps of Engineers established in the third 5-year review in 2014 that the remedy at this site is protective of human health and the environment. Operation and maintenance activities and monitoring are ongoing. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The 83-acre Pacific Sound Resources site, formerly known as the Wyckoff West Seattle Wood Treating facility, is located on the south shore of Elliott Bay on Puget Sound in Seattle, Washington. A wood-treating facility operated at the site between 1909 and 1994. Wood-preserving operations used creosote, pentachlorophenol, and various metal-based solutions of copper, arsenic, and zinc. Daily operations, as well as spills, leaks and storage of treated wood products resulted in soil and groundwater contamination. Direct discharge or disposal of process wastes and waste transport were the most likely sources of contamination to marine sediment. Over half of the site is located in either intertidal or subtidal lands. Cleanup actions included the placement of subtidal and intertidal caps over the 58-acre marine sediment area, including placement of at least 5 feet of cap material in the intertidal zone; dredging and removal of contaminated sediment for off-site disposal; and removal of marine pilings for off-site disposal. Construction of long-term cleanup remedies concluded in 2005 and, following cleanup, operation and maintenance activities, including periodic groundwater monitoring, are ongoing. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The Wyckoff Company / Eagle Harbor Superfund site is on the east side of Bainbridge Island in Central Puget Sound, Washington. The site was used for creosote wood treatment for more than 85 years, according to the Washington Department of Ecology. Environmental investigations revealed extensive contamination— including creosote, mercury, and other metals—in soils, groundwater, and in the sediment on the bottom of Eagle Harbor. EPA reports that extensive cleanup actions have been completed at the site, including operating a groundwater extraction and treatment system since 2012, capping sediment on more than 70 acres of Eagle Harbor, and hauling away contaminated soils and debris. Further cleanup actions are needed in the soil and groundwater at the former wood treatment facility and in adjacent beach sediment. In 2016 EPA released a proposed plan for additional cleanup actions at the site and, after a public comment period, divided the work into two cleanup decisions. The first was issued in May 2018 and the second is planned for issue near the end of 2018. The 10-acre Pesticide Lab site is an active agricultural research laboratory located at the Yakima Agricultural Research Laboratory in Yakima, Washington, and has been in operation since 1961.The site is leased by the U.S. Department of Agriculture (USDA). Wastes from the formulation, mixing, and storage of pesticide were discharged into a septic tank disposal system at the site from 1965 through 1985. USDA addressed cleanup under the Resource Conservation and Recovery Act. The site has been cleaned up and is no longer a threat to human health. Long-term monitoring is not required because cleanup left no contaminants of concern on the site. EPA deleted the site from the NPL in 1993. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The 92-acre Hidden Valley Landfill site is located in Puyallup, Washington. The site contains a former landfill and gravel pit that operated from 1967 through 1985. The landfill accepted liquids, solids, industrial wastes, and heavy metal sludge. Waste disposal activities contaminated groundwater with hazardous chemicals and heavy metals. The site’s long-term remedy included covering the waste with an impermeable barrier, collecting landfill gases, controlling surface water and soil erosion, and minimizing the lateral and vertical movement of contaminated groundwater. Remedy construction took place in 2000. Landfill gas and groundwater monitoring are ongoing. The Tulalip Landfill site, located within the boundaries of the Tulalip Indian reservation, is a former landfill located between Marysville and Everett, Washington. The site consists of a 147-acre landfill and 160 acres of wetlands. The Seattle Disposal Company operated the landfill from 1964 until 1979. The landfill received an estimated 3 million to 4 million tons of commercial and industrial waste. In 1979, landfill operators closed the landfill, added a soil cover, and constructed a perimeter barrier berm. However, insufficient grading of the soil cover resulted in poor drainage and allowed precipitation to collect and eventually infiltrate the landfill surface. As a result, the landfill contaminated groundwater, surface water and sediment with metals, pesticides, PCBs and polycyclic aromatic hydrocarbons. EPA’s interim remedy for the landfill included capping the landfill and installing a landfill gas collection and treatment system, among other actions. EPA continued the interim remedy for the landfill and included institutional controls for the wetlands, such as placing and maintaining signs to warn of potential risk from harvest and consumption of resident fish and shellfish. The tribe is responsible for maintenance of the remedy, inspections, and sampling at the site. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) Harbor Island is a 420-acre manmade island in Elliott Bay in Seattle Washington. The site includes the entire island and associated sediment. Built in the early 1900s, the island housed businesses that conduct commercial and industrial activities, including oil terminals, shipyards, rail transfer terminals, cold storage, and lumberyards. Site operations contaminated groundwater, sediment and soil with lead, PCBs, arsenic, mercury, and other contaminants. Remedial activities include removal and treatment of contaminated soil, treatment of groundwater, removal of approximately 6,000 creosote treated piles, and dredging sediment. Most portions of the site have been cleaned up and are undergoing long- term monitoring. The Commencement Bay, Near Shore/Tide Flats site is located in the City of Tacoma and the Town of Ruston at the southern end of Puget Sound in Washington. The site encompasses an active commercial seaport and includes 12 square miles of shallow water, shoreline, and adjacent land, most of which is highly developed and industrialized. EPA found widespread contamination of the water, sediment, and upland areas at the site and has divided the site into seven areas being managed as distinct cleanup sites. As part of this cleanup, EPA has remediated 2,436 properties with the worst contamination, restored 11 acres of shallow marine habitat, and restored 70 acres of estuarine habitat. The site’s long-term remedy includes demolishing remaining buildings and structures, excavating soil and slag from the five most contaminated source areas on the site, depositing demolition debris in an on-site containment facility, and monitoring the impacts of cleanup on groundwater and off-shore marine sediment. Investigations and remedy construction are ongoing at the site. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) Midnite Mine is an inactive former uranium mine in the Selkirk Mountains of eastern Washington. Located within the reservation of the Spokane Tribe of Indians, the mine was operated from 1955 until 1981. The site includes two open pits, backfilled pits, a number of waste rock piles, and several ore/protore stockpiles. The site contamination has resulted in elevated levels of radioactivity and heavy metals mobilized in acid mine drainage, both of which pose a potential threat to human health and the environment. The site drains to Blue Creek, which enters the Spokane Arm of Franklin D. Roosevelt Lake. Contaminated water emerging below the waste rock and ore piles is currently captured for treatment in an on-site treatment system. Cleanup includes consolidation of mine waste rock, protore, and contaminated soils; backfilling these materials in lined pits; covering these pits to prevent water infiltration; and ongoing water treatment. According to EPA, significant cleanup is planned to occur between 2017 and 2024. The 40-acre Lockheed West Seattle site is located in Elliott Bay near the mouth of the West Waterway in Seattle, Washington. The site includes about 7 acres of aquatic tidelands owned by the Port of Seattle and 33 acres of state-owned aquatic lands. Historic industrial practices at the former shipyard contaminated sediment with hazardous chemicals, including PCBs, dioxins, and furans. Industrial activities generated considerable quantities of sandblast grit and other industrial waste that discharged to sediment and accumulated beneath dry docks and shipways. The Lockheed Martin Corporation, as the potentially responsible party for the cleanup, will remove contamination from a 40-acre area in the northwest corner of the mouth of the West Waterway and north of the Port of Seattle’s Terminal 5. An estimated total of 167,000 cubic yards of contaminated material will be removed over the course of the cleanup. According to EPA, the cleanup was to begin in 2018 and is anticipated to be completed in the spring of 2019. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) Makah Reservation Warmhouse Beach Dump is located within the Makah Indian Reservation at the northwest tip of the Olympic Peninsula in Washington. The site includes a former open dump on top of a ridge about 3 miles northwest of Neah Bay and two streams that originate within the dump and flow to East Beach and Warmhouse Beach. Municipal and household solid and hazardous wastes were disposed of at the dump from the 1970s until 2012. Elevated levels of metals, perchlorate and PCBs have been found in soil at the dump and in the sediment of both creeks. Mussels at the beach also contain elevated concentrations of lead; however, EPA has not determined whether this is from the dump or creeks. EPA is in the remedial investigation stage of the cleanup. Bremerton Gas Works is a former manufactured gas plant located about a mile and a half north of downtown Bremerton, Washington. It occupies about 2.8 acres of property along the Port Washington Narrows in Puget Sound. Two species of fish that are listed as threatened under the Endangered Species Act (steelhead trout and chinook salmon) live near the site. This portion of Puget Sound is used as a sport and commercial fishery, as well as for subsistence fishing by the Suquamish Indian Tribe. EPA is in the early stages of the cleanup process, conducting the remedial investigation and feasibility study, which EPA expects to complete in spring 2019. The Hamilton/Labree Roads Groundwater Contamination site is located about 2 miles southwest of Chehalis, Washington. According to EPA, past site activities included spilling and dumping tetrachoroethene in Berwick Creek and burying drums and other containers of assorted hazardous chemicals on-site. The release at the site has contaminated soil, sediment, groundwater, and surface water. EPA’s selected interim remedy includes rerouting Berwick Creek around contaminated areas, thermally treating tetrachoroethene-contaminated soil and sediment, and treating contaminated groundwater. Remedial design is under way. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) Penta Wood Products site is located in the town of Siren in Burnett County, Wisconsin. A wood treatment facility operated at the site from 1953 until 1992, and used pentachlorophenol (PCP) to treat wood posts and telephone poles. Facility operations contaminated soil and groundwater with PCP and arsenic. During cleanup, EPA removed about 28 storage tanks containing liquid and sludge. Also, 43,000 gallons of a PCP/oil mixture and sludge were disposed of off-site. The treatment building was demolished and contaminated soil was cleaned on-site or disposed of off-site. Cleanup was completed in 2000, and operation and maintenance activities and monitoring are ongoing. In September 2014, the State of Wisconsin took over operations and maintenance activities at the site. Tribe or tribes with known interest in the site (CC) (HEUC) control (GWMUC) use (SWRAU) The Ashland/Northern States Power Lakefront site is located on the shore of Chequamegon Bay, which is part of Lake Superior, in northern Wisconsin. The site consists of several properties, including those owned by Northern States Power Co. of Wisconsin, Canadian National Railroad and the city of Ashland. 16 acres of contaminated lake sediment just off-shore are also part of the site. The near-shore portion of the site was formed by the placement of fill consisting of sawdust, wood, and wood waste; demolition debris; and other waste materials. Contaminants including tar, oil, PAHs, volatile organic compounds, and metals have been found in sediment, groundwater, and soil. Contamination has also been found in an adjacent residential area. Because groundwater is contaminated at levels of health concern, two artesian wells have been closed as a precautionary measure. Access to a portion of the bay and shore is restricted for boats and swimmers because when sediment is agitated, oil and tar can be released causing a slick to form. Cleanup at the site is ongoing and is being overseen by the Wisconsin Department of Natural Resources and EPA. Phase 1,soil and groundwater cleanup under portions of the site was completed in 2016. This entailed removing contaminated soil, covering the area with clean material, and installing barriers to stop groundwater from migrating. Phase 2, the full-scale wet dredge in the Chequamegon Bay, was completed in 2018. EPA is conducting the first five-year review of the site. In providing technical comments on a draft of this report, the Confederated Salish and Kootenai Tribes of the Flathead Reservation identified this additional site. State ID Site Overview Blackbird Mine is located 25 miles west of the town of Salmon in the Salmon-Challis National Forest in east- central Idaho. Cobalt, silver, and copper ore were extracted from underground and open-pit mining operations. Contaminated soil, sediment and tailings were released from the mine site during high water flows from thunderstorms and snowmelt. Acid rock drainage and leachate from the mining tunnels, waste piles, and tailings contaminated soil, sediment, surface water, and groundwater with heavy metals such as copper, cobalt, and arsenic. Affected surface waters include Blackbird Creek, the South Fork of the Big Deer Creek, Big Deer Creek, and Panther Creek. Since 1995, cleanup actions have collected contaminated runoff water in the mine area and treated it for copper and cobalt. Cleanup actions have also stabilized waste-rock piles at the mine. Remedy construction is complete except for determining whether to divert Bucktail Creek. Post-construction monitoring of these cleanup activities is ongoing. Since the early 1900s, General Electric operated a large- scale industrial facility that manufactured and serviced power transformers, defense and aerospace materials, and plastics, and used numerous industrial chemicals at its Pittsfield facility. Years of PCB and industrial chemical use, and improper disposal, led to extensive contamination around Pittsfield, Massachusetts as well as down the entire length of the Housatonic River, which is approximately 150 miles from its source on the East Branch in Hinsdale, Massachusetts and flows through Connecticut into Long Island Sound. After testing groundwater, river sediment, soil, and wildlife, EPA determined that the contamination needed to be addressed and that the greatest concern in the area is the possibility of direct contact or ingestion of PCB contamination. Since 1977, there has been a ban on fishing and consumption of fish from areas of the Housatonic River. These restrictions will remain in place until PCB levels decrease. Data are collected to ensure that the current restrictions protect human health. EPA collects information regarding PCBs in fish and shellfish. In addition to PCBs, other industrial compounds present at the site pose an unacceptable risk to people and the environment. Site Overview The Smurfit-Stone Mill Frenchtown site is located 11 miles northwest of Missoula, Montana. The 3,200-acre site formerly housed a pulp mill that operated from 1957 through 2010. The core industrial footprint of the mill site covers about 100 acres, and there are more than 900 additional acres containing a series of unlined ponds used to store treated and untreated wastewater from the mill, as well as sludge recovered from untreated wastewater. The site also includes landfills used to dispose of solid wastes, including general mill refuse and asbestos. Various hazardous substances were used or produced on-site, including bleaching chemicals that produced dioxins and furans that may have been released into the environment. A screening investigation by EPA determined that the site’s primary contamination sources include four sludge ponds, an emergency spill pond, an exposed soil pile adjacent to a landfill, a wastewater storage pond, and a soil land farming area. The results of the investigation will determine cleanup needs and identify potential cleanup options at the site. The Anaconda Copper Mine site covers more than 3,400 acres of the Mason Valley, near the city of Yerington, Nevada. Portions of the site are owned by a company, while other areas are public lands managed by the U.S. Bureau of Land Management. Nevada Department of Environmental Protection and EPA have conducted several emergency removal actions at the site to address immediate concerns. Remedial investigations and feasibility studies will be conducted to determine the extent of contamination and potential cleanup options for other areas at the site. Site Overview The Lower Fox River, located in northeastern Wisconsin, begins at the Menasha and Neenah channels leading from Lake Winnebago and flows northeast for 39 miles to where it discharges into Green Bay and Lake Michigan. The Fox River Natural Resource Damage Assessment / Polychlorinated Bisphenyls Releases site addresses releases caused by operations of several pulp and paper mills that, during the 1950s and 1960s, routinely used PCBs in their operations that resulted in contamination of the river. Samples from the site also indicate the presence of polycyclic aromatic hydrocarbons resulting from manufactured gas plant processes co-mingled or underneath the PCB contamination. Approximately 270,000 people live in the communities along the river. 2018 is the 10th year of dredging in the Lower Fox River, and EPA estimates 450,000 cubic yards of PCB- contaminated sediment will be removed before the end of the year. In addition, about 2.1 acres of sediment will be capped and 179 acres will be covered with sand. EPA plans to oversee a second 5 year review in 2019. Appendix II: Objectives, Scope, and Methodology This report (1) examines the extent to which the U.S. Environmental Protection Agency (EPA) has reliable data identifying National Priorities List (NPL) sites that are located on tribal property or that affect tribes, (2) examines the extent to which EPA has reliable data on the agency’s consultation with tribes and (3) describes what actions, if any, EPA has taken to address the unique needs of tribes when making decisions about cleanup actions at NPL sites. To examine the extent to which EPA has reliable data identifying NPL sites that are located on tribal property or that affect tribes, we reviewed relevant provisions of the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA) of 1980 as amended and policies and guidance regarding EPA’s identification and clean-up of NPL sites. We obtained and evaluated EPA data from the Superfund Enterprise Management System (SEMS) on proposed, final, and deleted NPL remedial sites that have tribes associated with them or that EPA has designated as having Native American Interest (NAI). We limited our review to examining proposed, final, and deleted NPL sites because they represent sites with the highest national priority due to the significance of releases, or threatened releases, of hazardous substances. EPA also indicated whether such sites may be located within 10 miles of known tribal property by comparing the sites’ coordinates to the tribal geographic location as recorded in publicly available EPA data. We also obtained information about whether a site was considered a federal facility because other federal agencies may have different consultation policies than EPA. We did not determine whether EPA has information about consultation with tribes for sites considered federal facilities. EPA initially identified 265 NPL Superfund sites that were on tribal property, had NAI, had a tribe or tribes with potential interest in the site, or may have been within 10 miles of tribal property. We then worked with EPA headquarters officials and each regional office to perform data quality checks and identify any errors or omissions, in order to develop a revised list of a total of 87 NPL sites—of which 11 were federal facilities— known to affect tribes or to be located on tribal property. As an example of the data quality checks, officials from each EPA regional office reviewed the list of sites for their respective regions and made corrections to the sites’ designation as having NAI or tribes with interest in the sites. As another example, we compared data from EPA’s Tribal Consultation Opportunity Tracking System (TCOTS) database with the list of sites EPA provided us and determined that a tribal consultation had occurred for a site that EPA had not identified as having NAI. We checked with officials from the appropriate EPA regional office and they told us that the site should have been designated as having NAI, so we added it to our list. We also interviewed officials from EPA’s headquarters and regional offices to better understand the agency’s management, use, and the reliability of these data. In providing comments on a draft of this report, the Confederated Salish and Kootenai Tribes of the Flathead Reservation identified an additional site that was not included in EPA’s data, which we reviewed with EPA and added to our list of NPL sites known to be on tribal property or that affect tribes, bringing the total to 88 sites. We recognize that there may be additional sites at which there is tribal interest but determined that the data were sufficiently reliable to provide information on NPL sites known to be on tribal property or that affect tribes, and to select six sites for nongeneralizable case studies for our work. We did not select case studies from sites located on federal facilities because federal agencies may have different tribal consultation policies. For the case studies, we selected sites based on geographic diversity, and in order to represent sites that have been listed since the publication of EPA’s tribal consultation policy in 2011. We also selected sites that had at least two assessments or inspections performed according to EPA data so the tribes would have sufficient information to share with us about their experiences. In one of the case studies, we had to change to a different site from the same region when the tribe associated with the site we had initially selected did not wish to participate. We chose a replacement site in the same EPA region that was at a similar point in the cleanup process as the site we originally selected. To examine whether EPA has reliable data regarding its consultation with tribes about NPL sites, we reviewed EPA-specific guidance that applies to tribal consultation on NPL sites. We evaluated data from EPA’s TCOTS, reviewed related agency documentation, interviewed knowledgeable agency officials, and compared TCOTS data with other information EPA provided. Specifically, we compared data from TCOTS with information that officials from EPA headquarters and each EPA region provided to us regarding consultation for each of the nonfederal sites that had NAI. In order to determine the frequency with which EPA consults with tribes on cleanup actions of NPL sites, we examined and compared available data on consultation from the TCOTS system with other information provided by EPA in light of EPA’s consultation guidance. We also interviewed officials from EPA and selected tribes from our six nongeneralizable case studies regarding consultation. While we selected case studies based on nonfederal NPL sites EPA has identified as being on tribal property or affecting tribes, our interviews with tribal and EPA officials covered a broader range of sites and included officials’ views about any Superfund activities in which they had been involved. For each case study, we requested information documenting EPA’s consultation with tribes as well as any materials that demonstrated whether and how agency decisions took into account unique tribal needs associated with the site. We also conducted semi-structured interviews with officials from the tribe or tribes involved at each of our case study sites, as well as EPA regional officials for the region in which the site is located. We visited the Jackpile-Paguate Uranium Mine site and conducted interviews with tribal officials in person. We evaluated EPA and tribal officials’ experiences with consultation at our selected case study sites based on EPA’s consultation policies. To describe what actions EPA has taken to address the unique needs of tribes when making decisions about cleanup actions at NPL sites, we interviewed EPA officials from the regional offices associated with our selected case study sites about consultation regarding our case study sites, as well as at other NPL sites that affect tribes in their region. We also conducted semi-structured interviews with tribal officials who had consulted or coordinated with EPA regarding each of the selected sites in our review. We asked the tribes to describe the effects of the site on any unique needs such as subsistence fishing and gathering, and whether EPA has explored or addressed these needs during the agency’s cleanup actions. We conducted this performance audit from May 2017 to January 2019 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix III: Description of Case Study Sites To analyze examples of consultation and better understand the tribal perspective on consultation with the Environmental Protection Agency (EPA), we conducted six nongeneralizable case studies of final or proposed National Priorities List (NPL) sites with Native American Interest (NAI). We selected these case studies on the basis of geographic diversity and in order to represent sites that have been listed since the publication of EPA’s tribal consultation policy in 2011. For each of these case studies, we collected documentation and interviewed the relevant tribal and EPA regional officials. Figure 2 provides an overview of these case studies. Case Study 1: Creese and Cook Tannery (Former)—EPA Region 1 General Information on the Site According to EPA, the Creese and Cook Tannery site is located on the Crane River in Danvers, Massachusetts. According to an October 2018 proposed cleanup plan, several businesses operated at the site, including leather tanneries that operated from the late 1800s until the early 1980s and a former railroad station. Use of arsenic and chromium at tanneries resulted in these chemicals contaminating soil at the site. Other soil contaminants include dioxins, furans, and polycyclic aromatic hydrocarbons from railroad operations, combustion, and use of asphalt pavement. In the mid-1980s, the Massachusetts Department of Environmental Protection conducted an initial investigation to determine the nature and extent of contamination and evaluate the potential remedial options under state law. The department then reviewed and approved, pursuant to state law, a plan for excavation of the waste and its placement in a containment cell. EPA began investigations in 2010 and found arsenic in surface soils. As a result, in 2012 EPA removed 450 tons of contaminated soil from the site. EPA conducted six site assessments, including an archaeological assessment, and placed the site on the NPL in 2013. Site Status in Cleanup Process The site is in the early stages of the cleanup process. The feasibility study for the site was completed in September 2018, and EPA issued a cleanup proposal for comment in October 2018. According to information provided by EPA, the site has not yet reached any Superfund site-wide milestones because the remedial action has not begun. Tribal Interest in the Site EPA officials stated that both the Mashpee Wampanoag Tribe and Wampanoag Tribe of Gay Head (Aquinnah) have expressed interest in the site due to possible adverse impacts on significant cultural resources in the contaminated area. EPA officials told us they notified both tribes of the site concurrently with notification to the Massachusetts Historical Commission in August 2014. In a consultation response form dated September 2014, the Mashpee Wampanoag Tribe indicated that the cleanup has the potential to have adverse effects on historical or cultural resources important to the tribe and requested that the tribe be notified prior to any archaeological activity on-site, and that they be provided any archaeological assessment documents. EPA’s Consultation and Coordination with the Tribes for the Site The National Historic Preservation Act requires federal agencies to take into account the effects of their undertakings on historic properties, including properties to which Indian tribes attach religious and cultural significance. According to EPA Region 1 officials, they are consulting with both tribes under the act. EPA sent an archaeological survey to the tribes in June 2017. Officials from the Mashpee Wampanoag Tribe indicated that they agree with the survey’s findings and required that consultation continue. EPA officials told us that the Wampanoag Tribe of Gay Head (Aquinnah) did not comment on the assessment. Both tribes have asked EPA to inform them of cleanup status for the site and share any reports. Perspectives of Tribal and EPA Officials on Consultation and Coordination for the Site EPA officials told us they were consulting with both tribes under section 106 of the National Historic Preservation Act. Officials also told us that EPA will negotiate a memorandum of understanding with both tribes once the final cleanup is selected, if it is determined that the selected remedy will have an adverse effect on any resources that are eligible for the National Register of Historic Places. With regard to coordination, both tribes noted that resource constraints prevent their further involvement with the site cleanup process. Officials from the Wampanoag of Gay Head (Aquinnah) tribe indicated that EPA has been available for discussions if the tribe raises an issue. Case Study 2: General Motors (Central Foundry Division)—EPA Region 2 General Information on the Site The General Motors (Central Foundry Division) site is located on the St. Lawrence River in Massena, New York, adjacent to the Saint Regis Mohawk Tribe’s reservation. According to an EPA document, General Motors operated an aluminum die casting plant on the site beginning in 1959 and used polychlorinated biphenyls (PCB) in the manufacturing process through 1980. EPA found contamination in soils and industrial lagoons on the General Motors site property, in groundwater, in the St. Lawrence and Raquette Rivers, in Turtle Cove, and in soils and sediment within the Saint Regis Mohawk reservation. After General Motors’ bankruptcy, ownership of the site was transferred to a trust. This General Motors site was placed on the Superfund NPL in September 1983. Site Status in Cleanup Process According to information provided by EPA, the cleanup of the General Motors site is ongoing, with the last substantial cleanup of the Remedial Design and Remedial Action phase focused on a 10-million-gallon industrial lagoon. To date, contractors have dredged sediment in the St. Lawrence River, Turtle Cove, and Raquette River systems. EPA officials told us that, in addition to these dredging activities, they have completed other significant cleanup work, including installation of a groundwater collection system, installation of a multi-layer cap on the industrial landfill on-site, and demolition of the 1-million-square-foot factory building, EPA officials stated that consultation with the tribe led to excavating a portion of the industrial landfill in order to establish a 150-foot buffer between a landfill on the site and the tribe’s reservation. EPA declared human exposure to contaminants at the site under control in 2008. EPA officials told us there is no requirement to consult with tribes to determine that site-wide milestones have been reached, and that the Saint Regis Mohawk Tribe was not consulted regarding the designation of human exposure under control. Tribal officials do not agree with this determination and stated that EPA has not asked the tribe for any input on this measure. EPA officials responded that while EPA did not consult with the tribe on the human exposure under control environmental indicator, they coordinated extensively with the tribe with respect to cleanup status, strategy, and site-wide milestones prior to making the designation. Tribal Interest in the Site Tribal officials noted concern regarding contamination of tribal property and the effect on subsistence fishing in the St. Lawrence River and tribal member health. The Saint Regis Mohawk Tribe is concerned that PCB contamination from the site is airborne and affecting the health of tribal members. Further, the tribe is concerned that PCB accumulation in fish tissue results in fish that are unsafe to eat in the quantities typically consumed by tribal members who rely on subsistence fishing. See figure 3 below for a fish consumption advisory issued by the tribe because of PCB contamination concerns. Tribal officials also told us the tribe is concerned that PCBs may be transferred through breast milk, exposing future generations to the contamination. Tribal officials told us that tribal members also complain of a strong odor emanating from the site, and have advocated for the tribe to take a more active role in the site cleanup. EPA’s Consultation and Coordination with the Tribe for the Site According to EPA, the agency sent an official consultation letter to the tribe in 2011, as directed by EPA’s 2011 Policy on Consultation and Coordination with Indian Tribes. Consultations with the tribe focused on the tribal role in the cleanup process at the General Motors (Central Foundry Division) site, as well as the Alcoa Aggregation and Reynolds Metals sites, which also affect the tribe. EPA officials told us they have responded to tribal concerns, in part, by agreeing to a stricter treatment threshold for maximum allowable PCB contamination (10 parts per million instead of 500 parts per million), based on the tribe’s objection to the originally-proposed plan. EPA officials also told us that they have responded to tribal concerns by adopting practices to mitigate air contamination during response activities, such as minimizing the size of excavation areas to reduce potential exposure and wetting contaminated soils before removal. EPA officials told us that coordination with the tribe began in the 1980s, and that the region coordinates extensively with the Saint Regis Mohawk Tribe. Additionally, these officials told us that, through annual meetings with tribes in the region and periodic visits to individual tribes, they coordinate with all tribes in the region, including the Saint Regis Mohawk Tribe, at least once a year. In technical comments provided in response to the draft of this report, EPA officials told us that the Saint Regis Mohawk Tribe has been treated as a support agency, equivalent to the state of New York, since 1995, and that the tribe has been asked to concur on all records of decision for the site as early as 1990, though they have not always concurred. Perspectives of Tribal and EPA Officials on Consultation and Coordination for the Site Tribal and EPA officials have differing perspectives on the effectiveness or utility of consultation. Saint Regis Mohawk Tribe officials noted that they have met repeatedly with EPA over the years but the consultation has felt perfunctory and like a “box checking exercise.” Tribal officials stated that EPA did not consider their input as seriously as General Motors’ input, and they believe that EPA is over-reliant on the initial research conducted by scientists from the company, and has not sufficiently considered updated and independent research. Saint Regis Mohawk tribal officials noted that EPA did not recognize tribal members’ stronger reliance on the environment and exposure to contamination. The tribe also provided us with examples of less formal coordination with EPA, including a letter from EPA responding to tribal officials’ requests for additional air monitoring at the site. EPA Region 2 officials stated that consultation with the Saint Regis Mohawk Tribe has become more extensive and sophisticated since the issuance of the 2011 tribal consultation policy. The region held a consultation with the tribe in 2011 to address coordination with the tribe about three Superfund sites. In a summary of that consultation, EPA noted that they will take steps to further the tribe’s partnership role with respect to the three sites by providing as much time and opportunity as feasible for consultation, consistent with the mutual desire to move the cleanups forward expeditiously; continuing to share, for advance review, drafts of pertinent documents; consulting with the tribe prior to taking actions or implementing decisions that may affect the tribe’s interests; inviting tribal officials to technical meetings where potentially responsible parties and other trustees are present; and informing the tribe of the results of meetings or substantive decisions with any potentially responsible party. Further, EPA officials noted that they cannot fulfill some requests made by the Saint Regis Mohawk Tribe; however, EPA officials stated that tribal activism led to a more stringent 10 parts-per- million treatment threshold for PCBs on the site, rather than the originally proposed 500 parts-per-million standard. EPA also provided documentation of less-formal coordination with the tribe, including correspondence regarding approaches to addressing the tribe’s concerns of PCB air impacts during cleanup. Case Study 3: Petoskey Manufacturing Company (PMC) Groundwater—EPA Region 5 General Information on the Site According to information provided by EPA, the PMC Groundwater site is located in a former industrial area on the shores of Lake Michigan’s Little Traverse Bay in Petoskey, Michigan. PMC was established in 1946 as a small fabricating and painting business that later produced parts for the automotive industry until 2000. During this period PMC improperly disposed of solvents used in plant operations, contaminating groundwater and Petoskey’s municipal well with volatile organic compounds and inorganic contaminants. Site Status in Cleanup Process According to EPA officials, the agency has gone through several rounds of cleanups at PMC Groundwater. EPA initially listed the PMC Groundwater site on the NPL in 1983. The City of Petoskey completed construction of a new municipal water source in 1996. EPA began cleanup in 1999 and declared the site as ready for anticipated use in 2007; the site was subsequently redeveloped with condominiums. In the site’s 2014 5-year review, EPA noted that the remedies they had put in place, including excavation and off-site disposal of contaminated soil, installation and operation of a system to remove volatile organic compounds from subsurface soil, and a groundwater monitoring plan, were protective of human health and the environment in the short term, but that an effective long-term remedy would require additional steps. According to EPA officials, EPA is conducting a remedial investigation and feasibility study to determine the nature and extent of soil and groundwater contamination, which is expected to be completed in 2019. According to EPA officials, in 2016, EPA fieldwork indicated that trichloroethene concentrations exceeded acceptable levels under some condominiums’ slab foundations, and in 2017, EPA conducted an emergency removal action to address the intrusion of the vapors. Tribal Interest in the Site Little Traverse Bay Bands of Odawa Indians officials told us the tribe’s interest in the site is due to potential exposure of tribal members and the effects on nearby surface waters. Tribal members rely on subsistence fishing in the Bear River in close proximity to the site. These officials also told us the tribe also conducts commercial fishing in Lake Michigan. Tribal members residing in Petoskey relied on the contaminated municipal well. Additionally, tribal officials told us that they want to understand the status of the site because they may be interested in future land acquisitions in the area and the U.S. Department of the Interior may not be willing to take contaminated land into trust for the tribe. EPA’s Consultation and Coordination with the Tribe for the Site According to tribal officials, the tribe contacted EPA officials in 2017 when local news reported vapor intrusion issues into condominiums built on the site. Neither tribal officials nor EPA have found any indication of previous consultation or coordination for the site. Since the tribe’s initial contact, EPA officials have shared relevant information and spoken with the tribe regarding the site. EPA officials told us that representatives from the tribe attended a public meeting about the site in June 2018 and that EPA is in close contact with an official from the tribe and will provide him with reports as appropriate. Perspectives of Tribal and EPA Officials on Consultation and Coordination for the Site According to EPA and tribal officials, EPA has not consulted with the tribe about the site. With respect to coordination, tribal officials told us that they were satisfied with EPA’s response following the tribe’s initial contact. EPA officials told us that the tribe is aware that consultation is available if the tribe desires it, and officials will coordinate with the tribe. EPA officials stated that the relationship with tribes in the region has evolved considerably since the 1990s and that coordination with tribes in the region has improved. Case Study 4: Jackpile-Paguate Uranium Mine— Region 6 General Information on the Site According to information provided by EPA, the Jackpile-Paguate Uranium mine is a 2,656-acre site located on the Pueblo of Laguna, New Mexico, about 40 miles west of Albuquerque. Anaconda Copper Mining and The Anaconda Company, predecessors to the Atlantic Richfield Company, moved more than 400 million tons of rock within the mine between 1952 and 1982 area in addition to 25 million tons of uranium ore off-site for additional processing. Mining operations contaminated surface water with hazardous substances. Additionally, according to a report by the U.S. Department of Health and Human Services, people living in villages near the site could be exposed to contamination through radioactive materials from the site being used in home construction, or through contact with mine contaminants suspended in air or present in dust blown or tracked from the mine. Reclamation of the mine began in 1990 and was closed out in June 1995; however, EPA was not involved in the initial reclamation prior to the site being listed on the NPL. Figure 5 is a photograph of Gavalon Mesa, one of the major mining areas at the site, and erosion typical to a previously reclaimed area. Site Status in Cleanup Process EPA listed the site on the NPL in 2013. EPA officials conducted four assessments at the site. The site is currently in its remedial investigation and feasibility study stage, and the site has not met any site-wide milestones. Tribal Interest in the Site The site is located within the boundaries of the Pueblo of Laguna’s reservation. Pueblo of Laguna officials stated that the site impacted the Pueblo in several ways, including radon contamination in homes due to use of contaminated mining debris in home construction, contamination of water sources, and dust from mining operations reaching homes and gardens. EPA’s Consultation and Coordination with the Pueblo for the Site EPA officials stated that neither EPA nor the Pueblo of Laguna have initiated consultation for the Jackpile-Paguate Uranium Mine under the 2011 consultation policy. EPA consulted with the tribe for the site in 2009, which resulted in a memorandum of understanding (MOU) to facilitate coordination in performing removals and site assessments for the site. According to EPA officials, once the remedial investigation and feasibility study is complete, they will seek to consult with the tribe before making a decision about cleanup goals. EPA officials noted that the agency has consistently coordinated with the tribe, including regular briefings to the tribe and working closely with the tribe’s Environmental and Natural Resources Department since EPA became involved at the site. In addition, the tribe is a support agency for the site—which means EPA must provide the tribe substantial and meaningful involvement in the initiation, development, and selection of the remedial action at the site. The Pueblo has a Superfund support contract with EPA to facilitate its support agency work helping EPA perform oversight of the response work, and reviewing and commenting on EPA documents, according to EPA officials. Perspectives of Pueblo and EPA Officials on Consultation and Coordination for the Site Pueblo officials told us that they have been satisfied with the coordination for the site, and they prefer that coordination be face-to-face when possible. Officials told us that consultation requires a senior EPA official to present in person to the Pueblo Council, and all other interactions are considered coordination. According to the Pueblo, coordination with EPA has been effective, in part, because EPA acknowledges that site contamination extends beyond the mine lease boundaries. EPA officials told us that they are in frequent communication with the Pueblo. EPA officials noted that they hold regular briefings with tribal officials, as well as through routine electronic and phone communication. EPA officials noted that coordination with the tribe early in the Superfund cleanup process facilitates their work. For example, since the site is on tribal property, EPA worked with the Pueblo to gain site access to investigate the extent of the contamination. Case Study 5: Smurfit-Stone Mill Frenchtown—Region 8 General Information on the Site According to information provided by EPA, the Smurfit Stone Mill- Frenchtown site is a 3,200-acre area located northwest of Missoula, Montana. The site was originally a pulp mill operated from 1957 through 2010. It includes more than 900 acres of unlined ponds that were used to store wastewater effluent from the mill, as well as sludge recovered from untreated wastewater. Contamination includes dioxins and furans produced through bleaching of pulp, as well as PCBs. Site Status in Cleanup Process EPA proposed to add the site to the NPL in 2013 and is evaluating public comments on the proposal before making a final decision. EPA negotiated an administrative settlement agreement and order on consent in 2015 with three potentially responsible parties to conduct a remedial investigation and feasibility study at the site. EPA officials told us that these parties have completed several site tasks contributing to the remedial investigation and feasibility study for the site. Tribal Interest in the Site Both the Confederated Salish and Kootenai Tribes of the Flathead Reservation and the Kalispel Indian Community of the Kalispel Reservation (hereafter Kalispel or Kalispel Indian Community) have interest in the site. Officials from the Confederated Salish and Kootenai Tribes of the Flathead Reservation stated that their interest in the site is drawn from the Hellgate Treaty of 1855. According to these officials, the site is located on land where the tribes retain treaty hunting, fishing, and gathering rights in portions of the Clark Fork River that are potentially contaminated by the site. The two tribes are concerned about adverse health impacts on tribal members due to exposure through consumption of fish from near and downstream from the site and ensuring that tribal cultural and historical resources are protected during cleanup activities. Officials from the Kalispel Indian Community believe that contaminants from the site and throughout the watershed have reached its reservation in Northeast Washington. These contaminants may affect tribal members’ nutrition and exercise of their culture. The tribe would like EPA to sample for contamination from Smurfit Stone Mill further down the Clark Fork River to the areas where the Kalispel have interest. EPA’s Consultation and Coordination with the Tribes for the Site According to EPA officials, EPA has not consulted with the tribes but has coordinated with the natural resource trustees, which include the Confederated Salish and Kootenai Tribes, and told us they have also coordinated with the Kalispel Indian Community. EPA officials told us that coordination with the Kalispel Indian Community differs from coordination with the Confederated Salish and Kootenai Tribes because the Kalispel do not have treaty rights at the site. Region 8 notified the Confederated Salish and Kootenai Tribes about the site in 2014, but told us they did not send corresponding notification to the Kalispel Indian Community because they had not been identified as having tribal interest during the preliminary assessment and site investigation. EPA officials told us the reason they have not yet consulted with the tribes under the 2011 policy is that the site is still being characterized. According to officials from the Confederated Salish and Kootenai Tribes, they were first informed of the site by the Missoula County Water Quality district in 2012. Officials from this tribe told us that in December 2012, they sent a letter to the state Governor supporting NPL listing for the site, and also indicated their support of NPL listing to EPA when responding to a Federal Register notification indicating EPA’s intent to add the site to the NPL. EPA officials told us that the agency wants to improve communication with the tribes by scheduling quarterly calls, site visits, and offering opportunities to review and comment on documents produced during the remedial investigation process. Perspectives of Tribal and EPA Officials on Consultation and Coordination for the Site Officials from the Confederated Salish and Kootenai Tribes have been dissatisfied with the extent of coordination with EPA. Specifically, they told us that EPA has not provided the tribes with sufficient information to engage in the cleanup process in a meaningful way. For example, officials stated EPA did not involve them when EPA entered into the administrative settlement agreement and order on consent to conduct the remedial investigation and feasibility study. Tribal officials told us that this experience is inconsistent with other Superfund sites where EPA has given the tribes greater opportunity for meaningful input. EPA officials told us they coordinated with the interested tribes through communications with the natural resources trustees in the region as a whole. EPA officials told us that they officially notified the tribes about the site after the preliminary assessment and site investigation, and that they typically do not issue a trustee notification letter or invite tribes to consult until after EPA completes a preliminary assessment. Officials told us that the Confederated Salish and Kootenai Tribes was notified at the same point as other natural resource trustees, and that this was sufficiently early to allow for meaningful input because it occurred prior to any major decisions. According to Kalispel tribal officials, coordination with EPA has been limited. Kalispel tribal officials told us that they have faced some difficulties coordinating with EPA about the site because they are located in EPA Region 10, while the site is managed by EPA Region 8. One tribal official we spoke with expressed that he felt EPA may be trying to exclude the Kalispel Indian Community from cleanup decisions at the site. For example, this official told us that the tribe had requested that EPA Region 8 extend their water sampling area further downstream on the Clark Fork River to determine the extent of releases from the site, but that EPA issued its sampling plan without taking the tribe’s concerns into account. However, these officials told us that they are developing their relationship with EPA region 8. They also told us that coordination with EPA is valuable, and that they consider consultation as a tool to be employed when coordination is insufficient. Region 8 officials acknowledged the letter from the natural resource trustees requesting a stronger role in decision-making and highlighted improvements EPA has made to communication. Further, officials cited several actions to demonstrate their commitment to working with the tribes: evaluating the berms at the site, as the Confederated Salish and Kootenai Tribes requested; evaluating contamination’s impact on tribal health through fish consumption patterns; and responding in writing to natural resource trustee letters. However, EPA considers the role of the Kalispel Indian Community in the cleanup to be different because that tribe does not have treaty rights within the site boundaries. EPA officials stated that they keep the tribe informed of meetings and invite them to site visits. Figure 6 shows the berms during a high-water event in 2011 and a portion of a berm indicated to be in poor condition by the work plan for the remedial investigation and feasibility study in 2017. Case Study 6: Midnite Mine—Region 10 General Information on the Site The Midnite Mine site is a former open-pit uranium mine located in eastern Washington state on the Spokane Indian Reservation, near Wellpinit, Washington. According to information from EPA, Dawn Mining Company and Newmont USA Limited operated an open-pit uranium mine intermittently between 1955 and 1981. During mining operations, over 33 million tons of rock was blasted and excavated to access uranium ore. The waste was dumped in piles, used to fill mine pits, or spread on the surface. About 2.4 million tons of ore and near ore-grade rocks were also stockpiled at the mine in anticipation of later processing. The former mine site includes approximately 350 acres directly affected by mine operations, as well as affected groundwater, surface water, and sediment. Hazardous substances released at the site as a result of mining include numerous metals and radio-nuclides. Key contaminants of concern that EPA identified in the human health risk assessment for the site include uranium, radium, lead, and manganese. Site Status in Cleanup Process According to EPA, construction of the remedies is currently under way for the site. EPA listed the site on the NPL in 2000 and performed the remedial investigation and feasibility study from 1998 through 2006. In 2012, the potentially responsible parties and the United States agreed to a consent decree that required the potentially responsible parties to develop a design for and implement the remedial action at the site. No site-wide milestones have been met. Tribal Interest in the Site According to tribal officials, the Spokane Tribe of Indians is interested in the effect of contamination from the site on subsistence hunting and fishing, particularly elk and rainbow trout, respectively. Tribal officials stated that contamination from the mine flows into Blue Creek, which impacts the tribe’s ability to conduct traditional practices such as sweat lodges. Tribal officials stated their ultimate goal would be for the site to be sufficiently clean for wildlife to safely live on the site, for fish to thrive in water adjacent to the site, and for the tribe to resume its traditional hunting and gathering activities in the area. EPA’s Consultation and Coordination with the Tribe for the Site EPA consulted with the Spokane Tribe of Indians in June 2013 regarding a potential change to water treatment practices. Tribal officials stated the tribe is pleased that the new water treatment plant will operate year-round and will discharge treated water via a pipe into Lake Roosevelt, which is a larger body of water with less direct impact on the tribe’s natural resources. In addition, tribal officials stated that EPA invited the tribe to consult at other times but the tribe did not think it was necessary. Perspectives of Tribal and EPA Officials on Consultation and Coordination for the Site Tribal officials told us that their coordination with EPA has resulted in more consideration of the natural resources and hopefully a fuller remediation of the site. For example, EPA applied the tribe’s more stringent water quality standards to discharge from the site, which EPA supported by providing technical assistance to the tribe during the development and approval processes. Spokane tribal officials stated that during the Remedial Investigation and Feasibility Study phase, EPA’s program manager offered to consult with the tribe at various points, which the tribe declined because the tribe felt they had sufficient interactions with EPA. The Superfund cleanup process has been a learning process for tribal officials but, overall, the tribe is pleased with the result and the open exchange of information with EPA. Speaking generally, EPA officials noted that the 2011 consultation policy has had a positive effect on the frequency of consultation with tribes in the region. The policy has led Superfund remedial project managers to more routinely invite tribes to consult. Appendix IV: Comments from the Environmental Protection Agency Appendix V: Comments from the Confederated Salish and Kootenai Tribes of the Flathead Reservation Appendix VI: Comments from the Pueblo of Laguna Appendix VII: GAO Contact and Staff Acknowledgements GAO Contact Staff acknowledgements In addition to the individual named above, Barbara Patterson (Assistant Director), Emily Norman (Analyst-in-Charge), Matthew Bond, John Delicath, Justin Fisher, Andrew Furillo, Jeanette Soares, Ruth Solomon, Sara Sullivan, and Kiki Theodoropoulos made significant contributions to this report.

Why GAO Did This Study Superfund is EPA's principal program to address sites with hazardous substances, and some of the most seriously contaminated of these sites are listed on the NPL. Many of these sites can affect Indian tribes or their land. EPA has a policy to consult with tribes when EPA actions or decisions may affect tribal interests, including on cleanup of NPL sites that are on tribal property or that affect tribes. GAO was asked to analyze NPL sites that are on tribal property or that affect tribes and EPA's consultation with tribes at these sites. This report: (1) examines the extent to which EPA has reliable data identifying NPL sites that are located on tribal property or that affect tribes, (2) examines the extent to which EPA has reliable data on the agency's consultation with tribes regarding NPL sites, and (3) describes the actions EPA has taken to address the unique needs of tribes when making decisions about cleanup actions at Superfund sites. GAO reviewed laws and policies, assessed EPA data on NPL sites, and interviewed EPA and tribal officials about cleanup actions and consultations at six non-generalizable NPL sites selected in part for their geographic diversity. What GAO Found The Environmental Protection Agency (EPA) does not have reliable data identifying National Priorities List (NPL) sites that are located on tribal property or that affect tribes. Specifically, EPA collects data on whether sites are on tribal property or have Native American Interest (a data variable indicating sites where tribal members or tribal land would be directly affected by the release of hazardous substances), as well as which tribes are associated with NPL sites. However, EPA's data are not always accurate or complete for a number of reasons. For example, EPA can have difficulty identifying some tribal property boundaries, and NPL site boundaries may evolve as the site is investigated and remediated. EPA does not have a regular review process for its data on whether an NPL site is on tribal property. In addition, EPA's guidance for determining whether a site has Native American Interest is unclear, and regions may not interpret it consistently. Without improving its review process and clarifying its guidance, EPA will not have reasonable assurance that its data on tribes that are affected by NPL sites are accurate or complete. EPA consults with tribes when actions at an NPL site may affect tribal interests, but the agency does not have reliable data on its consultations with tribes. Data from EPA's system for tracking consultation did not include documentation of some consultations that GAO confirmed had occurred. One possible reason that EPA data are incomplete is that the agency's policy is unclear on which interactions are considered consultation and are therefore to be documented in EPA's system of record, which is not consistent with federal standards for internal control. EPA's policy provides a broad definition of consultation and specifies which staff are responsible for determining when consultation may be appropriate. However, the policy does not provide further guidance on the circumstances under which consultation should be considered. For example, it does not specify any specific points in the hazardous substance cleanup process at which consultation should be considered or provide further detail on which tribal interests should be considered when determining if tribal interests on NPL sites are affected. Without clarifying guidance to clearly define circumstances under which consultation with tribes should be considered, EPA cannot have reasonable assurance that it is applying its consultation policy consistently. EPA has taken various actions to address the unique needs of tribes when making decisions about cleanup actions. These actions include minimizing tribal members' exposure to contaminants because of tribal lifestyle (e.g., greater consumption of local fish and game) and limiting potential damage to culturally important sites. For example, EPA officials said that at one site, they altered the design and route of the roads used to remove contaminated materials to minimize the impact of cleanup activities' on cultural resources. EPA also published a memorandum in 2017 with recommendations on considering tribes' traditional ecological knowledge in the cleanup process if tribes offer it. What GAO Recommends GAO is making four recommendations to EPA, including that it take actions to improve the data it collects and to clearly define circumstances under which consultation with tribes should be considered. EPA generally agreed with GAO's recommendations.

gao_GAO-19-126

gao_GAO-19-126_0

Background This section describes the (1) Kansas City site’s role in providing nonnuclear parts and components, and (2) current and planned nuclear weapons stockpile life extension and alteration efforts that drive workload. Kansas City Site’s Role in Providing Nonnuclear Parts and Components The Kansas City site is NNSA’s primary site for procuring or producing nonnuclear parts and components, providing over 80 percent of the parts and components that compose a typical nuclear weapon. The Kansas City site interacts with a number of other NNSA sites that comprise the nuclear security enterprise to support the nuclear weapons stockpile. For example, NNSA’s design laboratories develop precise parts or component specifications or requirements to which production sites, such as the Kansas City site, must conform in procuring or producing these items for use in the nation’s nuclear weapon stockpile. Figure 1 depicts how sites in the nuclear security enterprise interact with each other to design, produce, procure, and assemble nonnuclear components. Components procured or produced by the Kansas City site range from simple items such as nuts and bolts to more complex components such as radars, arming and firing mechanisms, and critical nuclear safety devices meant to prevent accidental detonation. The site delivers approximately 100,000 parts annually, according to our previous report. According to Kansas City site contractor documents, the primary mission of the site is keeping the nation’s nuclear stockpile safe, secure, and reliable by delivering mission-critical mechanical, electrical, and engineered material components and services. NNSA and the Department of Defense (DOD) jointly manage LEPs and Alts under a multi-step process known as the phase 6.X process (see fig. 2). Phase 6.4 of this process, or the production engineering phase, involves activities to adapt designs for production and prepare production facilities, including the Kansas City site. For example, according to a senior NNSA official, activities to adapt designs could include updating product specifications to make parts easier to produce, changing or refining tester limits, and substituting among commercial off-the-shelf parts. The B61-12 LEP and W88 Alt 370 are currently in phase 6.4 (production engineering) of this process and are approaching production. Other LEP efforts are in earlier phases. Current and Planned Nuclear Weapons Stockpile Life Extension and Alteration Efforts NNSA describes its plans to meet nuclear weapons stockpile life extension and alteration goals in two key documents that also describe NNSA’s operations and budget estimates for implementing these plans. These documents, which NNSA updates annually, constitute NNSA’s nuclear security budget materials. First, the Stockpile Stewardship and Management Plan is NNSA’s formal means of communicating to Congress information on modernization and operations plans and budget estimates over the following 25 years. Second, NNSA’s annual justification of the President’s budget provides program information and budget estimates for the following 5 years. This 5-year plan is called the Future-Years Nuclear Security Program (FYNSP), and the budget estimates in this plan reflect amounts approved by the Office of Management and Budget. These estimates align with those presented for the first 5 years included in the Stockpile Stewardship and Management Plan. According to the Fiscal Year 2018 Stockpile Stewardship and Management Plan, NNSA and its nuclear security enterprise are conducting a substantial level of activity to ensure the continued credibility of the nation’s nuclear weapons stockpile. Specifically, in fiscal year 2018 NNSA was executing three nuclear weapons LEPs and one major Alt, which are described in table 1. In addition, the 2018 Nuclear Posture Review calls for NNSA to resume a program to replace the W78 warhead in fiscal year 2019; produce a low- yield submarine launched ballistic missile warhead, known as the W76-2; and consider options for providing a nuclear warhead for a potential sea- launched cruise missile. According to NNSA officials and contractor representatives, NNSA developed an early production planning roadmap for implementing the Nuclear Posture Review in late 2018. The conference report accompanying DOE’s fiscal year 2019 appropriations act directed the agency to spend a specified amount on the W78 warhead replacement and W76-2 efforts. Projected Workload for the Kansas City Site Has Increased Significantly from 2012 Forecasts Projected workload for the Kansas City site has increased significantly, based on NNSA’s stockpile plan changes from 2012—when the new modern facility was built—to the 2018 stockpile plan update. A comparison between the 2012 and 2018 plans shows that the start of full production for the B61-12 LEP and the W88 Alt were delayed by approximately 2 years, and their completions were delayed by 3 years from initial schedule estimates in 2012. The 2018 plan also accelerates production of the W80-4 LEP by approximately 5 years. Figure 3 below shows the change in the full production timelines for key weapons systems. Using an enterprise risk management approach, the Kansas City site determined that this change in production schedule represented a significant challenge that needed to be better understood and regularly monitored. NNSA contractor representatives at the Kansas City site developed a strategy for analyzing workload to better understand the enterprise risk and ensure the site’s ability to provide an adequate supply of nonnuclear components under variable requirements scenarios. Specifically, in 2015, the Kansas City site increased the frequency of using its “what-if” approach that models standard production work and allows for an in-depth review of labor, equipment, and material capacity information, according to contractor representatives at the Kansas City site. This analytic capability is intended to help ensure that the site contractor can accurately predict future workload demand across multiple scenarios representing different production requirements. Contractor representatives update the model every quarter to reflect the current hardware schedules; testing requirements; and nuclear weapon scope, production quantities, and schedules. These representatives use the model to develop hourly staffing, equipment, and other capacity-related forecasts and plans. For example, contractor representatives evaluate capital equipment capacity quarterly for multiple programs, with a primary focus on equipment that is at or above a two-shift capacity. However, according to these representatives, this approach has not been in place long enough to allow comparison of historical data with forecasts from the model to assess their accuracy. According to site contractor documents and representatives, forecasting data from the what-if models project that, under the 2018 plans, the full- time equivalent workload for production of nonnuclear parts and components will continue to increase annually through 2020. Specifically, the number of production and administrative staff at the time of the relocation to the new facility in 2014 was almost 2,500, based on needs at that time. However, the fiscal year 2018 updates, based on “what-if” capacity analyses, now show that the headcount will need to almost double, growing to more than 4,900 administrative and production staff by 2020. For example, according to 2018 “what-if” capacity analyses prepared by site contractor representatives, personnel dedicated exclusively to two efforts—the B61-12 LEP and W88 Alt 370—will double from 251 full-time equivalents needed in fiscal year 2018 to over 500 during fiscal years 2020 through 2022, as shown in figure 4. Full-time equivalent reflects the total number of regular straight-time hours (i.e., excluding overtime or holiday hours) worked by employees divided by the number of compensable hours applicable to each fiscal year. Annual leave, sick leave, and compensatory time off and other approved leave categories are considered to be “hours worked” for purposes of defining full-time equivalent employment. In this figure, the full-time equivalents reflect workload forecasts for hourly production staff only for the B61-12 LEP and W88 Alt 370. Kansas City Site Has Identified and Begun to Mitigate Several Management Challenges Related to Forecasted Workload, Which May Further Increase NNSA officials and contractor representatives at the Kansas City site have identified and begun to mitigate several management challenges to meeting the forecasted workload for known future production requirements, but they face uncertainties about future workload demands. Specifically, current mitigation efforts should help the site meet currently forecasted increased workload and capacity demands, according to NNSA analysis and consistent with the program plan included in the Fiscal Year 2018 Stockpile Stewardship and Management Plan. However, the February 2018 Nuclear Posture Review, the results of which were not fully reflected in the Fiscal Year 2018 Stockpile Stewardship and Management Plan, may change requirements and add to the site’s workload because it calls for additional weapons efforts. Current Efforts to Mitigate Identified Challenges Should Support Meeting Forecasted Future Workload Increases at the Kansas City Site Kansas City site contractor representatives have identified management challenges that could affect the site’s ability to meet forecasted future workload increases based on 2018 analyses and its Enterprise Risk Management process, and NNSA officials agreed with the challenges the contractor representatives identified. These management challenges include ensuring that the site has (1) sufficient production and administrative office space, (2) up-to-date production equipment, (3) a sufficient workforce with necessary security clearances, (4) capable and reliable external suppliers, and (5) complete weapons designs early enough in development to minimize production changes and delays. The Kansas City site has identified strategies to mitigate the effects of each of these management challenges and has begun taking steps to implement these strategies. NNSA’s Enterprise Modeling and Analysis Consortium NNSA’s Enterprise Modeling and Analysis Consortium is composed of NNSA site representatives and program representatives from NNSA’s Defense Programs offices and is a principal source for NNSA model- informed analytics for decisions about stockpile stewardship program management, policy, and implementation. The consortium conducts modeling based on common data sets and assumptions of current and planned stockpile plans, design alternatives, commodity requirements, and nuclear security enterprise capacity. One of the consortium’s projects includes analyzing the nuclear security enterprise’s capacity to execute the nuclear weapon production program of record to identify any important issues or bottlenecks within or between sites. NNSA analysis concludes that current mitigation efforts initiated at the Kansas City site should support currently planned increased workload and an increased capacity to achieve the 2018 workload forecast. Specifically, according to analyses conducted by NNSA’s Enterprise Modeling and Analysis Consortium, the Kansas City site’s operations will be stressed above current capacity for multiple consecutive years in the future, and current mitigation efforts should reduce risk associated with the elevated workload. In addition, NNSA Kansas City Field Office and headquarters officials said that they have high confidence in the ability of the Kansas City site to forecast and manage infrastructure and staffing needs at the site to support currently planned nuclear weapon stockpile life extension needs over the coming decades. In particular, NNSA’s recent annual performance evaluation reports—which document the contractor’s overall performance for a fiscal year—show that the Kansas City site contractor has delivered the vast majority of hardware on time, within budget, and in a safe and secure environment. Production and Administrative Office Space Kansas City site officials indicated that ensuring adequate production and administrative office space at the site is a management challenge because the current facility is too small to accommodate future workload. Specifically, forecasted workload demand has grown significantly since the modern facility was built in 2012. The new facility, which accommodates both production and administrative staff, replaced a deteriorating World War II-era facility that was much larger and had significant maintenance and operations costs, according to site contractor representatives. For example, according to NNSA documents, the move reduced the footprint of the site’s production activities from about 3 million square feet to 1 million square feet. According to site contractor representatives, the modern facility was designed to be more flexible in accommodating changes in the production line. For example, equipment can more easily be removed or installed at any location in the facility, to accommodate increased workload, because there is ready access to electrical, ventilation, or other necessary hookups and connections. Figure 5 shows a photo of the new facility. The Kansas City site has identified that it needs an additional 250,000 square feet of production space in 2019 and ultimately a total of an additional 400,000 square feet to support the forecasted workload and associated staff increase. To mitigate the challenge of insufficient production and administrative space to support the forecasted increase in production staff, Kansas City site officials told us they are pursuing multiple short- and long-term strategies. With respect to production space, under the short-term plan the Kansas City site is pursuing a temporary lease of commercial space to allow for the offsite storage of unclassified materials that are currently at the production facility. According to site contractor representatives, this new lease would free up production space at the main site. Further, the site submitted a request to NNSA for leasing an additional 250,000 feet of production space—an increase of almost 30 percent over current production space in the modern facility. Kansas City site contractor representatives stated that the cost of this lease will be based on competitive offers, and they expect the lease to be awarded by summer 2019. With respect to administrative office space, the site has leased more than 150,000 square feet of space since 2014 for the short term at a cost of more than $3.5 million per year. Under the long-term plan, expected to take a minimum of 5 years to implement, the site will complete an analysis of alternatives and submit a combined office and production space expansion project plan to NNSA, which will determine final costs and timelines. Currently, the mission need statement for the project indicates the need for over 400,000 square feet of additional production and administrative space to accommodate the planned increased workload for known production and supporting administrative requirements—an increase of roughly 50 percent over current leased production space. According to Kansas City site contractor representatives, at this early stage, costs would be based on the current Kansas City site lease of $43 per square foot, or roughly $17 million per year. This long-term plan would include space for approximately 1,200 administrative personnel. Kansas City site contractor representatives told us in September 2018 that, depending on the selected long-term solution, the short-term leases for administrative space could either be terminated or modified into long- term arrangements. In 2017, we reported in our high risk list update that federal agencies have not demonstrated that they have the capacity to reduce their reliance on costly leases, particularly high-value leases—defined as $2.85 million and above per year in lease costs—where owning properties would be less costly in the long run. In particular, we reported that the General Services Administration had not implemented our 2013 recommendation to develop a strategy to increase ownership of investments for a prioritized list of high-value leases where ownership would be less expensive in the long run. The Kansas City site’s plans for significantly expanding its production and office space underscores the challenges that exist in meeting these space needs while at the same time limiting overall reliance on costly leases. Production Equipment NNSA and its contractor at the Kansas City site have identified challenges in ensuring that the plant has up-to-date production equipment. Recapitalizing equipment was not a significant part of the move to the new modern production facility, according to site contractor representatives. Information from NNSA’s Master Asset Plan 2017, for example, states that most of the equipment used for producing nonnuclear parts and components at the Kansas City site is nearing or past the end of its useable life—defined as 15 years. Specifically, as shown in figure 6 below, 39 percent of the equipment at the Kansas City site is from 6 to 15 years old, and 27 percent is 16 years old or more, according to the plan. Much of the oldest equipment is located in functional areas used for machining, refurbishment, and dismantlement operations, or for production functions using rubber and plastics. The oldest piece of equipment still maintained is more than 60 years old. In addition to age-related challenges, officials at the Kansas City site identified equipment challenges regarding capacity, based on an equipment workload forecast analysis performed in 2015. For example, according to this forecast, starting late in calendar year 2019, demand for vibration- and shaker-test equipment will become consistently greater than existing capacity, requiring additional equipment. To address these challenges, Kansas City site contractor representatives stated that they evaluate equipment needs across the facility at least annually, based on production and maintenance schedules. The representatives then develop a master list of equipment requests— weighted for risk, age and condition of existing equipment, and whether an external supplier can provide the functional need, among other factors—and ranked according to current and future business needs, according to these officials and contractor representatives. NNSA officials at the Kansas City site and senior contractor representatives then review the master list to determine priorities for equipment purchases. Site contractor representatives are developing a 10-year equipment strategy, expected to be completed in December 2019, to sharpen focus on the future needs of the production facility to support capacity and capability, according to NNSA officials at the Kansas City site. Budgets for equipment procurements at the Kansas City site vary from year to year and are subject to change. According to Kansas City site contractor representatives, the site is regularly adjusting and communicating its equipment needs to reflect the results of equipment evaluations to ensure that the funding NNSA will request for equipment procurement is adequate. For example, according to Kansas City site contractor representatives, the site originally received $4.5 million in fiscal year 2018 to fund planned equipment procurements and received an additional $13.4 million from NNSA in April 2018 to move fiscal year 2019 work scope into fiscal year 2018. The remaining funding available is $11.6 million, which covers the remaining fiscal year 2019 work scope. Site plans for fiscal year 2018 specifically included capital equipment replacement and upgrades needed for parts assembly, electronics and fabrication, and non-destructive testing of nonnuclear parts and components. For fiscal year 2019, planned procurements include equipment for testing of parts and components, rubber- and plastics-related production, precision milling, machining and welding, paint and heat treatment, fabrication, and chemical processing. NNSA officials at the Kansas City site stated that planned budgets for fiscal years 2019 through 2023— which currently include $8 million in equipment procurements and $2 million for area modifications for each of the 5 years—are subject to adjustment based on ongoing evaluation of site equipment needs. These estimates could change, depending on the outcome of the 10-year equipment strategy, according to NNSA Kansas City site officials and contractor representatives. In addition to new equipment procurements, the Kansas City site has developed other mitigation plans also focused on equipment capacity risks. For example, these plans include options such as better allocating equal workload amongst similar equipment, and additional batching of material, according to Kansas City site officials. The batching of material processed by a certain set of equipment increases efficiencies because it consolidates material into larger portions, which minimizes inefficiencies associated with starting and stopping the equipment multiple times, according to NNSA contractor representatives. Workforce Kansas City site officials and contractor representatives have identified three management challenges in ensuring the site can achieve a sufficient contractor workforce to meet forecasted future workload: (1) retention of existing staff, (2) recruiting skilled staff in a competitive job market, and (3) obtaining security clearances for new staff in a timely manner. To address these challenges, the site has been taking actions to retain existing staff, hiring and recruiting hundreds of new staff, and working to speed the security clearance process, according to site contractor representatives. Kansas City site contractor representatives said that retaining existing staff is challenging because the majority of the workforce falls into either of two categories: (1) recent, younger hires who have a high attrition rate, or (2) staff eligible to retire. More than half (53 percent) of all staff have 5 years or less of service working at the site (see fig. 7). In addition, approximately 32 percent of the Kansas City site’s contractor staff are eligible to retire. Figure 8 shows the distribution of staff by age at the Kansas City site, with the highest number in their late 50s and the next highest number in their late 20s. According to Kansas City contractor representatives and NNSA documents, site strategies for retaining newly hired and retirement-eligible staff include improvements in rewards and recognition programs, along with an emphasis on pay for performance. Contractor representatives also noted that the site offers telecommuting from a home office for those approved, flexible work hours—such as working 9-hour days to allow for a day off every 2 weeks—and flexible work options, including part-time employment. To better retain retirement-eligible staff the site has also created talking points to better prepare managers to discuss retirement and delayed retirement, covering topics such as the potential for reduced hours or returning to work after retirement, consistent with certain restrictions and policies. Because of these steps, according to Kansas City contractor representatives, many retirement-eligible staff are electing to continue to work; projected retirements are less than 20 percent of those eligible for retirement, based on actual retirement data for years 2013 to 2017. For example, although an employee may be eligible to retire at age 55 with at least 25 years of service, contractor representatives we interviewed noted that most retirements on average are at age 62 with 30 years of service. Kansas City site contractor representatives we interviewed have identified a management challenge in recruitment because of a gap between the critical technical skills needed at the site and those available in the local labor market. In particular, they cited high demand for skilled labor in the Kansas City area and low unemployment in the labor market at 4 percent, which can make it difficult to fill positions. Contractors at the site said that filling skilled positions can take an average of 58 days and that certain positions, such as electrical engineers and toolmakers, are particularly difficult to fill. Kansas City site contractors noted that they have taken actions to mitigate this challenge. These actions, which contractor site representatives have characterized as largely successful, include participation in and development of university relations programs, involvement in research and development partnerships and consortiums, recruitment from area trade schools and technical schools, and expanding the market area in which the site searches for recruits. For example, contractors noted the site’s participation in a service academy career conference in San Diego, California, in August 2018. They also said they are considering ways to recruit skilled positions that are in high demand, such as toolmakers, by offering to cover relocation expenses for newly hired workers. They further noted that the site maintains an internship program and has plans to double the number of interns, from 35 in 2018 to 71 in 2019, as a strategy to increase talent in critical areas. According to Kansas City site contractor representatives, the site increased the total number of contractor staff by about 65 percent in a 4- year period, from 2,492 in August 2014 to 4,134 in August 2018, and is expected to continue to increase to nearly 5,000 staff by August 2019. Figure 9 shows the change in number of Kansas City site staff during the last fiscal year for which data are complete, and the reasons for the changes, as reported to us by site contractor representatives. To meet forecasted workload increases, the site plans to continue to increase staff in each year through 2020, with the numbers of planned annual hires ranging from 800 to more than 1,000 staff, according to site contractor representatives. Obtaining Timely Security Clearances for New Staff Kansas City site officials identified a challenge in obtaining appropriate, high-level security clearances for new staff on a timely basis. Contractor representatives we interviewed noted that 100 percent of staff who directly contribute to the design, disposition, fabrication, inspection, scheduling, and protection of products and services related to nuclear weapons require a Q clearance. They further noted that the large majority of support functions also require a Q clearance. As we reported in March 2018, the National Background Investigation Bureau had a backlog of more than 700,000 investigations as of February 2018. As we reported, this backlog was caused in part by two 2015 breaches of Office of Personnel Management personnel records. We designated the government-wide personnel security clearance process as a high-risk area in January 2018. Of this national backlog, 3,609 were investigations of Q applicants. As of April 2018, over 790 Kansas City site personnel were awaiting Q clearances, according to Kansas City site contractor representatives. According to these representatives, historically, the Bureau took 80 days, on average, to investigate most Q applicants prior to the 2015 breaches; however, as of February 2018, the Bureau took 316 days, on average, to do so. According to Kansas City site contractor representatives, the Bureau is not projecting normal operations until late 2019 or early 2020. From fiscal year 2017 through March 2018, 778 Q clearances were granted for the Kansas City site, with an average of 335 days at the Bureau and another 27 days at NNSA to make a final determination. According to site contractor representatives, these long wait times may contribute to less than full employee utilization at the site. For example, they noted that fully cleared staff are able to perform roughly 38 percent more productive work than uncleared staff, and that difference amounts to approximately 695 direct labor hours of productive work per person in a year. The Kansas City site is taking steps to mitigate the challenges associated with the Bureau’s backlog. For example, the site is hiring hourly production factory staff well in advance of the full production schedule for the B61-12 and W88 Alt 370 weapons systems in fiscal year 2019, in part to ensure these staff will be cleared in time to meet workload demands, according to site contractor representatives. Site contractor representatives told us that they have also worked to expedite the issuance of clearances by working with local Office of Personnel Management officials on interviews for clearance cases. In addition, the site has worked to ensure that new staff can be trained and productive while awaiting clearances. Specifically, according to contractor representatives, the site has established segregated training space for uncleared workers; created security plans and escorting practices that allow uncleared staff supervised access into secure areas to perform unclassified work, where possible; and temporarily converted some production space into areas where uncleared staff can perform unclassified hand assembly work. In addition, the Kansas City site has requested 339 interim Q clearances, 267 of which had been approved, as of January 2018. DOE’s order that establishes requirements for processing and granting security clearances allows for interim security clearances to be issued under exceptional circumstances and when such action is clearly consistent with agency and national interests. DOE considers interim clearances to be temporary measures pending completion of the investigation, which must be in process when the interim clearance is granted. As of September 2018, less than 1 percent of interim clearances approved for the Kansas City site had been cancelled once full investigations were completed, according to site contractor representatives. External Suppliers Kansas City site contractor representatives identified challenges regarding the site’s monitoring and management of external suppliers’ capacity and skills, and other challenges—such as ensuring that suppliers are willing to establish long-term partnerships with the Kansas City site— that could affect supply chain risk. Since the site procures about 65 percent of its nonnuclear components from external suppliers, these management challenges are highly important, according to site contractor representatives. For example, disruption to the established supply chain due to insufficient capacity, skills, or a supplier’s decision not to do business with the Kansas City site can result in production delays. According to Kansas City site contractor representatives, delays in such instances are possible because site contractor representatives would need to take additional time to either replace the lost supplier or develop its own production line to produce the parts in-house at the Kansas City site. To help mitigate challenges regarding the site’s overall monitoring and management of suppliers’ capacity, skills, and other risks, the Kansas City contractor representatives said that they developed two key analytic tools. These tools are a Supplier Capacity Analysis Tool, developed in 2018, and a Supplier Overall Risk Tool, which has been evolving since 2015, according to these representatives. According to Kansas City site officials, contractor representatives use these analytical tools to evaluate over 230 suppliers on a quarterly basis and to evaluate the top 39 suppliers monthly. The evaluations assess factors such as operational performance and financial health, whether a supplier is the sole commodity supplier, and a supplier’s willingness to partner with the site. To help mitigate supplier capacity risks, the site develops plans, using information from the supplier evaluations, to ensure sufficient external supplier capacity, according to Kansas City site contractor representatives. For example, Kansas City site contractor representatives used the supplier capacity analysis tool to identify capacity gaps for at- risk commodities, including machine parts, and to develop gap-closure plans, according to these contractor representatives. As a result of these plans, contractor representatives certified two new suppliers and entered into agreements with several other suppliers to provide reserve capacity. In addition, NNSA’s Enterprise Modeling and Analysis Consortium conducted alternate analysis on the Kansas City site’s workload capacity that corroborated the Kansas City site’s conclusion that mitigation steps being taken at the Kansas City site, including ensuring adequate external supplier capacity, should address increased workload concerns. To help mitigate risks regarding suppliers’ skills in working with the Kansas City site, site contractor representatives also said that the site has taken steps to help train new suppliers. For example, site contractor representatives perform multiple on-site training exercises within the first 6 months of new supplier relationships. These exercises educate the suppliers on purchase order requirements, terms, drawing definitions, and quality expectations using a documented, comprehensive, nine-step process, according to site contractor representatives. To help mitigate risks regarding suppliers’ willingness to establish long- term partnerships with the Kansas City site, site contractor representatives told us that they have begun taking steps to encourage and foster long-term partnerships with suppliers. According to these representatives and a study NNSA conducted of lessons learned from an essentially complete warhead life extension program, facilitating effective supply chains for the nuclear enterprise requires enduring business relationships with suppliers of commercial off-the-shelf components. Because specifications for weapons components and materials are exacting and quantities required are frequently low, many potential suppliers are reluctant to expose themselves to the risk of production for a niche market, according to Kansas City site officials and contractor representatives. To mitigate reluctance to partner with the Kansas City site, contractor representatives stated that the site has developed points of contact with each supplier. These points of contact work toward establishing and maintaining a collaborative partnership in which production forecasts are routinely shared and performance metrics are discussed to foster continuous improvement when needed. In addition, Kansas City site contractor representatives stated that the site is taking steps to develop relationships with other sites to address site- wide challenges regarding supplier evaluations, which can contribute to risks such as lower efficiency and effectiveness and higher costs. The site is taking this action in response to a July 2018 DOE Office of Inspector General (OIG) report that identified the potential duplication of supplier evaluations among NNSA sites, including the Kansas City site, resulting in lower efficiency and effectiveness, and higher costs. The OIG report noted that the need to minimize duplication of efforts will become even more important when considering the additional demands on production related to upcoming weapon refurbishment efforts, which are expected to increase the number of supplier quality auditors needed by the Kansas City site. The OIG recommended that to maximize efficiencies and effectiveness, NNSA should work with contractors, including the Kansas City site, to assess ways to improve the efficiency of supply chain management activities, among other things. Steps the Kansas City site has taken in response to this OIG report include establishing a point of contact with Sandia National Laboratories, which is leading an overarching effort across the nuclear security enterprise to address duplication concerns, according to site contractor representatives. In addition, a December 2018 report to the President by DOD, in consultation with other agencies, identified supply chain risks in the government’s manufacturing and defense industrial base, including at DOE and NNSA sites, and recommended that DOE establish an Industrial Base Analysis and Sustainment program to address risks within the energy and nuclear sectors. According to NNSA officials at the Kansas City site, they are still determining how it will respond to this recommendation. Weapons Design Changes With increasing concurrency of production forecasted, Kansas City contractor representatives have identified challenges regarding their need to minimize weapons design changes during production, which in the past contributed to cost increases and schedule delays for the W76-1 life extension. According to Kansas City contractor representatives and NNSA officials, at least two general weapons design issues can contribute to overall schedule pressure at the Kansas City site. For example, delays due to design changes intended to make parts easier to produce can exacerbate schedule delays by compressing the overall weapons refurbishment schedule. In addition, design changes are undertaken for other reasons, such as in response to weapons testing results. First, according to NNSA’s B61-12 program manager, even though both design laboratories and production site team members advocate for the design changes that make parts easier to produce, the enterprise-wide impact of these changes late in the design process may, as site contractor representatives noted, impact the LEP’s schedule and may require more resources and plant/vendor capacity to meet the schedule. According to this official, given the resource demands of simultaneously occurring major weapons refurbishments, such as the B61-12 and W88 Alt 370, schedule impacts can be magnified and have caused justifiable concern with leadership at NNSA, the design laboratories, and the Kansas City site. Second, Kansas City site officials expressed concern that some component design requirements continue to change late in the production development phase, sometimes because of test results, which creates tension between improving the design and stabilizing production requirements and processes in preparation for full-scale production. Kansas City site officials stated that such design changes pose an ongoing management challenge. Specifically, time lost because of design delays in the earlier stages of weapons’ design and development often needs to be recovered later, during time allotted for production, to meet established delivery schedules, according to Kansas City site officials. Such delays have triggered the need for schedule recovery plans at the Kansas City site in the past. In response to the concerns, NNSA has led several mitigation steps to address schedule risk as both the B61-12 and W88 Alt 370 enter the final stages before full production begins, according to NNSA’s B61-12 program manager. For example, NNSA revised its baseline change process for the B61-12 and W88 Alt 370 to require all changes, including production-related changes, to be reviewed, according to NNSA’s B61-12 program manager. Specifically, NNSA implemented a change management board with several tiers for review and approval of proposed design changes based on the type of change, and potential impact to program milestones, cost, and risk. Varying levels of required review and approval, depending on the change, can include NNSA production and design agency officials, senior site managers, B61-12 or W88 Alt 370 project officers, or other senior managers at DOD and NNSA. The intent, according to this official, is to screen all the changes and determine if they are really needed and when, and if site-wide resources and schedules can support the changes. In addition, Kansas City site contractor representatives said that they have developed management strategies to help mitigate production- related impacts of design changes, such as adding work shifts to increase production output. For example, an August 2017 analysis by Kansas City site contractor representatives shows the use of three shifts—both partial and full shifts—to meet workload demand in multiple functional areas, including production of cables, high voltage assembly, encapsulation and welding, arming and firing mechanisms, machining, and environmental and pressure laboratories. Using additional shifts can help the Kansas City site recover from schedule delays that might result from late design changes, according to site contractor representatives. Moreover, lessons learned from the W76-1 LEP—which will complete production in 2019— are helping to improve coordination between production sites and design agencies, specifically through increased coordination earlier in the weapon development process, according to Kansas City site contractor representatives. Further Changes to Stockpile Requirements Are Anticipated, Which May Affect Existing Workload Plans at the Kansas City Site While current efforts to mitigate the challenges Kansas City site contractor representatives have identified are expected to help address the site’s anticipated future workload, as discussed previously, this workload could further increase if certain 2018 Nuclear Posture Review policy statements, based on nuclear weapons stockpile studies now underway in response to the review, result in changes to production requirements. For example, the Nuclear Posture Review called for modifying existing sea-launched ballistic missile warheads to provide a low-yield option; advancing a program to replace the W78 Intercontinental Ballistic Missile warhead by 1 year; the study of a sea-launched, nuclear-armed cruise missile; and sustaining the B83 strategic nuclear bomb past its currently planned retirement date. NNSA and DOD are developing studies and implementation plans for the 2018 Nuclear Posture Review, but it is too soon to know to what extent these studies and plans may affect the Kansas City site. One early indication of how implementing the 2018 Nuclear Posture Review may affect the Kansas City site is that, according to the Fiscal Year 2019 Stockpile Stewardship and Management Plan, concurrent production of the W80-4 LEP and the W78 replacement LEP is now expected to extend into the 2030s. In addition, the 2019 plan anticipates that alts may be needed to sustain the B83, if the weapon system remains in the stockpile for long enough. We concluded in an April 2017 report that the new Nuclear Posture Review comes during a particularly challenging decade for NNSA’s nuclear modernization efforts, as the agency plans to simultaneously execute at least four nuclear LEPs along with major construction projects, such as efforts to modernize NNSA’s uranium and plutonium capabilities. We further concluded that NNSA’s modernization budget estimates for fiscal years 2022 through 2026, which reflected past program plans, may exceed the funding levels programmed for modernization in future budgets, raising affordability concerns. Moreover, we concluded that NNSA had not addressed a projected “bow wave” of future funding needs—that is, an impending and significant increase in requirements for additional funds—or the mismatch between potential funding needs and potential funding available even before the Nuclear Posture Review was completed. We recommended that NNSA include an assessment of affordability of NNSA’s portfolio of modernization programs in future versions of the Stockpile Stewardship and Management Plan—for example, by presenting options NNSA could consider to bring its estimates of modernization funding needs into alignment with potential future budgets. NNSA did not explicitly agree or disagree with our recommendation, but we will continue to monitor any action NNSA takes in response to the recommendation. In addition to addressing affordability concerns, NNSA has been advised to stabilize long-term workload at operating sites. A congressional advisory panel examining the governance of the nuclear security enterprise issued a report in November 2014 recommending, among other things, actions intended to stabilize long-term workload at operating sites. In particular, it recommended that NNSA, working with DOD, create a long-term operating plan to support the nation’s warhead modernization strategy; it further specified that this plan should be designed to create a relatively stable, long-term workload. The panel’s report stated that a stable baseline of design, engineering, and production is needed to make effective use of the available capabilities in the weapons complex, provide the basis for sizing and modernization of the weapons complex, and identify potentially conflicting demands on available capabilities. While NNSA has taken some actions in response to this recommendation, an expert panel concluded in March 2018 that NNSA’s overall response had been inadequate and called for NNSA to develop, among other things, an integrated strategic plan for the entire nuclear security enterprise. The panel concluded that, given NNSA’s expected increase in workload across the nuclear weapons complex, and the new 2018 Nuclear Posture Review uncertainties, NNSA’s ongoing implementation of this and other recommendations made by the Panel over the next several years will take on additional importance. Agency Comments We provided a draft of this report to NNSA for its review and comment. NNSA provided technical comments, which we incorporated into this report, as appropriate. We are sending copies of this report to the appropriate congressional committees, the Administrator of the National Nuclear Security Administration, and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-3841 or bawdena@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix II. Appendix I: Objectives, Scope, and Methodology The Senate committee report accompanying a bill for the National Defense Authorization Act for Fiscal Year 2018 included a provision for us to review the Kansas City site’s staffing plans and capabilities to meet national security requirements. Our report examines (1) workload forecasts for the Kansas City site since 2012, and (2) management challenges the Kansas City site has identified for achieving the forecasted workload and actions the site has taken to mitigate these challenges. To examine workload forecasts for the Kansas City site since 2012, we visited the site, obtained and reviewed workload documents, and interviewed officials from the National Nuclear Security Administration’s (NNSA) Kansas City site office and headquarters offices, and NNSA contractor representatives at the site. In particular, we obtained information on the Kansas City site forecasted workload based on fiscal years 2012 and 2018 Stockpile Stewardship and Management Plans (SSMP), comparing full production schedules, including upcoming B61-12 Life Extension Program (LEP) and W88 Alteration (Alt) 370 work. Because the design and capacity of the modern production facility, completed in 2012, was based largely on the 2012 SSMP and previous plans, we used this as the baseline plan. We then compared nuclear weapons systems LEP and Alt schedules in the 2012 SSMP with the 2018 SSMP because Kansas City contractor representatives told us that plans and associated workload had changed significantly by 2018. In addition, we reviewed Kansas City contractor information provided by the “what-if” capacity analyses tool, including graphs and charts depicting workload for each weapons system undergoing LEPs or Alts. Whenever possible, we validated or corroborated contractor-forecasted data on workload and facility capacity by reviewing other sources such as NNSA’s Enterprise Modeling and Analysis Consortium analysis and conclusions and SSMP information. To examine management challenges the Kansas City site has identified for achieving the forecasted workload, and any actions the site has taken to mitigate these challenges, we visited the Kansas City site, obtained and reviewed documentation, and interviewed NNSA and contractor officials who identified management challenges in five areas: ensuring that the site has (1) sufficient production and administrative office space, (2) up-to-date production equipment, (3) a sufficient workforce, (4) capable and reliable external suppliers, and (5) complete weapons designs early enough in development to minimize production changes and delays. We selected these five areas for review based on NNSA officials’ and contractor representatives’ identification of such challenges as being the most significant at the Kansas City site. To corroborate information on management challenges and associated mitigation action(s) provided by the Kansas City site, we conducted interviews with additional sources, reviewed alternative documentation or analyses, and obtained examples of the specific action(s) being taken, when available. For example, regarding the first management challenge of ensuring adequate production and administrative office space, we reviewed Kansas City site information, including information on space in the modern facility, the mission needs statement for expanding the site’s space, and NNSA budget justifications for fiscal years 2018 and 2019. We also obtained information on short- and long-term plans for meeting forecasted workload demands. Regarding the second management challenge—ensuring it has up-to- date production equipment—we reviewed Kansas City site information and information from an alternative source. Specifically, we reviewed NNSA’s 2017 Master Asset Plan, which provided additional information and alternate analyses concerning the age of the Kansas City site’s production equipment. Regarding the third management challenge— ensuring a sufficient, capable, and security-cleared workforce—we reviewed both site-level information and information from other sources, including from NNSA and the Department of Energy (DOE). For example, we reviewed NNSA’s Fiscal Year 2018 Stockpile Stewardship and Management Plan, which also includes workforce information and analyses. In addition, we asked the Kansas City site contractor representatives and NNSA officials for additional clarification and detail concerning the management challenges and mitigation actions, as well as specific examples to support their statements. For issues related to the clearance process, we contacted DOE officials to obtain information on DOE supplemental guidance for interim clearance mitigation steps. To confirm the accuracy of staffing-related information provided by Kansas City site contractor representatives, we obtained information from these representatives on how the site performed certain calculations, such as determining the change in number of Kansas City site staff; number of Kansas City site staff, by years of service; and distribution of Kansas City staff, by age. We reviewed the various formulas Kansas City contractor representatives used in preparing its analyses in order to understand the logic used in making these determinations. Furthermore, we validated that these calculations were accurate by independently performing the calculations to see if our results matched the site’s results. For information concerning the fourth management challenge—ensuring capable and reliable external suppliers—we interviewed a senior NNSA headquarters official overseeing NNSA’s Enterprise Modeling and Analysis Consortium, which conducted alternate analyses on the Kansas City site’s workload capacity, equipment, and workforce. The Consortium corroborated the Kansas City site’s conclusion—that mitigation steps being taken at the Kansas City site should address increased workload concerns. Regarding the fifth and last management challenge—ensuring complete weapons designs early in development to ensure that production changes and delays are kept to a minimum—we reviewed the W76 lessons learned report, which also describes design completion issues affecting the Kansas City site. In addition, we interviewed NNSA’s B61 program manager to obtain additional perspective on design-related challenges facing upcoming B61-12 refurbishments. We conducted this performance audit from November 2017 to April 2019 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: GAO Contact and Staff Acknowledgments GAO Contact Allison B. Bawden, (202) 512-3841 or bawdena@gao.gov. Staff Acknowledgments In addition to the individual named above, Jonathan Gill (Assistant Director), Christopher Pacheco (Analyst in Charge), and Sophia Payind made significant contributions to this report. Also contributing to this report were Elizabeth Dretsch, R. Scott Fletcher, Thomas Gilbert, Richard Johnson, Cynthia Norris, Jeanette Soares, and Sara Sullivan.

Why GAO Did This Study Modernization of the nation's nuclear stockpile depends on timely procurement and production of nonnuclear parts and components. Such parts and components make up over 80 percent of the items in a nuclear weapon. The Kansas City site procures or produces most of these parts, under NNSA oversight. In fiscal year 2012, the site completed construction of a modern production facility. The new facility was expected to accommodate rising future workload demands, based on the forecasts that were current in 2012, according to Kansas City site contractor representatives. The Senate committee report accompanying a bill for the National Defense Authorization Act for Fiscal Year 2018 included a provision for GAO to review the Kansas City site's staffing plans and capabilities to meet national security requirements. This report examines (1) workload forecasts for the site since 2012, and (2) management challenges the site has identified for achieving the forecasted workload, and actions the site has taken to mitigate these challenges. GAO reviewed NNSA and contractor documents from 2012 through 2018 relevant to workload changes, and associated workload capacity, including information on infrastructure, equipment, and business processes—as well as personnel data. GAO also interviewed NNSA program and field officials and contractor representatives. What GAO Found Workload forecasts have significantly changed at the National Nuclear Security Administration's (NNSA) primary site for procuring or producing nonnuclear parts and components of nuclear weapons since the site's modern production facility was built in 2012. Specifically, workload projections made by the contractor operating the site, known as the Kansas City National Security Campus (Kansas City site), has increased significantly from forecasts used in planning the site's new production facility. More recent forecasts show that to meet workload requirements, production and administrative staff will need to almost double by 2020 compared to 2014 levels. For example, workload to modernize the B61-12 and W88 weapons systems will double during fiscal years 2020 through 2022. According to NNSA officials and contractor staff, the site has identified and begun to mitigate management challenges to meeting future workload, including: Ensuring sufficient production and office space. Because the current space is not sufficient for the increase in projected workload, the site is leasing additional space until long-term solutions, currently in planning, can be implemented. Updating production equipment. To update aging production equipment, the site is developing a 10-year equipment strategy, among other things. Retaining and recruiting a sufficient workforce. The site has offered rewards and benefits to retain existing staff, about a third of whom are eligible to retire. It is also recruiting skilled new staff in tight labor markets and seeking to expedite security clearances for them. Ensuring adequate external supplier capacity. The site procures about 65 percent of its nonnuclear components from external suppliers. The site is assessing capacity and risk of existing suppliers and developing new ones. Current mitigation efforts should help the site meet currently planned increased workload and capacity demands, according to contractor and NNSA analyses. However, the February 2018 Nuclear Posture Review—conducted by the Department of Defense under the direction of the President to determine the role of nuclear weapons in the nation's security strategy—may change requirements and add to the site's workload in ways not yet fully known because studies and plans in response to the review are not fully complete.

gao_GAO-18-68

gao_GAO-18-68_0

Background There are several aspects of individual market plans sold through the exchanges for consumers to consider when selecting a plan, including: (1) metal tiers; (2) premium variation and plan availability; (3) covered benefits; and (4) premium tax credits. Current exchange consumers who are eligible for continued health insurance and do not actively select and enroll in a health plan for the subsequent year may be automatically re-enrolled in the same or a similar crosswalked plan. Considerations for Selecting and Enrolling in Individual Market Health Plans through the Exchanges Plans sold through exchanges are offered at one of four levels of coverage, or metal tiers—bronze, silver, gold, and platinum—that reflect the out-of-pocket costs that may be incurred by a consumer. The four metal tiers correspond to the plan’s actuarial value—a measure of the relative generosity of a plan’s benefits that is expressed as a percentage of the covered medical expenses expected to be paid, on average, by the issuer for a standard population and set of allowed charges for in-network providers. The actuarial values of these metal tiers are as follows: bronze (60 percent), silver (70 percent), gold (80 percent), and platinum (90 percent). If an issuer sells a plan on an exchange, it must offer at least one plan at the silver level and one plan at the gold level. Issuers are not required to offer bronze or platinum plans. Premium Variation and Plan Availability As we have previously reported, the range of premiums for health plans offered through the exchanges can vary widely across counties and states, and the number and type of plans available in the health insurance exchanges vary from year to year. Issuers can add new plans and adjust or discontinue existing plans from year to year, or they can extend or restrict the locations in which plans are offered. As a result, the options available to consumers can change from year to year. PPACA requires that health insurance plans offered through the exchanges be certified as qualified health plans, meaning that they must provide essential health benefits, comply with cost sharing limits, and meet certain other requirements. Essential health benefits include items and services within ten categories. Some health insurance plans offered through the exchanges include benefits above and beyond the minimum requirements. For these plans, only the percentage of the plan premium that covers the essential health benefits is considered when determining the consumer’s benchmark plan. Certain consumers purchasing health insurance through the exchanges are eligible for and receive premium tax credits that may reduce their out-of-pocket costs for premiums. To be eligible for premium tax credits, individuals and families must generally have a household income of at least 100, but no more than 400, percent of the federal poverty level (FPL). Consumers who are eligible for premium tax credits and enrolled in the benchmark plan are responsible for paying premiums that are generally limited to a percentage of household income, such that individuals and families with lower household incomes contribute a smaller portion of their income toward the health plan premium than individuals and families with higher incomes, and premium tax credits may be applied to only the portion of the premium that covers essential health benefits. For example, in 2016, the percentage of household income that consumers who were eligible for premium tax credits and who lived in the United States were expected to pay toward the portion of their premiums for their benchmark plan that covered essential health benefits was 2.03 percent for those at 100 percent of the FPL, 8.18 percent for those at 250 percent of FPL, and 9.66 percent for those at 400 percent of FPL. A consumer’s required contribution to the premium is the amount of that benchmark plan premium that is not covered by the premium tax credit. (See table 1.) Although consumers’ premium tax credit amounts are determined in part based on the cost of premiums for their local benchmark plan, the credit can also be applied towards the premiums for other eligible exchange plans. However, the premium tax credit available to consumers does not increase if they enroll in exchange plans with higher premiums than the local benchmark plan. In such cases, consumers are responsible not only for their required contribution but also for the difference in premiums. Similarly, if a consumer chooses to enroll in an exchange plan with lower premiums than the local benchmark plan premium, then the consumer’s premium tax credit would also generally remain the same, so the consumer would pay less for that plan. The tax credit cannot, however, exceed the total value of the premium. Because most consumers enrolling in exchange plans are eligible for premium tax credits, most consumers’ out-of-pocket premium costs are lower than the advertised cost of premiums. (See table 2.) Process for Automatically Re-enrolling Consumers into Exchange Plans Re-enrollment in an exchange plan may occur through either an active choice by a consumer or through automatic re-enrollment by the exchange. Eligible returning consumers may enroll in a health insurance plan through the exchange each year during an open enrollment period. Federally facilitated exchanges automatically re-enroll eligible exchange consumers for the next year, unless their health insurance is terminated or the consumer makes an active plan selection. Through automatic re-enrollment a consumer is re-enrolled in the same plan for the next year if that plan remains available to him or her; if the same plan is no longer available (e.g., because the issuer decided to discontinue a particular plan or to stop offering the plan in certain locations), then the consumer is generally re-enrolled in a similar crosswalked plan. The criteria HHS established for identifying appropriate similar crosswalked plans have changed over time, but the similar crosswalked plan is typically the same metal tier level as the original plan. During the 2015 to 2016 transition, all similar crosswalked plans were plans offered by the same issuer as the original plan. If that issuer no longer offered an exchange plan, there was generally no crosswalked plan and automatic re-enrollment was not an option. Starting with the 2016 to 2017 transition, if the original issuer did not offer a similar plan, then automatic re-enrollment could be into a health plan offered by a different issuer, with plan similarity determined using established criteria. Both issuers and exchanges have had roles in informing consumers about the enrollment process. For example, prior to the start of the 2015 and 2016 open enrollment periods, both the exchange and health plan issuer were to provide current exchange consumers with general information about the upcoming enrollment period, including key dates and information regarding eligibility for re-enrollment. In addition, some consumers were also to receive special notices from the exchange that provided more detailed information regarding their application status, eligibility for enrollment and affordability programs, and potential effects on enrollment if they had not updated information about their income or eligibility or reviewed their re-enrollment options with the exchange prior to the end of the open enrollment period. Consumers who were automatically re-enrolled by an exchange were to receive an additional notice with updated information about their re-enrollment status. According to CMS officials, automatically re-enrolled consumers were provided with information about their new premium amount and any new advance premium tax credit amounts in a message confirming their enrollment. From 2015 to 2017, Most Plans Identified as Benchmark Plans Changed and Benchmark Plan Premiums Generally Increased In most of the nearly 2,600 counties included in our analysis, the plan that we identified as the benchmark plan changed from 2015 to 2017. For example, in 85 percent of the counties included in our analysis, the 2015 benchmark plans were not benchmark plans in either 2016 or 2017, the other 2 years we studied. The benchmark plan was the same plan in all 3 years in only 3 percent of counties. (See table 3.) In addition, benchmark plan premiums were more likely to increase than decrease from year to year, and increases were higher from 2016 to 2017 than they were from 2015 to 2016. Among all the counties in our analysis, the median change in monthly premiums for the benchmark plans was an increase of 11 percent from 2015 to 2016, and 28 percent 2016 to 2017. As shown in figure 1, the gross premiums for benchmark plans increased by more than 55 percent from 2016 to 2017 in 12.4 percent of the counties in our analysis but did not increase by more than 55 percent in any counties from 2015 to 2016. In contrast, although not particularly common, relatively stable or even decreasing premiums from year to year were more likely from 2015 to 2016 than from 2016 to 2017. Appendix I provides examples of median benchmark plan premiums for 2015, 2016, and 2017 for select groups of consumers. Because premium tax credits limit eligible consumers’ payments for benchmark plan premiums to a percentage of their income, an increase in premiums may not increase their financial responsibility. Instead, for eligible consumers, the amount of the tax credit would increase. According to HHS, most exchange consumers have been eligible for these tax credits; those who were not eligible for tax credits would not have this protection from premium increases. The premium increases for consumers who were not eligible for premium tax credits, or for those who were eligible but who chose plans that had higher premiums than their benchmark plan premiums, could have had a more substantial financial impact, because premium tax credits would not have offset, or fully offset, the higher premiums. Although gross premiums for benchmark plans were likely to increase from 2015 to 2016 and from 2016 to 2017, we found that in many counties, the implications for automatically re-enrolled consumers were modest because net premiums—after accounting for tax credits for those eligible for those credits—were limited. We compared the 2016 premiums for plans that had been benchmark plans in 2015 to the 2016 benchmark plan premiums, and we compared the 2017 premiums for plans that had been benchmark plans in 2016 to the 2017 benchmark plan premiums. To focus this analysis on the potential effects for those who were automatically re-enrolled, we limited our comparisons to plans that were available in both years, or plans for which a similar crosswalked plan had been identified for the second year. For this analysis, we excluded plans that were benchmark plans in one year and were also benchmark plans, or were crosswalked to a benchmark plan, in the following year. We found that in many counties, the new premiums for plans that had been (but were no longer) benchmark plans differed only modestly from the new benchmark plan premiums. For example, in 60 percent or more of the counties in our analysis, the premium for the previous benchmark plan was within plus or minus about 7.5 percent of the new benchmark plan premium. This finding indicates that automatic re-enrollment from a benchmark plan into a plan that was not a benchmark plan did not necessarily result in substantially higher premiums compared to the premiums for the new benchmark plans, and the same would be true for consumers who actively chose their same or similar crosswalked plan. While modest premium differences were not uncommon in either year, figure 2 also shows that some differences were substantial. (See fig. 2.) Although consumers who were eligible for premium tax credits were somewhat insulated from large differences in premiums, if they were automatically re-enrolled in a plan with a premium that was higher than their benchmark plan premium, no matter how great the difference, they would have been be required to pay a larger share of their incomes on those premiums. And, as already noted, the premium differences for consumers who were not eligible for premium tax credits, or for those who were eligible but who chose plans that had higher premiums than their benchmark plan, could have had a more substantial financial impact because premium tax credits would not have offset, or fully offset, the higher premiums. Thirty Percent of Consumers Who Re-enrolled in 2016 Were Automatically Re-enrolled, and the Remaining Consumers Actively Re-enrolled, Generally into Different Plans Among consumers who were enrolled in plans through the federal platform in both 2015 and 2016, 30 percent (about 1.7 million consumers) were automatically re-enrolled. Of those consumers who were automatically re-enrolled, 71 percent were re-enrolled in their same plan and 29 percent were re-enrolled in a similar crosswalked plan, because their 2015 plan had been discontinued or was no longer offered in the consumer’s local area. These data do not indicate whether these consumers explored their options for switching plans and made an active decision not to change plans. The remaining 70 percent of consumers who enrolled in exchange plans through the federal platform in both 2015 and 2016 (more than 3.9 million consumers) actively re-enrolled in 2016. Of these consumers, 39 percent chose the same plan in which they had been enrolled in 2015 or the similar crosswalked plan to which they would have been automatically re-enrolled. The majority of consumers who re-enrolled actively, 61 percent, switched to a plan that was neither their 2015 plan nor the similar crosswalked plan. (See fig. 3.) Of those consumers who actively switched plans, more than half (54 percent) would have been automatically re-enrolled in their same plan if they had not actively switched plans, indicating that plan discontinuation was not the only factor involved in consumers’ decisions to change plans. Consumers’ Financial Responsibility for Premiums Generally Increased Less with Active Re-enrollment than with Automatic Re-enrollment Consumers’ median net monthly premiums (after premium tax credits) generally increased less from 2015 to 2016 for those who actively re-enrolled ($5) than for those who were automatically re-enrolled ($22). As shown in table 4, consumers who actively re-enrolled had a lower median increase in their net monthly premiums than consumers who were automatically re-enrolled for both the same and similar crosswalked plans. Moreover, table 4 also shows that consumers who re-enrolled actively, and who switched plans from 2015 to 2016, enrolled in plans with median monthly net premiums that increased the least overall—a median net increase of $1 compared to $13 per month for those who enrolled in the same plan. In addition, the table shows that enrollment in a similar crosswalked plan did not generally result in a higher median net premium than enrollment in the same plan: whether enrollment was active or automatic, consumers’ median net monthly premiums increased less for those who enrolled in a similar crosswalked plan than for those who enrolled in the same plan. Changes in net monthly premiums varied around these medians, however, with some consumers facing large increases or, in some cases, large decreases in their net monthly premiums. Large increases or decreases in net monthly premiums could result from changes to eligibility for tax credits, selections of plans of different metal levels, or other circumstances. Our findings are consistent with other work by ASPE that suggested that consumers consider possible cost savings when deciding to switch plans. For example, ASPE found that average net monthly premium for the 61 percent of consumers who actively switched plans in 2016 was $132, which represented an average savings of $42 per month compared to what they would have paid if they stayed in their same or similar crosswalked plans. This work also found that the net monthly premiums of consumers who actively chose to remain in their same or similar crosswalked plans in 2016 were, on average, only $10 more than those for consumers who actively switched plans. In addition, ASPE found that consumers’ plan selections indicated sensitivity to net premiums. For example, ASPE found that consumers were much more likely to switch plans when the net premium of their 2015 plan increased than when the gross premium of their 2015 plan increased, but the net premium did not. Agency Comments We provided a draft of this report to HHS for review and comment. HHS provided technical comments, which we incorporated as appropriate. As agreed with your offices, unless you publicly announce the contents of this report earlier, we plan no further distribution until 30 days from the report date. At that time, we will send copies to the Acting Secretary of Health and Human Services and other interested parties. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staffs have any questions about this report, please contact me at (202) 512-7114 or dickenj@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix II. Appendix I: Median Benchmark Plan Premiums for Select Groups of Consumers, 2015 through 2017 Benchmark plan premiums generally increased from 2015 through 2017. Table 5 shows the median monthly gross benchmark plan premiums (exclusive of any applicable premium tax credits) for select consumer groups. Appendix II: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, Gerardine Brennan, Assistant Director; Kristen Joan Anderson, Analyst-in-Charge; Todd Anderson; and LaKendra Beard made key contributions to this report. Also contributing were Muriel Brown; Daniel Lee; Laurie Pachter; and Emily Wilson.

Why GAO Did This Study During open enrollment, eligible returning consumers may re-enroll in their existing health insurance exchange plan or choose a different plan. Those who do not actively enroll in a plan may be automatically reenrolled into a plan. According to the Department of Health and Human Services, automatic re-enrollment is intended to help ensure consumers' continuity in coverage. However, some have questioned whether automatic reenrollment could have unintended financial consequences for consumers. GAO was asked to review automatic reenrollment and benchmark plans. GAO examined 1) the extent to which plans identified as benchmark plans remained the same plans from year to year, and how premiums for benchmark plans changed; 2) the proportion of exchange consumers who were automatically re-enrolled into the same or similar plans, and how these proportions compared to those for consumers who actively re-enrolled, and 3) the extent to which consumers' financial responsibility for premiums changed for those who were automatically re-enrolled compared to those who actively re-enrolled. GAO reviewed relevant guidance and analyzed county-based data from the Centers for Medicare & Medicaid Services (CMS) for the 37 states that used the federal information platform, healthcare.gov, from 2015 through 2017. GAO also interviewed CMS and ASPE officials and analyzed information from ASPE on reenrollment from 2015 to 2016. What GAO Found Through the exchanges established under the Patient Protection and Affordable Care Act, consumers can directly compare and select among health plans based on a variety of factors, including premiums. Most consumers who purchase health plans through the exchanges receive tax credits to help them pay for their premiums. The value of a consumer's premium tax credit is based, in part, on the premium for the benchmark plan, which is the second lowest cost option available in the consumer's local area within the exchange's silver metal tier (one of four metal tiers that indicate the value of plans). Because plan premiums and plan availability can change over time, the benchmark plan in each local market can also change over time. GAO analyzed changes in benchmark plans and premiums from 2015 through 2017 and found: In most of the nearly 2,600 counties included in the analysis, the plans identified as benchmark plans, and the premiums for these plans, changed from year to year. For example, in 85 percent of counties, the 2015 benchmark plans were not benchmark plans in either 2016 or 2017. Gross benchmark premiums (exclusive of tax credits) increased from year to year, and increases were higher from 2016 to 2017 than they were from 2015 to 2016. Premium tax credits would limit the costs of increasing premiums for most consumers, though some consumers, including those not eligible for premium tax credits, would have incurred more or all of the higher premium costs. During the annual open enrollment period, consumers who do not make an active plan selection are automatically reenrolled into their existing plan or, if that plan is no longer available, they are generally re-enrolled into a similar plan if one has been identified. GAO analyzed information from the Office of the Assistant Secretary for Planning and Evaluation (ASPE) for consumers enrolled in both 2015 and 2016 and found: About 30 percent of consumers were automatically re-enrolled in 2016, while the remaining 70 percent chose to actively re-enroll. Median net monthly premiums—what consumers paid after premium tax credits—increased less from 2015 to 2016 for those who actively enrolled ($5) than for those who were automatically reenrolled ($22), although there was variation. Our findings are consistent with other work by ASPE that suggests that consumers consider possible cost savings when deciding to switch plans. The Department of Health and Human Services provided technical comments on a draft of this report, which GAO incorporated as appropriate.

gao_GAO-18-184T

gao_GAO-18-184T_0

Background According to NRC’s website, the higher the radiation dose, the sooner the effects of radiation will appear, and the higher the probability of death. Radiation doses such as those received by survivors of the atomic bombs in Japan can cause cancers such as leukemia and colon cancer and, if levels are high enough, acute radiation syndrome. The symptoms of this syndrome range from nausea, fatigue, and vomiting to death within days or weeks. In contrast, the effects of low-dose radiation are more difficult to detect. In particular, below about 100 millisieverts (mSv) (10 rem)—the level below which the National Academies of Sciences, Engineering, and Medicine’s (National Academies) 2006 report on radiation and human health considered radiation to be low dose—data do not definitively establish the dose-response relationship between cancer and radiation exposure. Selected Agencies Generally Used Advice from Scientific Advisory Bodies to Develop and Apply Radiation-Protection Requirements and Guidance In developing and applying radiation protection requirements and guidance for workers and the public—specifically, limits on dose or increased health risk and guidance levels on exposure—EPA, NRC, DOE, and FDA have generally taken the advice of scientific advisory bodies. In particular, they have relied on the advice of the International Commission on Radiological Protection, the National Council on Radiation Protection and Measurements, and the National Academies’ Nuclear and Radiation Studies Board. This advice includes the use of the linear no-threshold model, which assumes that the risk of cancer increases with every incremental increase in radiation exposure. For example, the National Academies published a report in 2006 stating that the balance of evidence from various types of studies tends to favor a simple proportionate relationship between radiation at low doses and cancer risk. According to the National Academies, the availability of new and more extensive data since the publication of its previous report in 1990 strengthened confidence in the 2006 report’s estimates of cancer risk. The advisory bodies have recognized challenges in accurately estimating cancer risks from very low doses of radiation exposure when using the linear no-threshold model. For example, much of the data on health effects of radiation exposure come from non-U.S. populations, such as Japanese atomic bomb survivors. These individuals received a large exposure to radiation over a short period of time (an acute exposure), and there is uncertainty about the extent to which the health effects for these populations can be extrapolated to a U.S. population that is regularly (chronically) exposed to low-dose radiation. Nevertheless, NRC officials told us that, in the absence of convincing evidence that there is a dose threshold below which low levels of radiation are beneficial or not harmful, NRC will continue to follow the recommendations of scientific advisory bodies to use the linear no- threshold model. Similarly, officials from EPA told us that they would consider changing the use of the linear no-threshold model as the basis of their requirements and guidance only if there were a strong recommendation from scientific advisory bodies on radiation protection as well as an endorsement of the change by the National Academies. Under this model, federal regulations set dose limits for radiation exposure that are below the level in the National Academies’ 2006 report on radiation and human health for defining low-dose radiation. For example, NRC’s annual dose limit for members of the public (excluding natural, or background, sources of radiation) from operation of nuclear power plants is a hundredth of the level the National Academies considers low dose. NRC based the dose limit on an advisory body recommendation that the cancer risk to the general public from exposure to radiation should be comparable to the public’s risk from everyday activities, such as taking public transportation. The low-dose radiation limits and guidance that federal agencies have developed and applied vary depending on the settings in which exposure can occur. For example, NRC has established limits on occupational dose that apply to nuclear power-plant workers; these limits are higher than NRC’s annual dose limit for members of the public but are still below the level the National Academies considers low dose. In keeping with advisory body recommendations, NRC also applies the principle that doses should be kept as low as reasonably achievable (ALARA). NRC defines ALARA to mean making every reasonable effort to maintain exposures to radiation as far below dose limits as is practical. At a nuclear power plant we visited as part of our work, representatives told us that under their ALARA plan, the plant set its own dose limit for workers at 40 percent of the NRC’s regulatory limit. Moreover, officials at the plant told us that they have been able to keep exposures below the plant’s own limit by continuously seeking opportunities to reduce unnecessary worker exposure to radiation, such as using robots to perform maintenance work in radiation areas. In contrast to radiation exposure received from nuclear power plants, FDA officials stated that the agency regulates the maximum radiation output of medical equipment, instead of setting limits on the total amount of radiation exposure to patients. According to FDA officials, FDA does not generally have the authority to regulate the total amount of radiation exposure a patient receives from medical imaging equipment. However, in keeping with the principle that radiation exposure should be kept as low as reasonably achievable, FDA encourages voluntary measures by health care providers, such as to investigate and determine whether it is possible to reduce radiation exposure to patients from the use of medical- imaging equipment. Seven Agencies Have Funded Research on the Health Effects of Low-Dose Radiation but Have Not Collaborated on Overall Research Priorities From fiscal year 2012 through fiscal year 2016, seven federal agencies obligated $209.6 million for research on the health effects of low-dose radiation, but they did not use a collaborative mechanism to address overall research priorities in this area. DOE and NIH accounted for most of the funding, with DOE obligating $116.3 million and NIH obligating $88.6 million, or about 56 percent and 42 percent of the total, respectively. The five other agencies—NRC, NASA, DOD, EPA, and CDC—obligated the remaining $4.7 million, or about 2 percent of the total. DOE has two offices that have funded research on the health effects of low-dose radiation—the Office of Science and the Office of Environment, Health, Safety and Security—according to funding information DOE provided. The Office of Science established the Low Dose Radiation Research Program in 1998 and funded it through fiscal year 2016. A primary focus of this program was radiobiological research, which examines molecular and cellular responses to radiation exposure. According to DOE’s website for the program, the program provided data and information about the low-dose range of exposure, producing 737 peer-reviewed publications as of March 2012. The Office of Environment, Health, Safety and Security provided funding for epidemiological studies, including studies involving Japanese atomic bomb survivors. NIH has funded and conducted both epidemiological and radiobiological studies on low-dose radiation, according to NIH officials. The officials stated that the studies are conducted through the National Cancer Institute’s internal research program for radiation epidemiology, as well as through NIH’s research programs for external funding of investigator- initiated research. Other institutes of NIH, including the National Institute of Environmental Health Sciences, also fund research related to the health effects of radiation exposure as part of NIH’s overall mission to fund medical research. Among the other agencies that provided some funding to low-dose radiation studies, several provided funding to the Epidemiological Study of One Million U.S. Radiation Workers and Veterans (Million Person Study)—an ongoing study headed by the National Council on Radiation Protection and Measurements. DOE also provided funding for this study. In fiscal years 2012 through 2016, the seven agencies who provided funding for research on health effects of low-dose radiation collectively decreased their annual funding obligations in this area by 48 percent, from $57.9 million in fiscal year 2012 to $30.4 million in fiscal year 2016. DOE accounted for a large portion of this overall decrease in annual funding. Specifically, over this 5-year period, DOE reduced its annual funding obligations for this area of research by 45 percent—from $32.6 million in fiscal year 2012 to $18.0 million in fiscal year 2016. According to DOE, the decrease was primarily due to DOE’s reduction in funding for its Low Dose Radiation Research Program. According to DOE officials, decreases in funding for the program reflected a shift toward bioenergy and environmental research. Similarly, over the 5-year period, NIH’s funding for low-dose radiation research decreased by 48 percent—from $23.1 million in fiscal year 2012 to $12.0 million in fiscal year 2016. NIH officials explained that funding levels for a particular disease or research area can fluctuate depending on several factors, including the number and quality of research proposals submitted and the outcome of NIH’s peer reviews of the proposals, as well as the overall research budget. The seven agencies that funded research on health effects of low-dose radiation for fiscal years 2012 through 2016 collaborated on particular research projects through various mechanisms, including joint funding of individual projects, but they did not use a collaborative mechanism to address overall research priorities. As previously noted, the 2016 report of DOE’s Biological and Environmental Research Advisory Committee provided information about research needs in low-dose radiation and found that further research could decrease uncertainty in predicting cancer risk from low-dose radiation. The report stated that other agencies—including NRC, NIH, EPA, DOD, and NASA—could benefit from the reduction in uncertainty that could be obtained by this research. In our September 2017 report, we recommended that the Secretary of Energy lead the development of a mechanism for interagency collaboration to determine roles and responsibilities for addressing priorities related to research on the health effects of low-dose radiation. We made this recommendation because our previous work has shown that collaborative mechanisms can serve multiple purposes, such as leading interagency efforts to develop and coordinate sound science and technology policies across the federal government. Although collaborative mechanisms differ in complexity and scope, they all benefit from certain key features, such as leadership. We directed this recommendation to DOE for several reasons. In the past, DOE took a leading role in advocating for greater communication and coordination between the fields of radiation biology and epidemiology. In addition, DOE is the federal agency that currently has primary responsibility under the Atomic Energy Act of 1954 for research related to the protection of health during activities that can result in exposure to radiation. DOE is well positioned to lead an effort to ensure that federal agencies have a mechanism for interagency collaboration to address overall research priorities related to low-dose radiation health effects because of the agency’s past experience as a leader in this area of research. Such an effort could help DOE and the collaborating agencies determine roles and responsibilities, including leadership when addressing shared research priorities. DOE did not agree with our recommendation. In particular, DOE stated that EPA and NRC also have legal mandates to research low-dose radiation exposure and that these agencies establish their research priorities in accordance with their respective budget authorities and recommendations from independent advisory bodies. DOE stated that as a result, it would not be appropriate for DOE to lead the development of a mechanism for interagency collaboration. We believe that DOE’s concerns stem from a misinterpretation of our recommendation, and we made several changes to our report and our recommendation to clarify DOE’s role. We noted that we did not recommend that a mechanism for interagency collaboration serve as a replacement for agencies’ legal mandates, budget authorities, and recommendations from independent advisory bodies. Instead, this mechanism would help agencies address shared research priorities. In making our recommendation, we did not specify the coordinating mechanism that agencies should use and instead left it to DOE to lead the development of an appropriate mechanism. We continue to believe that an interagency coordination mechanism for low-dose research is needed and that DOE is in the best position to lead agencies in developing the most appropriate mechanism. Chairman Weber, Ranking Member Veasey, and Members of the Subcommittee, this concludes my prepared statement. I would be pleased to respond to any questions that you may have at this time. GAO Contact and Staff Acknowledgments If you or your staff have any questions about this statement, please contact John Neumann at (202) 512-3841 or neumannj@gao.gov. In addition, contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this statement. Individuals who made key contributions to the report on which this testimony is based include Allen Chan, Kendall Childers, Joseph Cook, Richard Johnson, Cynthia Norris, Josie Ostrander, Amber Sinclair, and Jack Wang. This is a work of the U.S. government and is not subject to copyright protection in the United States. The published product may be reproduced and distributed in its entirety without further permission from GAO. However, because this work may contain copyrighted images or other material, permission from the copyright holder may be necessary if you wish to reproduce this material separately.

Why GAO Did This Study This testimony summarizes the information contained in GAO's September 2017 report, entitled Low-Dose Radiation: Interagency Collaboration on Planning Research Could Improve Information on Health Effects ( GAO-17-546 ). What GAO Found The Department of Energy (DOE), Nuclear Regulatory Commission (NRC), Environmental Protection Agency (EPA), and Food and Drug Administration generally used the advice of scientific advisory bodies to develop and apply radiation protection requirements and guidance for workers and the public in the radiation exposure settings that GAO reviewed. These settings were: (1) the operation and decommissioning of nuclear power plants; (2) the cleanup of sites with radiological contamination; (3) the use of medical equipment that produces radiation; and (4) accidental or terrorism-related exposure to radiation. Specifically, the agencies relied on the advice of three scientific advisory bodies that supported the use of a model that assumes the risk of cancer increases with every incremental radiation exposure. Accordingly, the agencies have set regulatory dose limits and issued guidance to confine exposure to levels that reduce the risk of cancer, while recognizing that scientific uncertainties occur in estimating cancer risks from low-dose radiation. For example, NRC requires nuclear power plants to consider measures for limiting workers' exposure below NRC's regulatory dose limit, such as by using robots for maintenance work in radiation areas. GAO identified seven federal agencies that funded research on low-dose radiation's health effects. In fiscal years 2012 to 2016, DOE, NRC, EPA, and four other federal agencies obligated about $210 million for such research . Although the agencies have collaborated on individual projects on radiation's health effects, they have not established a collaborative mechanism to set research priorities. GAO's previous work has shown that federal agencies can use such mechanisms to implement interagency collaboration to develop and coordinate sound science policies. In the past, DOE took a leading role in this area because DOE provided stable funding and advocated for greater coordination on research on low-dose radiation's health effects. However, since fiscal year 2012, DOE has phased out funding for one of its main research programs in this area. This has created a void in coordination efforts among federal agencies, and no other agency has stepped forward to fill this void. Because of DOE's prior experience as a leader in this area of research and its research responsibility under the Atomic Energy Act of 1954, it could play an important role in helping federal agencies establish a coordinating mechanism for low-dose radiation research. Dollars are in millions and have not been adjusted for inflation Source: GAO analysis of agency data. | GAO-17-546

gao_GAO-18-666

gao_GAO-18-666_0

Background The HUBZone Act of 1997 (which established the HUBZone program) identified HUBZones as (1) qualified census tracts, which are determined generally by area poverty rate or household income; (2) qualified nonmetropolitan counties, which are determined generally by area unemployment rate or median household income; and (3) lands meeting certain criteria within the boundaries of an Indian reservation. Congress subsequently expanded the criteria for HUBZones to add former military bases, counties in difficult development areas outside the continental United States, and certain areas affected by disasters. According to SBA officials, 5,306 firms were HUBZone certified as of July 1, 2018. As of that date, there were 20,154 HUBZone qualified census tracts, 834 HUBZone qualified non-metropolitan counties, 125 HUBZone base realignment and closure areas, 593 HUBZone Native American lands (Indian lands), and 95 HUBZone qualified disaster areas (87 qualified disaster census tracts and 8 qualified disaster non-metropolitan counties). The HUBZone program provides certified small businesses located in designated areas with contracting opportunities in the form of set-asides, sole-source awards, and price-evaluation preferences. A set-aside restricts competition for a federal contract to specified contractors. For example, competition may be restricted to SBA-certified HUBZone businesses if there is a reasonable expectation of at least two SBA- certified HUBZone bidders and a fair market price. A sole-source award is a federal contract awarded, or proposed for award, without competition. Also, in any full and open competition for a federal contract, the HUBZone price evaluation preference allows the price that a HUBZone firm offers to be deemed lower than the price of another offeror (if the HUBZone firm’s offer is not more than 10 percent higher than the other offer and the other offeror is not a small business concern). HUBZone Certification and Recertification Processes and Related GAO Recommendations To be certified to participate in the HUBZone program, a firm must meet the following criteria: when combined with its affiliates, be small by SBA size be at least 51 percent owned and controlled by U.S. citizens, or owned by an Indian Tribal Government, Alaska Native Corporation, Community Development Corporation, or small agricultural cooperative; have its principal office—the location where the greatest number of employees perform their work—in a HUBZone; and have at least 35 percent of its employees reside in a HUBZone. In a 2008 report, we found that SBA had relied on information that firms entered into an online application system called the HUBZone Certification Tracking System to indicate they met the size, ownership, location, and employee residence standards. At the time, SBA performed limited verification of the self-reported information. Although agency staff could request additional supporting documentation, SBA did not have specific guidance or criteria for such requests. Thus, we recommended that SBA develop and implement guidance to more routinely and consistently obtain supporting documentation from applicant firms. SBA agreed with the recommendation and changed its certification procedures. Since fiscal year 2009, SBA has required all firms applying for HUBZone certification to provide supporting documentation about their size, ownership, location, and employees’ residence, which the agency then is to review to verify the firm’s eligibility for the program. According to SBA officials, the current initial certification process has two components: (1) submission and review of an online application, which is processed by SBA through the HUBZone Certification Tracking System; and (2) submission of corroborative documentation, which SBA processes through a document review. The tracking system contains information on firms that apply to the HUBZone program, as well as on certified firms that apply for recertification. Firms wishing to remain in the program must recertify their continued eligibility to SBA every 3 years. The tracking system automatically identifies firms that are due for recertification and sends notifications to those firms. In 2015, we reported that SBA implemented controls for certification, but generally did not require firms seeking recertification to submit any information to verify continued eligibility. Instead, the agency relied on firms’ attestations of continued eligibility. According to SBA officials at the time, they did not believe they needed to request supporting documentation from recertifying firms because all firms in the program had undergone a full document review. We noted that SBA could apply a risk-based approach to its recertification process to review and verify information from firms that appeared to pose the most risk to the program. We concluded that recertification essentially remained a self-certification process. We recommended that SBA reassess its recertification process and add additional controls, such as developing criteria and guidance on using a risk-based approach to requesting and verifying firm information. SBA agreed with our recommendation and noted it would assess the process. In March 2017, we reported that SBA planned to address our recommendation using a technology solution. GAO’s Fraud Risk Framework and Related Legislation According to federal internal control standards and GAO’s fraud risk framework, managers in executive branch agencies are responsible for managing fraud risks and implementing practices for combating those risks. When fraud risks can be identified and mitigated, fraud may be less likely to occur. Federal internal control standards call for agency management officials to assess the internal and external risks their entities face as they seek to achieve their objectives. The standards state that management should consider the potential for fraud when identifying, analyzing, and responding to risks. Risk management is a formal and disciplined practice for addressing risk and reducing it to an acceptable level. In July 2015, we issued the fraud risk framework, which provides a comprehensive set of key components and leading practices that serve as a guide for agency managers to use when developing efforts to combat fraud in a strategic, risk-based way. The Fraud Reduction and Data Analytics Act of 2015 required the Office of Management and Budget to establish guidelines for federal agencies to create controls to identify and assess fraud risks and design and implement antifraud control activities. In July 2016, the Office of Management and Budget issued guidelines that, among other things, affirm managers should adhere to the leading practices identified in GAO’s fraud risk framework. National Defense Authorization Act for Fiscal Year 2018 The National Defense Authorization Act for Fiscal Year 2018 includes a number of provisions relating to the HUBZone program. For example, the act requires SBA (by January 1, 2020) to verify the accuracy of documentation provided by a HUBZone firm seeking recertification to determine whether the firm remains qualified for HUBZone certification. The act also requires SBA to begin conducting examinations of qualified HUBZone firms by January 1, 2020, using a risk-based analysis to select firms to be examined. According to the act, these risk-based examinations are intended to ensure that each firm examined meets the program requirements for certification. The act also specifies that any small business that SBA determines to have misrepresented its status as a qualified HUBZone firm be subject to liability for fraud. HUBZone Expansion in Puerto Rico On June 16, 2016, SBA announced it had revised the definition of qualified census tracts eligible to be designated as HUBZones to provide more opportunities for firms in Puerto Rico. Previously, in addition to poverty rate and income, SBA applied a statutory population cap that limited the number of eligible census tracts. According to the announcement, SBA had determined that the 20 percent population cap was not in keeping with the spirit and intent of the HUBZone program. On June 30, 2016, PROMESA authorized an exemption to the 20 percent cap for HUBZone designations in Puerto Rico for a limited time (10 years or until the date on which the Financial Oversight and Management Board for Puerto Rico ceased to exist, whichever came first). It also required SBA to promulgate regulations to implement the exemption. SBA promulgated the regulations, which became effective December 22, 2017. By lifting the population cap in June 2016, SBA increased the number of eligible census tracts in Puerto Rico by 516 (from 260 to 776). As a result, nearly all of Puerto Rico now qualifies as a HUBZone (see fig.1). 2017 Hurricanes In 2017, two major hurricanes (Irma and Maria) hit Puerto Rico. Hurricane Irma skirted Puerto Rico and left more than 1 million people without power. Hurricane Maria, a Category 4 hurricane, made landfall and caused catastrophic damage. For instance, Hurricane Maria wiped out the power grid, resulting in outages across Puerto Rico that continued for months after the storm. The majority of the island had its power restored by April 4, 2018, according to the Department of Energy. However, the power grid remains fragile. For example, Puerto Rico experienced an island-wide outage on April 18, 2018. SBA Adopted Criteria and Guidance for a Risk-Based Approach to HUBZone Certification and Recertification but the Extent to Which It Conducted a Risk Assessment of the Recertification Process Is Unclear SBA Adopted New Documentation Requirements for Recertifying Certain HUBZone Firms In response to the PROMESA provision to implement a risk-based approach to HUBZone certification and recertification, SBA adopted certification and recertification criteria and guidance on March 27, 2017, for requesting and verifying information. Since 2009 and in response to a prior GAO recommendation, SBA has required and reviewed documentation from all small businesses seeking initial certification to show compliance with HUBZone eligibility requirements. The criteria and guidance adopted by SBA in March 2017 did not change the HUBZone certification process. SBA’s internal guidance directs staff to review and confirm the accuracy of all documentation provided by the firms applying for HUBZone certification. However, the 2017 guidance and criteria introduced some documentation requirements to the recertification process. (Before March 2017, firms seeking HUBZone recertification were not required to submit documentation to demonstrate continued compliance with eligibility requirements.) Currently, any certified firm seeking recertification that received $1 million or more in HUBZone contract dollars since its last certification or recertification has to demonstrate compliance with the 35 percent HUBZone employee residency and principal office requirements. More specifically, under the March 2017 guidance and criteria, certified HUBZone small business concerns that have received $1 million or more in HUBZone contract dollars since their initial certification or most recent recertification must submit (1) a list of all current employees, identifying the name of the employees, their addresses, the number of hours they worked per month, and the location at which they performed their work; and (2) payroll documentation. HUBZone firms seeking recertification that have not received $1 million in HUBZone contract dollars are not required to submit documentation supporting continual eligibility for the program. Extent to Which SBA Conducted a Risk Assessment of the Recertification Process Is Unclear SBA officials stated that they had completed a risk assessment of their HUBZone recertification process to develop the March 2017 guidance and criteria. In our 2015 report, we found SBA did not require firms to submit supporting documentation as part of the recertification process—in effect, firms self-certified. We recommended that SBA conduct an assessment of the recertification process and implement additional controls, such as developing criteria and guidance on using a risk-based approach to requesting and verifying firm information. However, as of July 31, 2018, SBA had not provided documentation of when the risk assessment was performed or of what risks were identified and considered in developing the criteria and guidance. SBA officials stated that out of necessity, given their current information system and staffing resources, they chose the $1 million threshold as the only criterion for their risk-based approach. As of July 31, 2018, SBA also had not provided documentation of what analysis was performed to establish the $1 million threshold. According to SBA officials, the threshold was determined based on the belief that, when a firm received contract awards over $1 million, it increased opportunity for the firm to fall out of compliance with HUBZone requirements. SBA officials noted that they plan to review the current threshold at the end of fiscal year 2018 and that the threshold amount will likely decrease because a relatively small number of firms have exceeded the $1 million threshold. Based on our analysis of Federal Procurement Data System- Next Generation data, we estimated that from fiscal year 2015 through fiscal year 2017 about 9 percent of HUBZone firms nationally exceeded the $1 million threshold. SBA officials also noted that the agency has been developing a new information system to replace the current certification tracking system. According to SBA officials, the new system will help simplify the application process and allow firms to submit additional documentation more easily. SBA officials said the 2017 criteria and guidance will be built into the parameters of the new information system, which will allow SBA to expand the review of documentation submitted by firms seeking recertification. Officials have stated that subject to funding and resources, the HUBZone Certification Tracking System is scheduled to be decommissioned in fiscal year 2019. In the past, SBA has implemented a number of actions to better ensure that only eligible firms participate in the HUBZone program and to address internal control weaknesses that we have identified in previous GAO reports. However, in 2015, we found that SBA lacked key controls for its recertification process. Specifically, SBA did not require firms to submit supporting documentation as part of the recertification process—in effect, firms self-certified. We reported that by not routinely requiring and reviewing key supporting documentation from recertification applicants, SBA was missing an additional opportunity to reduce the risk that ineligible firms obtain HUBZone contracts. We recommended that SBA conduct an assessment of the recertification process and implement additional controls, such as developing criteria and guidance on using a risk-based approach to requesting and verifying firm information. Standards for Internal Control in the Federal Government state that management should identify, analyze, and respond to risks related to achieving the defined objectives of the program. Additionally, the standards state that management should use a risk assessment to identify and analyze risks related to achieving the defined objectives to form a basis for designing risk responses. Once identified, management can analyze the identified risks to estimate their significance and design responses to the analyzed risks. Similarly, according to GAO’s fraud risk framework, leading practices for managing fraud risk include identifying inherent fraud risks affecting the program, assessing the likelihood and impact of inherent fraud risks, and examining the suitability of existing fraud controls. The Fraud Reduction and Data Analytics Act of 2015 states that agencies shall conduct an evaluation of fraud risks using a risk-based approach, and then design and implement financial and administrative control activities to mitigate identified fraud risks. Additionally, as discussed above, the National Defense Authorization Act for Fiscal Year 2018 requires that SBA perform examinations of HUBZone firms using a risk-based selection process. As we reported in 2015, the characteristics of firms and the status of HUBZone areas—the bases for program eligibility—can often change, and need to be monitored. We continue to believe that conducting a risk assessment of the recertification process would help inform a risk-based approach to reviewing and verifying information from firms that appear to pose the most risk to the program. For example, a risk assessment could help inform SBA’s planned review of the current threshold for requesting and verifying information from firms seeking HUBZone recertification. SBA Did Not Consistently Document or Follow Its Policies and Procedures for Certification Reviews for Firms in Puerto Rico That We Reviewed SBA Had Complete Documentation for Office Location but Not Employee Residency for Firms in Puerto Rico We Reviewed and Had Not Updated Its Policies Our review of 12 case files for Puerto Rican firms that recently received HUBZone certification found incomplete documentation for certification reviews (and by extension, recertification reviews) and undocumented procedures. SBA also did not consistently follow its procedures to complete three distinct levels of review when approving the 12 firms for certification. Representatives of firms in Puerto Rico with whom we spoke said certification was generally straightforward, but identified some challenges, including with providing documentation. We found that all 12 cases we reviewed in a non-generalizable sample of Puerto Rican firms that had received HUBZone certification between March 2017 and March 2018 had complete documentation to demonstrate that the primary office was located in a HUBZone. However, we also found that documents were missing, illegible, or did not corroborate the information claimed regarding employees’ residency for 9 of 12 firms that we reviewed. We focused on the document review process for the principal office location and the employee residency requirements because, according to SBA’s March 2017 guidance for the certification and recertification processes, firms must corroborate their compliance with these requirements at both certification and recertification. In addition, SBA officials told us that the document review process is essentially the same for firms seeking certification as it is for firms seeing recertification. We also reviewed the files of two HUBZone firms in Puerto Rico that were subject to recertification between March 2017 and March 2018, but neither met the threshold to trigger a document review under the updated criteria and guidance. SBA’s guidance for the certification process requires firms to submit documentation so that SBA can verify that a firm meets the employee residency and principal location requirements. The documentation must show that each individual is an employee of the firm (payroll documentation), each individual asserted to be a HUBZone resident lives at the address they claim (proof of address, such as a driver’s license), and the address is in a HUBZone (copy of a HUBZone map indicating the employee and his or her address), and the firm’s primary office is located in a HUBZone (such as a copy of a lease or rental agreement and list of employees who work there). Additionally, SBA’s internal policy manuals describe procedures that analysts are to follow in reviewing and verifying these documents, including the types of documents that firms may submit to demonstrate eligibility, and scenarios in which analysts should request additional information or clarification from the firm (for example, if a driver’s license is expired). For 9 of 12 cases we reviewed, SBA lacked complete documentation to verify that every employee asserted to be a HUBZone resident lived in a HUBZone. That is, although SBA analysts contacted firms during the certification process to obtain additional documentation, and firms responded to the requests, SBA analysts still had incomplete information for 9 of the 12 cases at the time they approved the firms. In 5 of these 9 cases, the firm would not be eligible for HUBZone certification if the SBA analyst had not counted employees for which documentation was missing or did not corroborate the information claimed regarding an employee’s address. Specifically, in 3 cases, the address on the employee’s identification for at least one employee did not match the address at which the firm claimed the employee lived. in 3 cases, the identification for at least one employee was expired. in 1 case, the copies of the HUBZone maps did not indicate to which employees they referred. in 1 case, payroll documentation demonstrating that individuals were employees of the firm was missing. in the other 4 cases, the firm still would be eligible for HUBZone certification (would meet the 35 percent residency requirement) even if the analyst had not counted the employee for which documentation was missing or illegible (1 case) or for which identification did not match the claimed residency address (4 cases). We also identified an inconsistency when reviewing the case files for the only Puerto Rican HUBZone firm recertified between June 2017 and May 2018. Specifically, the firm reported on its application that it had zero employees, but also claimed 19 of its employees lived in a HUBZone. The firm did not provide any supporting documentation because it did not meet the threshold in the new recertification criteria to require a document review. SBA officials said this likely resulted from a display error in the online system. SBA officials said that they did not contact the recertifying firm because the firm was not required to submit corroborating documentation. However, without confirmation of the correct total number of employees, SBA analysts would not be able to determine the firm’s compliance with the employee residency requirement (35 percent living in a HUBZone) and therefore would not be able to determine the firm’s eligibility to continue HUBZone participation. SBA officials explained how they handled incomplete information provided by firms by describing other procedures that analysts followed in these cases. For example, if any employee’s address did not match across the firm’s employee list, the employee’s identification, or the HUBZone map, the analyst reviewing the firm’s application would request clarification and enter the address from the identification into the current HUBZone map. Or if the address could not be plotted, the analyst would presume the employee lived in a HUBZone if the broader geographic area on their identification (such as zip code) was clearly in a HUBZone. SBA officials said this approach is useful for applications from firms in Puerto Rico, because of difficulties with mapping addresses in the territory. SBA officials also said that if a firm did not provide documentation for an employee or SBA could not verify it, but the firm clearly met the 35 percent resident requirement, the analyst would not follow up with additional documentation requests and not count that employee as a HUBZone resident. However, these procedures are not documented in SBA’s internal written policies for certifying HUBZone firms and analysts did not document their use of these procedures in the case files we reviewed. Specifically, the policy manuals do not include certain procedures SBA officials explained to us they used for the document review process, such as assumptions of eligibility for addresses located in a broader geographic area that is clearly a HUBZone. SBA has not updated its three internal policy manuals since 2014, 2010 and 2007, respectively. According to SBA officials, analysts rely on oral direction from the HUBZone program director or deputy to review and process HUBZone certifications and recertifications to supplement outdated policy documents. Internal control standards state that management should document in policies the internal control responsibilities of the organization. Because SBA has not updated its internal policy manuals, analysts who review applications for HUBZone certification and recertification may not be consistently following applicable internal policies and procedures. As a result, management does not have reasonable assurance that analysts are following procedures correctly to obtain and review all of the required documents from firms. Without such review, SBA may not have reasonable assurance that firms meet the eligibility criteria for HUBZone participation—which increases the risk of ineligible firms participating in the program. Because SBA officials told us that firms seeking recertification that meet the threshold for additional document review undergo essentially the same reviews as firms seeking certification, the gaps in certification cases that we observed suggest that gaps are possible in recertification as well. SBA Did Not Consistently Follow Its Quality Review Procedures When Certifying Firms In Puerto Rico That We Reviewed SBA did not consistently follow its quality review procedures to complete three distinct levels of review when approving the 12 Puerto Rican firms in our non-generalizable sample for certification. In 4 of 12 cases, one person reviewed the application for two levels of review; in the other 8 cases, three different analysts reviewed the application. We found that SBA notified all 12 firms of their approval in writing, in accordance with its policy. SBA’s internal policy states that the certification process has three levels of review: (1) an analyst reviews the application documents and makes a recommendation to approve or deny the firm, (2) a senior analyst reviews the application and the first analyst’s recommendation, and (3) the program director or deputy finalizes the approval or denial and notifies the firm of the decision. The three levels of review are intended to provide quality assurance. For example, the second analyst reviews the case file and the first analyst’s recommendation for completeness, accuracy, and consistency with eligibility criteria. SBA officials told us that, in some cases, one individual’s review may count as two of the three levels of reviews in order to make the most efficient use of available analyst capacity. SBA officials said the deputy director reviewed several applications from Puerto Rican firms, including the four cases we identified, as part of work to develop an application screening tool and to engage the Puerto Rico district office when processing applications from firms in Puerto Rico. These reviews were counted as two of the three levels of review. Officials said that in these cases, the program director completes the final review and should annotate the case file notes about the quality assurance actions that were taken. In three of the four cases, the program director noted that she completed the final review and approval but not what quality assurance actions were taken; the deputy director completed both the second and final review in the other case. SBA officials described the third-level review (final approval or denial) as a high-level review to ensure no questions or unresolved matters are outstanding in order to approve or deny the application. Therefore, it is not likely that the program director would be reviewing supporting documents to identify any problems that could be identified if an additional analyst conducted a second-level, quality assurance review. Internal control standards state that management should design control activities at various levels with a segregation of duties, and periodically review its procedures and associated internal control activities for effectiveness. Management should also monitor its internal control system through ongoing monitoring to assess the quality and effectiveness of the internal control system’s performance over time. SBA provided an assurance letter prepared in response to the Federal Managers Financial Integrity Act that stated the agency evaluated the Office of HUBZone’s internal controls and concluded the controls were effective. However, it is not known to what extent SBA reviewed staff’s compliance with certification and recertification quality review procedures as part of this assessment, because the letter does not describe what steps SBA took to conduct the evaluation. Without reviewing staff compliance with certification and recertification procedures, SBA lacks reasonable assurance that analysts follow such procedures and that internal controls function effectively. This increases the risk that ineligible firms could receive HUBZone certification and thus contracting preferences to which they are not entitled. Representatives of Firms in Puerto Rico We Interviewed Said Certification Process Generally Was Straightforward, but Documentation Requests Presented Some Challenges Representatives of nine HUBZone firms we interviewed in Puerto Rico were not aware of changes made to the recertification process, but said that the certification process they followed was straightforward and generally easy. However, they reported some challenges, including with documentation. Representatives of most of the firms said that the most difficult part of the certification process was documenting the address for their primary office location and employees’ residences to show they are located in a HUBZone. They said this is time-consuming and tedious, especially for firms with many employees. Although they felt that SBA’s HUBZone mapping software had improved, they said the formatting of addresses in Puerto Rico creates a challenge, consistent with what we reported in 2017. One representative described having to pinpoint locations manually in Google Maps to obtain geographic coordinates, and then enter the coordinates into SBA’s HUBZone map instead of street addresses. Other representatives noted that some of their employees have informal living arrangements and cannot easily provide proof of address. Some representatives said that they struggled with the specific time frames that payroll documentation must cover. SBA requires firms to submit payroll documentation for the pay period that includes the date of their application and a sufficient number of preceding payrolls to cover a 4-week period, but firms said it was difficult to submit payroll reports at exactly the right time to meet this requirement. According to representatives from HUBZone firms and two economic development organizations we interviewed, some challenges with certification may be unique to firms in Puerto Rico, including the address format issue described above. Representatives from one firm and the economic development organizations said that some firms in Puerto Rico may face language barriers if Spanish is their primary language, or if they lack formal documentation required for certification, such as not having computerized records. The 2017 hurricanes also created several challenges for firms. For example, some firms did not have electricity and closed for several months. Firms that closed while trying to obtain HUBZone certification could not respond to follow-up requests from SBA immediately or provide documentation of their business operations for the time period in which they were affected. Representatives from firms we interviewed said they had little contact with SBA other than in relation to compliance. Although some have attended SBA events, representatives said they generally worked with SBA partner organizations, such as the Federal Contracting Center and Small Business and Technology Development Center to obtain assistance with gaining HUBZone certification and applying for federal contracts. Representatives from three firms we interviewed said that officials from SBA visited their offices to provide technical assistance support. According to SBA officials in Washington, D.C., they have improved their communications with firms and increased outreach for the HUBZone program in recent years. For example, the HUBZone tracking system automatically generates emails to firms, such as reminders that they are due for recertification, which according to SBA officials, has eliminated the backlog of recertifications on which we reported in February 2017. Representatives from the Puerto Rico District Office also said they increased promotion of the HUBZone program in recent years through monthly events and seminars on all SBA programs across Puerto Rico, including a seminar on obtaining HUBZone certification. Representatives from the district office said they are not involved with the certification and recertification processes, but provide support by conducting from three to six site visits to HUBZone firms annually in Puerto Rico and the U.S. Virgin Islands. HUBZone Set-Aside Contracting Was Minimal in Comparison to Other Small Business Contracting in Puerto Rico; However, Assessment of HUBZone Expansion Impacts Might Be Premature An increase in HUBZone set-aside contracts could theoretically help deliver economic impact to Puerto Rico, but it is likely too soon to assess larger-scale economic benefits resulting from HUBZone program expansion in Puerto Rico as of June 2018. From fiscal years 2006 through June 2018, the share of HUBZone set-aside contracts as a percentage of small business contracts in Puerto Rico has remained low despite sharp increases in the number of HUBZone firms and overall federal contracting in Puerto Rico. Over this period, small businesses in Puerto Rico have been winning a large and increasing percentage of total federal contracting obligations in Puerto Rico. Firms, economic development organizations, and SBA representatives said any impacts of HUBZone expansion in Puerto Rico may not yet be observable due to hurricane-related setbacks. They also identified longer-standing challenges that Puerto Rican firms have faced in obtaining HUBZone set- aside contracts consistent with those we identified in 2017. In 2018, SBA established a procurement center representative (PCR) in Puerto Rico, which the agency expects will help address some of those challenges. Economic Impacts of HUBZone Expansion in Puerto Rico Not Yet Apparent Based on our analysis as of June 30, 2018, it appears that larger-scale economic benefits from HUBZone expansion have not been realized (it may be too early to assess the impact of program expansion). While HUBZone certification alone likely would not have a direct economic impact (such as on job creation) in Puerto Rico, the expansion of HUBZones in Puerto Rico, which now cover nearly the entire island, gives opportunities to more firms to qualify for and pursue preferential contracting opportunities. The number of HUBZone firms has risen since the expansion—from 20 in September 2016 to 101 as of June 30, 2018. The program could help deliver economic impacts if HUBZone firms received contract awards as a result of contract preferences such as set- aside or sole-source contracts. For instance, employees hired to fulfill HUBZone set-aside contracts could represent jobs created in Puerto Rico (if those contracts otherwise would have been awarded to a firm outside Puerto Rico). However, if the contract otherwise would have been awarded to another Puerto Rican firm, the new jobs would represent an economic transfer, not jobs created. Our analysis of HUBZone set-aside contracts in Puerto Rico indicated that a few, newly certified HUBZone firms received set-aside contracts, which can be a source of job creation. Six of the nine firms that received HUBZone set-aside contracts in fiscal years 2017 and 2018 (through June 2018), totaling $5.1 million, were certified after the 2016 program expansion. One newly certified HUBZone firm that we interviewed said that program expansion resulted in the firm becoming HUBZone- eligible. The firm’s representative said that it obtained a 5-year HUBZone set-aside contract for $700,000, which resulted in the firm hiring six employees to fulfill the contract. But HUBZone set-aside contract obligations have remained largely unchanged in recent years, and firms face both temporary and longer- standing challenges in accessing and winning contract awards, as we discuss later in this section. For a description of contracting trends in Puerto Rico, see appendix II. Federal Contracting to Small Businesses in Puerto Rico More Than Doubled Over the Past 3 Years Based on our analysis, federal contracts to businesses in Puerto Rico increased from $355 million in fiscal year 2015 to $841 million in fiscal year 2018 (through June 2018). Similarly, federal contracts to small businesses in Puerto Rico increased from $244 million to $688 million in the same period (see fig. 2). Although HUBZone set-aside contract obligations have remained largely unchanged in recent years, small businesses in Puerto Rico also have been winning a large and increasing percentage of total federal contracting obligations in Puerto Rico. Specifically, federal contract obligations to small businesses located in Puerto Rico as a percentage of total federal contract obligations in Puerto Rico increased from 69 percent in fiscal year 2015 to 82 percent in fiscal year 2018 (through June 2018). A significant portion of the increase in overall contracting and small business contracting to Puerto Rican firms in fiscal year 2018 (through June 2018) was related to hurricane relief. Specifically, we calculated that 55 percent of total federal contracting obligations in Puerto Rico and 64 percent of obligations awarded to Puerto Rican small businesses in fiscal 2018 were associated with hurricane relief. HUBZone Set-Aside Contracts as a Share of Federal and Small Business Contracts in Puerto Rico Remained Relatively Low in Recent Years While the obligations for federal contracting to small businesses in Puerto Rico sharply increased in recent years, as did the number of HUBZone firms, the share of HUBZone set-asides as a percentage of small business contracts in Puerto Rico remained low and relatively unchanged (see fig. 3). HUBZone set-aside contract obligations increased from negative $20,881 in fiscal year 2016 to $3.8 million in fiscal year 2017 (1 percent of the total value of small business contracts in Puerto Rico). The set-aside obligations dropped to $1.7 million through June 2018 (0.2 percent of the total value of small business contracts in Puerto Rico). Representatives of two economic development organizations that we interviewed stated that continued low use of HUBZone set-aside contracts by contracting agencies could subsequently decrease firms’ willingness to participate in the HUBZone program in Puerto Rico. Representatives of firms and two economic development organizations noted that the process of monitoring, identifying, and applying for HUBZone set-aside contracts is time consuming and without the availability of set-aside contracts program participation may not be worthwhile for firms in Puerto Rico. Temporary Challenges May Hinder Use of HUBZone Set-Aside Contracts in Puerto Rico in the Short Term Representatives of firms and two economic development organizations and SBA officials in Washington D.C. and the Puerto Rico District Office told us that use of HUBZone set-aside contracts and any resulting economic impacts of HUBZone expansion in Puerto Rico may not yet be observable because of temporary challenges, such as advance contracts and outmigration due to the 2017 hurricanes. Advance Contracts One temporary challenge they cited was the use of advance contracts by agencies in response to the 2017 hurricanes. In advance contracts, an agency establishes contracts before a disaster for goods and services typically needed during a disaster response. Such contracts may prevent contracting officers from establishing a HUBZone set-aside on contracts awarded for disaster response and recovery. For example, a representative from one firm we interviewed said that in the wake of the 2017 hurricanes it had ample supplies of a vaccine in high demand on the island. The firm’s representative thought that contracting officers previously established an advance contract with a company in the continental United States to supply the vaccine. Therefore, contracting officers were unable to contract the Puerto Rican HUBZone firm to supply the vaccine to the island. However, SBA officials in the Puerto Rico District Office told us that once advance contracts are completed, the contracting agency can replace the contract with another contract to perform those services, which could be a HUBZone set-aside contract. Outmigration from 2017 Hurricanes Representatives of firms and one economic development organization and SBA officials in the Puerto Rico District Office told us that outmigration from Puerto Rico after the 2017 hurricanes reduced the amount of talent and workers available to fulfill contracts. Representatives of one firm said that contracting officers may be concerned that the defection of talent from Puerto Rico could limit firms’ abilities to complete contracts, limiting the contracting officers’ willingness to set aside contracts. However, representatives of firms and one economic development organization said that contracts can be fulfilled by multiple Puerto Rican firms if necessary. Also, representatives of firms and one economic development organization said that people who left the island have started returning and replenishing the workforce. Longer-Standing Challenges Also May Hinder Use of HUBZone Set-Aside Contracts Representatives of firms and two economic development organizations and SBA officials in Washington D.C. and the Puerto Rico District Office also identified longer-standing challenges to increased use of HUBZone set-aside contracts in Puerto Rico, including difficulty meeting procurement requirements, limited knowledge of the federal contracting process, lack of access to contracting officers, and award of contracts to firms outside Puerto Rico. These concerns are similar to those we recognized in our 2017 report on SBA contracting program in Puerto Rico. Difficulty Meeting Procurement Requirements Representatives of some firms we interviewed stated that procurement requirements for federal contracts (such as performance history for construction contracts) posed challenges for small businesses in Puerto Rico. Representatives of some firms said that as an island, Puerto Rico faces specific challenges in meeting procurement requirements. For example, one firm’s representative said that the construction of a school required a business to demonstrate experience in developing several schools in the past. According to this representative, opportunities to construct schools are limited in Puerto Rico and many Puerto Rican firms would be capable of fulfilling these contracts, but relevant experience constructing similar buildings is not considered as meeting this requirement. In 2017, we reported that the experience of construction businesses in Puerto Rico does not match procurement requirements, which are often standardized to mainland building standards and do not consider unique conditions in Puerto Rico. As a result, firms and associations that we interviewed in 2017 said that agencies’ contracting officers may not consider the experience of Puerto Rican businesses as qualifying. Representatives from four associations that we interviewed in 2017 stated that construction businesses in Puerto Rico demonstrate in their construction plans greater understanding of building requirements in Puerto Rico, such as accounting for tropical climate or the risk of seismic activity, but these factors were not incorporated into federal procurement requirements. Limited Knowledge of the Federal Contracting Process Representatives of firms and two economic development organizations and SBA officials in the Puerto Rico District Office said that limited knowledge of the federal contracting process can be a challenge for HUBZone firms in Puerto Rico. One firm’s representative explained that responding to agencies’ requests for information was important because to create a HUBZone set-aside contract a federal agency must demonstrate that at least two firms could compete for the contract. Another firm’s representative said that because it was unaware of the importance of responding to the requests for information, it had not responded to requests for information for any contract opportunities. Firms’ representatives also noted that the process of responding to requests for information and requests for proposals is time consuming and often infeasible because they lack the financial and personnel resources of larger businesses. Firms can obtain information about federal and small business contracting from several sources. Firms’ representatives we interviewed said that they generally seek assistance from SBA partner organizations, such as the Federal Contracting Center and Small Business and Technology Development Center, to gain a better understanding of the contracting process. Officials from SBA’s Puerto Rico District Office said that they hold regular training on contracting programs and how to navigate the contracting process. The District Office also holds one-on-one appointments with businesses to help them navigate the federal contracting process and has offered training on proposal writing. Lack of Access to Contracting Officers Representatives of firms we interviewed said that obtaining HUBZone set- aside contracts was difficult for small businesses in Puerto Rico partially because of a lack of access to contracting officers. Most firms’ representatives we interviewed said that, similar to obtaining assistance in navigating the contracting process, they seek assistance from SBA partner organizations, such as the Federal Contracting Center and Small Business and Technology Development Center, to identify HUBZone contract opportunities and connect with contracting officers. Firms’ representatives also said that they attend conferences and matchmaking events at which they meet contracting officers; however, they said that this approach has not yet helped them to obtain increased HUBZone set-aside contracts in Puerto Rico. One firm’s representative said that the advice at networking events is that firms should approach contracting officers, but small businesses do not have the contacts to approach contracting officers. Representatives of some firms said that SBA should act as an advocate or facilitator to provide connections between small businesses and contracting officers. Some of the firms’ representatives noted that in the absence of HUBZone set-aside contracts, they tend to pursue subcontracts with companies that have prime federal contracts. Award of Contracts to Firms Outside of Puerto Rico Representatives of two economic development organizations we interviewed said that most HUBZone set-aside contracts and most of the hurricane relief-related contracts performed in Puerto Rico were awarded to firms outside Puerto Rico. In 2017, we similarly reported that representatives from four associations stated that challenges such as lack of access to contracting officers and difficulty meeting procurement requirements led to concerns about contracts being awarded to businesses located outside of Puerto Rico for work to be performed in Puerto Rico. Representatives of firms and one economic development organization also pointed out that for some industries, work does not need to be physically performed at the contract location; therefore, Puerto Rican HUBZone firms could be competitive for contracts located in other states. According to firms and one economic development organization, this is especially true in the technology industry, which is well represented in Puerto Rico with multiple HUBZone firms. However, representatives of the Puerto Rican technology firms we contacted noted that they have experienced great difficulty obtaining HUBZone contracts. One firm’s representative suggested that this challenge might be partially due to Puerto Rican firms having less experience than firms outside of Puerto Rico in competing for HUBZone set-aside contracts. To compete for contracts, one firm said that its chief executive officer permanently moved to Washington, D.C., and another firm said its staff travels to Washington, D.C., once a month. Procurement Center Representative Based in Puerto Rico May Help Address Some Challenges Faced by HUBZone Firms in Puerto Rico In 2018, SBA established a PCR in Puerto Rico and SBA officials said the agency expects that the PCR will help address some of the challenges discussed previously. SBA PCRs work with federal agencies and small businesses to identify contracting opportunities for small businesses. The Puerto Rico PCR said that she plans to hold events for HUBZone firms to help them better understand and navigate the federal contracting process. For example, she plans to hold a Federal Acquisition Regulation “boot camp,” a 40-hour training to educate firms on the federal contracting process. She also stated that through regular interaction with other PCRs and with contracting officers, she can identify and discuss opportunities for HUBZone set-aside contracts for Puerto Rican small businesses. One economic development organization we interviewed said that it is too early to see an effect of the new Puerto Rico PCR, but that they have met with her and have been encouraged by her efforts to date. In 2017, we reported that several stakeholders identified the lack of an SBA PCR in Puerto Rico as a disadvantage for small businesses in Puerto Rico seeking contracts with the federal government. According to a representative from the Federal Contracting Center that we interviewed in 2017, having a PCR in Puerto Rico is important because the PCR can advocate for small businesses there. For example, the PCR can work with contracting officers to determine small business set-asides, make adjustments to procurement requirements, and make agency contracting officers more aware of businesses in Puerto Rico. According to one economic development organization’s representative, the PCR could assist local businesses and promote businesses located in Puerto Rico to federal agencies. Conclusions Our review of a sample of 12 case files for Puerto Rican firms certified between March 2017 and March 2018 found that SBA lacked complete documentation for one of two requirements we reviewed and did not consistently follow its own procedures for quality control reviews when approving firms. This suggests potential gaps in internal controls for both the certification and recertification processes, as SBA reviews firms’ compliance with program requirements at both certification and recertification. Although SBA policy includes documentation reviews and a quality control process, its internal policy manuals do not reflect all document review procedures, and analysts did not always follow procedures for quality control reviews. SBA has not updated its policy manuals and, it is not known to what extent it has reviewed staff compliance with the quality review procedures. Although SBA provided an assurance letter stating its internal controls are effective, the letter did not describe steps taken in the evaluation. Documenting all procedures would help ensure that analysts consistently follow policy when certifying and recertifying firms. Documented reviews of staff compliance also could serve to identify and remediate any noncompliance with certification and recertification processes. Both actions would serve to strengthen the verification function in the HUBZone program, which is necessary to help ensure that only eligible firms participate in the program. Recommendations for Executive Action We are making two recommendations to SBA. Specifically: The Administrator of SBA should update the agency’s internal policy manuals for certification and recertification reviews to reflect existing policies and procedures not currently in written guidance. (Recommendation 1) The Administrator of SBA should conduct and document reviews of staff compliance with procedures associated with HUBZone certification and recertification. (Recommendation 2) Agency Comments and Our Evaluation We provided a draft of this report for review and comment to SBA. In response, SBA provided written comments, which are reproduced in appendix III. SBA also provided technical comments, which we incorporated as appropriate. SBA generally agreed with both of our recommendations. However, SBA disagreed with three of our findings in the draft report. In the draft report, we stated that SBA’s guidance for the certification process requires firms to submit documentation so that SBA can verify that a firm meets the employee residency requirement, including documentation showing that each employee’s address is located in a HUBZone, for employees claimed to be HUBZone residents. We noted that documents were missing, illegible, or did not corroborate the information claimed regarding at least one employee’s residency for 9 of 12 certifying firms that we reviewed. We also stated that in 5 of these 9 cases, the firm would not be eligible for HUBZone certification if the SBA analyst had not counted such employees. In its comments, SBA stated that it is only required to verify that no fewer than 35 percent of a HUBZone firm’s employees reside in a HUBZone. SBA asserted that it had sufficient documentation to conclude that at least 35 percent of each firm’s employees resided in a HUBZone for each of the 12 firms we reviewed. While SBA analysts may have addressed insufficient documentation by following additional procedures, which are described in the report, we found that these procedures were not documented in SBA’s internal written policies for certifying HUBZone firms and analysts did not document their use of these procedures in the case files we reviewed. Therefore, we were not able to verify that SBA took such steps to verify the employees’ addresses. In the draft report, we identified an inconsistency when reviewing the case files for the only Puerto Rican HUBZone firm recertified between June 2017 and May 2018. Specifically, the firm reported on its application that it had zero employees, but also claimed 19 of its employees lived in a HUBZone. In its comments, SBA stated that these numbers came from a program examination system that is no longer used by the agency and therefore would not have been administratively correct to use in the 2017 recertification process. However, the file we received from SBA’s HUBZone Certification Tracking System asks for such information and had a response date noted as December 26, 2017. We recognize and noted that the firm was not required to submit corroborating documentation to verify employee information, because it was under the $1 million threshold. However, SBA policy states that recertifying firms must represent that the circumstances relative to their eligibility at the time of certification have not materially changed. In this instance, SBA recertified the firm when basic information obtained for the firm appeared to be erroneous and did not indicate that the firm was in compliance with the employee residency requirement. In the draft report, we stated that SBA did not consistently follow its quality review procedures to complete three distinct levels of review when approving the 12 Puerto Rican firms in our non-generalizable sample for certification. In its comments, SBA stated that while its legacy system, the HUBZone Certification Tracking System, requires three levels of review, its internal policies do not require that each level of review be performed by different staff. However, SBA internal policies that we reviewed state that the three levels of review should be conducted by different people, specifically an analyst, a senior analyst, and the program director. Furthermore, internal control standards state that management should design control activities at various levels with a segregation of duties, and periodically review its procedures and associated internal control activities for effectiveness. We are sending copies of this report to congressional committees, the Small Business Administration, and other interested parties. In addition, this report will be available at no charge on GAO’s website at http://www.gao.gov. If you or your staff members have any questions about this report, please contact me at (202) 512-8678 or shearw@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix IV. Appendix I: Objectives, Scope, and Methodology Our objectives in this report were to (1) examine the Small Business Administration’s (SBA) development of criteria and guidance on using a risk-based approach for certifying and recertifying Historically Underutilized Business Zone (HUBZone) firms, (2) examine SBA’s implementation of the revised policies and procedures for firms located in Puerto Rico, and (3) describe any economic impacts of HUBZone expansion and recent trends in federal small business contracting in Puerto Rico. To examine SBA’s development of its criteria and guidance for HUBZone certification and recertification, we reviewed SBA’s policies and procedures for certifying and monitoring HUBZone firms. To learn about SBA’s risk-based approach, we interviewed officials at the headquarters level responsible for certifying and recertifying HUBZone firms. We reviewed prior GAO and SBA Office of Inspector General reports, applicable statutes and regulations, and SBA documents. We also compared the development of SBA’s certification and recertification processes with federal internal control standards and relevant federal guidance and statutes for managing fraud risk. To examine SBA’s implementation of the revised recertification processes in Puerto Rico, we used SBA’s Dynamic Small Business Search database to identify HUBZone firms located in Puerto Rico that received HUBZone certification after March 27, 2017. We examined SBA’s document review process for Puerto Rican firms that received HUBZone certification or were due for recertification between March 2017 and March 2018. We reviewed the case files for a non-generalizable sample of 12 firms that received initial certification and two firms that were due for recertification during this time period. We reviewed documents submitted by firms to SBA as part of the initial certification process. But we were unable to examine the document review component of recertification directly. According to SBA officials, only two firms in Puerto Rico recertified between March 27, 2017, and May 30, 2018, when we received the applicant case files from SBA, and neither met the threshold of $1 million in contract awards to trigger a full document review. From SBA’s small business database, we identified 443 firms in Puerto Rico that had had HUBZone certification at any point as of March 12, 2018, and removed 350 firms that exited the program in order to review only currently certified firms (leaving 93 certified firms). From this universe, we identified 47 HUBZone-certified firms located in Puerto Rico with a certification date later than March 27, 2017, after removing duplicates. We chose a judgmental sample of 12 firms that we randomly selected from the total of 47 firms using a random number generator. We checked the city and industry of each of the 12 firms to confirm the firms provided some geographical and industry variation. Results from this sample are not generalizable to all HUBZone-certified firms. We examined the contents of the case files for each of the 12 firms, which included the application the firm submitted through SBA’s HUBZone Certification Tracking System and supporting documents that firms submitted to corroborate their eligibility, including proof of office location, payroll documentation, HUBZone maps, and employees’ identification such as driver’s licenses. We compared the case file contents to SBA’s guidance on which documents firms must submit and internal guidance for analysts reviewing applications. To determine which documents to review, we compared SBA’s list of required documents for certification to those required at recertification and identified which are similar. We reviewed SBA analysts’ notes, which are recorded in the HUBZone tracking system to verify that they reviewed the documents to determine the firms’ eligibility. We also reviewed any emails exchanged between SBA and the applicant to identify cases in which SBA requested follow-up from the firm. We reviewed the name of the SBA analyst who completed each level of review, as indicated in the tracking system file, to determine whether SBA followed its policy of completing three levels of review. We also reviewed the tracking system files to verify the final approval and whether SBA sent the certification notice to the firm. We compared these files to SBA’s internal policy manuals. To describe any economic impacts of the expansion of the HUBZone program in Puerto Rico, and to describe recent federal contracting trends, we used SBA’s small business database and federal procurement data from the Federal Procurement Data System-Next Generation. We analyzed the number of firms in designated HUBZones in Puerto Rico and the amount of federal contract obligations awarded to the firms from fiscal year 2006 through June of fiscal year 2018—that is, from the starting point for the analysis we used in our 2017 report to the most recent available data at the time of our current review. We assessed the reliability of the two databases by reviewing database guides and prior GAO work, and determined them to be reliable for the purposes of that analysis. For all of the objectives, we also interviewed SBA officials in Washington, D.C., and Puerto Rico and representatives from the Puerto Rico Chamber of Commerce and the Federal Contracting Center. In addition, we conducted a site visit to San Juan, Puerto Rico, in May 2018. There, we conducted two discussion groups and one interview with representatives from nine HUBZone firms located in Puerto Rico to obtain their perspectives on the HUBZone certification process, federal contracting opportunities, economic impacts of HUBZone expansion in Puerto Rico, and economic impacts of the 2017 hurricanes. We invited every Puerto Rican HUBZone firm to participate in the discussion groups and met with the nine firms that responded to our email. The views of representatives from these firms are not generalizable. We conducted this performance audit from February 2018 to September 2018 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Appendix II: Federal Contracting Trends and Use of Socioeconomic Set-Asides in Puerto Rico, Fiscal Years 2006–2018 This appendix provides information, based on our analysis of federal procurement data, on small business contracting in Puerto Rico, including the use of socioeconomic set-aside contracts and contracting by sector and agency. Small Business Contracting in Puerto Rico Annual federal contract obligations to small businesses in Puerto Rico more than doubled from fiscal year 2015 through June 2018. Small businesses in Puerto Rico also won a larger percentage of total federal contracting obligations in Puerto Rico. Use of Socioeconomic Set-Asides in Puerto Rico Although federal contract obligations to small businesses in Puerto Rico sharply increased in recent years, obligations through set-aside contracts for the Historically Underutilized Business Zone (HUBZone), 8(a) Business Development, Women-Owned Small Business, and Service- Disabled Veteran-Owned Small Business programs remained relatively unchanged (see fig. 4). The increases in federal contract obligations to small businesses in fiscal years 2017 and 2018 came mostly through free and open competition and awards reserved for small businesses that excluding those for the 8(a), HUBZone, Women-Owned Small Business, and Service-Disabled Veteran-Owned Small Business programs. HUBZone set-aside contract obligations increased from negative $20,881 in fiscal year 2016 to $3.8 million in fiscal year 2017 (1 percent of the total value of small business contracts in Puerto Rico), but dropped to $1.7 million through June 2018 (0.2 percent of the total value of small business contracts in Puerto Rico). Furthermore, 17.3 percent of the $21.9 million in contract obligations awarded to Puerto Rican HUBZone firms in fiscal year 2017 were through HUBZone set-aside contracts, and 3.3 percent of the $50.5 million awarded in fiscal year 2018 (see fig. 5). Other socioeconomic set-asides accounted for the highest share of obligations awarded (76.6 percent in 2017 and 80.1 percent in 2018) and free and open competition accounted for the rest (6.1 percent in 2017 and 16.6 percent in 2018). The fiscal year 2017 increase in federal contracting obligations in Puerto Rico were not substantially tied to hurricane relief since Hurricane Irma and Hurricane Maria hit Puerto Rico in September, the last month of the fiscal year. One percent of total federal contracting obligations in Puerto Rico and 1 percent of obligations awarded to Puerto Rican small businesses were associated with hurricane relief (see table 1). However, a significant portion of the increase in overall contracting and small business contracting to Puerto Rican firms in fiscal year 2018 (through June 2018) was related to hurricane relief. Specifically, 55.3 percent of total federal contracting obligations in Puerto Rico and 64.3 percent of obligation awarded to Puerto Rican small businesses were associated with hurricane relief. Furthermore, 16.1 percent of HUBZone set-aside contract obligations awarded to Puerto Rican firms were hurricane-related through June 2018. Contracting in Puerto Rico by Sector In fiscal years 2017 and 2018, overall federal prime contracting obligations in Puerto Rico and those awarded to small businesses in Puerto Rico were concentrated in the construction sector (about 25 percent in 2018), the manufacturing sector (about 20 percent in 2018), and the administrative and support and waste management sector (about 36 percent in 2018). For contracts awarded using HUBZone set-asides in Puerto Rico, the construction sector (about 52 percent in 2018), the information sector (about 11 percent in 2018), and the health care sector (about 37 percent in 2018) were most represented (see table 2). Contracting in Puerto Rico by Agency Among federal agencies, the Department of Defense has awarded the greatest percentage of overall federal prime contracting obligations, obligations to small businesses, and HUBZone set-aside contract obligations to firms located in Puerto Rico. Specifically, in fiscal year 2018, Department of Defense contract obligations represented 58 percent of total obligations, 64 percent of obligations to small businesses, and 84 percent of HUBZone set-aside obligations to Puerto Rican firms (see table 3). Although, Department of Homeland Security contract obligations in Puerto Rico typically have represented less than 3 percent of total obligations and less than 5 percent of obligations to small businesses in Puerto Rico, those percentages increased to 21 percent and 23 percent, respectively, in fiscal year 2018 due to hurricane-related contracts awarded to Puerto Rican firms. Appendix III: Comments from SBA Appendix IV: GAO Contact and Staff Acknowledgments GAO Contact William B. Shear, (202) 512-8678 or shearw@gao.gov. Staff Acknowledgements In addition to the contact named above, Harry Medina (Assistant Director), Chris Ross (Analyst in Charge), Tarik Carter, Lilia Chaidez, Pamela Davidson, Erika Huber, Julia Kennon, John McGrail, John Mingus, Barbara Roesmann, and Jena Sinkfield made key contributions to this report.

Why GAO Did This Study The HUBZone program is intended to stimulate economic development in economically distressed areas. Certified HUBZone firms are eligible for federal contracting benefits, including limited competition awards such as set-aside contracts, and are required to be recertified every 3 years. The Puerto Rico Oversight, Management, and Economic Stability Act of 2016 required SBA to develop criteria and guidance for a risk-based approach to verify firm eligibility for the program and included a provision for GAO to review SBA's development and implementation of the required criteria and guidance. This report examines, among other objectives, (1) SBA's development of criteria and guidance on using a risk-based approach for certifying and recertifying HUBZone firms, and (2) SBA's implementation of the revised policies and procedures for firms located in Puerto Rico. GAO analyzed SBA documents and reviewed files of a non-generalizable sample of 12 firms located in Puerto Rico that received certification between March 2017 and March 2018. GAO also interviewed SBA officials, representatives from HUBZone-certified firms in Puerto Rico, and local economic development agencies in Puerto Rico. What GAO Found The Small Business Administration (SBA) adopted criteria and guidance for a risk-based approach to certifying and recertifying firms for the Historically Underutilized Business Zone (HUBZone) program in March 2017, but the extent to which it conducted a risk assessment to inform its approach is unclear. In 2009, in response to GAO's prior recommendations to address weaknesses in the HUBZone certification process, SBA increased documentation requirements for certification, but not recertification (which every 3 years determines continued program eligibility). In March 2017, SBA changed its recertification criteria and guidance to require firms with $1 million or more in HUBZone contract awards to provide documentation to support continuing eligibility. SBA officials stated they completed a risk assessment of the HUBZone recertification process, but as of July 2018, had not provided GAO with documentation on when they performed the risk assessment, which risks were identified and considered, or what analysis established the $1 million threshold. GAO previously found SBA lacked key controls for its recertification process and recommended in 2015 that SBA assess the process. GAO continues to believe that an assessment of the recertification process would help inform a risk-based approach to reviewing and verifying information from firms that appear to pose the most risk to the program. Based on GAO's review of case files for a non-generalizable sample of 12 firms in Puerto Rico that received HUBZone certification in March 2017–March 2018, SBA did not consistently document or follow its policies and procedures for certification reviews. SBA did not have complete documentation in 9 of 12 cases. SBA officials described alternative procedures they used to determine firms' eligibility, but SBA has not updated its internal policy manuals to reflect these procedures and analysts did not document use of such procedures in the files GAO reviewed. In 4 of 12 cases, SBA did not follow its policy to conduct three levels of review (by an analyst, a senior analyst, and the program director or deputy) when determining to approve or deny a firm. It is not known to what extent SBA reviewed staff compliance with certification and recertification review procedures. SBA provided an assurance letter stating it evaluated the Office of HUBZone's internal controls and concluded the controls were effective, but the letter did not specify what steps SBA took for the evaluation. Standards for internal control state management should document its control policies and conduct periodic reviews to ensure controls are effective. Because SBA has not updated its internal policy manuals or conducted a documented review of staff compliance with its quality review procedures, it lacks reasonable assurance that firms are eligible or its review process is effective. In turn, this increases the risk of ineligible firms participating in the HUBZone program and receiving contracting preferences to which they are not entitled. What GAO Recommends GAO recommends that SBA (1) update internal policy manuals to reflect current policies and procedures, and (2) review and document staff compliance with procedures for certification and recertification reviews of firms. SBA generally agreed with GAO's recommendations.

gao_GAO-18-324

gao_GAO-18-324_0

Background MDA is responsible for developing a number of systems, known as elements, with the purpose of defending against ballistic missile attacks. MDA’s mission is to combine these elements into an integrated system- of-systems, known as the Ballistic Missile Defense System. Specifically, the goal of the BMDS is to combine the abilities of two or more elements to achieve objectives that would not have been possible for any individual element. These emergent abilities are known as “integrated capabilities” or “BMDS-level capabilities.” Table 1 provides a list and description of elements included in our review. MDA’s Acquisition Flexibilities and Steps to Improve Traceability and Oversight When MDA was established in 2002, it was granted exceptional flexibilities to set requirements and manage the acquisition of the BMDS—developed as a single program—that allow MDA to expedite the fielding of assets and integrated ballistic missile defense capabilities. These flexibilities allow MDA to diverge from DOD’s traditional acquisition life cycle and defer the application of acquisition policies and laws designed to facilitate oversight and accountability until a mature capability is ready to be handed over to a military service for production and operation. Some of the laws and policies include such things as: obtaining the approval of a higher-level acquisition executive before making changes to an approved baseline, reporting certain increases in unit cost measured from the original or current baseline, obtaining an independent life-cycle cost estimate prior to beginning system development and/or production and deployment, and regularly providing detailed program status information to Congress, including specific costs, in Selected Acquisition Reports. In response to concerns related to MDA’s flexibilities, Congress and DOD have taken a number of actions. For example, Congress enacted legislation in 2008 requiring MDA to establish cost, schedule, and performance baselines—starting points against which to measure progress—for each element that has entered the equivalent of system development or is being produced or acquired for operational fielding. MDA reported its newly established baselines to Congress for the first time in its June 2010 BMDS Accountability Report. Since that time, Congress has provided more detailed requirements for the content of these baselines. Additionally, to enhance oversight of the information provided in the BMDS Accountability Report, MDA continues to incorporate suggestions and recommendations from us. However, not all of our recommendations have been fully implemented. MDA’s Process for Delivering Capabilities Because MDA is not a military service, it does not abide by the same policies that the services use for delivering capabilities. Instead, a process exists whereby MDA declares an asset or capability ready for delivery for potential operational use. During this process, MDA communicates the capabilities and limitations of its delivery, and provides evidence supporting these assertions. Representatives from the receiving military service or combatant command then have the ability to assess this evidence and decide whether to accept the new capability. Because the military services conduct minimal missile defense testing of their own, this process is one of the only ways to convey vital performance information. The accuracy of this information is especially important as it informs training materials, doctrine, and deployment decisions. Typically, MDA makes capability deliveries through approved changes to its Operational Capacity Baseline (OCB). Proposed changes to the baseline are coordinated with the warfighter, including the affected combatant commands. Subsequently, the combatant commands assess these element capabilities to determine whether to accept them. This process is used for the vast majority of deliveries, including relatively minor ones such as software patches and updates. In recent years, MDA has declared major capabilities ready for delivery through a process that culminates in the issuance of a Technical Capability Declaration (TCD). According to MDA officials, the primary purpose of a TCD is to allow MDA’s senior management to manage the delivery of integrated, BMDS-level capabilities that require more than one element to function; however, TCDs have also been issued in response to mandates from the President. MDA’s Contracting Practices Though MDA has flexibilities in managing the acquisition process, it must follow the same contracting regulations that apply to DOD, including the Federal Acquisition Regulation and the Department of Defense Federal Acquisition Regulation Supplement (DFARS). We reviewed MDA’s use of a particular type of contract action that authorizes a contractor to begin work before contract terms, specifications, or price have been agreed upon. These “undefinitized contract actions” are permitted by the DFARS, with certain limitations. Undefinitized contract actions are generally used when negotiation of a definitive contract action is not possible in sufficient time to meet the government’s requirements and the government’s interest demands that the contractor be given a binding commitment so that contract performance can begin immediately. Under the DFARS, undefinitized contract actions must include a specific “not-to-exceed” price. Once the action’s terms, specifications, and price have been agreed upon or determined, a process known as definitization, the contract action converts to a “definitive” contract. Under the DFARS, undefinitized contract actions must contain definitization schedules that provide for definitization by the earlier of (1) 180 days after issuance or (2) the date on which the amount of funds obligated under the action is more than 50 percent of the not-to-exceed price. Once the government has received a qualifying proposal from the contractor, however, the government can extend the undefinitized period another 180 days. Similarly, the government may obligate up to 75 percent of the not-to-exceed price, if the contractor submits the qualifying proposal before 50 percent of the not-to-exceed price has been obligated. The amount of funds obligated should be consistent with the contractor’s requirements for the undefinitized period. Figure 1 shows the expected time frame and amount the government should spend within a specified period. Models and Simulations Used in Operational Testing of the BMDS The BMDS is a system of systems that cannot be completely assessed using intercept flight tests that are operationally representative because of the system’s scope and complexity and safety constraints. Consequently, MDA, independent DOD testing organizations, and the warfighter must rely heavily on representations of the integrated BMDS called models and simulations in ground testing, rather than live tests, to test the operational performance of the whole BMDS against attacks with more threats represented. In ground testing, each BMDS element is represented by a model and connected to a computer framework. During ground test execution, a model of threat ballistic missiles is applied to the framework and stimulates the modeled representations of BMDS elements to react. The resulting simulation models a BMDS engagement. Figure 1 illustrates the BMDS ground test sequence. To ensure that BMDS models and simulations accurately represent the real-world operational BMDS capabilities and that the limitations of the model are understood, they are verified, validated, and accredited. The verification, validation, and accreditation process is designed to identify and gather evidence needed to certify that the model and its associated data used in ground testing are acceptable for operational testing. No model is completely representative of the real world so the verification, validation, and accreditation process is used to assess the extent to which it reflects the operational performance of the BMDS in the real world, and how any modeling deficiencies impacted ground test results. Any modeling limitations identified in the verification, validation, and accreditation process restrict the extent to which ground test data can be used for BMDS assessment. For example, limitations in modeled sensor tracking of the threat restrict the extent to which tracking data can be relied on for interpreting operational real-world performance. Figure 2 illustrates the verification, validation, and accreditation process. The BMDS Operational Test Agency (OTA) is responsible for analyzing the verification and validation data for the models used in operational BMDS tests and provides accreditation recommendations to the Commanding General, Army Test and Evaluation Command, an independent accreditation authority for operational testing. In this role, the BMDS OTA develops accreditation criteria and assesses if the model can be used for operational assessments against these criteria. The BMDS OTA is also responsible for analyzing the extent to which the threat model, once it is applied to the ground testing framework, can be traced back to the threat model that MDA developed and the intelligence community’s description of the threat. MDA Made Some Progress, but Did Not Meet Many of Its Acquisition Goals, and Has Inconsistently Applied Its Capability Delivery Processes In fiscal year 2017, MDA made some progress delivering assets, including BMDS-level capabilities and conducting tests. However, MDA did not meet many of its goals as expressed in the Ballistic Missile Defense System Accountability Report for fiscal year 2017, its integrated master test plan, and master integration plan. Specifically, MDA continued to deliver interceptors for three elements and successfully conducted its first test against an intercontinental ballistic missile target. In addition, MDA announced the delivery of one package of integrated BMDS-level capabilities through a technical capability declaration (TCD), which had been delayed from the previous year, and planned to complete the delivery of another set of capabilities by March 2018. MDA, however, did not complete its goals for delivering assets, specifically for the THAAD interceptors or conducting planned testing for Aegis BMD. We also identified several deficiencies in MDA’s processes for communicating progress in delivering integrated capabilities. MDA Achieved Mixed Results in Delivering Assets and BMDS-Level Capabilities, Adhering to the Planned Test Schedule MDA made progress delivering assets against its backlogs from fiscal year 2016, while its test program achieved several notable milestones. MDA also delivered several new integrated capabilities, though not always on time and often with reduced content compared to what was planned to be delivered. In addition, not all deliveries and testing objectives were met, and MDA made a number of changes, additions, and deletions to its test and capability delivery schedule during the year. Elements: While BMDS elements made progress delivering assets, including some that were delayed from fiscal year 2016, MDA did not meet all of its asset delivery goals as planned. For a summary of MDA’s major asset deliveries for fiscal year 2017, see table 2 below. Both the Aegis Standard Missile-3 (SM-3) Block IB and Ground-based Midcourse Defense (GMD) programs succeeded in achieving their asset delivery goals for the fiscal year, although both included acceptance of assets delayed from prior fiscal years. Specifically, due to quality issues and design problems discovered during testing, production on the Aegis SM-3 Block IB interceptor was temporarily halted in fiscal year 2016, and as a result MDA fell short of its deliveries for that year by 15 interceptors. To make up for this, MDA rolled over an additional 15 interceptor deliveries into fiscal year 2017, for a total delivery of 55 interceptors. In addition, MDA achieved its goal of delivering 44 ground-based interceptors by the end of calendar year 2017. However, some programs that achieved their milestones continued to employ high-risk approaches to acquisition, which we have recommended MDA reduce in previous reports. In addition, MDA maintains an ambitious schedule for key programs, such as for GMD’s Redesigned Kill Vehicle program. For more information regarding specific programs, see appendixes II through X. Other MDA elements missed asset delivery milestones. The Command, Control, Battle Management, and Communications (C2BMC) software spiral (or version) 8.2-1 was previously due to be delivered in October 2017, but was delayed again from its new date of December 2017 to second quarter of 2018. This spiral will play an important role in several tests of integrated capabilities, such as FTM-29, which was executed in January 2018. The Terminal High Altitude Area Defense (THAAD) program’s delivery of interceptor Lot 6 was scheduled to be delivered by the end of June 2017, but has since been delayed to the second quarter of 2018. THAAD officials stated this delay was due to a component production issue as well as the addition of 12 additional interceptors to the fiscal year 2017 procurement. Additionally, the Army and MDA have reached an impasse regarding the transfer of the THAAD program from MDA to the Army. MDA and the Army have been directed by the Deputy Secretary of Defense to develop a memorandum of agreement that would guide the transfer of the THAAD and AN/TPY-2 programs to the Army, and the National Defense Authorization Act for fiscal year 2018 requires the Secretary of Defense to transfer the acquisition authority of all missile defense programs that have received full-rate production authority, which includes THAAD, to the military departments not later than the date the President’s fiscal year 2021 budget is submitted. The Army, however, has identified a $10.1 billion requirements gap, and the Secretary of the Army issued a memo that he would non-concur with the transfer of the THAAD program in its current state. There is currently no plan or timeline to resolve the issue. We will continue to follow this issue in our future work. Finally, additional delays to the construction of the Aegis Ashore facility in Poland resulted in significant schedule compression, reducing the time allotted for installation and checkout activities from 16.5 months to 9.5 months. MDA initially maintained that the site would be delivered on schedule, but early in fiscal year 2018 the agency announced that the site would not be delivered until at least December 2019. Integrated BMDS Capability Increments: MDA also encountered challenges delivering packages of integrated capabilities, which it refers to as “increments.” Increment deliveries signify delivery of integrated BMDS-level capabilities, which are designed to significantly improve effectiveness and efficiency of the BMDS over its constituent elements working independently. MDA planned to deliver two increments in 2017, but both were delayed, and some constituent capabilities were removed and are planned to be delivered in future increments. For instance, MDA was late in delivering Increment 3, known as “Discrimination Improvements for Homeland Defense – Near Term.” We previously reported on schedule slips to this increment from its initial September 2016 delivery date to December 2016. However, program documentation indicates that MDA encountered further challenges in fiscal year 2017 that required an additional delay to March 2017. According to MDA officials, this most recent delay was driven by additional time needed to analyze testing results. However, we found that GMD had experienced development delays for some software upgrades leading up to assessment and integration activities. Moreover, MDA’s Increment 4, known as “Enhanced Homeland Defense,” was not completed in December 2017 as planned, because a C2BMC and a key GMD upgrade initially planned to support four BMDS-level capabilities intended for this increment would not be available until the second quarter of fiscal year 2018. MDA officials told us that they will rely on the current GMD software version, which lacks some key improvements, until this upgrade is delivered. Additionally, MDA significantly reduced the content of its BMDS cyberdefense capability planned for Increment 4. MDA documentation originally planned to deliver this capability with 10 elements and, prior to testing, the BMDS OTA declared four elements to be priorities. Of these four, MDA has conducted the assessment for only three. The remaining BMDS elements will deliver cyberdefense capabilities in future increment deliveries. MDA’s plans for delivery of future capabilities continue to be volatile. For example, plans for Increment 6 in fiscal year 2021, which will include delivering a new radar and kill vehicle for GMD, now require its capabilities to be broken up into three sub-increments delivered across several years, some as late as 2023, with multiple new capabilities added and several others deferred to Increment 7. Many of these delays continue to postpone achievement of BMDS integration, needed to improve performance against realistic attacks with multiple ballistic missiles. Most recently, MDA again delayed a capability designed to improve automated coordination between regional BMD shooters—that is, Aegis BMD, THAAD, and Patriot. While initially planned for delivery in 2015 with Increment 2, in fiscal year 2017, the capability was further delayed, from 2020 to 2023. In addition, a further integration capability that would centralize and automate command decisions across the BMDS will not be available until December 2025. See figure 3 for more information on how capabilities have been delayed within and across increments. Testing: MDA successfully completed most of its planned tests in fiscal year 2017 and achieved several notable milestones, though MDA continued to add, alter, delete, or delay parts of its test schedule throughout the year. Within the elements included in this report, MDA had nine tests in its fiscal year 2017 test plan, of which it conducted six as planned. MDA also added three additional tests to its plan over the course of the year. A summary of these tests can be found in table 3. Many of these tests are notable firsts for MDA, though others indicate continuing challenges. FTG-15 was a success, in which a Ground-Based Interceptor with a Configuration-2 booster and a CE-II Block I Exo-atmospheric Kill Vehicle intercepted for the first time an intercontinental ballistic missile with threat representative characteristics. In addition, this was the first use of the new booster avionics and upgrades to the software. The success of this test was necessary to deliver Increment 4’s requirements for Enhanced Homeland Defense. However, Department of Defense operational testing officials stated that the complexity and objectives of the test had been scaled back from what MDA originally planned. SFTM-01 was a success, in which an Aegis BMD SM-3 Block IIA missile intercepted a medium-range ballistic missile target. This was the first intercept test for the Aegis BMD SM-3 Block IIA. SFTM-02 was a failure, as the Aegis BMD SM-3 Block IIA interceptor failed to intercept its medium-range ballistic missile target. MDA officials stated that the interceptor acted “as designed” during the test, and the Navy is considering whether changes to its tactics, techniques, and procedures may be warranted. MDA officials maintained that this developmental test existed in part for risk- reduction ahead of fiscal year 2018’s FTM-29, in which the Aegis BMD SM-3 Block IIA would have to intercept an intermediate-range ballistic missile for the first time. Despite the failure, MDA has chosen not to reschedule and has instead re-assigned SFTM-02’s objectives to FTM-29. FTT-18 was a success, in which a THAAD battery intercepted an intermediate-range ballistic missile target. This test was originally planned for several years ago, as part of the 2015 delivery of Increment 2, and has been delayed in part due to issues with range availability. This is the first demonstration of THAAD against an intermediate-range ballistic missile target despite a THAAD battery having been delivered to Guam for this mission in 2013. FET-01, previously known as FTT-15, was a success, demonstrating THAAD’s ability to intercept a target in the endo-atmospheric stage of flight. MDA re-classified the test a “Flight Experiment” midway through fiscal year 2017 to reflect its more observational and experimental nature. The test objectives for FET-01 have changed several times, and while the final iteration of test objectives did not include intercept as an objective, an intercept against a medium-range ballistic missile target was achieved nonetheless. MDA’s Process for Managing the Delivery of BMDS-Level Capabilities Is Not Applied Consistently and Has Unclear Requirements When MDA declares a capability ready for delivery to warfighters, it communicates the capabilities and limitations of the delivered asset. This information is critical for allowing warfighters to make informed decisions about whether to accept the capability, how to prepare for its deployment, and how to plan for its use. Typically this process occurs through the Operational Capacity Baseline (OCB) change process, which is structured around the delivery of new capabilities to individual elements. Alternately, as noted above, when MDA declares a key integrated, BMDS-level capability ready for delivery, it does so through a process which culminates in the issuance of a Technical Capability Declaration (TCD). The TCD is a memorandum signed by the Director, MDA and is usually reserved for significant new capabilities such as: those mandated by the President; or delivery of integrated BMDS-level capabilities that require more than one element to function. This last category of capabilities is especially important as, according MDA’s charter, the BMDS is intended to be an integrated and interoperable system. Integration is important in order to counter the larger-scale and more complex attacks that are likely to occur during a conflict. We have reported since 2014 that MDA has taken steps to improve the management and reporting of integrated capabilities, and to increase the level of BMDS integration. While MDA has recently made some progress in the area of integrated capabilities, the majority of MDA’s capability deliveries continue to be made at the element level. Until recently, MDA has done little to document the requirements and process for issuing a TCD, resulting in an inconsistent and, at times, ad- hoc process. We found inconsistencies in MDA’s decisions regarding which integrated, BMDS-level capabilities MDA would deliver through a TCD, and which it would not. For example, since 2015, the agency planned to deliver 14 integrated, BMDS-level capabilities, but delivered only 7 through the TCD process. According to MDA’s prior capability delivery documents, several of these excluded capabilities were intended to be part of the formal TCD delivery during the planning stage, but were dropped at some later point. According to MDA officials, those deliveries were made when all their constituent elements were delivered via the OCB process. MDA officials acknowledged that distinctions between requirements for element-level deliveries and BMDS-level capabilities were not readily apparent in their policy and took steps in fiscal year 2017 to do so. MDA issued a memorandum on Technical Capability Declaration Planning and Definitions in June 2017 to help distinguish element-level OCB deliveries and deliveries of integrated BMDS capabilities that would occur via TCD. This document established several definitions and requirements such as assigning responsibilities, establishing lines of authority, and defining some requirements that are not found in the other guidance document that MDA uses to govern TCD. The June 2017 memorandum also identified which capabilities through 2023 that MDA will deliver via a TCD, and identified some ways to add a new capability to the list of those receiving a TCD. While MDA’s new policy represents a substantial improvement in the management of the TCD process, it does not address several important problems with the TCD process. Specifically, although MDA has identified capabilities that it plans to deliver using a TCD, it does not identify any criteria or reasoning that guided this determination. It also does not explain the criteria MDA will apply to future capabilities under consideration for a TCD, leaving open the possibility of the same inconsistent application MDA has used in the past. Moreover, the capabilities it identified for a TCD are only a subset of all planned integrated, BMDS-level capabilities. Consequently, only some integrated capabilities are currently planned to be delivered to the warfighter with comprehensive information about their performance and limitations at the BMDS level. Unless MDA requires that all integrated capabilities are delivered via the TCD process, as the BMDS becomes more integrated, military services and other decision makers will have reduced insight into the capabilities and limitations of the BMDS as a whole. MDA’s June 2017 policy also establishes some processes governing the requirements for, and development of, test plans in support of a TCD, but it does not address some of the most problematic aspects of this process to date. Specifically, the new policy requires convening an Assessment Requirements Review board to develop a baseline for a planned TCD, determine what capabilities will be included, and identify what test plans will be necessary to generate the “body of evidence” that MDA will provide in support of the TCD’s assertions regarding capabilities and limitations. However, we found that Assessment Requirements Reviews can be held shortly before the planned delivery date—which affords no opportunity to build the test plan around the requirements identified in the review. MDA held Assessment Requirement Reviews in preparation for two of the previous three TCDs. The timing of these reviews in relation to the date of the TCD’s issuance suggests that they had little influence on MDA’s actual test plans. MDA officials stated that an Assessment Requirement Review is ideally held 18 months to 2 years prior to the issuance of the related TCD. However, we found that, for recently issued TCDs, the reviews were held much closer to the beginning of testing and the planned TCD delivery. For example, for the TCD issued in December 2017 that delivered 44 ground-based interceptors, MDA held this review less than 8 weeks in advance. Figure 4 depicts the timeline of the Assessment Requirements Review as compared to the start of testing for the TCD and the TCD delivery date. Because these reviews identify requirements that must be tested, the Assessment Requirements Review would ideally inform MDA’s test plans so that each component of the integrated capability could be adequately tested by the planned delivery date. But because the policy does not give exact requirements, process, and key milestones necessary to issue a TCD, MDA is able to hold an Assessment Requirements Review that merely acknowledges the results of tests already completed. These practices are consistent with our prior findings on MDA, which identified a lack of a management framework for delivering integrated capabilities, and showed that the lack of this framework resulted in concurrency, fragmentation of development activities, and delays for some originally planned capabilities. According to DOD’s guidance on acquisition and testing, a program’s test and evaluation strategy should begin with a review of requirements so that management can devise a test and evaluation strategy that generates the knowledge necessary to inform the acquisition and operational decisions of a program. Holding the Assessment Requirement Review so close to the planned delivery date affords no opportunity to build the test plan around the requirements identified in the review, and instead only ratifies the results of a test plan that was not necessarily developed with these requirements in mind. MDA’s Use of Undefinitized Contract Actions Poses Cost and Schedule Risks to the Government Undefinitized contract actions are authorized when the negotiation of a definitive contract is not possible in sufficient time to meet the government’s requirements and government interests demand that the contractor be given a binding commitment so that contract performance can begin immediately, and are subject to certain limitations. Our analysis of MDA contracting from fiscal year 2013 to fiscal year 2017 shows that the combined not-to-exceed price of all undefinitized contract actions entered in a given year, and the average time it takes to definitize undefinitized contract actions, have increased. GAO has reported that while this type of contract action may be necessary under certain circumstances, it is considered risky in part because the government may incur unnecessary costs if requirements change before the contract is definitized. Though MDA reports on its contracting activities in its annual BMDS Accountability Report, its reporting on details unique to undefinitized contract actions is often inconsistent or even absent. MDA’s Acquisition Management Instruction 5013.02-INS states that undefinitized contract actions will be used only on “an extremely limited basis” and only when negotiating contract terms before the contractor begins work is not feasible, such as when delay “would adversely impact mission accomplishment.” Our prior work, as well as that of the DOD inspector general, has found that this type of contract action is considered risky in part because the government may incur unnecessary costs if requirements change before the contract is definitized. Under undefinitized contract actions, substantial funds may be obligated before essential questions of contract scope and system design have been settled. Over the past 5 years, the average length of the undefinitized period and not-to-exceed price for MDA’s undefinitized contract actions have increased. Since 2013, MDA has entered into 11 undefinitized contract actions as shown in table 4. MDA’s use of undefinitized contract actions has fluctuated between one and five instances per year. The combined not-to-exceed price of all such contract actions entered into each year has increased, however, from $2.5 million in fiscal year 2013 to $1.4 billion in fiscal year 2017 as shown in figure 5. The average time to definitize these contract actions has steadily increased as well, from 78 days in fiscal year 2013, to over 600 days in fiscal year 2016 (see figure 6). Two undefinitized contracts were awarded in fiscal year 2017 and both exceeded 180 days without definitization. The value of MDA’s undefinitized contract actions entered into in a given year, as measured by their combined not-to-exceed prices, has risen significantly. The length of the undefinitized period has also risen on average. Together, these figures show that MDA may be initiating contractor work with incomplete knowledge of the requirements or costs involved. With regard to the increasing duration of the undefinitized period, MDA contracting officials told us that when they do not achieve definitization within 180 days, it is often because the contractor’s proposal is not adequately supported by a sound estimate, and negotiation past 180 days is necessary to achieve a fair and reasonable price. They added that the task of making this determination is made more complicated by the highly developmental nature of the work that MDA often conducts. For example, the 2015 undefinitized contract action for Aegis BMD SM-3 Block IIA test interceptors remained undefinitized for 629 days. According to MDA officials, this delay was due in part to the difficulty of accurately estimating costs on a highly developmental project. MDA officials reported having to develop a substantial knowledge base and consult closely with other DOD entities that would have insight into the costs of similar projects, after the undefinitized contract action was entered into. Using an undefinitized contract action in this case, however, was not without risk to the government. MDA made major financial commitments to a program before it fully understood the requirements or the costs. To mitigate the risks related to these contract actions, MDA’s Instruction requires all undefinitized contract actions to be supported by a determination and findings that articulates the requirement to begin performance prior to a negotiated agreement, the not-to-exceed price and the definitization schedule. The DFARS and MDA instruction require all undefinitized contract actions to be approved by the Director, MDA. MDA officials told us that they interpret the MDA Instruction to require that the Director, MDA, sign determination and findings documents in support of undefinitized contract actions. In addition, MDA contracting officials stated that to further mitigate the risks related to undefinitized contract actions, they, as a matter of practice, strive to obligate only the minimum amount of funding necessary to achieve definitization. Officials indicated that doing so limits the cost risk for the government, and forces programs to think carefully about what work needs to be done prior to definitization and its likely costs. While the Director, MDA is required to sign the determination and findings document, in one instance, this document specifically authorized the program to amend the requirements and not-to-exceed price without further formal approval from the Director, MDA. This specific undefinitized contract action was the largest MDA has entered into since fiscal year 2013. MDA entered into the undefinitized contract action in May 2017, authorizing the design, development, and initial production of the GMD program’s Redesigned Kill Vehicle (RKV), with a not-to-exceed price of $1.088 billion. This undefinitized contract action will allow MDA to continue with the RKV program despite significant cost, schedule, and performance risks, some of which the determination and findings document for the RKV undefinitized contract action acknowledged. When MDA released its acquisition strategy for the RKV in 2015, it predicted the phase covered by this contract action would cost approximately $800 million, covering initial testing and production of up to eight RKVs for initial fielding. Officials stated that the current contract action, with a not-to-exceed price of $1.088 billion, is for only four interceptors, although since it is undefinitized, that is subject to change. If the RKV program definitizes this contract action according to its schedule in May 2018, after 12 months, this will result in the definitization of the contract action with less than a year remaining before the program’s critical design review. In other words, the government will have agreed on contract terms, including costs, after much of the design work and related costs have been incurred. As of February 2018, MDA reports obligating $324 million, or 30 percent of the not-to-exceed price, to this undefinitized contract action. This is in excess of the $244 million planned for the undefinitized period at the time of award. As part of MDA’s annual BMDS Accountability Report, MDA reports on its planned performance and schedule for the coming fiscal year across several baselines, one of which is dedicated to contracting performance. MDA provides these baselines in response to statutory requirements. By establishing these baselines and then reporting any deviations in cost, schedule, or performance as a program proceeds, the BMDS Accountability Report provides information for oversight by identifying areas of program risk and their causes to decision makers. Baselines also help ensure that the full financial commitment is considered before embarking on major development efforts. These reports contain some information on undefinitized contracts. However, the information is often inconsistently presented and distributed throughout the report. Information specific to undefinitized contract actions is often absent, such as the following: the definitization schedule (that is, the expected time frame for finalizing contract terms); the amount of funds obligated to the action for the undefinitized period; or any changes to the above that have occurred since award of the action. As a result, decision makers in Congress have limited insight into how MDA is handling the risks that come with undefinitized contract actions, or how the programs enacting these contracts are performing. For example, these reports do not typically disclose how much has been obligated under an undefinitized contract action, or if this amount has increased since the contract was awarded. They do not report if the not-to-exceed value has been revised, or if the current definitization schedule adheres to the schedule determined at the time of award. Despite Steps Taken to Improve BMDS Modeling Capabilities, Modeling Challenges Limit the Credibility and Accuracy of BMDS Performance Data Despite taking steps to improve the realism of the models it uses for ground testing, MDA continues to face challenges with its models. As a result, decision makers lack key information about BMDS performance, which could lead to miscalculations about how best to employ the BMDS and where to focus future capability development and investment. Specifically, MDA continues to encounter challenges with ensuring that its models and simulations are accredited for operational testing when they are used to test BMDS capabilities, resulting in uncertain performance outcomes in assessments supporting BMDS deliveries. Additionally, accreditation status and modeling limitations for these assessments are not communicated to most decision makers in Congress and some in the DOD and executive branch, limiting their insight into the data limitations underlying their decisions to make investments in and employ the BMDS. Finally, MDA’s assessment of the resources needed to validate and accredit its current models does not match requested funding for this effort. MDA Has Taken Steps to Improve Its Modeling Capabilities, but Most Delivered BMDS Capabilities Were Tested Using Unaccredited Models Since MDA cannot conduct enough system-level flight testing of the entire BMDS to completely assess BMDS performance, BMDS decision makers within MDA, DOD, Congress, and the executive branch use information from model-based ground tests to evaluate the operational effectiveness of the BMDS. The results from these model-based operational tests inform many acquisition and operational decisions, including: capability delivery, asset fielding, and interceptor inventory. Model-based testing also informs the warfighter’s tactics, techniques, and procedures to maximize BMDS effectiveness such as how many interceptors they will fire at a threat; and the capability gap analysis, the basis for warfighter requests for new capabilities. Recognizing the importance of models and simulations, MDA has taken steps to improve its ability to provide realistic modeled representations of the integrated BMDS necessary to assess operational performance. For instance: In 2009, MDA adjusted its test baseline, known as the Integrated Master Test Plan, and refocused its testing on collecting data needed for model development and accreditation. In 2016, MDA developed an update to a framework that is used to integrate the modeled representations of BMDS elements for assessments, and in 2017 continued an effort to develop digital end- to-end models and simulations to increase modeling capabilities and to expand the scope of BMDS assessments in the future. In 2017, MDA increased its collaboration with BMDS OTA to prioritize modeling needs and to address them. Despite these steps, MDA continues to deliver assets and capabilities using models that have not been accredited. In April 2016 and May 2017, we found that MDA had delivered EPAA Phase 2 capabilities in December 2015 using models that have not been accredited to support the delivery. MDA continued this practice by delivering two sets of BMDS-level capabilities since 2015, relying on operational tests conducted with models that were not accredited for use in such an assessment. The next delivery, expected at the end of the second quarter of fiscal year 2018, has also been tested using mostly unaccredited models. Relying on models that are not accredited for operational assessment increases the risk that modeling errors are not discovered, and a single undetected modeling error can distort the assessment results for the entire BMDS. DOD’s acquisition instruction requires that models and simulations used in operational assessments be verified, validated, and accredited. Although, as noted above, MDA is generally exempt from DOD acquisition policies, its own modeling and simulation policy requires that models and simulations used in operational assessments be verified, validated, and accredited for that use. Moreover, experts at DOD, MDA, and other institutions we interviewed agree that models should be verified, validated, and accredited to ensure that decisions based on models are informed by the correct data, and that the limitations of that data are understood. Additionally, according to DOD officials, defense acquisition programs that follow DOD acquisition regulations verify, validate, and accredit their models before operational assessments. However, our analysis indicates that the accreditation of many MDA models for operational assessment is, in most cases, not completed in time to support testing. In fact, many of them are not complete even after a capability has been delivered. Additionally, BMDS OTA officials said that models that are not accredited before delivery are not generally accredited later on. Figure 7 shows the percentage of accredited models that were used in the operational assessment of each BMDS capability delivery in 2015 through 2017. BMDS models are not accredited for operational assessment in large part for three reasons: (1) MDA does not provide sufficient evidence to the BMDS OTA for accreditation, (2) some models do not accurately represent BMDS performance in the real world, and (3) the threat model used to stimulate the test cannot be traced to the original intelligence community assessment. These challenges affect assessments across the entire BMDS engagement, from detection and processing of the threat to the intercept. While modeling uncertainty in any one of these areas affects uncertainty for the BMDS as a whole, factored together this uncertainty is magnified. Lack of Data: In some cases, MDA did not provide the BMDS OTA data needed to accredit the models used in operational ground testing, even though it is a signatory to the BMDS OTA’s accreditation plan. This plan identifies the data needed to achieve accreditation and directs that these data should be provided at least 60 days prior to official operational ground testing. MDA officials noted that the BMDS OTA recently changed its data requirements for accreditation and that they were unable to meet the new requirements in time to inform the capability deliveries shown above. However, we have found that MDA has encountered similar challenges since 2009. In fact, according to BMDS OTA officials, MDA has never completely provided the needed data on time and often missed numerous subsequent deadlines. In many cases, MDA failed to deliver the required data even after it tested and delivered its capabilities, and in some instances the data MDA provided did not meet the BMDS OTA’s requirements. As we have previously reported, disruptions to MDA’s testing program—such as flight test failures and delays—reduce the amount of real-world data that is available to accredit models. We also found that MDA proceeded with model-based ground tests and capability deliveries without leveraging the knowledge it planned to obtain from these tests. For example, in 2016 and 2017, we found that MDA delivered the European Phased Adaptive Approach Phase 2, even though key models, such as the model for Aegis Ashore, were unaccredited. Additionally, in other instances, MDA lacks technical data and other model information that is needed for accreditation, especially for models representing older systems. In 2017, as noted above, MDA and the BMDS OTA increased their collaboration to improve model accreditation status and, in 2017, co- developed a list of prioritized modeling deficiencies. Additionally, MDA is making progress in gathering and providing model data for operational assessment accreditation. MDA officials reported that based on this increasing collaboration, they expect that more models will be accredited in 2018. It is unlikely, however, that all models will achieve accreditation prior to the upcoming December 2018 delivery of the European Phased Adaptive Approach Phase 3. Modeling Deficiencies: Another reason that some models are not accredited for operational use is that certain models contain deficiencies, such as optimistic representations of BMDS performance and simplistic representations of BMDS environments. In these cases, while MDA initially supplied BMDS OTA with the relevant data, the model’s performance failed to meet the criteria for accreditation. Subsequently, MDA did not provide supporting rationale to explain these failures, or to explain how the modeling issues skewed the overall performance results. For example, in 2016, the BMDS OTA compared modeled sensor tracking data used in recent ground tests to real-world sensor tracking data and found that the models representing some radars performed better than the real-world radar. These modeling deficiencies can affect other BMDS elements that rely on sensor data and can artificially inflate BMDS performance. In one case, Aegis BMD’s launch-on-remote capabilities were over-estimated. As a result, the BMDS OTA could not accredit the models, and thus verify that ground test results that support Aegis’s launch-on-remote capability and other tested capabilities are credible and reliable. MDA is working to address this issue and it is too early to assess progress. Additionally, some models used in operational assessments are overly simplistic. For example, modeled representations of the battle scene in moments after intercept do not display the resulting complex scene that is caused by the large quantity of missile and interceptor debris. This deficiency limits insight into how the BMDS will perform during realistic ballistic missile attacks that could require follow-on interceptors to be launched, and how the BMDS will determine that the incoming threats have been destroyed. According to BMDS OTA and MDA officials, MDA’s efforts to develop digital models can help in this area, by providing more processing power and great scalability for engagement complexity; however, the capability is not expected to be mature until 2021 or later. Threat Models Cannot Be Traced Back to Underlying Threat Assessments: The value of ground test-generated data is dependent on the quality of the threat model that stimulates the test. However, the BMDS OTA has never been able to accredit threat models before operational testing, and in some cases, after testing. As is the case with other models, in some cases, the BMDS OTA does not receive data needed to accredit the models from MDA in a timely manner. Additionally, the BMDS OTA cannot trace the threat model used in ground testing to the threat model that MDA developed based on the intelligence community’s threat assessment. For example, according to BMDS OTA officials, during a past ground test event, a model representing a BMDS element rejected the intended threat model and instead ran its own internal threat model. As a result, the test did not reflect real world conditions where the entire BMDS would be exposed to the same threat stimulus. BMDS OTA officials said that MDA’s ground test architecture is not designed to generate the data needed to confirm that all elements are reacting to the same model during ground testing, meaning that unbeknownst to testers, other BMDS elements could also reject the approved threat model during testing. These deficiencies introduce ambiguity into the test results including the extent to which the BMDS operated as an integrated system of systems against a common threat set. BMDS OTA officials said that MDA is currently working on a pathfinder activity to help understand and rectify the traceability issue. Information about the Accreditation Status and Limitations of Models Used in Operational Assessments Is Not Communicated to Decision Makers Although the warfighter and other decision makers inside DOD, Congress, and the executive branch rely on models to provide information about BMDS effectiveness, MDA’s capability delivery documentation does not include information about the quality of modeling data. Specifically, MDA’s TCD memos and OCB change packages, which describe technical capabilities delivered to the warfighter and their limitations, do not discuss the extent to which the models used to assess the new capability are verified, validated, and accredited for assessment, or how ground test results were affected by model limitations. As a result, decision makers do not have complete information about the validity of the capability assertions in these documents and how much confidence should be placed in reported BMDS performance. According to Standards for Internal Control in the Federal Government, decision makers need access to reliable and timely information to make operational decisions. Additionally, according to DOT&E guidance, in cases where models and simulations cannot be validated and accredited, any modeling results should be caveated with a clear explanation of which areas of performance assessment could be affected by the lack of accreditation. Lack of such information could lead to miscalculations about how best to employ the BMDS or uninformed decisions about where to focus future capability development and investment. While the BMDS OTA has recently begun to brief some combatant commands on how modeling limitations impact the warfighters’ understanding of delivered capabilities, these briefings are not readily available to other stakeholders and decision makers, such as cognizant congressional committees or others in DOD and the executive branch. In its report accompanying a bill for the National Defense Authorization Act for Fiscal Year 2017, the House Armed Services Committee requested that MDA brief the House and Senate Armed Services Committees on the accreditation status of models used in testing indicating that congressional decision makers benefit from such information. Including information about model accreditation and limitations in TCD and OCB packages would ensure decision makers in DOD, Congress, and the executive branch have the same necessary information to inform their decisions. Funding Decisions May Delay Some Modeling Capability Development Moving forward, the Director, MDA will have to make difficult decisions on balancing funding priorities, including the need to adequately fund the validation and accreditation of models. MDA has started to make progress validating and accrediting existing models using DOT&E and OTA recommended criteria. However, MDA’s assessment of the resources needed to validate and accredit its current models and simulations does not match funding levels it requested for this effort. MDA determined that it needs an additional $99 million for fiscal years 2017- 2022 to accredit BMDS models and simulations. MDA requested $395.7 million from 2017-2022 to meet modeling and simulation needs. Figure 8 shows MDA’s fiscal year 2018 funding request for model development and the additional funding, over the 5 year period, that would be required to verify, validate, and accredit its models. Additionally, funding is not requested for the verification, validation, and accreditation of some models used in BMDS assessments because MDA officials said that they do not have written agreements with the military services that operate these elements defining funding and technical requirements for this purpose. Specifically, while the Army and the Air Force develop and accredit models to support their missions for the Patriot, the Space-based Infrared System, and the BMDS communication systems, these models have to be modified to accurately represent their BMDS roles for BMDS operational assessments. While MDA does fund the development of the Space-based Infrared System and BMDS communication models for use in BMDS assessment, it does not fund the verification, validation, and accreditation of these models or the Patriot model. Additionally, MDA officials report that it currently has no written agreements with the Army or the Air Force to define funding and technical requirements for these models for BMDS assessment. Because these requirements are not formally agreed upon and communicated between MDA and the Services, the verification, validation, and accreditation of these models is often unfunded, further complicating MDA’s and the BMDS OTA’s verification, validation, and accreditation analyses. Standards for Internal Control in the Federal Government states that organizations should assign responsibility and delegate authority to achieve their objectives. Additionally, in our prior work we found that all acquisitions efforts should have well defined roles and responsibilities for all stakeholders. Although MDA and the BMDS OTA were able to accredit the Space-based Infrared System and BMDS communications models in 2017, future upgrades to these BMDS elements will require verification, validation, and accreditation to ensure that they continue to accurately reflect the real-world system. Moreover, DOD and Congress have instructed the transfer of missile defense programs that have received full-rate production authority, which would include THAAD and Aegis BMD, to the military services for operations, which may increase the scope of this issue. Even though these systems will no longer be under MDA management, they will still be part of the BMDS and, like the Space-based Infrared System and Patriot, will require model updates to reflect changes to the tactical systems. However, as noted above, there are currently no agreements between MDA and the services to fund these modeling requirements, increasing the risk that model upgrades will not be implemented, thus preventing their verification, validation and accreditation for operational testing. If MDA and the services do not agree to the technical and funding requirements for models of elements used in BMDS testing but operated by the services before the elements are transferred, disagreements will likely continue to impede the verification, validation, and accreditation of those models, decreasing confidence in test results and understanding of how the real-world BMDS will operate. Conclusions MDA continues to make mixed progress in delivering assets and integrated capabilities. Moreover, its processes for communicating the extent and limitations of these capabilities can be improved. While MDA met several significant milestones in fiscal year 2017, MDA failed to deliver either of its two most recent packages of integrated capabilities on time, and its plans for future capabilities, even in the near term, continue to be characterized by a high degree of fluidity. MDA has recently taken steps to document in policy its processes for communicating the extent and limitations of deliveries of integrated capabilities. However, these policies still do not clearly specify the exact requirements, process, and key milestones needed to complete some capability deliveries. Moreover, they do not require that all integrated BMDS capabilities are delivered using a process that describes their performance and limitations at the level of the BMDS, rather than at the element level, increasing the risk of delivered capabilities not being communicated properly to their end users: the warfighter. In addition, while no contracting strategy can be completely risk-free, trends in the not-to-exceed prices and duration of MDA’s undefinitized contract actions indicate a troubling pattern. Making major commitments to large developmental programs before important questions of scope and price have been determined exposes the government to increasing amounts of risk. MDA’s inconsistent and incomplete reporting on its use of undefinitized contract actions makes it even more difficult for Congress and decision makers to exercise oversight and track these risks. Finally, deficiencies and limitations in the models used to support operational testing of the BMDS, including the lack of accreditation, provides decision makers with some flawed information about BMDS performance. Because flight tests cannot provide complete information on BMDS performance, it is important that ground tests can be relied upon to provide accurate and representative data. This flawed information could lead to miscalculations about how best to employ the BMDS and uninformed decisions about where to focus future capability development and investment. If steps are not taken to improve BMDS models and to communicate their status and limitations clearly to decision makers, there is a risk that the BMDS will not perform as expected when needed to defend the United States at home, its regional allies, and deployed forces. Recommendations for Executive Action We are making the following six recommendations to the Under Secretary of Defense for Research and Engineering: The Under Secretary of Defense for Research and Engineering should ensure that the Director, MDA, takes the following actions: The Director, MDA should revise MDA policies to require that all integrated capabilities—capabilities that require integration of two or more elements—be included in a Technical Capability Declaration. (Recommendation 1) The Director, MDA should clarify, in written policy, the exact requirements, process, and key milestones necessary to issue a Technical Capability Declaration, including a requirement that the Assessment Requirements Review be held in such a time frame that it can provide meaningful input to MDA’s test plans. (Recommendation 2) The Director, MDA should include information on current undefinitized contract actions in the BMDS Accountability Report, including the not-to- exceed price, the definitization schedule, the amount of funds obligated for the undefinitized period, and any changes since the contract action was entered into. (Recommendation 3) The Director, MDA should ensure that models used for operational tests are validated and accredited for such assessments. To help achieve this, MDA should provide the BMDS Operational Test Agency all evidence previously agreed to and needed to accredit models before ground testing events, as specified in the BMDS OTA accreditation plan. (Recommendation 4) The Director, MDA should include in capability delivery packages, such as the Technical Capability Declaration memos and Operational Capability Baseline change packages, the following: a. The verification, validation, and accreditation status of the models used in operational ground tests; and b. Modeling and simulation limitations that affect operational ground test results. (Recommendation 5) The Director, MDA and the Secretaries of the Armed Services responsible for operating BMDS elements should develop written agreements as soon as feasible for modeling and simulations technical and funding requirements for any BMDS elements that are service- operated but represented in BMDS performance assessments. (Recommendation 6) Agency Comments and Our Evaluation DOD provided written comments on a draft of this report. DOD’s comments are reprinted in Appendix I and summarized below. DOD and MDA also provided technical comments which were incorporated as appropriate. In its response, DOD concurred with five out of six of our recommendations, and partially concurred with one. In addition, DOD recommends the closure of five recommendations. However, we believe that it is premature to close out four of the five recommendations until all of its planned actions are fully implemented. For the remaining recommendation, we agree with DOD and will undertake the steps to close out the recommendation. DOD partially concurred with our first recommendation to revise MDA policy to require all integrated capabilities—capabilities requiring the integration of two or more elements— be declared and delivered via the Technical Capability Declaration (TCD) process. While DOD agreed with the intent of this recommendation, DOD stated that the Director, MDA will determine which major integrated capabilities should be delivered via the TCD process. The Department also noted that the agency developed a list of such capabilities that it will update annually. These actions are an improvement over the current process, but they do not meet the full intent of our recommendation. Specifically, the list of future TCDs that MDA produced is not inclusive of all future integrated capabilities. In addition, MDA’s policy does not articulate definitive standards for identifying capabilities requiring a TCD and leaves this decision to the discretion of the Director, MDA. As we’ve identified in this report, some capabilities have been deleted from or added to planned TCDs without explanation. The new policy leaves open the possibility of continued inconsistent application of the TCD process. This poses the risk that not all integrated capabilities will be delivered to warfighters with comprehensive information about their performance and limitations at the BMDS level. We continue to believe that in order for the agency to meet the full intent of our recommendation, it should establish in policy a clear, definitive standard for which capabilities require a TCD for delivery. In addition, DOD recommends the closure of the first two recommendations to (1) revise MDA’s policies to require that all integrated capabilities be included in a TCD; and (2) clarify the exact requirements, process, and key milestones necessary to issue a TCD as it contends that its new Policy Memorandum 90 meets the intent of our recommendation. This memorandum is dated March 28, 2018 and was provided to us on May 8, 2018. As such, we have not had an opportunity to fully assess the memorandum and the process laid out in it. However, as noted above, this new Policy Memorandum 90 leaves open the possibility of continued inconsistent application of the TCD process. This poses the risk that not all integrated capabilities will be delivered to warfighters with comprehensive information about their performance and limitations at the BMDS level. In order for the agency to meet the full intent of our recommendation, MDA should establish in policy a clear, definitive standard for which capabilities require a TCD for delivery. In addition, DOD writes that the same Policy Memorandum 90 satisfies the second recommendation to clarify the exact requirements, process, and key milestones necessary to issue a TCD. We believe it necessary to wait until MDA delivers a TCD in accordance with the new parameters set out in the memorandum before this recommendation can be closed. For the third recommendation to include information on current undefinitized contract actions in the BMDS Accountability Report, DOD states that the BMDS Accountability Report for 2018, approved by the Director, MDA on March 9, 2018 provides the information necessary for closure. We concur with this assessment will take the necessary steps to close this recommendation. In responding to our fourth recommendation requiring the Director, MDA to ensure that models used for operational tests are validated and accredited for such assessments, DOD states that MDA is actively working with the BMDS Operational Test Agency (BMDS OTA) to resolve any issues associated with, and the reporting of, modeling limitations. However, as we found in this report, according to BMDS OTA officials, MDA has never completely provided the needed data on time and often missed numerous subsequent deadlines to support the validation and verification of its models from BMDS OTA. Consequently, we believe it is premature to close out the fourth recommendation, but we will continue to track MDA’s progress and timeliness in providing the evidence previously agreed to and needed to accredit models before ground testing events. In responding to our fifth recommendation to include the verification, validation and accreditation status used in operational ground tests in capability delivery packages, such as TCDs and Operational Capability Baseline change packets, DOD states that MDA has made significant progress over the last year in achieving the BMDS OTA accreditation of MDA’s models and simulations. In addition, it states that the addition of MDA policy will ensure the verification, validation and accreditation status of each model will be discussed and assessed by the Operational Capability Baseline Working Group for each capability delivery package. We agree that MDA has made significant progress over the last year, however, we believe it premature to close out the recommendation until BMDS OTA can ensure that the status of the models used, as stated in our recommendation, are included in subsequent capability delivery packages such as the Technical Capability Declaration memos and Operational Capability Baseline change packages. We are sending copies of this report to the appropriate congressional committees, the Secretary of Defense, the Undersecretary of Defense for Research and Engineering, and to the Director, MDA. In addition, the report is available at no charge on the GAO website at http://www.gao.gov. If you or your staff have any questions about this report, please contact me at (202) 512-4841 or chaplainc@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix XI. Appendix I: Comments from the Department of Defense Appendix II: Aegis Ballistic Missile Defense (BMD) Weapons System Program Overview Aegis Ballistic Missile Defense is the naval component of the Missile Defense Agency’s (MDA) Ballistic Missile Defense System. It consists of the Aegis Ballistic Missile Defense Weapon System (AWS), including a radar and Standard Missile-3 (SM-3) interceptors. MDA is developing the Aegis BMD weapons system in versions called spirals that expand on preceding capabilities. Deliveries of the spirals are planned to support MDA’s capabilities for Regional and Homeland defense. Specifically, MDA delivered Aegis BMD 5.0 Capability Upgrade (5.0CU) in fiscal year 2016 for the European Phase Adaptive Approach (EPAA) Phase 2, but had not verified its full capability before delivery. In fiscal year 2017, the program delivered Aegis BMD 4.1 on ships with legacy hardware in order to provide similar ballistic missile defense capabilities to those of Aegis BMD 5.0 CU. MDA plans to deliver additional upgrades for such ships in 2019 and 2023. Additionally, the program is developing Aegis BMD 5.1 with upgrades for EPAA Phase 3, planned for December 2018. The Aegis BMD program also plans to deliver additional upgrades in 2023, called Aegis BMD 6.0, capitalizing on Navy’s upgrades to the Aegis radar. For specifics on Aegis Ashore and the Aegis BMD SM-3 interceptors, see appendixes III, IV and V, respectively. Table 5 provides key fiscal year 2017 AWS program facts. Aegis BMD resolved some prior challenges and delivered capabilities initially planned for delivery with European Phased Adaptive Approach Phase 2 MDA resolved software challenges and testing delays for Aegis BMD 5.0 CU and delivered Aegis BMD 4.1, expanding the number of ships with EPAA Phase 2 missile defense capabilities. While MDA delivered initial Aegis BMD capabilities for EPAA Phase 2 with AWS 4.0.2 prior to the December 2015 Technical Capability Delivery (TCD), planned capabilities would not be available until the subsequent versions—Aegis BMD 5.0CU and 4.1—completed development and fielding. However, both encountered technical challenges and schedule slips, as well as testing delays. In fiscal year 2017, MDA continued work on Aegis BMD 5.0 CU and 4.1 and overcame some of these challenges. Specifically: Aegis BMD 5.0 CU: MDA completed Aegis BMD 5.0 CU certification late in fiscal year 2017, resolving prior technical challenges and testing delays. Specifically, MDA implemented fixes to significant defects that were in the software at the time of initial delivery. Additionally, in December 2016 and August 2017, MDA flight tested fleet and ship self-defense capability against medium-range ballistic missiles in terminal phase of flight –a capability initially planned for December 2015. Aegis BMD 4.1: MDA also delivered Aegis BMD 4.1 in August 2017, after multiple schedule slips. While initially scheduled for delivery in support of the EPAA Phase 2 TCD, the spiral was first delayed to the middle of fiscal year 2016 due to technical and cost challenges. Subsequently, activities for Aegis BMD 4.1 were suspended in 2016 to reassess the program and delivery was delayed to September 2019, to align it with a related Navy effort. In fiscal year 2017, MDA resumed activities for Aegis BMD 4.1, and certified the delivery of ballistic missile defense capabilities in August 2017. These ballistic missile defense capabilities are currently being integrated with the Navy’s larger Aegis combat system, into a single computer program called Aegis Baseline 5.4, which is still scheduled for delivery in September 2019. MDA mitigated key Aegis BMD Weapons System challenges for EPAA Phase 3, but they will not be verified until 2018 According to MDA’s program management documentation, Aegis BMD 5.1 is on track for delivery in support of EPAA Phase 3 in December 2018, as the program overcame or reduced key risks. For example, despite a lack of schedule margin, the program met a key software development milestone in June 2017, and delivered it for system-level ground tests, which will assess integrated BMDS capabilities for EPAA Phase 3. It also met all objectives in a fiscal year 2017 flight test. Additionally, the program reduced the ongoing programmatic risk to Aegis BMD 5.1 that could affect its interoperability with other elements. However, testing to demonstrate the risk has been resolved is not yet complete. According to the Aegis BMD program management documentation, upgrades to the Aegis communication management system, which are managed by the Navy, lag behind MDA’s Aegis BMD 5.1 development schedule. The lag in development could result in integration challenges between these upgrades, and could impact Aegis integration with other BMD elements, including the capability to intercept threats entirely on tracks from forward based radars – called Engage on Remote. In fiscal year 2017, MDA and the Navy took steps to mitigate this risk. However, MDA has yet to demonstrate the fixes in a flight test. Moreover, MDA documentation indicates that if issues are discovered, they could impact the Aegis BMD 5.1 mission and could result in interoperability restrictions against Aegis BMD 5.1. Lastly, Engage on Remote could also be affected if development challenges with C2BMC, which forwards threat track data from forward based sensors to Aegis BMD, are not mitigated. For more information on the C2BMC element, see Appendix VI. Aegis BMD is developing additional capabilities for deployment in 2023 and beyond, leveraging Navy’s Aegis upgrades In fiscal year 2017, MDA continued to develop Aegis BMD capabilities that are expected to be deployed in 2023. Specifically, MDA continued developing and maturing capabilities for an effort it started at the end of fiscal year 2016 called Aegis BMD 6.0. Aegis BMD 6.0 is planned to provide capabilities against more threat types, larger raids, better discrimination, and improved communication with its interceptors. Additionally, it takes advantage of the Navy’s effort to replace the Aegis SPY-1 radar with a more capable SPY-6, and to overhaul the entire Aegis combat system. While it is early in development, MDA has begun identifying knowledge gaps that could diminish planned capabilities and took initial steps to address disconnects between Navy’s effort and its own. According to program management documentation, MDA plans to develop an Aegis BMD 6.0 acquisition baseline late in fiscal year 2018. The acquisition baseline is expected to detail Aegis BMD 6.0 planned capabilities, its schedule, and cost. MDA is also planning additional upgrades to Aegis BMD 4.1, called Aegis BMD 4.2. Specifically, MDA plans to collaborate with the Navy to integrate and field refurbished and upgraded SPY-1 Antennas onto legacy ships. This modification improves radar sensitivity resulting in improved tracking capabilities and is planned for delivery in fiscal year 2023. MDA plans to begin developing and maturing technologies for this upgrade in fiscal year 2019, and baseline the effort at the end of fiscal year 2020. Appendix III: Aegis Ashore Program Overview Aegis Ashore is a land-based, or ashore, version of the ship-based Aegis Ballistic Missile Defense (BMD). Aegis Ashore is designed to track and intercept ballistic missiles in the middle of their flight using Aegis BMD Standard Missile-3 (SM-3) interceptors. Key components include a vertical launching system, interceptors, and an enclosure, called a deckhouse, that contains the SPY-1 radar and command and control system. DOD deployed an Aegis Ashore test facility in Hawaii in April 2014. The test facility has been used to flight test Aegis Ashore, and in some cases, Aegis BMD SM-3 interceptors. MDA deployed its first operational site in Romania in fiscal year 2016 as part of the European Phased Adaptive Approach (EPAA) Phase 2, and is currently constructing a second site in Poland for delivery in 2018 as part of EPAA Phase 3. Both operational sites are intended to provide additional coverage for the defense of Europe. Aegis Ashore will share many components with the sea-based Aegis BMD and will use future versions of the Aegis weapon system currently in development, including the Aegis BMD SM-3 Block IIA interceptor. The Missile Defense Agency (MDA) plans to equip Aegis Ashore with a modified version of the Aegis weapon system software that will share many components with the sea-based Aegis BMD. For further details on the Aegis Weapon System and Aegis BMD interceptors, see appendixes II, IV, and V. Table 6 provides key fiscal year 2017 Aegis Ashore program facts. The Aegis Ashore facility in Poland became increasingly reliant on concurrency to meet its schedule, but construction issues eventually forced a delay of at least one year Construction of the Aegis Ashore site in Poland has not overcome an initial delay that was largely due to contractor performance issues. MDA and the Army Corps of Engineers, which manages military construction at the site, took a number of measures to mitigate or reverse these delays, including working to modify the Army Corps of Engineers’ contract to permit joint occupancy of the site for a longer duration, and for the contractor to provide more granular project data to the Army Corps of Engineers. Also, the contractor has moved key personnel on site, and added a second shift. Program officials stated that they have withheld some award fees from the contractor over these issues. Program documentation states the contractor continues to be late in submitting documentation needed to move forward. If this and other issues are not corrected, it will increase the risk of further schedule slips. To make up for these delays, MDA introduced increasing levels of concurrency into its schedule, and shortened key phases of the delivery process. MDA has reduced the time allotted for Installation and Checkout activities from 16.5 months to 6.5 months. These activities would occur concurrently with the final phases of construction at the site. For example, installation of the deckhouse at the Poland site was scheduled for the end of the fourth quarter, fiscal year 2017, but was delayed to the end of the first quarter, fiscal year 2018. Despite this, installation and checkout activities still began in the fourth quarter of fiscal year 2017. The Navy’s systems testing procedures, which are needed prior to operational acceptance of the site, will have occurred mostly concurrently with the final stages of MDA’s construction and installation work on the site. MDA maintained through all of fiscal year 2017 that the site would be ready for delivery in December 2018 as scheduled. Program documentation noted that further program concurrency presented risks not only to the Aegis Ashore program, but to multiple elements relying on timely delivery of the site, up to and including the scheduled EPAA Phase 3 declaration. Early in fiscal year 2018, MDA announced that construction of the Poland site would not be completed until at least December 2019. The Poland and Romania sites both experienced unforeseen program challenges Both Aegis Ashore sites in Europe have faced continuing challenges in several areas. For example, attrition problems have complicated efforts to keep the Poland site’s construction on schedule. These problems led to several persistent vacancies in important positions during the fiscal year. At one point in fiscal year 2017, the program lacked a full-time onsite program manager or dedicated government safety engineer, as well as other important positions. These roles had been, up to that point, filled by deputies in an acting capacity or were divided among others. MDA officials have also pointed to morale problems at the Poland site, where conditions for sailors are relatively austere. The Romania site has required more post-delivery support from MDA than was originally planned, largely due to quality and design issues in several areas. This post-construction wrap-up work was accounted for in MDA’s plans, but was originally planned to be complete by late fiscal year 2016. However, MDA has continued to provide warranty-like support in areas such as water supply, seismic-activity certification, and compatible electrical systems. Program officials stated that many of these issues arose from having to adapt Aegis systems to Romanian infrastructure, which in some cases proved to be a more complicated task than expected. Despite the issues encountered, the Romania site has remained operational throughout all of this work. Appendix IV: Aegis Ballistic Missile Defense (BMD) Standard Missile-3 (SM-3) Block IB Program Overview The Aegis BMD Standard Missile-3 (SM-3) Block IB is a ship- and shore- based missile defense system interceptor designed to intercept short- to intermediate-range ballistic missiles during the middle stage of their flight. The Aegis BMD SM-3 interceptor has multiple versions in development or production: the SM-3 Blocks IA, IB, and IIA. Compared to the Aegis BMD SM-3 Block IA, the Block IB features an enhanced target seeker for increased discrimination, an advanced signal processor for engagement coordination, an improved throttleable divert and attitude control system for adjusting its course, and increased range. The Aegis BMD SM-3 Block IB interceptor is linked with the Aegis Ballistic Missile Defense (BMD) Weapons System and Aegis Ashore. For additional information about the Aegis Weapon Systems, see Appendix II and for Aegis Ashore, see appendix III. Recent technical and production problems have continually delayed a decision to authorize full production of the Aegis BMD SM-3 Block IB due to reliability concerns. Since fiscal year 2015, Aegis BMD SM-3 Block IB production has been delayed by several technical issues. In response to a GAO recommendation, program officials in 2015 delayed the decision to enter full-rate production until they could implement further testing and design changes. In fiscal year 2016, two failures during testing forced a suspension of interceptor deliveries, causing the program to miss its delivery target for the year. Table 7 provides key fiscal year 2017 Aegis BMD SM-3 Block IB program facts. The Aegis BMD SM-3 Block IB program made progress against its delivery backlog from the previous year, and mitigated some technical risks, though others remain The Aegis BMD SM-3 Block IB experienced two failures in fiscal year 2016, the investigation of which forced a temporary suspension of interceptor deliveries. As a result, MDA delivered only 33 interceptors out of a planned 47 for the year. MDA added the remaining interceptors to its planned delivery for fiscal year 2017, resulting in a target of 54 interceptors. The program successfully delivered 55 interceptors over the course of the year, and thus made up for the existing backlog. The program tracked two technical risks during fiscal year 2017, one of which it succeeded in removing, and another which will not be implemented into the production process until the third quarter of fiscal year 2018. According to MDA officials, the program successfully managed the transition of the production of the system’s Throttleable Divert and Attitude Control System to a new facility without experiencing significant delays or quality issues. In the other case, a component that was implicated in a previous test failure is currently undergoing a redesign. Program officials stated that they plan to have the new design certified by the second quarter of fiscal year 2018, and incorporated into the production line by the end of the third quarter. As we reported last year, problems testing a redesigned third-stage rocket motor on the Aegis BMD SM-3 Block IB forced the program to postpone its planned full production decision until the second quarter of fiscal year 2017, and successive delays have ensued. Though the tests validating the redesign were successful, the Undersecretary of Defense for Acquisition, Technology, and Logistics issued an Acquisition Decision Memorandum in February 2017 requesting an additional flight test in fiscal year 2017 as well as supporting analyses from the Director, Operational Test and Evaluation and the office of Cost Assessment and Program Evaluation. The memorandum issued these requirements in support of a planned full production decision in the first quarter of fiscal year 2018. Full-rate production for the Aegis BMD SM-3 Block IB was initially scheduled for fourth quarter, fiscal year 2012. MDA had one Aegis BMD SM-3 IB flight test scheduled for fiscal year 2017 at that time (FTM-24), and added another (FTM-26) in response to the Acquisition Decision Memorandum’s requirement, but neither were held as scheduled. MDA delayed FTM-24 to fiscal year 2020, in order to first analyze the new target missile’s performance to ensure it would work within the parameters of the test. While FTM-24’s delay was due to its very specific test design, its timing in fiscal year 2017 would have afforded additional information about the reliability of the interceptor that will not now be available before the full production decision. MDA deleted FTM-26 several months after adding it, and moved its objectives to coincide with NATO’s Formidable Shield – 17 naval exercises which took place in early fiscal year 2018 (wherein the system did achieve a successful intercept). As a result of the delay in conducting a test for production-readiness, the program is currently planning on a production decision in second quarter, fiscal year 2018. Appendix V: Aegis Ballistic Missile Defense (BMD) Standard Missile-3 (SM-3) Block IIA Program Overview The Aegis Ballistic Missile Defense (BMD) Standard Missile-3 (SM-3) interceptor has multiple versions in development or production: the Aegis BMD SM-3 Blocks IA, IB, and IIA. The latest version, the Aegis BMD SM- 3 Block IIA interceptor, provides increased speed and range, more sensitive seeker technology, and an advanced kinetic warhead than previous versions. It is expected to defend against short-, medium-, and intermediate-range ballistic missiles. Additionally, most of the Aegis BMD SM-3 Block IIA components will differ from other the prior versions, and therefore requires new technology to be developed specifically for it. For additional information on the Aegis BMD SM-3 Block IB interceptor, see appendix IV. Initiated in 2006 as a cooperative development program with Japan, the Aegis BMD SM-3 Block IIA program was added as a capability to support the European Phased Adaptive Approach (EPAA) Phase 3 architecture to defend against longer-range threats. The Aegis BMD SM-3 Block IIA interceptor is planned to be fielded with Aegis Weapons System 5.1. For additional information on Aegis Weapons System, see appendix II. Table 8 provides key fiscal year 2017 Aegis BMD SM-3 Block IIA program facts. The Aegis BMD SM-3 Block IIA program has experienced mixed results in testing performance and problems with program execution, with consequences for cost and schedule The first intercept flight test using the Aegis BMD SM-3 Block IIA interceptor, SFTM-01, was conducted in February 2017. It was originally scheduled for fiscal year 2016, but was delayed to evaluate technical issues discovered during previous tests. During this test, the Aegis BMD SM-3 Block IIA interceptor successfully engaged a medium-range ballistic missile (MRBM) target. The next intercept flight test, SFTM-02, occurred 4 months later, in June 2017. However, the interceptor failed to reach its MRBM target during this test. MDA convened a failure review board to identify the cause of the failure, and concluded that the failure was not attributable to a fault in the design or performance of the interceptor itself. The Navy is currently considering changes to its tactics, techniques, and procedures to address the findings from the failure review board. Two prior non-intercept tests using the Aegis BMD SM-3 Block IIA interceptor, although considered successful, showed potential design issues with the missile’s guidance system, which steers the interceptor to the target. Consequently, the program decided to develop a replacement component. The redesigned component passed initial acceptance testing and the program plans to employ it during FTM-29, which is scheduled for the second quarter of fiscal year 2018. The program continues to experience unit cost growth due to several factors, including decreases in the total amount being procured and increases in shipping costs. According to MDA officials, shipping costs grew because MDA underestimated the cost to ship missile components manufactured in Japan to the US on US-flagged ships. MDA officials stated that they did not adequately account for these costs when establishing the original baseline cost. Since 2014 the program’s unit cost has increased by almost 60 percent, from $24 million in fiscal year 2014 to $39 million in fiscal year 2017. Program officials stated that they do not expect either of these issues to lead to further cost growth in the future. Further delays or technical challenges within the Aegis BMD SM-3 Block IIA program could impact the EPAA Phase 3 declaration The Aegis BMD SM-3 Block IIA program has limited schedule margin to address any issues prior to operational testing to meet the EPAA Phase 3 declaration by the first quarter of fiscal year 2019. For the EPAA Phase 3 declaration, the Aegis BMD SM-3 Block IIA interceptor must demonstrate the ability to intercept an intermediate-range ballistic missile (IRBM) target using remote sensor data. The program has one flight test, FTM-29, prior to its operational flight test. This test was originally scheduled for the first quarter of fiscal year 2018, but was delayed to the second quarter, and the launch site for the test was moved to the land-based Aegis Ashore facility in Hawaii. Adapting the Aegis BMD SM-3 Block IIA interceptor for a land-based test delayed this test further, from the first quarter to the second quarter of fiscal year 2018. Despite these delays, the dates for the operational test of the Aegis BMD SM-3 Block IIA—FTO-03 E1—and the EPAA Phase 3 declaration remain unchanged: the third quarter of fiscal year 2018 and first quarter of fiscal year 2019, respectively. That leaves the program approximately 3 to 5 months to resolve any issues discovered during FTM-29, prior to the operational test, which is needed to support the EPAA Phase 3 declaration. In addition, FTM-29 will be the Aegis BMD SM-3 Block IIA interceptor’s first test against an IRBM, first test of its ability to engage a target using remote sensor data, and the first test with to incorporate the new missile guidance system component incorporated. As a result of the complex test environment and limited time between tests, any significant failure during FTM-29 could lead to a delay in the EPAA Phase 3 declaration. Appendix VI: Command, Control, Battle Management, and Communications (C2BMC) Appendix VI: Command, Control, Battle Management, and Communications (C2BMC) Program Overview C2BMC is a global system of hardware—workstations, servers, and network equipment—and software that integrates all missile defense elements of the Ballistic Missile Defense System (BMDS). Specifically, it allows users to plan operations, see the battle develop, and manage BMDS sensors. As the integrator, C2BMC enables the defense of a larger area than the individual BMDS elements operating independently and against more missiles simultaneously, thereby potentially conserving interceptor inventory. MDA is developing C2BMC in spirals, or software and hardware upgrades designed to improve various aspects of the integrated BMDS performance. MDA fielded Spiral 6.4 in 2011 and plans to complete the fielding of Spiral 8.2-1 by March 2018. The program is working on efforts for additional capabilities in the future. Table 9 provides an overview of C2BMC spiral development and table 10 provides key fiscal year 2017 C2BMC program facts. C2BMC Spiral 6.4 supported delivery of discrimination upgrades but cyber vulnerabilities continue to place the BMDS at risk At the beginning of 2017, MDA completed the Spiral 6.4 assessment, which was designed to enable capabilities for Increment 3, Near Term Discrimination Improvements for Homeland Defense. The spiral performed nominally during testing, providing discrimination tasking from a forward-positioned radar for long-range threats, multiple-radar discrimination tasking of a threat, and several fixes related to sequencing and timing of messages. These tests provided performance data, which informed MDA’s Technical Capability Delivery for Increment 3 in March 2017. Despite this success, however, the spiral continues to have cyber vulnerabilities that place the BMDS operations in certain geographic areas at risk. For example, Spiral 6.4 has been in use since 2011, and its operating system (Windows XP) as well as other supporting software products will remain in the field well past their end of life cycle and support by vendors. According to program documentation, upgrading these systems before they are replaced by subsequent spirals is cost prohibitive, but program documentation does not indicate the cost. While MDA is in the process of fielding Spiral 8.2-1 to replace Spiral 6.4 in the Strategic, Northern and Pacific Commands by March 2018, Spiral 6.4 will remain operational at the European and Central Commands until the delivery of Spiral 8.2-3 in early fiscal year 2019. According to fiscal year 2017 MDA program reviews, the likelihood that critical cyber vulnerabilities are discovered is low for the remaining two years, and, according to MDA, no fielded system has been exploited to date. However, known vulnerabilities have been exploited in lab experiments. Moreover, MDA program documentation from fiscal year 2017 acknowledges that new security deficiencies could still be discovered, and if those or known deficiencies are exploited, mission capabilities like BMD planning, radar control, track reporting, and situational awareness may be significantly degraded. MDA collaborated with Combatant Commands to monitor and minimize the risks. MDA Continued its Development of C2BMC Spiral 8.2-1 and expects its fielding in Fiscal Year 2018 In fiscal year 2017, MDA mitigated developmental risks necessary to complete the development and testing of C2BMC Spiral 8.2-1 in fiscal year 2018. Spiral 8.2-1—planned to support Enhanced Homeland Defense capabilities—was initially planned for delivery by December 2017, but, according to MDA officials, the delivery was delayed to allow additional time for assessment of results from BMD system-level ground test campaign called Ground Test (GT)-07a. Prior to GT-07a, the program identified risks that could affect interoperability with other elements and threat tracking, but, according to recent program documentation, MDA implemented fixes to many of them before the testing began. At the time of our assessment, MDA’s analysis was ongoing. However, MDA plans to complete its fielding by March 2018. Spiral 8.2-3 continues to face technical challenges and cost increases MDA has begun testing Spiral 8.2-3, which is planned for fielding throughout fiscal year 2019, but it continues to face technical challenges and cost risk. This spiral is to replace Spiral 8.2-1 at the Strategic, Northern and Pacific Command, and Spiral 6.4 at European and Central Commands. According to MDA, the spiral is designed to enable a five-fold increase in the size of area that can be defended by the BMDS, and is an integral part of EPAA Phase 3. However, the program continues to track a prior risk and identified a new risk to an element level C2BMC capability needed for EPAA Phase 3 called Engage on Remote. Specifically, program documentation indicates that processing of data about threat missile flight paths, known as threat tracks, has issues that could reduce the likelihood of the successful engagements utilizing Aegis Ballistic Missile Defense in Engage on Remote scenarios. C2BMC has faced similar challenges with threat tracking capabilities for prior spirals, which required delays certain aspects of integration with Aegis BMD until fixes were implemented. MDA is implementing fixes to these issues in Spiral 8.2-3, which once fielded should resolve these integration issues, but it still needs to assess them in the ongoing test campaign for EPAA Phase 3. Since 2016, MDA Spiral 8.2-3 costs have increased by about 20 percent, from $68 million to $82 million. According to MDA documentation, the increase is due to several factors, including higher than expected costs for architecture and system engineering, as well as testing and integration requirements, and additional requirements for cybersecurity, which increased algorithm complexity required for Engage on Remote. MDA officials stated that some of the cost increases for cybersecurity were driven by DOD-wide cyber requirements, implemented in March 2014. Further cost increases, according to MDA, were driven by a warfighter request for geographic redundancy. While the original concept for 8.2-3 had the suites for Central and European Command at each location, MDA met the warfighter request by installing the suites at different locations so that losing a single node would not result in the loss of all capability for the warfighter. According to the C2BMC program, implementation of this requirement cost about $6.4 million. MDA identified requirements for Spiral 8.2-5, but it is already facing potential technical, as well as schedule and cost challenges MDA identified element requirements for Spiral 8.2-5, which is planned for delivery in fiscal year 2021. This Spiral will integrate the Long Range Discriminating Radar and provide additional BMDS-level planning, track processing, and battle management capabilities. While MDA currently plans to hold the Preliminary Design Review by March 2018 and may report its acquisition baseline for the first time in the subsequent BMDS Accountability Report, program management documentation has already identified two specific challenges: Program documentation indicates that the Northern Command has concerns about performance issues associated with threat track processing, called System Track, for GMD engagements. While this is a key C2BMC function, track processing has been a challenge for other spirals supporting prior and upcoming regional and homeland defense capabilities. MDA is currently working with stakeholders to address this issue. The program also identified disconnects between LRDR, GMD and C2BMC, which are driving up element development and test costs, and delayed some capabilities initially planned to be delivered along with the LRDR. MDA developed a mitigation plan and established a working group to coordinate with stakeholders to address these issues. Appendix VII: Ground-based Midcourse Defense (GMD) Appendix VII: Ground-based Midcourse Defense (GMD) Program Overview The GMD system is a missile defense interceptor system designed to defend the United States against a limited intermediate and intercontinental ballistic missile attack from countries such as North Korea and Iran. To counter such threats to the homeland, GMD, in conjunction with a network of ground-, sea-, and space-based sensors, launches interceptors from missile fields based in Fort Greely, Alaska and Vandenberg Air Force Base, California. After launching from in-ground silos, the interceptor boosts towards the incoming enemy missile and releases an Exoatmospheric Kill Vehicle (EKV) to find and destroy the threat. GMD also has ground support and fire control systems that the warfighter relies upon to operate the system. Since the program’s initiation in 1996, DOD has spent over $45 billion developing, operating, and maintaining the GMD system, including: fielding ground station assets and a fleet of 44 interceptors; upgrading, redesigning, refurbishing, and retrofitting the system; successfully performing 5 out of 9 intercept tests and 3 out of 3 non-intercept tests; and developing Multi Object Kill Vehicle technology. Three of the intercept tests failed because of problems with the EKV while one of the tests failed because of a target failure, which is not associated with the GMD system. MDA has efforts ongoing to address concerns with the existing fleet of interceptors and increase protection to the U.S. homeland. In March 2013, the Secretary of Defense directed MDA to increase the number of fielded GMD interceptors from 30 to 44 by the end of 2017. To achieve this fielding goal, MDA performed a limited redesign of the CE-II, called the CE-II Block I, to fix known issues, address obsolescence, and improve producibility and cost. MDA also performed an extensive upgrade to the boost vehicle to improve reliability and address obsolescence issues. Although the CE-II Block I will address some concerns with the CE-II design, MDA determined a more complete redesign of the CE-II was needed. MDA subsequently developed an acquisition strategy and began developing the new kill vehicle, called the Redesigned Kill Vehicle (RKV). The RKV is intended to be more reliable, producible, testable, and cost effective. Table 11 provides key fiscal year 2017 GMD program facts. Fiscal Year 2017 was one of GMD’s most successful years for results achieved Fiscal year 2017 was a seminal year for the GMD program, as it achieved a number of major accomplishments. Over the past several years, the GMD program developed the newest interceptor version equipped with the CE-II Block I EKV and C2 boost vehicle. The program conducted its first successful flight test of this interceptor in May 2017 when it successfully intercepted a target representative of an intercontinental ballistic missile—another first for the GMD system. MDA proceeded to produce and field eight of these new interceptors and complete the refurbishment of Missile Field 1 in Fort Greely, Alaska, enabling the program to meet its directive from the Secretary of Defense to field 44 interceptors by the end of 2017. The program also fielded a software upgrade to the fire control segment of the GMD ground station, which included some improvements for battle management and discrimination. In addition, the program completed a preliminary design review for the RKV in March 2017. The program was able to execute all of these activities while also maintaining 24/7 availability of the system to the warfighter during a heightened period of North Korean missile testing. GMD’s cost now exceeds $67 Billion, the fourth highest among DOD’s weapon systems In total, the GMD program’s total cost has increased to over $67 billion and that total is likely to continue to increase as MDA defines future capability increments. In March 2013, we reported the total cost of the GMD program was estimated to be approximately $41 billion. Since that time, MDA defined new capability increments that included major GMD initiatives, such as the RKV and Multi Object Kill Vehicle efforts, which increased the program’s total cost. GMD is now the fourth most expensive DOD weapon system among a portfolio of 78 major defense acquisition programs, totaling approximately $1.5 trillion. As seen in table 12 below, only the F-35 and two naval programs are projected to cost more than the GMD system, demonstrating the department’s level of resources committed to defending the U.S. homeland against a long range ballistic missile attack. In November 2017, the President submitted to Congress an amendment to the fiscal year 2018 budget request for DOD to, among other things, increase current missile defense capacity, expand the sensor network, and accelerate technology development efforts. According to MDA, the request for additional funds was in direct response to recent demonstrations of advanced and accelerated capabilities by North Korea. MDA’s justification materials for the budget amendment includes an addition $774 million for GMD to build a new, 20-silo missile field at Fort Greely, begin procuring four additional interceptors, continue booster development, accelerate RKV development, and to add a non-intercept target to an initial RKV flight test. In total, MDA now plans to spend over $14 billion on GMD over the next six years with 64 total interceptors fielded by 2023. New Director, MDA revised the GMD acquisition strategy to keep the current prime contractor in place, reversing plans for MDA to lead system integration The new direction of the GMD program reflects a decision by the Director, MDA to set aside a strategy approved in 2016 by the prior Director for the government to take on the role of system integrator. Since the late 1990s, Boeing has been the GMD prime contractor, performing the role of system integrator. In 2011, Boeing competitively won a follow-on GMD development and sustainment contract that runs through December 2018. According to MDA, the government serving as the system integrator provides several benefits, such as eliminating organizational conflicts of interest issues—where industry tests and delivers assets based on requirements it wrote—and providing an unbiased assessment of system performance. However, a subsequent review team identified gaps and risks with implementing the strategy and the agency determined that transitioning to the new strategy at a time of heightened threat activity created unacceptable levels of risk for defending the U.S. homeland. On January 31, 2018, MDA awarded a sole-source contract modification to Boeing to extend the current development and sustainment contract. The contract modification has a total value of $6.56 billion and includes the accelerated delivery of a new 20-silo missile field, development of a new boost vehicle and the RKV, procurement of 20 new interceptors, and ground system upgrades. MDA faced a difficult choice, as both options included advantages and disadvantages. Under the prior approved strategy, MDA expected to achieve cost savings through competition. According to MDA, the sole- sourced labor rates for new development efforts under the recently modified development and sustainment contract have proven to be significantly higher than originally planned. In addition, MDA stated that the contract modification process is also often very lengthy, making it difficult for the agency to respond to the rapidly changing threat environment. MDA also stated that the lack of competition makes it challenging for the government to achieve favorable contract outcomes. Conversely, the government taking on the role of system integrator would make it responsible for managing multiple contracts. MDA plans to implement measures to mitigate some of the current challenges with extending the development and sustainment contract and ultimately provide the GMD program with a level of continuity during the current period of heightened threat activity. MDA’s plan to accelerate the Redesigned Kill Vehicle effort may instead prolong it In October 2017, MDA informed the Under Secretary of Defense for Acquisition, Technology and Logistics (USD(AT&L)) that it had revised the RKV acquisition plan that was previously established in 2015 and approved by the USD(AT&L). This revision, in response to the advancement of the North Korean missile threat, accelerates the RKV’s development by concurrently performing development and production and reducing the number of necessary flight tests. MDA removed the previously-established alignment between flight tests and production decisions, which enables the program to begin production well before the system’s design is stabilized. In addition, MDA now plans to contract for production, on a sole source basis with the current GMD prime contractor, rather than through a full and open competition. According to MDA, the acceleration plan does not change the content of the RKV’s development plan and the program will continue to execute the same engineering processes including hardware qualifications essential to delivering the RKV. However, MDA’s revision of the RKV acquisition plan is more likely to prolong the effort rather than accelerate it. Our prior best practice work has shown that finding a balance between resources available (i.e., time and funding) and needed operational attributes (i.e., reliability and effectiveness) and obtaining buy-in from across the department is essential for program success. Although some risk may be necessary, programs that rely on heightened levels of concurrent development and production, starting production before stabilizing the design, and other risky practices greatly increase the likelihood a program will fail to deliver reliable, effective capabilities in an accelerated manner. The revised RKV plan no longer includes some of the key best practices, such as alignment between testing and production decisions included in the 2015 RKV plan. In addition, MDA has already experienced development delays and was operating on the threshold schedule of the prior acquisition plan, with no additional margin for delays. Moreover, MDA did not vet the revised plan in a similar manner to that of the 2015 RKV acquisition plan, which Congress required to be subject to approval by the USD(AT&L) and include rigorous elements for systems engineering, design, integration, development, testing and evaluation. The revised plan is also inconsistent with the acquisition best practice to “fly before you buy”, as MDA will begin production based on the results of design reviews rather than flight testing. In May 2017, we recommended that the Secretary of Defense require the Director of DOD’s Office of Cost Assessment and Program Evaluation (CAPE) perform a comprehensive review of the RKV acquisition strategy and provide any recommendations to the Secretary of Defense that the Director deems necessary and appropriate to obtain CAPE’s concurrence for the RKV program’s acquisition strategy. DOD did not concur with our recommendation, stating that CAPE and other organizations had previously reviewed the strategy prior to USD(AT&L)’s approval. As we noted in our report, CAPE raised serious concerns about the plan and expected MDA would encounter development delays. MDA justified the prior RKV plan, in part, so that it could begin urgently replacing the less reliable CE-Is as expeditiously as possible, which were fielded between 2004 and 2007. Under the newly accelerated plan, MDA does not plan to begin replacing the CE-I interceptors until after it has fielded the additional 20 RKV-equipped GBIs in 2024. However, GBIs only have an initial service life of 20 years and MDA previously decided not to make any upgrades to the CE-I because of initial plans to begin replacing them with RKVs in 2020. We continue to believe that DOD should implement our recommendation in order to ensure that MDA’s plans for the RKV are viable and meet the needs of the warfighter. Appendix VIII: Sensors A family of satellite-, sea-, and land-based radars provides worldwide sensor coverage to enable the Ballistic Missile Defense System (BMDS) to effectively detect and track threat missiles through all phases of their trajectory. Land-based BMDS sensors include the Army/Navy Transportable Radar Surveillance and Control Mode-2 (AN/TPY-2), Upgraded Early Warning Radars (UEWR), and the future Long Range Discrimination Radar (LRDR). Figure 16 below illustrates the locations of select BMDS sensors world-wide. AN/TPY-2 is a transportable X-band high resolution radar that is capable of tracking all classes of ballistic missiles. AN/TPY-2 in the forward-based mode is capable of detecting and tracking missiles in all stages of flight to support Aegis BMD and GMD engagements and provides threat missile data to C2BMC. AN/TPY-2 in the terminal mode can track missiles in the later stages of flight to support THAAD engagements. Five AN/TPY-2 radars for use in forward-based mode are deployed to support regional defense: two in U.S. European Command, two in U.S. Pacific Command, and one in U.S. Central Command. Two AN/TPY-2 radars for use in terminal mode is also deployed to U.S. Pacific Command. UEWR are U.S. Air Force early warning radars that are upgraded and integrated into the BMDS to provide sensor coverage for critical early warning, tracking, object classification, and cueing data. Upgraded Early Warning Radars are located in Beale, California; Fylingdales, United Kingdom; and Thule, Greenland. MDA awarded a contract to upgrade the early warning radars in Clear, Alaska and at Cape Cod, Massachusetts, and both of these assets are approaching their operational acceptance for use in the BMDS. The upgrades to the Clear and Cape Cod Early Warning Radar sites are joint MDA / Air Force projects. Both organizations are contributing funding to these sites. LRDR is being designed as an S-band radar intended to address the need for persistent, precision tracking and discrimination capability in the Pacific sensor architecture. MDA anticipates the addition of LRDR will optimize the employment of the Ground-based Midcourse Defense (GMD) interceptors and address evolving threats. The radar will be located at Clear Air Force Station, Alaska with initial operational capability planned for 2020. Table 13 provides key fiscal year 2017 Sensors program facts. AN/TPY-2 Program transitions to a new development phase To address future requirements and as part of its spiral development process, AN/TPY-2 transitioned from its Increment 2 software development phase to its Configuration 3 software development phase. The transition results in Configuration 3 subsuming all unfinished Increment 2 content including 44 percent of development costs ($60 million), 31 percent of productions costs ($61 million), 88 percent of operations and support costs ($2,281 million), and 100 percent of disposal costs ($30 million). Four Knowledge Points and Technical Performance Metrics for the program were also carried over from Increment 2. New capabilities were also added in Configuration 3 including electronic protection and discrimination improvements. Additionally, the Conditional Materiel Release of software upgrade CX 2.1.0 was delayed from the first quarter of fiscal year 2017 to the first quarter of fiscal year 2018. To mitigate this delay, MDA executed an Urgent Software Release for CX 2.1.1 to support the fielding of Command, Control and Battle Management (C2BMC) S6.4-3 in December 2016. UEWR operational acceptances delayed for Beale, Clear, Cape Cod, and Fylingdales Sites The UEWR is executing a concurrent development approach to improve UEWR Object Classification (OC), Data Processor/Signal Processor (DP/SP), and Bias Correction capabilities, and to certify the UEWR Clear and Cape Cod sites for use in the BMDS. Because of this concurrent development, a delay in the Beale UEWR’s operational acceptance for the OC and DP/SP program has had cascading effects on the same upgrades for the Clear, Cape Cod, and Fylingdales UEWRs in addition to the BMDS Certification for the Clear UEWR, delaying the use of these key radar capabilities. These delays are shown in figure 17 below: The delay in Beale’s Operational Acceptance was due in part to the following: The contractor, Raytheon, delivered unacceptable UEWR technical orders that required rework. Development and operational testing supporting the operational acceptance were delayed because the operators required remediation of all emergency operational maintenance issues found on the operational UEWRs. Some UEWR software required fixes to address deficiencies. Other programs were competing for test time on needed equipment. The delay in operational acceptance will affect the delivery of Bias Correction for the Clear and Cape Cod UEWRs in addition to the delivery of and Data Processor/ Signal Processor improvements to support the missile defense mission of Beale, Clear, Cape Cod, and Fylingdales UEWRs. It has also delayed the BMDS Certification of the Clear UEWR. Because the program currently has sufficient schedule margin before the Cape Cod BMDS Certification, the delays have not yet affected the missile defense mission for that radar. The program office is working with Raytheon on a recovery plan to address the Technical Order issues and other issues that arose from the developmental and operational testing conducted in July 2017. We have previously reported that concurrent development increases program risk for cost and schedule delays caused by redesigns and retrofits needed after testing has occurred. In fiscal year 2017, MDA made progress towards stabilizing LRDR’s design, by completing a preliminary design review in March 2017 and a critical design review in September 2017. The program also began production of long lead radar electronic components and awarded a military construction contract for the Mission Control Facility. However, the program has experienced challenges integrating multiple facilities- related projects, which require synchronization between MDA, the U.S. Air Force, the U.S. Army Corps of Engineers, and contractors. For example, in fiscal year 2017, the LRDR program began demolishing a decommissioned, Cold War-era radar, which sits on the planned LRDR site at Clear Air Force Station. The program discovered that the radar’s foundation and surrounding soil contained steel and concrete coated with polychlorinated biphenyl (PCB), which was a common industrial material used at the time of the radar’s construction in the late 1950s. PCBs do not readily break down once in the environment and have been demonstrated to cause a variety of adverse health effects. In April 2017, the U.S. Army Corps of Engineers modified its contract for the removal of the PCB- contaminated foundation and soil and plans to complete excavation and removal by early fiscal year 2018. Demolition is now expected to be completed in 2019 and the additional costs for these complications are not covered under the program’s resource baseline. In June 2017, the LRDR program initiated a power study with the commercial power supplier for the LRDR radar. The program expects to complete the study on LRDR’s power demands on the commercial electrical grid, as well as assess updated U.S. Northern Command concept of operations to determine the extent, if any, of system capabilities, limitations, and mitigations. During the LRDR critical design review, MDA officials stated that U.S. Air Force informed the agency that it required 24/7 availability of the radar if it is to become operational. According to MDA officials, the current LRDR design, with its reliance on commercial power and limited back-up generators, would not provide that capability. MDA officials stated the program plans to increase the number of back-up generators, which may increase the military construction costs and annual operational expense of the radar. A November 2016 study of LRDR’s power system performed for MDA by a contractor indicated that agreements with the commercial power provider place limitations on the warfighter’s ability to operate the radar without consulting the commercial power provider in advance and that emergency activation of the radar could result in other commercial power provider customers having their power supply temporarily switched off if the generators were not brought online in time. The LRDR program has also encountered design challenges with the radar’s circuit card assemblies, as the planned design included the use of pure tin parts, which are susceptible to corrosion. Lockheed Martin, the prime contractor for the LRDR program, plans to replace some of the pure tin parts with parts that have a lead-based finish, as available. The program does not anticipate there to be enough of these parts available and estimated that redesigning the pure tin parts would result in an approximate 9-month delay. For those parts that cannot be readily replaced, Lockheed Martin plans to use corrosion mitigation techniques, such as applying conformal coating to the circuit card assemblies and applying lead solder. Although MDA maintains that these mitigation techniques will ensure corrosion-free operations, government and industry studies show that such mitigations reduce, but do not eliminate the risk, Lockheed Martin is conducting on-site inspections and providing additional information on the historical use of pure tin parts in similar systems and anticipates being able to clear the unmitigated, pure tin parts through the MDA’s Parts, Materials, and Processes control board. Appendix IX: Targets and Countermeasures Program Overview The Missile Defense Agency’s (MDA) Targets and Countermeasures procures missiles to serve as targets during the developmental and operational testing of independent or integrated ballistic missile defense system (BMDS) elements. Specifically, this program supplies MDA with short-, medium-, intermediate-, and intercontinental-range targets to test, verify, and validate the BMDS elements’ performance in threat relevant environments. As targets are solely test assets, they are not operationally fielded. The number of targets that the program supplies vary based on each element’s requirements and testing schedule. While some targets have been used for years, others have been recently added or are now being developed to more closely represent current and future threats. The quality and availability of these targets is instrumental to the execution of MDA’s flight test schedule. Table 14 provides key fiscal year 2017 Targets and Countermeasure program facts. First successful intercept test using an ICBM was achieved during GMD flight test, FTG-15 Despite challenges MDA has previously experienced using new targets during intercept flight tests, in fiscal year 2017, the program successfully flew the first intercontinental ballistic missile (ICBM) range target to support a critical intercept test for the GMD element. The GMD element provides the warfighter capability to engage and destroy intermediate- and intercontinental-range ballistic missile threats for the protection of the United States. In March 2013, the Secretary of Defense announced plans to increase the number of deployed GMD interceptors called Ground- based interceptors (GBI) from 30 to 44 by the end of 2017. To do this, a test—FTG-15—was needed to collect data on the GBI’s new booster design and demonstrate its performance against a target at the ICBM threat range before completing this mandated fielding goal. The successful flight of the ICBM target, the GBI’s performance against the target, and other information gathered during this test will provide the warfighter with a better understanding of the GBI’s capabilities and limitations. For further details on the GMD element, see appendix VII. Program planning to award contract for additional targets despite cost growth, schedule delays, and unproven performance The Targets and Countermeasures program is planning to contract for up to 12 additional medium range ballistic missile (MRBM) T1/T2 targets despite cost growth, schedule delays, and the lack of demonstrated performance. In fiscal year 2014, the program competitively awarded the initial contract for 6 MRBM T1/T2 targets with an option for an additional 12, for a total of 18. According to program officials, the contract was structured with a fixed price for the target and incentives to ensure successful execution during testing. However, the contractor has been underperforming since the award. First, this target’s costs have continued to significantly increase as some MDA officials originally warned. One of MDA’s reasons for selecting the current MRBM T1/T2 contractor was because it offered a lower price. However, some officials within MDA objected to this award due to the near certainty that the contractor would overrun costs. Since then, both MDA and Defense Contract Management Agency (DCMA) officials have acknowledged that the contractor did not adequately account for the costs associated with this target. Consequently, this target’s costs have been volatile, and despite changes and rebaselines, the contractor has been unable to meet projections. In fiscal year 2017, the program conducted another review to address significant cost growth and set new projections, and despite a relatively steady period of performance against these new projections, DCMA officials believe that this contractor will continue to have increasing costs. In addition, the first delivery of this target has been delayed almost five years beyond the original plan primarily due to contractor performance issues. There was an initial delay because the contract was awarded later than planned due to an investigation of an unsubstantiated procurement integrity allegation. However, since then, contractor performance issues have further delayed the first target delivery, necessitating several substitute targets for tests in the interim. Finally, since the program will not fly the first target in a test until the second quarter of fiscal year 2019, the target’s performance has yet to be demonstrated. Hence, buying an additional 12 targets without confirmation of the target’s performance is a significant risk for the program, as even one failure would delay all future tests with this target, and ultimately, the entire test program. Appendix X: Terminal High Altitude Area Defense (THAAD) Appendix X: Terminal High Altitude Area Defense (THAAD) Program Overview THAAD is a rapidly-deployable ground-based system able to defend against short-, medium-, and intermediate- range ballistic missile attacks during the middle and end stages of a missile’s flight. THAAD is organized as a battery that consists of interceptors, launchers, an Army Navy / Transportable Surveillance (AN/TPY-2) radar, a fire control and communications system, and other support equipment. The first two batteries were originally conditionally accepted by the Army for operational use. Since then, THAAD received urgent materiel release approval from the Commanding General of the United States Army Aviation and Missile Command to enable an earlier delivery of equipment for THAAD batteries one through six for operational use to meet the Army’s request to support urgent warfighter needs. The MDA plans to continue THAAD production through fiscal year 2024, for a total of 7 batteries, 503 interceptors, and 7 radars. MDA has two THAAD acquisition efforts—THAAD 1.0 and THAAD 2.0. THAAD 1.0 is for the production of the batteries, interceptors, and supporting hardware and provides the warfighter with initial integrated defense against short- and medium-range threats in one region. Appendix X: Terminal High Altitude Area Defense (THAAD) THAAD 2.0 is primarily software enhancements that expand THAAD’s ability to defend against threats in multiple regions and at different ranges, and adds debris mitigation and other upgrades. Table 15 provides key fiscal year 2017 THAAD program facts. THAAD Successfully Completed Two Flight Tests, but Delayed Several Hardware and Software Deliveries, Impacting Warfighter Capabilities THAAD successfully completed two tests. In FTT-18 (previously scheduled for fiscal year 2015), THAAD successfully intercepted an Intermediate Range Ballistic Missile (IRBM)-representative target, demonstrating THAAD’s capability against IRBM threats. THAAD has been deployed to Guam since 2013 to defend against IRBM threats, but this is the first time it has demonstrated that capability in a flight test. According to program officials, for the second planned flight test originally named FTT-15, MDA changed the name to Flight Experiment THAAD (FET)-01 to more accurately reflect the experimental purpose of the test. However, an intercept was formerly a primary test objective in FTT-15, but this objective was removed before the test name was changed to FET-01. In FET-01, although not a primary objective, THAAD did complete an intercept of a medium-range ballistic missile target with countermeasures. Despite the intercept, the test revealed significant operational limitations. THAAD delayed the delivery of several key hardware and software deliveries that will impact warfighter capabilities. Figure 20 shows the delayed hardware and software deliveries. Appendix X: Terminal High Altitude Area Defense (THAAD) Appendix X: Terminal High Altitude Area Defense (THAAD) interceptors due to a production issue that had cascading schedule effects on interceptor production and delivery. Parts Quality Issues Were Resolved but Delayed Interceptor Deliveries In May 2017, we found that THAAD interceptor production was halted due to a parts quality issue discovered when a connector in the interceptor failed multiple testing iterations. Upon investigation, the contractor learned that one of its sub-contractors changed the manufacturing process on the connector without informing Lockheed Martin. According to program officials, Lockheed Martin halted interceptor delivery but continued interceptor production. The connector was redesigned and incorporated into 20 interceptors, which again failed testing before being deployed. After a second redesign the connector passed testing and interceptor delivery resumed in April 2017. As of December 2017, there were 16 interceptors that had been produced but not yet fitted with the redesigned connector. Program officials report that the delay should result in about 2 months of delivery delays of the last interceptor lot currently under contract. According to program officials, to prevent similar problems from occurring again, the government revised the Parts Materials and Processes Control plan to provide improved guidance and clarity related to parts selection and change control; added additional criteria to annual audits to enhance review of supplier parts management, materials, and processes; and tightened controls on suppliers to report any significant changes. MDA and Army are at an Impasse Regarding Transfer of THAAD Program to the Army Appendix X: Terminal High Altitude Area Defense (THAAD) THAAD, that would be the Army) not later than the date the President’s fiscal year 2021 budget is submitted. However, in a memo from the Secretary of the Army, the Army said that it would non-concur with a transfer of the THAAD program in its current state because it cannot meet the Army’s global mission requirements. To meet global mission requirements for the THAAD mission, the Army requires about $10.1 billion of additional hardware, life-cycle sustainment funding, and AN/TPY-2 upgrades. MDA is willing to transfer to the Army the THAAD program of record as is. An official from the Army said that this impasse has existed before, but that the recent reprioritization of the THAAD mission contributed to it. For further details on the AN/TPY-2 program, see appendix VIII. Table 16 below shows the difference between the THAAD program of record and the Army’s requirements. Appendix X: Terminal High Altitude Area Defense (THAAD) MDA’s requirements and warfighter requirements exist and can lead to situations such as this impasse. Consequently, we recommended that the Secretary of Defense require MDA to develop a plan to transition operational requirements analysis currently performed within MDA’s Achievable Capabilities List to the U.S. Combatant Commanders, with U.S. Strategic Command as the lead entity and, in the interim, require MDA to obtain their concurrence of the Achievable Capabilities List prior to its release. The Department of Defense (DOD) did not agree with our recommendation. However, as evidenced by the discrepancy between the Army’s and MDA’s requirements for the THAAD and AN/TPY-2 program, the difference between MDA’s requirements and those of the warfighter will continue to present substantial problems to DOD in executing the missile defense mission, and we continue to believe that our recommendation should be implemented. Appendix XI: GAO Contact and Staff Acknowledgments GAO Contact Staff Acknowledgments In addition to the contact named above, LaTonya Miller, Assistant Director; Matthew Ambrose; Kristine Hassinger; Helena Johnson; Joe Kirschbaum; Wiktor Niewiadomski; Steven Stern; Brian Tittle; Hai V. Tran; Alyssa Weir; Tonya Woodbury; and Robin Wilson made key contributions to this report. Related GAO Products Missile Defense: Some Progress Delivering Capabilities, but Challenges with Testing Transparency and Requirements Development Need to Be Addressed. GAO-17-381. Washington, D.C.: May 2017. Missile Defense: Opportunities Exist to Reduce Acquisition Risk and Improve Reporting on System Capabilities. Washington, D.C.: Mar. 2015. Missile Defense: Mixed Progress in Achieving Acquisition Goals and Improving Accountability. GAO-14-351. Washington, D.C.: Apr. 2014. Missile Defense: Opportunity to Refocus on Strengthening Acquisition Management. GAO-13-432. Washington, D.C.: Apr. 2013. Missile Defense: Opportunity Exists to Strengthen Acquisitions by Reducing Concurrency. GAO-12-486. Washington, D.C.: Apr. 2012. Missile Defense: Actions Needed to Improve Transparency and Accountability. GAO-11-372. Washington, D.C.: Mar. 2011. Defense Acquisitions: Missile Defense Transition Provides Opportunity to Strengthen Acquisition Approach. GAO-10-311. Washington, D.C.: Feb. 2010 Defense Acquisitions: Production and Fielding of Missile Defense Components Continue with Less Testing and Validation Than Planned. GAO-09-338. Washington, D.C.: Mar. 2009 Defense Acquisitions: Progress Made in Fielding Missile Defense, but Program is Short of Meeting Goals. GAO-08-448. Washington, D.C.: Mar. 2008 Defense Acquisitions: Missile Defense Acquisitions Strategy Generates Results but Delivers Less at a Higher Cost. GAO-07-387. Washington, D.C.: Mar. 2007 Defense Acquisitions: Missile Defense Agency Fields Initial Capability but Falls Short of Original Goals. GAO-06-327. Washington, D.C.: Mar. 2006. Defense Acquisitions: Status of Ballistic Missile Defense Program in 2004. GAO-05-243. Washington, D.C.: Mar. 2005 Missile Defense: Actions Are Needed to Enhance Testing and Accountability. GAO-04-409. Washington, D.C.: Apr. 2004.

Why GAO Did This Study Since 2002, MDA has been developing a Ballistic Missile Defense System that can identify and intercept enemy threats. MDA has received approximately $132 billion and is planning to spend an additional $47.8 billion through fiscal year 2022 to continue its efforts. The National Defense Authorization Act for Fiscal Year 2012 included a provision that GAO annually assess and report on the extent to which MDA has achieved its acquisition goals and objectives. This report addresses (1) the progress MDA made in achieving fiscal year 2017 goals; (2) the extent to which MDA uses contracting vehicles known as undefinitized contract actions; and (3) the extent to which models provide credible information about the system's operational performance. To do this work, GAO reviewed planned fiscal year 2017 baselines and other documentation and assessed them against baseline reviews and GAO's acquisition best practices guides. In addition, GAO interviewed relevant officials. What GAO Found In fiscal year 2017, the Missile Defense Agency (MDA) made mixed progress in achieving its delivery and testing goals. MDA continued to deliver assets to the military services. However, system-level integrated capabilities, such as some discrimination and integrated cyber defense improvements, were delayed and delivered with performance limitations. Several programs achieved notable firsts, including the first intercept of an Intercontinental Ballistic Missile. However, one program experienced a failure, and other tests were delayed or deleted. Moreover, GAO found challenges in MDA's processes for communicating the extent and limitations of integrated capabilities when they are delivered. As a result, warfighters do not have full insight into the capabilities MDA delivers. GAO found that the average length of the undefinitized period and the not-to-exceed price of MDA's undefinitized contract actions, which authorize contractors to begin work before an agreement on terms, specifications, or price have been agreed upon, have increased over the past 5 years. While MDA policy permits use of undefinitized contracts on a limited basis, GAO and others have found that they can place unnecessary cost risks on the government. MDA does not completely assess BMDS performance using traditional flight tests. Instead, MDA relies on models, some of which produce data with limited credibility. According to Department of Defense and MDA policy, models used to operationally assess weapons systems must be accredited to ensure they reflect the real-world system. In addition, using unaccredited models increases the risk that test results could be distorted, and leaves decision makers without key information on how the system will perform. While MDA has taken steps to improve its models, it has used many models in system operational ground tests that were not certified for that use (see figure). Additionally, MDA does not communicate model limitations to some decision makers. Percentage of Accredited Models Used in Operational Assessments of Ballistic Missile Defense System Capability Deliveries, 2015 through 2017 What GAO Recommends GAO is making six recommendations to, among other things, improve the way MDA communicates capability deliveries; better report information about MDA's use of undefinitized contract actions; and address the challenges MDA has encountered with certifying its test models and communicating limitations of those models. DOD partially concurred with the first recommendation and concurred with the other five. GAO continues to believe the recommendations are valid as discussed in the report.