context stringlengths 100 4.73k | A stringlengths 100 2.88k | B stringlengths 100 3.62k | C stringlengths 100 3.51k | D stringlengths 100 2.86k | label stringclasses 4
values |
|---|---|---|---|---|---|
The reward function captures the multiple, and sometimes competing, objectives of the planner. In this study, we consider the following objectives: | In our model, the state represents the current configuration of the energy system across a set of nn cities. Each region is described by its energy demand, renewable (RE) and non-renewable (NRE) supply levels, population, and income classification. The state also includes a global budget variable which represents the a... | Low-income population without sufficient energy (negative weight), which penalizes inequitable access. | While traditional power grid infrastructure perpetuates energy access inequities—with low-income communities experiencing disproportionately longer outages and higher energy burdens—existing renewable energy planning models have largely failed to integrate social equity considerations into their optimization frameworks... | where bb is the remaining budget at time tt, did_{i} is the energy demand in region ii, rir_{i} is the renewable energy supply in region ii, nin_{i} is the non-renewable energy supply in region ii, pip_{i} is the population in region ii, IiI_{i} is the income indicator (1 = high/medium income, 0 = low income), and ℐ={1... | B |
For example, investors may be subject to fixed or dynamic electricity prices. Electricity tariffs provided by energy suppliers typically include system charges, i.e. network charges and environmental levies which can account for up to 45% of total LCOH in the UK [9]. Different system configurations can determine exempt... | In this benchmark case, a 10 MW electrolyser draws all its power from the public grid under a standard agreement with an energy supplier offering typical electricity tariffs, which could be fixed, dynamic, based on time-of-use (ToU) or green energy (when purchase of electricity is accompanied by REGOs) that could be us... | This paper presented a techno-economic framework for the evaluation of 5 wind-electrolyser use cases. Our work highlights the crucial role that location, PPA arrangements and control strategies play in the economic feasibility of deploying green hydrogen systems. We show that behind-the-meter systems with an electrolys... | In addition, we perform a sensitivity analysis on the main parameters affecting the LCOH i.e. the electricity price (between developers and the grid electricity price), the PEMEL costs, future improvements on the electrolyser efficiency and lifetime, and discount rate to account for different economic parameters. Due t... | The main results of the analysis are summarised in Table II, where LCOH Free sets pPPA=0p_{\text{PPA}}=0 and serves as a benchmark for best-case scenarios. The configuration achieving the lowest LCOH is use case V-b-i (behind-the-meter, electrolyser-first, without grid back up, followed by II (Off-grid wind-PEMEL syste... | A |
Lorie, Mark, James E. Neumann, Marcus C. Sarofim, Russell Jones, Radley M. Horton, Robert E. Kopp, Charles Fant, et al. 2020. | An open-access cmip5 pattern library for temperature and precipitation: description and methodology. | An open-access cmip5 pattern library for temperature and precipitation: description and methodology. | An open-access cmip5 pattern library for temperature and precipitation: description and methodology. | An open-access cmip5 pattern library for temperature and precipitation: description and methodology. | A |
To illustrate their result, consider a finite state space Θ\Theta and signal space SS, and let μ=(μi)i∈Θ\mu=(\mu_{i})_{i\in\Theta} denote a generic experiment, where μi\mu_{i} is the signal distribution conditional on state i∈Θi\in\Theta. | In contrast to Proposition 3.1, KL cost functions can also be characterized by Blackwell monotonicity, additivity, and dilution linearity (see Appendix B.3.3). This characterization is analogous to Theorem 1 in Pomatto, Strack, and Tamuz (2023), except that they impose a continuity axiom (which they conjecture to be un... | They show that cost functions satisfying these axioms, along with Blackwell monotonicity and a form of continuity, can be represented by the KL cost function of the form: | Theorem 3.1 characterizes cost functions that satisfy mixture convexity, sub-additivity, identity additivity, and Blackwell monotonicity. Building on the techniques developed in Mu, Pomatto, Strack, and Tamuz (2021) and Farooq, Fritz, Haapasalo, and Tomamichel (2024), we show that such cost functions can be represented... | Observe that any such cost function satisfies independence. Therefore, by the second part of Theorem 3.2, Rényi cost functions can be characterized as the intersection of Max-Rényi costs and cost functions of the form (7). | B |
Taking into account the optimal matching between top workers and others, we estimate the production function for Canadian firms over the period 2003-2015, from which we obtain the firm-level total factor productivity (Hicks-neutral technology). We then obtain the measured total factor productivity which consists of the... | The basic facts concerning the top workers at the aggregate level are: First, aggregate top-worker quality declined over the period 2003-2015 but mostly since 2008; Second, the person fixed effect of top workers in AKM estimation accounted for most of the declines of top-work quality. Reallocation of top-worker quality... | Two important issues are left for future research. First, in our model, the coefficient for the elasticity of substitution between worker types is not identified in the structural model because the matching function does not involve the elasticity of substitution. How to fully estimate the production function with opti... | Second, the negative contribution of the top-worker quality to the slowdown of measured productivity growth is associated with the falling aggregate top-worker quality since the output elasticity θ^\widehat{\theta} does not vary over time. The quality of top workers dropped more than that of non-top workers. On average... | The equilibrium matching is linear in worker types. The shape of the matching function does not depend on the elasticity of substitution between top worker and non-top workers. The following condition is sufficient for PAM. For any σ≠1\sigma\neq 1, | B |
A significant portion of CID data sets contains rumours or fake information, which seriously hinders the further processing of extracted knowledge or data items [26, 29, 48]. As a result, problems related to rumour propagation are increasingly recognised as important, see e.g., [8, 52]. | The main output of trend-based modelling is the set of possible scenarios and possible transitions between them. This set of scenarios and transitions can be seen as the solution to the trend model. Finding this solution represents a combinatorial task and is not studied in this paper; for details, see [20]. There are ... | In summary, this paper provides a qualitative framework for integrating rumour dynamics into CID modelling under severe information constraints, using scenario-based trend reasoning and transition graphs to formalise and simulate possible system behaviours. | The presence of rumours – typically unverified, ambiguous, and uncertain – brings additional subjectivity into CID tasks. A real-world CID model integrated with a rumour component often suffers from severe shortages of information and knowledge. However, traditional statistical algorithms rely on extensive informationa... | The system of numerically based differential equations (8) is translated into the trend-based RRM (11) using the conversion rules given in (9) and (10). | B |
Such identity and registration infrastructure are currently missing for AI agents (Hadfield, 2025; Chan et al., 2025). Building them out will be essential, but their design raises questions around legal accountability. One possible route is to require that any AI agent entering into a contract or transaction be registe... | Such identity and registration infrastructure are currently missing for AI agents (Hadfield, 2025; Chan et al., 2025). Building them out will be essential, but their design raises questions around legal accountability. One possible route is to require that any AI agent entering into a contract or transaction be registe... | It is important to recognize that the design and deployment of AI agents will be driven by market forces. How might market incentives shape the pricing and design of AI agents? | Outline. The rest of this chapter is organized as follows. In Section 2 we outline questions around how (i) AI agents deployed in markets might shape prices, search, bargaining, and finance; and (ii) market forces in turn shape the design and proliferation of AI agents. In Section 3 we turn to AI agents within organiza... | A second possible route for AI agent accountability is to follow the model of the emergence of the corporation, which is another artificial entity that participates in the economy. AI agents could be accorded legal personhood, meaning they could sue and be sued in their own ‘name’ in court. Clearly such an approach wou... | D |
In this section, we verify our theoretical findings and demonstrate the prevalence of the bias-variance trade-off for data with non-negative outcomes. | Our next set of simulation results in Figure˜6 reports the bias, standard deviation, and RMSE of generalized PML estimators for κ∈[−1,1]\kappa\in[-1,1], under censoring probability 11+(τexp(θ0Tx))β\frac{1}{1+(\tau\exp(\theta_{0}^{T}x))^{\beta}} with τ=1\tau=1, β=2\beta=2, and α∈[0,1,2]\alpha\in[0,1,2]. We find that ... | Our simulation design follows the setup in Santos Silva and Tenreyro (2006), combined with the censoring model introduced in this paper. | The rest of the paper is organized as follows. In Section˜2, we provide the econometric background on non-negative data modeling and introduce the formal setup and key concepts of the paper. In Section˜3, we propose a class of moment estimators that we call generalized pseudo maximum likelihood estimators. We character... | in XX, i.e., log(𝔼(Y∣X))=θTX\log(\mathbb{E}(Y\mid X))=\theta^{T}X. The implications of this distinction, including the robustness of PPML to heteroskedasticity, are discussed extensively in Santos Silva and Tenreyro (2006). | B |
The key insight from our analysis is that successful variable pricing requires coordinated policy packages that address both individual flexibility constraints and the systemic factors that concentrate benefits among already-advantaged consumers. The distributional challenges identified in our analysis stem from a fund... | where the weights are the relative flexibilities of aggregate demand in each period (shown in C). This expression demonstrates that the flat price lies closer to the marginal cost of the period with higher consumer flexibility. As this is evaluated at the flat price rather than variable prices, we further examine the r... | We thank the MIT Portugal Program and the IDSS Initiative for Combatting Systemic Racism for financial support of this research. We also gratefully acknowledge support from the MIT UROP program, which enabled undergraduate research contributions to this project. | Together, these contributions connect empirical research on energy poverty and consumption with theoretical work on electricity market design, providing analytical foundations for designing variable pricing policies that balance efficiency with equity in demand-responsive systems. | The key insight from our analysis is that successful variable pricing requires coordinated policy packages that address both individual flexibility constraints and the systemic factors that concentrate benefits among already-advantaged consumers. The distributional challenges identified in our analysis stem from a fund... | B |
Define Γ∗=Γ∗({(Xi′,Di)′}i∈𝒮)≡|𝒮∗|−1∑i∈𝒮∗ZiZi\Gamma^{*}=\Gamma^{*}(\{(X_{i}^{\prime},D_{i})^{\prime}\}_{i\in\mathcal{S}})\equiv|\mathcal{S}^{*}|^{-1}\sum_{i\in\mathcal{S}^{*}}Z_{i}Z_{i}. | For a symmetric positive semi-definite matrix AA, let λmin[A]\lambda_{\min}[A] denote its smallest eigenvalue. For some fixed positive constant λ¯>0\underline{\lambda}>0, | lim inf|𝒮|→∞λmin[Γ∗]≥λ¯\displaystyle\liminf_{|\mathcal{S}|\rightarrow\infty}\lambda_{\min}[\Gamma^{*}]\geq\underline{\lambda} | lim inf|𝒮|→∞λmin[Δ∗]≥λ¯\displaystyle\liminf_{|\mathcal{S}|\rightarrow\infty}\lambda_{\min}[\Delta^{*}]\geq\underline{\lambda} | the index of the closest unit in 𝒮∗\mathcal{S}^{*} to ii, where κ>0\kappa>0 is a large, fixed positive constant. Our proposed estimator for Δ∗\Delta^{*} is then | A |
Table 3: Prevalence of identity slippage by discipline, framework and whether estimand was explicitly specified | Here, we provide a more detailed comparison of the different disciplines and application types. Moreover, we provide sensitivity analyses where we consider alternative specifications of potentially controversial classification criteria. | We distinguish between three different types of IV applications: RCTs, Mendelian Randomization (MR)121212I.e. genetic variants were used as instruments, and all other applications. Since MR application occurred almost exclusively in epidemiology and medicine, we limited the discussion in the main article to RCTs and na... | Figure 7 illustrates the distribution of pragmatic and strict studies across application types. Across types of application, ideal studies were relatively rare. The highest share of ideal studies was observed for RCTs (n = 6, 18%), followed by MR (n = 9, 8%), and other applications (n = 13, 8%). Figure 8 illustrates th... | Application of MR accounted for 110 studies (35%), applications in randomized controlled trials (RCTs)131313I.e. the investigators controlled the assignment of the instrument. accounted for 33 studies (11%), and other applications accounted for 166 studies (54%). Application type was a strong predictor of discipline. M... | B |
Now, what information do we need to identify the revision plan generated by a probability measure? Since we want each selection function σ\sigma to represent one particular revision plan (equivalently, consequence relation), this means that we want axioms for selection functions such that, for any σ\sigma satisfying th... | Thus we have characterised the way in which the probabilistically stable revision plans divide the probability simplex: each is determined by hyperplane equations of the form given above.121212Regions of hyperplane arrangements (chambers) do not contain points on the hyperplanes themselves (they are open regions). What... | Consider the simplex Δn−1\Delta^{n-1} of all probability distributions over Ω\Omega, conceived of as the set of vectors {(x1,…,xn)|xi≥0,∑i=1nxi=1}\{(x_{1},\dots,x_{n})\,|\,x_{i}\geq 0,\sum^{n}_{i=1}x_{i}=1\}, each measure μ\mu corresponding to the vector xi=μ(ωi)x_{i}=\mu(\omega_{i}). Say that two measures are equival... | Given a measure μ\mu and a selection function σ\sigma that μ\mu represents, the range of thresholds that generate the exact same belief revision policy as σ\sigma is given exactly by the interval [α,β)[\alpha,\beta) where | We first need a criterion for identifying when two distributions generate the same revision plan (i.e., when two distributions generate the same consequence relations). We begin with the following observation: | D |
Once the approximated normalizing constant cMF(X;θ)c^{MF}(X;\theta) is acquired, the Metropolis-Hasting method can be applied to sample pairs (θ,g)(\theta,g) from the posterior distribution given by Equation (2.7). The procedure is outlined as follows: | The parameter of interest is θ\theta. The primary goal of our study is to infer the value of θ\theta after observing ψ0\psi_{0} and XX. The estimation of θ\theta will provide insights into the underlying structure and individuals’ preferences. | Iterate the steps for a total of TT times. At the tt-th iteration with current parameter value θ\theta and network gg: | While the numerator of (2.7) eliminates the need for summation over g∈𝒢g\in\mathcal{G}, it presents computational challenges when estimating both θ\theta and gg in the MCMC framework. Specifically, the persistent issue of the intractable normalizing constant c(X;θ)c(X;\theta) complicates the sampling procedure for bo... | Recalling that the normalizing constant is essentially the sum of exponentiated potential functions, let us begin by defining the statistics TT and ϕ\phi as: | B |
Even if firms observe nn i.i.d. noisy signals from repeated queries to the pricing interface 𝒜\mathcal{A}, the resulting sample mean estimator s¯j\bar{s}_{j} retains residual noise. As a result, cartel convergence remains delayed in finite time. | Distorted competition (Information distortion): In Section 5, we quantify how imperfect information alters equilibrium outcomes: price distortions (Theorem 1) that cascade into profit (Theorem 2) and consumer welfare (Theorem 3). The core insight is that when pricing algorithms transmit distorted estimates, market outc... | Tradeoff Analysis (Section 7): Finally, we evaluate the use of controlled information friction as a regulatory lever to fight against (tacit) price coordination. We derive upper bounds on noise levels that limit harm to non-collusive firms and consumers, and quantify how such noise delays cartel consensus. This yields ... | Regulatory noise can indeed slow down cartel coordination, even under repeated queries, making controlled information quality an effective policy lever (Theorem 6, Proposition 1). But how much noise is too much (e.g., noise scale σ→∞\sigma\to\infty), given that legitimate market participants are subject to the same inf... | Noise as a policy lever (Tradeoff analysis): In Section 7, we evaluate controlled noise as a regulatory tool. The first-order concern is validity: should collusion under our dynamics be taken seriously? We show it should. Collusion benefits leaders while harming followers. Further, contrary to common intuition, leaders... | C |
Educators have already begun to shift in this direction: In response to too many AI generated research papers, one Northeastern professor indicated that | and the threshold will tilt in favor of which utility grows more (as students who were previously indifferent at TT will shift to that side). In particular, TT grows when pp increases (at the margin) if and only if | I will redesign the assignment so it can’t be done with AI next time. I had one student complain that the weekly homework was hard to do and they were annoyed because Claude and ChatGPT was useless in completing the work. I told them that was a compliment, and I will endeavor to hear that more from students (Anthropic ... | Instead, ability is determined at an earlier stage: during a future worker’s education. At many educational institutions across all levels, AI (and large language models in particular) has already begun to play a significant role. It has led to many positive impacts, such as helping students learn new concepts by provi... | Many open questions remain. Most interesting would be to take the model to data. Different models and domains have different values of d,pd,p which could influence how students use AI tools. Students may directly ask for solutions more often in foundational classes, characterized by higher d,pd,p, than in more advanced... | B |
Classical consumption theory, grounded in the life cycle and permanent income hypotheses, posits that rational consumers smooth consumption over time by allocating resources in accordance with expected lifetime income (Modigliani and Brumberg,, 1954, Hall,, 1978). Within this framework, credit services merely facilitat... | In line with previous studies, our results demonstrate a positive correlation between credit limits and consumer spending, underscoring the role of credit as a catalyst for consumption (Aydin,, 2022). Our analysis reveals a heterogeneous effect across different spending quantiles. In particular, as credit limits increa... | A growing body of empirical research has established that increases in credit availability—whether through higher card limits, relaxed lending terms, or digital financing options—tend to elevate aggregate consumer spending (Gross and Souleles,, 2002, Soman and Cheema,, 2002, Agarwal et al.,, 2007, Wilcox et al.,, 2011,... | However, a substantial body of empirical evidence challenges this neutrality by demonstrating that consumption is often excessively sensitive to credit conditions (Bacchetta and Gerlach,, 1997, Breza and Kinnan,, 2021). These findings suggest that many consumers face binding liquidity constraints or behavioral deviatio... | Understanding how variations in assigned credit limits influence consumer spending behaviors is fundamentally important, as platforms can identify optimal credit levels that stimulate consumption without inducing excessive risk. This insight enables BigTech firms to strategically allocate credit to maximize transaction... | C |
The first term in the objective represents revenue maximization, while the second term is an augmented penalty term. The penalty serves to dampen oscillations by constraining abrupt changes in agents’ local dispatch decisions across iterations. This helps to drive the iterative process towards an equilibrium state by p... | Given an electricity price forecast, each satellite model solves the following optimization problem. | Each prosumer maximizes revenues from electricity generated locally, or minimizes the consumption costs of energy carriers, subject to its technical constraints. The following problem is established for each prosumer. | Each generator agent maximizes its revenues from electricity generation within its technical limits by solving the following optimization problem. | The objective 1 minimizes the cost of electricity generation and the consumption costs of externally priced energy carriers (e.g., methane). Electricity, heat, and hydrogen are denoted by p,q,hp,q,h, respectively. Electricity generation and consumption must be balanced at all time steps 2. The prosumer electric power 3... | B |
In order to illustrate finite sample behavior and the applicability of the provided modeling approach, we present two different simulated examples. We provide one example where Θ\Theta is a diagonal matrix and one with a non-diagonal Θ\Theta. | The processes were simulated with 1000 different realizations of fBm, resulting in 1000 sample paths for each set of parameters in both examples. | Figure 3. Diagonal values of the componentwise differences σ^−σ\hat{\sigma}-\sigma in the diagonal case with two different Hurst indices. | In the non-Gaussian case, Assumption 2.3 (i) and (ii) are also typically valid provided that the memory vanishes as the lags grow without a limit. One can again compute the rates l1(T)l_{1}(T) and l2(T)l_{2}(T) explicitly from ‖Cov(b^U,T)‖\left\|\mathrm{Cov}(\hat{b}_{U,T})\right\| and ‖Cov(vec(γ^U,T(t)))‖\left\|\... | In both examples, two-dimensional interest rates with fBm as the random process XX were simulated using the Euler-Maryama method. The model parameters and the Hurst indices HH for the simulations are given in Table 1. | D |
Subsample Analysis. To test robustness, we replicate the analyses in subsections 5.1 through 5.3 using only later periods of each treatment. Specifically, for each treatment we restrict the data to observations from period 11 onward, ensuring that all subjects had a ten-period learning phase. Table 13 reports three out... | We now compare treatments across quality supports, l=50l=50 and l=80l=80. Products in D80 have a higher average quality of 90 than in D50, which has an average quality of 75, so, all else equal, one would expect higher prices and profits in D80. Strikingly, the opposite holds in D1 and D2. Prices and profit shares in D... | With risk-neutral buyers, sellers’ profits are lower in noisy certification (D3) than in accurate certification (D2). | Trends in Data. Figure 4 traces the dynamics of three variables: sellers’ posted prices, sellers’ profits, and buyers’ profits. Prices and sellers’ profits decline over time, while buyers’ profits rise. The price decline is consistent with experimental evidence on convergence toward more competitive pricing in Bertrand... | Since differences in paid certification costs cannot fully explain the observed profitability ranking, we compare the competitiveness of the environments. As discussed in the theoretical section, noisy certification introduces endogenous product differentiation by adding randomness to buyers’ WTP, which can reduce comp... | C |
We also extended the model to Calvo pricing, where each firm receives random opportunities to reset price. A stationary equilibrium in Markov reset policies exists (Proposition 7.1). More frequent resets discipline dispersion and accelerate learning by letting disadvantaged firms re‐price aggressively when beliefs move... | Methodologically, the paper offers a compact toolkit—posterior formulas, a one‐line threshold, and closed‐form cascade boundaries—that makes learning–pricing feedback analyzable in continuous time with random arrivals. The same primitives drive firms’ mixed supports and the belief process, making the comparative static... | The patterns line up with our theory. In the top panel, paths move quickly to the correct absorbing region because actions are informative and consumers are willing to check the alternative (low κ\kappa). In the bottom panel, actions are weak signals and search is costly, so beliefs drift slowly and wrong cascades are ... | Conceptually, the paper offers a clean, tractable way to study social learning and price competition in the same continuous-time environment with random arrivals. Technically, a one-line consumer cutoff and closed-form cascade boundaries make the feedback loop analyzable without heavy machinery and pin down tight compa... | In the static pricing game with observational learning and random arrivals, a (possibly mixed) Bayesian Nash equilibrium exists. In the symmetric environment (same primitives for both firms), there exists a symmetric BNE.777Learning makes payoffs discontinuous in prices at knife-edge beliefs. We use payoff security and... | A |
Research on multi-stage revolutionary movements has further enriched our understanding of coordination under uncertainty. Early works by Kuran, (1989, 1991) and Lohmann, (1994) explore how incomplete information, preference falsification, and cascading behavior shape the timing and likelihood of revolutions. More recen... | In our model, all payoffs depend on the fundamental state θ\theta, which measures the regime’s strength. We focus on a signaling equilibrium in which the policymaker’s intervention rr influences the agents’ beliefs about θ\theta. | Moreover, de Mesquita, (2010) develops a formal model in which regime transitions hinge on how individual incentives align with collective revolutionary action, emphasizing the role of private information and threshold-based coordination. In our setting, we focus on how a policymaker’s uninformed status influences thes... | Further contributions examine the effects of information precision and signaling on coordination. Iachan and Nenov, (2015) investigate how the quality of private signals affects the probability of attacks and welfare outcomes in a static global game, while Kyriazis and Lou, (2023) propose a signaling game in which a le... | Research on multi-stage revolutionary movements has further enriched our understanding of coordination under uncertainty. Early works by Kuran, (1989, 1991) and Lohmann, (1994) explore how incomplete information, preference falsification, and cascading behavior shape the timing and likelihood of revolutions. More recen... | B |
Consequently, assuming a constant birth rate b>0b>0, and normalizing the size of the genetic lineage to Γ0=1\Gamma_{0}=1, we posit that the “selfish gene” spreads geometrically from t=0t=0 until time TT when the extinction risk materializes: | This calculation implicitly assumes smooth temporal evolution of the dynasty’s size, thereby excluding the scenario where the dynasty randomly dies off before TT. This is an innocuous assumption as long as the dynasty’s size is sufficiently large. | Consequently, assuming a constant birth rate b>0b>0, and normalizing the size of the genetic lineage to Γ0=1\Gamma_{0}=1, we posit that the “selfish gene” spreads geometrically from t=0t=0 until time TT when the extinction risk materializes: | How should we think about the perspective of human extinction? On the one hand, human extinction should intuitively imply a greater loss than the combined loss of all lives, because it would also entail the loss of human history and heritage, as well as our potentially glorious future. But, on the other hand, perhaps t... | The exogenous parameter α∈(0,1)\alpha\in(0,1) captures the loss of genetic code due to recombination and mutation, and depends on the average age of human reproduction. By assuming that the gene is sufficiently widespread, we exclude the scenario in which the given genetic lineage randomly dies off before TT, which is ... | D |
We note that our paper offers several directions for future work. Firstly, we believe that it is interesting to further analyze the axiomatic properties of position-threshold rules. This may help to identify new desirable voting rules on the interval domain or to strengthen the argument for the endpoint-median rule. Mo... | I thank Evghenii Beriozchin for helpful discussions and Felix Brandt for valuable feedback. This work was funded by the NSF-CSIRO grant on “Fair Sequential Collective Decision-Making” (RG230833). | Such processes of collective decision-making are formally studied in the field of social choice theory, where researchers analyze voting rules from a mathematical perspective. While this research has led to significant advances in the understanding of voting rules (e.g., Arrow et al., 2011; Brandt et al., 2016), social... | In this paper, we will study voting rules on the interval domain and introduce the class of position-threshold rules. Intuitively, these rules determine for every alternative a collective position, which quantifies the voters’ relative position regarding this alternative, and choose the left-most alternative whose coll... | A ubiquitous phenomenon in today’s societies is collective decision-making: given the possibly conflicting preferences of multiple agents, a joint decision should be reached in a fair and principled way. | A |
This example illustrates that Cramér–Rao–type arguments alone are insufficient to rule out super-efficient points, indicating that the existence of general lower bounds for estimators cannot be derived solely from such inequalities. | Le Cam (1953) showed that the set of super-efficient points has Lebesgue measure zero, with subsequent extensions by Bahadur (1964) and Pfanzagl (1970). | If θ1\theta_{1} and θ2\theta_{2} are distinct elements of Θ\Theta, the set {ω∈[−π,π]:sθ1X(ω)≠sθ2X(ω)}\{\omega\in[-\pi,\pi]:s_{\theta_{1}}^{X}(\omega)\neq s_{\theta_{2}}^{X}(\omega)\} has a positive Lebesgue measure. | To overcome this limitation, one needs the LAN property together with the Hájek–Le Cam’s local asymptotic minimax theorem (Hájek, 1972; Le Cam, 1972), which ensures that no estimator can asymptotically achieve a smaller risk than the bound determined by the Fisher information in shrinking neighborhoods of the true para... | Such finite-sample inequalities control pointwise variances but do not rule out the existence of super-efficient points. | A |
=max(β,γ)∈[0,1]2[200p+100q−50−(25−100p)β+(25−100q)γ]=50+100p.\displaystyle=\max_{(\beta,\gamma)\in[0,1]^{2}}[200p+100q-50-(25-100p)\beta+(25-100q)\gamma]=50+100p. | if \Dirac[f(ω)]≿\Dirac[f(ω′)]\Dirac[f(\omega)]\succsim\Dirac[f(\omega^{\prime})] is equivalent to | f2f_{2} is more ambiguous than f1f_{1}—ambiguity aversion offsets the objective advantage of f2f_{2}. | The environment is informationally symmetric—there is no evidence to believe one color is more likely than another. | the core of ν\nu is the set \setμ∈Problem 515151Problem 51Problem 51.(Ω)\mvertμ(E)≥ν(E)\tforeachE∈2Ω\set{\mu\in\prob(\Omega)\mvert\mu(E)\geq\nu(E)\tforeach E\in 2^{\Omega}}, and | C |
This parameterization allows us to impose the identification constraints in ℐK1\mathcal{I}^{1}_{K} directly on the coefficients θK\theta_{K}, converting the infinite-dimensional constraint system into a finite-dimensional one. To implement this, we sample a finite set of vv values from its distribution. These sampled v... | Computing the feasible set ΘK\Theta_{K} may be computationally demanding due to the potentially large number of constraints. To improve numerical feasibility and avoid empty identified sets, researchers can introduce slack terms into both equality and inequality constraints, which allow for small tolerances. Monte Carl... | This parameterization allows us to impose the identification constraints in ℐK1\mathcal{I}^{1}_{K} directly on the coefficients θK\theta_{K}, converting the infinite-dimensional constraint system into a finite-dimensional one. To implement this, we sample a finite set of vv values from its distribution. These sampled v... | where {bkt(⋅)}t=1Tk\{b_{kt}(\cdot)\}_{t=1}^{T_{k}} are known basis functions from [0,1]J[0,1]^{J} to ℝ\mathds{R}, Tk∈ℕ+T_{k}\in\mathds{N}^{+} is the number of basis functions, and θk≡(θk1,…,θkTk)∈ℝTk\theta_{k}\equiv(\theta_{k1},...,\theta_{kT_{k}})\in\mathds{R}^{T_{k}} parameterizes the function E[Yk|V=v]E[Y_{k}|V... | ΘK={θK:{∑t=1Tkθktbkt(v)}k∈K∈ℐK1}\Theta_{K}=\left\{\theta_{K}:\left\{\sum_{t=1}^{T_{k}}\theta_{kt}b_{kt}(v)\right\}_{k\in K}\in\mathcal{I}^{1}_{K}\right\} | A |
In this section, we implement the impulse response decomposition and explore the dynamic attributions of investor sentiment in shaping the macroeconomic effects of monetary policy shocks. | Figure 2. Decomposition of impulse responses to a monetary policy shock. Structural vector autoregression impulse response functions to a 25-basis-point monetary policy shock, identified using high-frequency IV around Federal Open Market Committee announcements (IV: “MPS_ORTH” by \citeasnounbauer2023reassessment). Samp... | Figure 2 presents the decomposition of the impulse response function of industrial production to monetary policy shock. In each panel, the solid black line represents the total impulse responses, while the stacked bars illustrate the contributions of individual macroeconomic variables, as derived from our impulse respo... | The initial response is primarily driven by personal consumption expenditures, consumer prices, and the excess bond premium, with consumption expenditures exhibiting the strongest contemporaneous contribution to the impulse response of the monetary policy shock. At the 3-month horizon, the contribution of the excess bo... | We perform our impulse response decomposition analysis using a monetary dataset consisting of 383 monthly observations from February 1988 to December 2019. The data set includes eight key U.S. macroeconomic indicators: Industrial Production (IP), Consumer Price Index (CPI), Excess Bond Premium (EBP), Expected Default R... | D |
Part (i) states that the per-period payoff is increasing in the current state, and allows for a finite number of "kinks" in the first derivative. However, the right and left partial derivatives of the per-period payoff function w.r.t. the first argument are restricted to ensure monotonicity and differentiability of the... | Even though there are no restrictions on concavity and s↦π(s,s′)s\mapsto\pi(s,s^{\prime}) is not continuously differentiable, the value function still inherits the smoothness properties of π\pi. As the next proposition shows, the left and right derivatives exist and can be characterized in terms of those of π\pi. | Linear approximation methods require additional assumptions, such as strict concavity of the per-period payoff, and are both local in nature and uninformative about the precise shape of the basin of attraction.262626To our knowledge, global results based on linear approximation have been obtained only for the Cass–Koop... | Our analysis of the planner’s dynamic problem relies on establishing monotonicity and differentiability properties of the value function and the optimal policy correspondence. We establish these results in this subsection. However, due to the potential non-concavity and non-smoothness of π\pi, standard “textbook” argum... | Parts (i) and (iii) are both critical assumptions of our framework, and can be seen as "replacements" of the standard strict concavity of π\pi — while not innocuous, they still encompass many applications of interest. We refer the reader to Section 6 for a more thorough discussion, including ways in which | D |
Thus, iridium-specific power density ωi\omega^{i} emerges as the most critical factor influencing iridium demand from PEMEL electrolysis. As previously discussed, literature values for current ωi\omega^{i} vary significantly [23]. Eikeng et al. report a state-of-the-art value of 750 [kg⋅GW−1][kg\cdot GW^{-1}] [39], a... | Thus, iridium-specific power density ωi\omega^{i} emerges as the most critical factor influencing iridium demand from PEMEL electrolysis. As previously discussed, literature values for current ωi\omega^{i} vary significantly [23]. Eikeng et al. report a state-of-the-art value of 750 [kg⋅GW−1][kg\cdot GW^{-1}] [39], a... | While a 90% reduction is considered technically feasible, comparable to past platinum reductions in PEMEL fuel cells [9, 39, 23, 40], the decisive factor is the speed at which such improvements can be realized. Reducing ωi\omega^{i} early is essential not only to meet short-term capacity targets but also to ensure that... | Conversely, several studies suggest that overall platinum demand may continue to rise due to its broad range of applications, including jewelry, glass manufacturing, electrolysis, and fuel cells [21, 36]. In particular, the latter is expected to play a critical role in sustaining PGM production, highlighting the interd... | If iridium-specific power densities do not decline rapidly enough, one remaining, but questionable, strategy to reduce initial iridium demand would be to artificially shorten PEMEL lifetimes. As shown in the sensitivity analysis in Section 3, setting the average lifetime to τ=5\tau=5 years can significantly lower prima... | B |
In the case of a joint distribution over three variables, 𝒳={a,t,z}\mathcal{X}=\{a,t,z\}, each with two discrete values, the joint distribution is a 2×2×22\times 2\times 2 tensor array with 88 occupation numbers, qa,t,zq_{a,t,z} that we refer to as the ”more informative” case and present in Table 3. If we have informa... | The reasoning above reveals that there are at least three ways we might estimate the conditional frequency of outcome conditional on sex and treatment, depending on what level of data is available. If the data on clinical trials includes the sex of the subject (Table 2), we would want to use that information by basing ... | Each conditional or marginal depends on a single free parameter in [0,1][0,1]. With three binary variables, there are 77 independent parameters, matching the degrees of freedom of the normalized joint distribution. Table 4 shows the parameterization explicitly. | The ATE for the example in Table 1 is 0.10.1, indicating that the treatment has on average a positive effect on the rate of recovery. The ATE for the data in Table 2 conditional on sex is −0.2-0.2 for both males and females. While the ATE is a point estimate, we are really interested in a full posterior description of ... | In the case of a joint distribution over three variables, 𝒳={a,t,z}\mathcal{X}=\{a,t,z\}, each with two discrete values, the joint distribution is a 2×2×22\times 2\times 2 tensor array with 88 occupation numbers, qa,t,zq_{a,t,z} that we refer to as the ”more informative” case and present in Table 3. If we have informa... | B |
ROC AUC: The area under the receiver operator characteristic curve quantifies how well a model ranks the observations correctly. Specifically, if we are to randomly present both a positive instance and a negative instance, how often would the model rank the positive higher? A high AUC value indicates good discriminatio... | Table 6 presents the search space and final values determined through these steps. We selected hyperparameters using the one-standard-error rule333The one-standard-error rule is a heuristic for choosing a parsimonius model, which has statistically similar predictive performance to the most optimal model determined thro... | Results: Our findings reveal that the model is highly stable. The rolling Brier score stays extremely low (near 0.00) for the vast majority of the test period, suggesting consistently measured probabilities that are accurate and well-calibrated during a stable market. The brier score shows two elevated periods that map... | Brier Score: The Brier score measures the accuracy and calibration of the probability forecast itself. It is the mean-squared error of each of the predicted probabilities against the truth of what happened (0 or 1). A lower Brier score means that the model’s probability output is more trustworthy and closer to the true... | Notes: This figure plots the Brier score of the primary SVM model’s calibrated probability forecasts, calculated over a 63-day rolling window. The sample is the hold-out test set, covering the period from July 2023 to June 2025. The Brier score measures the mean squared error between predicted probabilities and actual ... | C |
Externalities are increasing [decreasing] if ∂2uA∂a∂r≥0\frac{\partial^{2}u_{A}}{\partial a\partial r}\geq 0 and ∂2uO∂a∂r≥0\frac{\partial^{2}u_{O}}{\partial a\partial r}\geq 0 [∂2uA∂a∂r≤0\frac{\partial^{2}u_{A}}{\partial a\partial r}\leq 0 and ∂2uO∂a∂r≤0\frac{\partial^{2}u_{O}}{\partial a\partial r}\leq 0] for all (... | Our leading example of pure externalities has ranked agent incentives because increasing [decreasing] externalities make ≿AI\succsim_{AI} equivalent to the increasing [decreasing] order of outsider actions. Another case of ranked incentives that matches applications with non-pure externalities (see Section 2.2) is sep... | With increasing [decreasing] externalities, each player’s selection of a higher action enhances [reduces] the opponent’s marginal payoff from choosing a higher action. Given the twice-differentiability assumption, our definitions here coincide with those of Segal, which are based on the payoff functions having increasi... | Our model consists of a principal, an agent, and an outsider. Principal contracts with the agent à la Segal (2003) by offering a public bilateral contract specifying transfers that only vary with the agent’s action. This contract is thus followed by a game in which the agent and the outsider act simultaneously. We allo... | Our results depend on two additional assumptions, Assumption 1 and 2, that generalize our setting beyond pure externalities, which may be the most prevalent family of strategic structures that people have studied. In particular, pure externalities include environments where players’ decisions are either supermodular (s... | B |
We consider a N×(T+1)N\times(T+1) data set represented by columns XtX_{t}, 0≤t≤T0\leq t\leq T. These columns are interpreted as observations of an NN-dimensional vector at T+1T+1 time points or as NN scalar time series. XtX_{t} is said to be cointegrated if there exists a linear combination with coefficients β\beta of ... | The test implemented in the Largevars package is based on the squared sample canonical correlations between transformed past levels (lags) and changes (first differences) of the data, as outlined in Section 2.2. Its asymptotic distribution (derived under N,T→∞N,T\to\infty jointly and proportionally) is given by the par... | It is well known (see Engle and Granger (1987); Johansen (1995)) that, under technical conditions, the process XtX_{t} is cointegrated if and only if Π≠0\Pi\neq 0. In particular, testing for the absence of cointegration can be recast as testing the hypothesis Π=0\Pi=0 in model (1). | Cointegration has been extensively studied in econometrics, beginning with seminal works by Granger (1981); Engle and Granger (1987). Many variables in macroeconomics and finance, such as price levels, consumption, output, trade flows, and interest rates, may exhibit cointegration. A classic example is the relationship... | Various R packages have been developed for cointegration testing. For instance, the urca package Pfaff (2008) includes functions such as ca.jo (implementing the procedure of Johansen (1988, 1991)), ca.po (for the test of Phillips and Ouliaris (1990)), and cajolst (implementing the procedure of Lütkepohl et al. (2004)).... | C |
Table 9: Evaluation of the proposed NARFIMA model’s performance relative to baseline forecasters across all forecast horizons for China (‘best’ and ‘second-best’ results are highlighted). The MASE metric is not defined for a forecast horizon of length one. | Figure 1: Multiple comparisons with the best (MCB) plot for BRIC nations based on (a) RMSE, (b) MAE, (c) SMAPE, and (d) MAPE metrics. In the plots, ‘NARFIMA-1.58’ indicates that the average rank of the NARFIMA model is 1.58, based on the RMSE metric. A similar interpretation holds across different models and metrics. | Fig. 2 presents the Murphy diagrams comparing NARFIMA with the top-performing ARIMAx and BSTSx frameworks, as identified by the RMSE-based MCB test results, for each BRIC nation over a 48-month-ahead forecast horizon. The diagram plots the extremal scores for competing models, where a lower score indicates better model... | [NARFIMA(x~t)±CQtΨ^(x~t)].\left[\operatorname{NARFIMA}\left(\tilde{x}_{t}\right)\pm{\operatorname{CQ}}_{t}\;\widehat{\Psi}\left(\tilde{x}_{t}\right)\right]. Fig. 3 represents the 95% conformal prediction intervals of the NARFIMA framework for the long-term 48-month-ahead forecast horizon. The figure depicts the poin... | In this section, we assess the robustness of our empirical results by evaluating the performance of different forecasting models based on differences in measurement errors, using multiple comparisons with the best (MCB) test. The model-agnostic MCB test is a nonparametric method that ranks each forecaster based on its ... | D |
Assuming conditional independence, our Stata implementation of the first-best policy learning in a multi-action setting incorporates both linear and quadratic risk-adjusted preferences using the first-best rule as reference (optimal) decision algorithm. | e(N_V_opt_new) is the number of observations for computing the value function in the new dataset for the optimal policy. | The command opl_ma_fb implements Optimal Policy Learning (OPL) in a multi-action treatment setting computing the first-best policy using the RA approach for estimating the value function under different risk preferences. It allows for training and evaluation of optimal treatment policies based on observational data. Th... | The command opl_ma_vf estimates the value function for multi-action optimal policy learning using three different methods: | This paper presents the command opl_ma_fb (and the companion command opl_ma_vf), implementing the first-best optimal policy learning (OPL) algorithm to estimate the best treatment assignment given the observation of an outcome, a multi-action (or multi-arm) treatment, and a set of observed covariates (or features). It ... | B |
are drawn from global (Normal) distributions:222Formally, 𝜷j≡(βj1,…,βjK)⊤.\boldsymbol{\beta}_{j}\equiv(\beta_{j1},\ldots,\beta_{jK})^{\top}. | are drawn from global (Normal) distributions:222Formally, 𝜷j≡(βj1,…,βjK)⊤.\boldsymbol{\beta}_{j}\equiv(\beta_{j1},\ldots,\beta_{jK})^{\top}. | yi∼𝒩(αj+𝐗i⊤𝜷j,σ)y_{i}\sim\mathcal{N}\big{(}\alpha_{j}+\mathbf{X}_{i}^{\top}\boldsymbol{\beta}_{j},\,\sigma\big{)} | ∼𝒩(μβk,σβk),k=1,…,K,\displaystyle\sim\mathcal{N}(\mu_{\beta_{k}},\,\sigma_{\beta_{k}}),\quad k=1,\ldots,K, | income of ∼0.34\sim 0.34 on the eleven-point scale. the negative impact of being unemployed is enormous — | C |
Likewise, if the actual GG is asymmetric and the sample reports multiple, unsymmetrized HH, one can recover MRs using Section 3.4.1 and construct IVs using Section 3.3.2. | Likewise, if the actual GG is asymmetric and the sample reports multiple, unsymmetrized HH, one can recover MRs using Section 3.4.1 and construct IVs using Section 3.3.2. | If the actual GG is symmetric and the sample reports a single, unsymmetrized HH, one can recover MRs using Section 3.4.2 and construct IVs using Section 3.3.1. | If the actual GG is symmetric and the sample report multiple, unsymmetrized HH, then one can recover MRs using either approach in Section 3.4, and construct IVs using either approach in Section 3.3. | The identification method of the previous section can be readily modified to recover the misclassification probabilities in the case with a single, unsymmetrized measure HH when the actual GG is known to be symmetric with undirected links. | C |
By using this reward function, the model learns to manage straddle option positions effectively, maintaining stability while responding to market fluctuations. The delay-reward and stop-loss mechanisms benefit in filtering out noise and focusing on significant market movements. | It is found that the Transformer-DDQN proposed in this paper outperforms the baseline methods based on transaction price direction on various performance indicators. Due to the high volatility of the A-share market, once the trading direction is wrong, large losses are encountered. Therefore, Transformer-DDQN turns to ... | It is also found that the performance of Transformer-DDQN varies across different datasets. In the A-share market, its profitability for the SSE 50 and CSI 300 is weaker compared to that for the CSI 500. This is because the primary components of the SSE 50 and CSI 300 are large-cap blue-chip stocks with lower volatilit... | The 15-minute candlestick data from January 1, 2018, to December 31, 2021, is designated as the training set, while the data from January 1, 2022, to March 31, 2024, is used as the testing set. The model is configured to look back at historical data over a period of 20 days, and historical volatility is calculated base... | The data used in this paper are the main broad-based indices traded on the Shanghai Stock Exchange(SSE), specifically the SSE 50, CSI 300, and CSI 500. To test the generalizability of the method, experiments are also conducted on Brent crude and Bitcoin datasets. We collect 15-minute candlestick data for these assets f... | D |
Third, the Optimal-α\alpha LRP calculates αt\alpha_{t} prices assuming that customers can accurately follow their target load profiles provided by the DSO. If customers have binding constraints in their optimization preventing this, the DSO’s optimization should be updated to endogenize these constraints and recalculat... | Finally, since the DSO calculates prices based on the optimal load profile for each customer, the DSO could measure a customer’s deviation from this ideal profile. This deviation could be the basis for performance metrics that compensate the customer for participating in the LRP tariff. | The optimal-α\alpha LRP calculates the minimum 𝜶\boldsymbol{\alpha} required at each time period for customers to shift load to follow a target profile determined through the DSO’s optimization. However, there are several issues that can arise with the optimal-α\alpha LRP. First, as seen in Case Study II, optimal-α\al... | If the DSO can accurately forecast each customer’s total energy demand over an optimization horizon, they can use the results to calculate α\alpha coefficients in the LRP. For example, the DSO use an optimal powerflow analysis to determine the optimal load profile for controllable loads on a distribution feeder. This “... | In contrast to the IR-LRP where the DSO creates prices to curtail load at its peak, the Optimal-α\alpha LRP allows a DSO to incentivize a customer to follow a specific load profile. In this case, we assume the DSO knows the customer’s energy needs and wants the customer to consume and discharge energy at the levels in ... | A |
Specifically, it employs the correlation integral C(r)C(r), which measures the fraction of point pairs | For comparison, we computed rolling volatility estimates and standardized returns using fixed windows | To extend this to rare-event detection, we incorporate volatility diagnostics based on recursive Fibonacci windows (21,34,55,8921,34,55,89). | with tail-sensitive volatility detection.This confirmed that the geometric approach anchored inference in baseline attractor structure. | Model B (Correlation–Integral with Fibonacci diagnostics) emphasizes recurrence statistics and volatility bursts, | B |
Outcome: Fast adoption among unskilled workers (i.e. non-artists) at the risk of professional artists. | We first consider a short-run problem, investigating how the worker’s level of investment in the creation of their AI twin depends on intrinsic task properties—specifically, (i) how easily a human worker can modify AI output to reach a desired quality level, which we term the editability of the task, and (ii) whether t... | Our results from a single period model show that a strategic worker will invest in training their AI twin only when this training increases the AI twin’s efficacy growth faster than that of the human and AI working in conjunction and will underinvest in other settings, trading off improvements in output quality and eas... | We show that the tradeoffs associated with a strategic human’s incentives to actively contribute to the creation of better generative AI systems are real. Even when one’s AI twin enables the production or higher-quality output or a reduction in effort, incentives to improve this twin are often constrained by the risk o... | We show that the tradeoffs associated with a strategic human’s incentives to actively contribute to the creation of better generative AI systems are real. Even when one’s AI twin enables the production or higher-quality output or a reduction in effort, incentives to improve this twin are often constrained by the risk o... | B |
Our paper joints these two avenues of research about leadership and organizations by showing that the desired characteristic of a leader should facilitate information transmission. The leader (Sender) has relevant information about the payoff from a team project while the followers (receivers) individually decide wheth... | Our results show that being too benevolent as a leader is suboptimal for the team because the concern of maximizing team welfare exactly prevents herself revealing information to her followers. In fact, our model provides a rationale for organizations to choose a leader who shares a different objective with them, since... | Among the vast literature on leadership and organizations272727See Bolton and Dewatripont, (2011) for a survey., one avenue of research studies how a leader should strategically convey her information to followers to build a more efficient team (See Dewan and Myatt, (2008), Majumdar and Mukand, (2007) and Ferreira and ... | Our paper joints these two avenues of research about leadership and organizations by showing that the desired characteristic of a leader should facilitate information transmission. The leader (Sender) has relevant information about the payoff from a team project while the followers (receivers) individually decide wheth... | Though concealing bad information could be an optimal action for the leader ex post, followers will respond to leader’s strategic communication and are hardly benefited from it ex ante due to insufficient information provision in equilibrium. Interestingly, among the famous U.S. Army’s Eleven Leadership Principles, the... | D |
Intuitively, this happens for the following reason: For low beliefs, both players are almost certain that they are facing a strategic dishonest partner, and in that situation the structure of the game is the same as in the classical Prisoner’s Dilemma. In particular, the players’ strategies (as described by their thres... | Finally, we compute the probability of cooperation from the viewpoint of someone who cannot observe the players’ beliefs. This case allows us to compute and compare the unconditional probabilities that the strategic player chooses to cooperate under the two scenarios, with common and diverse beliefs, respectively. Prop... | Since by definition of π†\pi^{\dagger}, the range of beliefs for which common knowledge of π\pi increases the likelihood of cooperation is given by (0,π†)(0,\pi^{\dagger}), the decrease (increase) in π†\pi^{\dagger} can be interpreted as greater (lower) benefits to belief diversity—in terms of increasing the scope for ... | This raises the question whether belief diversity raises or lowers the scope for cooperation from an ex ante perspective—that is, from the viewpoint of an outside observer who does know π\pi. To that end, we consider the ex-ante probability of cooperation in the environment k∈{common,diverse}k\in\{\text{common},\text{d... | Based on that, we analyze the likelihood of cooperation, both from an ex post and an ex ante perspective. From the viewpoint of someone who can observe the players’ beliefs, we claim that belief diversity raises the likelihood of cooperation whenever those beliefs are higher than a certain value (Proposition 4.2). The ... | D |
In our data application, we observed that only disadvantaged patients are more likely to use ambulatory care as their disease severity increases, in sharp contrast with the non-disadvantaged. The most significant difference in ambulatory care use is among patients with mild conditions, with this gap narrowing as the se... | In this section, we present a model that explains a patient’s decision to utilize ambulatory care. The core of this decision lies in balancing the relative benefits and costs of seeking ambulatory care early versus opting directly for inpatient care. We will examine how this choice is influenced by the patient’s trade-... | The observed difference in utilization between disadvantaged and regular patients suggests they may prioritize treating current conditions and disease prevention differently. Specifically, disadvantaged patients might be more focused on financial savings and therefore may only use ambulatory care when their conditions ... | However, the design of these programs may lead to unintended consequences due to cross-price effects. Reducing cost-sharing for inpatient care can inadvertently discourage the use of other essential medical services. For example, ChandraAER2010 demonstrated that increasing cost-sharing for outpatient care led to fewer... | In our framework, patients weigh the trade-off between using ambulatory care early versus opting directly for inpatient care. We conceptualize individuals as having two primary motivations for accessing ambulatory care: reducing medical costs and preventing disease progression. Patients may prioritize disease preventio... | D |
The full set of forecast performance measures is reported in Table 2. Two main results emerge. First, our forecasts perform at least as well as standard reduced-form alternatives. More importantly, our methodology outperforms a range of physical models from the WMO, while additionally offering the advantage of incorpor... | Given the apparent gains from achieving conditions similar to those outlined under the optimistic scenario, we complement the real-time forecasting exercise with a counterfactual analysis that aims to examine how temperatures would have evolved if this optimistic scenario was in fact implemented earlier. Given the avai... | Taking the aforementioned into consideration, the primary objective of this study is to demonstrate the utility of VAR models in (ex-ante) forecasting temperature changes across different Shared Socioeconomic Pathways (SSPs) scenarios. Traditional forecasting methods rely on “internal” model-based assumptions about the... | Building on the pseudo-out-of-sample evaluation, we next turn to a real-time forecasting exercise, where in this instance we will be considering both the adverse and optimistic forecasting scenarios. In both cases we estimate our model from 1850 to 2023 and then impose the corresponding paths on the three emission vari... | On the other hand, if the world turned into “Taking-the-Green-Road” scenario instead – by following an optimistic scenario where ambitious climate mitigation strategies could be adopted with rapid global action to reduce emissions, transition to renewable energy, and implementation of sustainable policies – this would ... | C |
At the subnational level within the United States, Figure 2 (right) shows that the distribution is skewed as well. California dominates with nearly 30% of US-based entries, followed by New York (11.1%) and Michigan (8.2%). This reflects the strength of California’s high-tech clusters [30] and Detroit’s automotive indus... | First, we analyze the temporal development of Green AI patenting, examining trends in the number of filings over time. Second, we investigate the main assignees, focusing on their evolution, the concentration of patenting activity through top-10 shares and Gini index, and the geographic distribution of assignees at bot... | Instead, the evolution of Green AI patent assignees over different time windows reveals a clear shift from traditional industrial control and combustion technologies toward data-centric and energy-efficient domains. This shift was accompanied by the growth of a larger innovation ecosystem, with a few players holding in... | In conclusion, the country-level map highlights the strong presence in East Asia and Europe alongside the US, while the US state-level heatmap reveals a clear clustering along the coasts and in key industrial corridors. To deepen the analysis of international leadership in Green AI, we build country-level collaboration... | Figure 2: (left) Country-level distribution of Green AI patents, using assignee location as a proxy. The United States is obscured in grey to better visualize activity in other countries. The heatmap reflects the relative share of non-US patents across countries. (right) US-level distribution of Green AI patents, using... | C |
R(Ei)R(E_{i}): The risk cost associated with the actor’s effort, including operational security, physical danger, doxxing, or platform bans. | where f(A)f(A) is the benefit from attention (e.g., a concave function to capture diminishing returns [40]), T(E)T(E) is time cost, and R(E)R(E) is risk cost (which may grow rapidly with increased effort). We assume α,β,γ>0\alpha,\beta,\gamma>0. | where βi\beta_{i} and γi\gamma_{i} vary depending on actor role, with higher values for in-theater contributors. This refinement acknowledges that soldiers and civilians face distinct constraints and rewards in the OSINT ecosystem. | Further, the risk function R(Ei)R(E_{i}) should be segmented by actor type. A soldier uploading a drone video from the front line incurs a radically different risk than a remote analyst reposting that clip hours later. We revise the utility function to reflect actor asymmetry: | γ\gamma: A coefficient for risk aversion. Front-line OSINT producers (e.g., soldiers filming GoPro footage) likely have higher γ\gamma than remote analysts. | D |
Our result in the TU market applies to a duopoly membership market. For example, consider a Chinese consumer who rarely cooks and thus subscribes to a meal-delivery service from one of the two dominant platforms, Meituan or Ele.me. In this scenario, memberships function as complements rather than substitutes for the pl... | From the two examples above, we can conclude that stability and setwise stability are independent concepts. Stability is a suitable solution criterion for economies in which agents make decisions independently, whereas setwise stability is better suited for environments in which agents are more cooperative. | If each central agent is required to pick one side in the NTU market, the outcome produced by our algorithm is also setwise stable. Setwise stability—introduced by Sotomayor (1999)—provides a criterion for many-to-many and multilateral matching that precludes blocking coalitions in which the members are more cooperativ... | A group of agents can implement a setwise block if they can renegotiate to a new outcome that is better and individually rational for all participants. In this new outcome, the agents involved in the renegotiation may not obtain their best choices from the newly signed contracts and the original contracts, as illustrat... | This result and Theorem 1 imply that a stable outcome—which is also setwise stable—exists when contracts are same-side complementary for all agents and each central agent has to pick one side. | B |
This is useful in case this information is already known. In Section 4 we use them to replicate analyses from the literature. | To illustrate transient analysis, we consider the “Artificial Anasazi” model introduced by Dean et al. (2000); Axtell et al. (2002) and further explored by Janssen (2009). We use the NetLogo implementation from Stonedahl & Wilensky (2010). The model simulates population dynamics in northeastern Arizona between 800–1350... | In this section,, we apply our transient analysis techniques to the “Artificial Anasazi” model originally proposed by Dean et al. (2000) and Axtell et al. (2002), and further explored by Janssen (2009). For all our applications, we use the NetLogo implementation provided by Stonedahl & Wilensky (2010).555The model is a... | Examples of transient properties are given in Section 3, where we study the average number of households alive for each of the 550 time points of interest for the artificial Anasazi model (Dean et al. 2000; Axtell et al. 2002). Here, a transient analysis is necessary because the model was proposed to replicate pointwis... | More in detail, the integration of MultiVeStA and NetLogo has been made possible using the NetLogo Java APIs 333https://ccl.northwestern.edu/netlogo/2.1/docs/controlling.html. These APIs allow MultiVeStA to access NetLogo programmatically, enabling the support for reset, next and eval. Any NetLogo model is now natively... | B |
The economic effects of natural disasters, and hurricanes especially, are well documented. Belasen and Polachek [1,8] establish the initial evidence, showing how hurricanes substantially change local labor markets, inducing long-lasting wage and employment changes. Their results were reinforced by follow-up research re... | Along with labor market repercussions, hurricanes impose deep effects on household finances and fiscal stability. Gallagher and Hartley [3] study the wake of Hurricane Katrina, indicating major and lasting instabilities in individual household finances. Deryugina [2] augments these conclusions, illustrating the fiscal ... | Gallagher, J., and Hartley, D. Household Finance after a Natural Disaster: The Case of Hurricane Katrina. | Collectively, this existing body of studies show clear patterns and persistent impacts of hurricanes on local labor markets, household finances,and broader economies, while also illustrating the ever changing nature and complexity of recovery processes. My study adds to this literature by giving detailed sector-specifi... | Natural disasters impose serious and often long-lasting changes on local labor markets [1, 2]. Previous studies show that hurricanes can lead to wage and employment volatility [1,4]. Understanding how local economies respond to these shocks is essential for designing effective disaster recovery policies. In particular,... | A |
The rapid advancement of artificial intelligence and automation technologies has led to increasing integration of service robots into traditional offline service environments. [88] reports that autonomous delivery, automated and intelligent assistants are rapidly entering industries such as retail, logistics and food s... | The increasing integration of service robots into traditional offline environments raises important questions about how consumers make service type choices in human-robot coexistence settings. Based on prior research in service marketing and technology adoption, we propose that consumer choice between human and robot s... | Prior research has primarily examined human-robot interaction in task execution, focusing on the roles of experts, workers, or consumers in conjunction with AI systems [59, 79, 68, 87] as well as human responses to AI-powered robots relative to traditional human or rule-based service providers [70, 54, 48, 83]. However... | Much of this literature has studied digital or simulated chatbots, recommender systems, or automated work assignment tools, where users passively receive outputs [70, 54, 71]. Relatively little is known about how users make active choices between robot and human service options. [71] is closely aligned with our researc... | The rapid advancement of artificial intelligence and automation technologies has led to increasing integration of service robots into traditional offline service environments. [88] reports that autonomous delivery, automated and intelligent assistants are rapidly entering industries such as retail, logistics and food s... | B |
Such multiplicity of beliefs has been extensively studied in decision theory; moreover, it is well known to be closely related to violations of rationality conditions—completeness and transitivity—assumed in the SEU model. | Indeed, Bewley’s (2002) seminal paper showed that violations of completeness can explain the multiplicity of probability distributions in DMs’ minds. | Lehrer and Teper showed that the key difference from the standard SEU model can be explained by violations of transitivity. | Rather, DMs facing ambiguity have multiple probability distributions in their minds and compare ambiguous prospects using them. | Indeed, if ℙ={𝒫}\mathbb{P}=\{\mathcal{P}\}, the generalized Bewley preference (u,ℙ)(u,\mathbb{P}) can be viewed as the Bewley preference (u,𝒫)(u,\mathcal{P}). | A |
(i) Implementation by efficient defection. With bounded actions, the max-claim action is a weakly dominant strategy; the induced outcome reproduces the cooperative payoff vector (Lj)j(L_{j})_{j} and is budget balanced when X≥IX\geq I. (ii) Robustness to collusion. At the dominant-strategy profile, no coalition can Pare... | Coalition-proofness. We adopt the coalition-proof Nash framework of Bernheim et al. [1] and show the dominant-strategy outcome is Strong Nash under TU, hence coalition-proof, because coalition-generated surplus is diluted proportionally to overage, limiting the coalition’s net gain. | The paper intersects three literatures. First, in the claims/rationing tradition (bankruptcy and uniform rationing), proportional rules are classically justified by axioms such as anonymity, consistency, and resource monotonicity [see, e.g., 2, 3, 4, 5, 6]. Our mechanism is noncooperative, budget balanced under scarcit... | Congestion/CPR and networks. Proportional sharing appears in congestion control and progressive-filling allocations [e.g., 7, 8]; those models rely on prices and potential-game structures. We instead give a direct, price-free mechanism with dominance under bounds and coalition-proofness at equilibrium. | Claims, bankruptcy, and rationing. Classical bankruptcy/claims problems allocate a fixed estate to claimants under axioms such as anonymity, consistency, and resource monotonicity; proportional and related rules are characterized in this tradition [4, 5, 6, 3, 2]. Our setting differs: actions are strategic, cooperators... | B |
The results in Section 3 can be extended to mi>1m_{i}>1 peers. However, if we let the number of peers an endogenous choice, identification becomes infeasible for two reasons. First, although the treatment assignment Z={Zi}i=1nZ=\{Z_{i}\}_{i=1}^{n} to all individuals is random, individuals with more peers are expected t... | for di=0,1d_{i}=0,1, and d(i)∈{0,1}mid_{(i)}\in\{0,1\}^{m_{i}}. This implies that individuals who are more likely to receive treatment also have better potential outcomes. | Because of the challenging identification problem, this assumption is common for the identification of spillover effects with multiple peers (Kang and Imbens, 2016; DiTraglia et al., 2023; Kormos et al., 2025; Vazquez-Bare, 2023). One-sided noncompliance is also common in treatment evaluation, with many empirical setti... | The results in Section 3 can be extended to mi>1m_{i}>1 peers. However, if we let the number of peers an endogenous choice, identification becomes infeasible for two reasons. First, although the treatment assignment Z={Zi}i=1nZ=\{Z_{i}\}_{i=1}^{n} to all individuals is random, individuals with more peers are expected t... | This violates 1.2 if individuals with more peers systematically differ in their outcomes from individuals with less peers. Borusyak and Hull (2023) discuss this identification problem in more detail. | D |
Figure 2 compares the expected payoff of attacking (Equation (1)) in the group stage and the incomplete round-robin league phase of the UEFA Champions League. In both tournament designs, two different prizes (see Section 3.2) and 12 different types of matches (see Section 3.3) are considered. Games played by teams draw... | Crucially, ℐ8\mathcal{I}^{8} remains consistently above ℐ2\mathcal{I}^{2}, the incentives are stronger by 70% to 200% with respect to the first prize. Similarly, ℐ24\mathcal{I}^{24} is higher than ℐ3\mathcal{I}^{3} for all types of matches, although the advantage of the league phase is smaller as it varies between 14% ... | According to the numerical results, offensive play is more encouraged in the new competition format. Compared to the old design, the incentives are stronger by 119% (58%) on average for obtaining the first (second) prize. The rise always exceeds 13% and can approach 200% depending on the type of match. | Analogously, ℐ24\mathcal{I}^{24} in the league phase consistently exceeds ℐ3\mathcal{I}^{3} in the group stage. Therefore, the same relation holds with respect to the second prize, and the advantage of the league phase varies between 13% and 106%. | ℐ8\mathcal{I}^{8} in the league phase is consistently higher than ℐ2\mathcal{I}^{2} in the group stage: it is more beneficial to play offensively in the league phase if the aim is to obtain the first prize. The difference is quite high; the incentives are stronger by 57% to 153% depending on the match type. | D |
The benchmark sample includes five Canadian variables and four foreign variables that together capture economic activity, prices, monetary policy, and financial conditions. For Canada, we use the nominal exchange rate (against the U.S. dollar when analyzing Fed shocks and against the euro when analyzing ECB shocks), th... | As a robustness check, we re-estimate the IRFs using the “Poor Man’s Sign Restriction” (PMSR) of Jarocinski [2020], which classifies policy announcements solely by the sign of the co-movement between interest rate and equity surprises: negative co-movements are interpreted as monetary policy shocks, while positive co-m... | Because this procedure delivers a range of admissible decompositions, we follow Jarociński [2022] in using the median rotation as our benchmark definition of a pure monetary policy shock. Figures 3 and 4 display the resulting shock series for the ECB and the Fed. As a robustness check, we also implement the “Poor Man’s... | Our benchmark identification strategy follows the high-frequency approach of Jarociński and Karadi [2020], which separates monetary policy shocks from central bank information effects by examining the joint behavior of asset price surprises around policy announcements. The logic is straightforward: a contractionary mon... | To recover the underlying structural shock, the identification strategy uses the rotational–angle decomposition proposed by Jarociński [2022], which separates monetary policy shocks from central bank information shocks using high-frequency asset price surprises around policy announcements. The intuition is that movemen... | C |
Bias(δ~j)=ρY𝔼[DitjYi,t−1]=ρY𝔼[Yi,j+1]={0,j=0,ρYδ0,j=1,ρYδ1+ρY2δ0,j=2.\mathrm{Bias}(\widetilde{\delta}_{j})=\rho_{Y}\mathbb{E}\left[D_{it}^{j}Y_{i,t-1}\right]=\rho_{Y}\mathbb{E}[Y_{i,j+1}]=\begin{cases}0,&j=0,\\ | Thus, even if the true effect at j=0j=0 is identified without bias, biases accumulate at longer horizons, distorting the entire treatment path. | A simple, yet revealing, illustration shows why excluding lagged dependent variables from an event study regression generates omitted variable bias in the estimated treatment effect path. Consider a panel with five periods (t=0,1,2,3,4t=0,1,2,3,4) and a common treatment occurring at t=2t=2, so that Ditj=𝟏{t−j=2}D_{i... | Notes: The black dashed line shows the true treatment effect path (δ0,δ1,δ2)=(1,1.2,0.5)(\delta_{0},\delta_{1},\delta_{2})=(1,1.2,0.5), while the blue solid line shows the estimated treatment effect path of {δ~j}\{\widetilde{\delta}_{j}\} from a naive event study regression without lagged dependent variables. The blue ... | The second class of estimators allows for unit-specific dynamic responses as in (1). We consider the following four heterogeneous treatment effect estimators, which differ in how they recover the marginal density of the sufficient statistics p(λ^∣Y0)p(\widehat{\lambda}\mid Y_{0}) in Tweedie’s formula (7). The oracle e... | A |
Ξ(μ)=q+H0+ΔH⋅P(μ),ΔH≡H1−H0≥0.\Xi(\mu)\;=\;q\;+\;H_{0}\;+\;\Delta H\cdot P(\mu),\qquad\Delta H\equiv H_{1}-H_{0}\ \geq 0. | The analytical structure that underpins the results is deliberately light: engagement outcomes during testing are i.i.d. conditional on quality, the graduation event is defined by a count threshold, and the private and planner problems depend on the binomial tail P(μ)P(\mu) and its slope P′(μ)P^{\prime}(\mu) through ... | While the model is stylized, the prescriptions are robust and managerially legible. The results extend to over-dispersed or correlated outcomes as long as the graduation statistic admits a smooth tail and density; they accommodate exploitation engines implemented by indices or sampling by reinterpreting HH as an expect... | Expression (21) nests the block model with H0=0H_{0}=0 and H1=HH_{1}=H. It also nests index or Thompson engines in which “failure” still earns some tail exposure, e.g., through long-tail sampling or periodic resets, provided that admission to a promising pool (or to a higher-priority band) is governed by a threshold on... | Two practical remarks help with calibration. First, H0H_{0} and H1H_{1} are not free parameters; they are induced by the chosen index or sampling policy and can be estimated ex ante by simulating the engine on historical posterior paths. The resulting (H0,ΔH)(H_{0},\Delta H) summarize the continuation landscape for po... | C |
Notes: Statistics computed over N=11,463 households. Statistics weighted by household and purchase period-specific sample weights - see Appendix Section A.3.2 for the exact formula. Implicit tax rate = 100 x tax revenue / (sales - tax revenue). | Finally, the combination of substitution effects and competition effects explains the changes observed in the contribution of each alcohol category to total household purchase volumes (in g of ethanol), displayed in the lower panel of Table 9. Unsurprisingly, the share of spirits increases sharply in the first three sc... | We consider several reforms of price regulation policies. In the current tax system, wines are almost exempt from specific taxes, taxes on beers and ciders are low, and spirits bear most of the burden. We test replacing current taxes by two alternative volumetric tax systems as in [20]: a uniform tax rate based on the ... | The differences between scenarios can be explained by changes in the market size of the overall alcohol market and by substitutions between alcohol categories, and thus transfers from one market to another. The current tax system is highly favorable to wines (see Table 2). Any reform based on the alcohol content of pro... | In general, the current price regulation system does not align with public health objectives. The third panel of Table 2 also shows that the different alcohol categories are subject to very different tax burden due to the differences in the specific volumetric taxes currently applied. The implicit tax rate, correspondi... | D |
Fτ(p)=cl{(v¯I,v¯)∈Cτ(p):∄(v¯I′,v¯′)∈Cτ(p) s.t. (v¯I′,v¯′)≻(v¯I,v¯)}F_{\tau}(p)=\mbox{cl}\{(\underline{v}_{I},\overline{v})\in C_{\tau}(p):\nexists(\underline{v}_{I}^{\prime},\overline{v}^{\prime})\in C_{\tau}(p)\mbox{ s.t. }(\underline{v}_{I}^{\prime},\overline{v}^{\prime})\succ(\underline{v}_{I},\overline{v})\}... | Unlike in the previous sections, here the frontier is itself endogenous. That is, the designer chooses both the frontier’s shape (via τ\tau) and also the point to implement on it (via μ\mu). We show that no matter the designer’s α\alpha, they will always choose a τ\tau that leads to the frontier taking the form of a si... | We have so far supposed that the designer believes that any relationship between the additional attributes and the need for treatment is possible. Section 6 extends our model to allow the designer to choose bounds on how predictive the additional attributes are. We show that our original characterization of the frontie... | We show that the τ\tau-efficient frontier qualitatively resembles the previous frontier, with a few key differences. To avoid repetition, we present results here for p<1/2p<1/2 (with the mirrored case reported in Appendix E.5). | The last step is to derive the frontier by considering all possible Bernoulli variables X1X_{1} and the resulting best-case and worst-case payoffs as derived in Step (1). We show that when p<1/3p<1/3, the point implemented by X1∼Ber(q)X_{1}\sim\mbox{Ber}(q) is undominated if and only if q∈[0,p]q\in[0,p], while if p∈(1... | C |
Matching Bandit problem: liu2020competing aimed at developing a regret minimization scheme to achieve player-optimal matching. In regret minimization, several extensions have been studied, decentralized algorithms in which agents make decisions only based on local information liu2021decentralized ; maheshwari2022decen... | We introduced a stable matching identification problem and designed computationally efficient Top-Two algorithms. We considered two-learning setups, one-sided learning in which we presented fluid dynamics that the algorithm tracks and prove that it is also asymptotically optimal. For two-sided learning, we proposed an ... | Matching Bandit problem: liu2020competing aimed at developing a regret minimization scheme to achieve player-optimal matching. In regret minimization, several extensions have been studied, decentralized algorithms in which agents make decisions only based on local information liu2021decentralized ; maheshwari2022decen... | Novel Algorithms: We design two practical Top-Two-style algorithms: (i) the Anchored Top-Two (ATT) algorithm, which leverages the Karush-Kuhn-Tucker (KKT) conditions of the optimal allocation, and (ii) the β\beta-Top-Two algorithm, applicable to both one-sided and two-sided learning settings. | One-Sided and Two-Sided learning: As in the existing literature, we study two problem settings: one-sided learning in which one side knows the preferences, and two-sided learning where both sides are unaware of their preferences. Based on practical applicability, either a one-sided or a two-sided learning model is vali... | D |
As business leaders look to the future, the pace of change will only accelerate. The next frontier of AI will further empower agile businesses: | 24/7 Intelligent Support: Modern AI-powered chatbots can handle a significant portion of routine customer inquiries—reports suggest up to 80%—without human intervention ColorWhistle (2025). They can answer frequently asked questions, track order statuses, and book appointments around the clock. This frees up human agen... | As business leaders look to the future, the pace of change will only accelerate. The next frontier of AI will further empower agile businesses: | The fundamental shift for business leaders to grasp is the move from a world of explicit programming to one of probabilistic training. Traditional software required developers to write precise ‘if-then‘ rules to cover every conceivable situation, a process that is rigid and cannot scale to handle the complexity of real... | AI Agents: The evolution from simple chatbots to autonomous AI agents is already underway. These agents will be capable of performing complex tasks on behalf of employees—proactively scheduling meetings, managing sales outreach, and even ordering supplies—further augmenting the capabilities of small teams Microsoft (20... | D |
In parimutuel betting markets, final odds and interim odds are calculated in the same way—based on the cumulative distribution of wagers—but they serve markedly different functions. Final odds are determined at the close of betting and directly govern payout calculations, thereby playing a dual role as both a summary o... | The contribution of our study is to move beyond this convention by testing whether expected returns depend not only on final odds but also on the trajectory of odds movements during the betting period. To this end, we extend the standard regression framework used to detect the FLB—where realized returns are regressed o... | In summary, the regression results demonstrate that expected returns are systematically related not only to the level of final odds but also to the trajectory of interim odds, particularly in the final minutes before post time. While the conventional specification in Column (1) detects the favorite-longshot bias, incor... | To address our research question, we empirically test the validity of inferences that rest on the assumption that final odds serve as sufficient statistics for bettors’ decisions. More specifically, we analyze the entire evolution of odds throughout the betting period to assess whether outcomes depend not only on final... | This specification highlights the three parameters of interest. The intercept, α\alpha, represents the baseline expected return in the absence of variation in odds, while β\beta traces how expected returns vary with the level of final odds, thus testing for the conventional FLB. The coefficient δ\delta isolates the cor... | A |
At the one-year horizon the picture flips. Horseshoe takes over. In 2020-21, its RMSE falls to 1.693 (gain 18.2%) and its CRPS improves upto 20.6%, while FAVAR slips back toward baseline. By 2022-24 the pattern is even clearer when RMSE records of UH-HS is 2.809 (outperforms the benchmark upto 29%) with DFM second at 3... | As the horizon lengthens the advantage of iterating fades and shrinkage gains importance. At h=12h=12 the ultra-high-dimensional horseshoe is the most reliable tail-risk forecaster. It posts the best left QWS at 2.013 with a 0.187 skill gain and the best right QWS at 2.059 with a 0.155 gain. It also leads on the tails ... | The density results deserve emphasis because they confirm that the ranking is not driven only by point targeting. CRPS is a strictly proper scoring rule that integrates the distance between the predictive distribution and the outcome. Lower is better because the score rewards both sharpness and calibration (Gneiting an... | Taken together the numbers support a pragmatic division of labor that lines up with economic theory. When common components dominate and information flows quickly through the macro system, factor compression with iterated dynamics wins. When the horizon stretches and the risk of overfitting rises, sparse priors that ke... | Next we move the interpretation to Quantile-weighted scores, where reinforce this long-horizon hand-off: UH-HS posts left- and right-tail improvements around 0.18-0.23 during 2020-21, exactly when risk calibration matters most. This horizon-specific performance fits both the Bayesian shrinkage literature (Carvalho et a... | D |
Bill differences are decomposed for different consumer groups, namely BEV consumers (by charging strategy), electrolyzers and other consumers (Germany [DE] and other countries [non-DE]). For each group, the difference is calculated relative to the bill that this group would have to pay in the 100% inflexible setting. F... | Our findings reveal that moderate shares of bidirectional charging (V2G) of less than 30% can lead to lower system costs than a fully smartly charging BEV fleet in a central European 2030 scenario with 15 million BEVs in Germany. At a fleet share of 50% and beyond, V2G even leads to overall system costs savings compare... | Further analyses of model results show that the overall system costs savings of bidirectional charging are unevenly distributed across different types of electricity consumers. Interpreting the dual of the model’s hourly energy balance as the wholesale price[50], we compute the average yearly amount paid for electricit... | To further illustrate the effects of BEV flexibility on different consumer groups, Figure 8 shows for selected cases how the aggregate electricity bill of each group changes when part of the fleet switches to more flexible operations, compared to a setting where the whole fleet charges inflexibly. It illustrates how th... | We further find that when half of the BEV fleet engages in V2G, the overall system costs are even lower than in the case where there are no BEVs at all. With 50% of BEVs charging bidirectionally and the remaining 50% charging inflexibly (our setting labeled as “0.5-0.0-0.5”), the cost difference to the reference is neg... | C |
Likewise, the requirement that hh is monotone can be relaxed. A voter will never be interested in investing effort if its cost is higher than her value of information. Hence, the equilibrium amount of effort of voter ii will not exceed c−1(vi)c^{-1}\left(v_{i}\right). At the same time, if ψ′(x)c′(x)\frac{\psi^{\prim... | In a classic game known as the volunteer’s dilemma, several players choose whether to make a contribution to a public good by paying some cost bb. The public good is provided if and only if at least one player contributes. The Nash equilibrium makes two predictions, fairly robust to various perturbations of the game, a... | Intuitively, because information is a public good, an increase in connectivity causes each voter to invest less in acquiring information. As a result, each voter has more peers from whom to receive a signal, but each of these peers is less likely to acquire the signal. Which of these opposing effects is stronger? The p... | In the model developed in this paper, investing effort in acquiring information is similar to contributing to a public good, which is provided if at least one individual in the social network of a given voter succeeds in acquiring a signal. In a standard volunteer’s dilemma, the public good is provided with certainty i... | Finally, the paper is also related to the literature on threshold public goods, that is, public goods that are provided if the amount of contributions reaches a certain exogenous threshold (see Palfrey and Rosenthal, 1984, for a classic reference). A specific type of a threshold public good game is the volunteer’s dile... | A |
We study the impact of LLMs on employment and earnings through a panel of occupations by month-year. Our strategy begins with the canonical difference-in-differences (DiD) design and extends it with the synthetic difference-in-differences (SDiD) estimator of Arkhangelsky et al. (2021) to address potential deviations fr... | Formally, let Y¯𝒯,t\bar{Y}_{\mathcal{T},t} denote the average outcome of the treated group at time tt, and Y¯𝒞,t(ω)=∑j∈𝒞ωjYjt\bar{Y}_{\mathcal{C},t}(\omega)=\sum_{j\in\mathcal{C}}\omega_{j}Y_{jt} the weighted average of controls. The unit weights are chosen by solving | In line with the raw trends in Figure 2, the post-treatment period shows a widening earnings gap between high-exposure and low-exposure occupations. Estimates from the baseline DiD specifications are presented in Table 1 in the Appendix. The results show sizable positive effects of LLM exposure on weekly earnings and u... | with occupation fixed effects αi\alpha_{i} and time fixed effects δt\delta_{t}. Let Yi(0)Y_{i}(0) and Yi(1)Y_{i}(1) be the potenital outcome for occupations with low exposure and high exposure respectivlty, then The coefficient β\beta identifies the average treatment effect on the treated (ATT), 𝔼[Yi(1)−Yi(0)|i∈�... | Let ii index occupations and tt time periods. Outcomes of interest YitY_{it} are either weekly earnings or the unemployment rate. We define treatment as | D |
Φ(ν):=∫T∫X^IF^(t,x)ν(t,dx)𝑑λ(t),ν∈ℛℱ(X^).\Phi(\nu)\;:=\;\int_{T}\!\int_{\widehat{X}}I_{\widehat{F}}(t,x)\,\nu(t,dx)\,d\lambda(t),\qquad\nu\in\mathcal{R}^{\mathcal{F}}(\widehat{X}). | is lower semicontinuous for the weak topology of ℛℱ\mathcal{R}^{\mathcal{F}} [2, Theorem 2.2 (a)]. Hence, by Lemma 8, | Then (μ^α)α(\widehat{\mu}_{\alpha})_{\alpha} is a net in ℛ𝒢(X^)\mathcal{R}^{\mathcal{G}}(\widehat{X}). | Since ℛ𝒢(X^)\mathcal{R}^{\mathcal{G}}(\widehat{X}) is compact in the weak topology [2, Theorem 2.3 (a)], there exists a subnet | Then Φ\Phi is weakly lower semicontinuous on ℛ𝒢(X^)\mathcal{R}^{\mathcal{G}}(\widehat{X}); see [2, Theorem 2.2 (a)]. | D |
We focus on the case where society consists of two999We view the case of two groups of agents with different models as | Definition 2 to the case of more than two groups. observably distinguishable groups of agents, A and B, who may behave | the stage game but may have misspecified beliefs about others’ strategies. We will no longer work in the special case of strategic | resolve the same fundamental uncertainty about the environment.141414We note that play between two groups gg and g′g^{{}^{\prime}} is not a Berk-Nash | We focus on the case where society consists of two999We view the case of two groups of agents with different models as | A |
End of preview. Expand in Data Studio
README.md exists but content is empty.
- Downloads last month
- 58