← Back to research
Submitted · Journal of Housing Economics · March 2026

Persistent Racial Disparities in U.S. Mortgage Approval

Evidence from 42.3 million HMDA applications, 2020–2024. Five independent identification strategies — DFL decomposition, within-lender fixed effects, regression discontinuity, difference-in-differences, and partial identification bounds — all reach the same conclusion.

ProblemBlack mortgage applicants are approved far less often than White applicants — and a single regression can always be dismissed as missing some hidden variable.
ApproachTest the same question five independent ways, each with different assumptions and different ways it could fail, across 42.3 million applications.
ResultAll five agree: most of the gap is not explained by creditworthiness, and it's largest inside the same lender, not between lenders.
Significance≈126,000 fewer approvals a year for Black applicants than the data can otherwise account for.
42.3MHMDA applications
14.95 ppRaw Black–White gap
74.6%Within-lender variation
126,000Fewer approvals/year (est.)

Why This Question Matters

This paper measures how much of the racial gap in mortgage approvals survives once creditworthiness is accounted for, and tests the answer five independent ways. A mortgage is the primary vehicle for wealth accumulation in most American households — who receives credit shapes financial trajectories, neighbourhood composition, and the persistent racial wealth gap. A substantial literature has documented for three decades that Black households face systematically higher denial rates than observably similar White households. But a single regression can always be dismissed as an omitted-variable artifact.

This paper asks three questions: How large is the within-lender racial approval differential — and has it changed over a macroeconomically turbulent five-year period? Does it concentrate at specific institutional rule boundaries? How does it behave under conservative, partial-identification assumptions about unobserved creditworthiness?

Five Strategies at a Glance

01Compare what similar applicants got — controlling for income, loan amount, property value, and loan-to-value ratio.
02Compare Black and White applicants evaluated by the same lender, in the same year — to rule out applicants simply choosing stricter lenders.
03Look for a sharp jump in the gap right at a rule boundary (mandatory insurance above 80% loan-to-value) — a discontinuity that's hard to explain any other way.
04Check whether the gap moved when the Fed sharply tightened credit in 2022 — a real-world stress test of the lending system.
05Assume the most generous possible case for hidden creditworthiness factors, and check how much gap survives even then.

Key Result: Five Methods, One Answer

The raw Black–White approval gap is 14.95 percentage points. Even after reweighting on every characteristic HMDA observes, 98.6% of this gap remains. The estimated scale: 126,000 fewer annual approvals for Black applicants.
0 pp 4 pp 8 pp 12 pp 16 pp RAW GAP 14.95 pp DFL RESIDUAL 14.75 pp WITHIN-LENDER FE 11.14 pp PMI RDD EFFECT +2.0 pp MANSKI LOWER BD. ≥6.37 pp DiD WIDENING +1.5 pp post-2022

Five independent strategies, five different failure modes if any one were wrong — all point the same direction

Key Takeaways

14.95 pp
Raw Black–White approval gap
Persistent across the entire 2020–2024 window, surviving both pandemic boom and tightening cycle.
74.6%
Within-lender share
Most of the gap is within the same institution — not explained by sorting of Black applicants toward stricter lenders.
+2.0 pp
PMI threshold discontinuity
The gap is 2.0 pp larger above the 80% LTV institutional boundary, concentrated in purchase loans. Absent in refinancings.
126K
Fewer approvals per year (est.)
The scale of the gap translates to an estimated 126,000 fewer annual approvals for Black applicants — a figure that motivates documenting conditions under which the gap is largest.

The Data

The analysis covers a near-universe of mortgage applications filed with HMDA-reporting lenders between January 2020 and December 2024 — 42.3 million records spanning more than 5,500 lending institutions. This five-year window captures extraordinary macroeconomic variation: a pandemic-driven refinancing boom during 2020–2021 (federal funds rate at zero lower bound), followed by the sharpest monetary tightening cycle in four decades (rates rising to 5.5% by 2023, cutting application volume by more than half).

2020 2021 2022 2023 2024 PANDEMIC REFI BOOM · RATES AT ZLB FED TIGHTENING · RATES → 5.5% · VOLUME −50% 42,361,440 APPLICATIONS · 5,500+ LENDERS · ALL 50 STATES

Figure 1 — Sample coverage: 2020–2024, spanning two distinct macro regimes

The 14.95 pp raw gap is not a new finding — but its persistence across a five-year window spanning a pandemic boom and the sharpest credit tightening in 40 years is new. This paper's contribution is documenting the institutional structure of the gap in detail: where it concentrates, how it responds to rule boundaries, and how it behaves when unobservable creditworthiness is bounded.

Five Identification Strategies, In Full

The headline estimate is stress-tested from five independent directions. Each has different identifying assumptions and different potential failure modes — when they all agree, the result is no longer attributable to any single one.

01
DFL DECOMPOSITION (DINARDO–FORTIN–LEMIEUX)
Constructs a counterfactual approval rate for White applicants holding their observable characteristics at the Black applicant distribution. Controls for income, loan amount, property value, LTV ratio.
→ 98.6% of the 14.95 pp gap not captured by HMDA-observable characteristics
02
WITHIN-LENDER FIXED EFFECTS
Compares Black and White applicants evaluated by the same institution in the same year. Tests whether disparities arise from sorting of Black applicants toward stricter lenders — or from differential approval standards within the same institution.
→ 74.6% of the gap is within-lender variation; this share grew from 66.8% (2020) to 78.3% (2024)
03
REGRESSION DISCONTINUITY AT 80% LTV THRESHOLD
Exploits the sharp institutional rule at 80% LTV, above which private mortgage insurance (PMI) becomes mandatory. A valid RD compares applicants just below and just above this cutoff to detect a discrete change in outcomes for Black applicants at an institutional boundary.
→ Gap 2.0 pp larger above the cutoff; concentrated entirely in purchase loans, absent in refinancings
04
DIFFERENCE-IN-DIFFERENCES: 2022 TIGHTENING CYCLE
Exploits the Federal Reserve's 2022 tightening as a macroeconomic shock. DiD framework compares the evolution of the within-lender racial approval difference before and after this regime shift, with HonestDiD sensitivity analysis for pre-trend violations.
→ Within-lender differential widened by 1.5 pp after the tightening; structural break of −1.541 pp
05
MANSKI PARTIAL-IDENTIFICATION BOUNDS
Makes no functional-form assumption. Calibrates bounds to Survey of Consumer Finances (2022) data, simultaneously bounding FICO scores, DTI ratios, liquid asset reserves, and employment stability at their maximum plausible values.
→ At least 44–55% of the gap (6.37–7.97 pp) not explained even under maximally adversarial selection

Within-Lender Gap: Growing Over Time

Within-lender fixed effects reveal that nearly three-quarters of the gap (74.6%) is associated with variation within lending institutions — a share that has grown monotonically from 66.8% in 2020 to 78.3% in 2024. This pattern implies the gap is not primarily a story of Black applicants sorting toward stricter lenders. It is concentrated inside the same institution, often in the same year.

100% 80% 60% 40% 66.8% 2020 68.5% 2021 71.2% 2022 75.1% 2023 78.3% 2024 WITHIN-LENDER SHARE

Figure 2 — Share of racial approval gap attributable to within-lender variation, 2020–2024. Growing monotonically from 66.8% to 78.3%.

Institutional Rule Boundary: The 80% LTV Threshold

Above the 80% loan-to-value ratio, private mortgage insurance (PMI) becomes mandatory — a sharp institutional rule that creates a valid regression discontinuity design. Black applicants face an additional 2.0 pp approval differential above this threshold. This effect is concentrated entirely in purchase loans and is entirely absent in refinancings, providing a compelling within-sample test of the PMI mechanism.

60% 70% 80% 90% 100% 20% 10% 0% 80% LTV · PMI THRESHOLD +2.0 pp PURCHASE LOANS REFINANCINGS (NO EFFECT) LOAN-TO-VALUE RATIO RACIAL APPROVAL GAP

Figure 3 — Regression discontinuity at the 80% LTV threshold. Purchase loans (green) show a 2.0 pp jump; refinancings (grey, dashed) show no discontinuity. P-values supported by density tests and exact permutation inference.

Geographic and Institutional Structure

State-level approval gaps range from 3.6 to 23.5 percentage points, with regional means spanning 9.3 pp (West) to 16.1 pp (Midwest). County gaps are spatially correlated with HOLC redlining coverage — counties with greater historical D-grade ("redlined") coverage show larger contemporary approval gaps, conditional on observed characteristics.

MIDWEST 16.1 pp SOUTH 14.5 pp NORTHEAST 13.0 pp WEST 9.3 pp STATE RANGE: 3.6 pp – 23.5 pp

Figure 4 — Regional mean racial approval gaps. State-level gaps range 3.6–23.5 pp. Larger gaps spatially correlated with historical HOLC redlining coverage.

Critically, the within-lender racial differential does not vary with lender size or market concentration — it is essentially flat across lender-size quartiles and non-monotonic across HHI quartiles. This pattern is more consistent with underwriting practices that are relatively uniform across market structures than with individual taste discrimination, which Becker's (1957) competitive-discipline model predicts competitive markets should eliminate.

Partial Identification: What Remains Under Conservative Assumptions

The most credible existing estimate — Bhutta, Hizmo, and Ringo (2025) in the Journal of Finance — conditions on actual FICO scores, DTI, liquid assets, and automated underwriting recommendations and finds residual gaps of 1–2 pp. Our estimates using public HMDA data are larger, and we address this directly through partial-identification bounds.

Even when FICO scores, DTI ratios, liquid asset reserves, and employment stability are simultaneously bounded at their maximum plausible values calibrated to SCF 2022 data, the analysis finds that a gap of at least 6.37 to 7.97 percentage points44 to 55% of the observed gap — is not captured by these variables in this specification.

≥44%
Share of gap not explained under maximally adversarial assumptions about unobservables
Even simultaneously bounding FICO scores, DTI ratios, liquid asset reserves, and employment stability at their maximum plausible values calibrated to Survey of Consumer Finances (2022) data — a deliberately conservative exercise — at least 6.37–7.97 pp of the gap remains unexplained.

Paper Details

Full title: Persistent Racial Disparities in U.S. Mortgage Approval: Evidence from 42 Million Applications, 2020–2024

Venue: Submitted to Journal of Housing Economics, March 2026. JEL codes: G21, J15, G28, R21, R31.

Author: Rajveer Singh Pall (submitted anonymously via journal system)

Data: HMDA public use files (FFIEC), 2020–2024; Survey of Consumer Finances 2022 (partial ID bounds calibration); HOLC Mapping Inequality database (geographic analysis).

↗ GitHub (Rajveer-code)