method string | state_count int64 | rows list |
|---|---|---|
CFPB HMDA 2025; FHA loan_type 2; decisioned = action_taken 1,2,3; denial = action 3. Small <$150K, large >=$250K, min 100 per cell. | 52 | [
{
"state": "TX",
"decisioned_applications": 131577,
"denials": 20711,
"denial_rate_pct": 15.7,
"small_loan_denial_pct": 25.5,
"big_loan_denial_pct": 14.9,
"small_loan_penalty_x": 1.71
},
{
"state": "FL",
"decisioned_applications": 120403,
"denials": 29257,
"denial_rat... |
FHA Denial Rates by State — with loan-size splits (2025)
All 50 states plus DC and PR, from the complete 2025 CFPB HMDA record: FHA denial rates, and the split that almost nobody publishes — small loans versus large loans, state by state.
The finding this dataset carries
Small FHA applications (under $150K) are denied more often than large ones ($250K+) in every state measured. The gap is widest where it is least expected:
| State | Small-loan denial | Large-loan denial | Penalty |
|---|---|---|---|
| Idaho | see data | see data | 3.19× |
| New Hampshire | see data | see data | 2.74× |
| Utah | see data | see data | 2.72× |
In the ten states requiring the highest incomes to buy, 42.1% of small FHA applications were denied, versus 35.1% in the ten cheapest — the opposite of the common assumption. Story: https://financeratecalc.com/salary-vs-denial-risk-by-state.html
Files
fha_denial_rates_by_state_2025.csv— one row per state, withsmall_loan_penalty_xstate_fha_2025.json— same data plus method string
Methodology
FHA = loan_type 2. Decisioned = action_taken {1,2,3}; denial = action 3. Small <$150K, large ≥$250K; a cell requires at least 100 applications. Full rules: https://financeratecalc.com/methodology.html
Denial rates partly reflect applicant mix. Historical observations, not predictions and not evidence of wrongdoing.
Citation
Primary data: CFPB HMDA 2025 Snapshot. Derived rates: Yetiş, Z. (2026), FHA Denial Rates by State, FinanceRateCalc. https://financeratecalc.com — companion paper: SSRN 7156938.
CC BY 4.0 · Not a lender · No ads · A denial is a data point, not a verdict on the applicant.
- Downloads last month
- -