Key findings

• Black-white mortgage denial rate gaps are higher in lower-income, less-expensive housing markets.
• A meaningful share of the variation across markets is associated with black-white differences in applicant income and in the share of applicants carrying very high debt burdens.

Black applicants are denied mortgages at higher rates than white applicants in every one of the 50 largest U.S. metro areas, according to a Zillow analysis of Home Mortgage Disclosure Act data. But that gap is not necessarily the largest where home prices are highest. Instead, across markets, wider denial gaps are associated with larger financial differences between Black and white applicants before they ever walk into a lender’s office.

The expensive metros aren’t the worst offenders

The intuitive assumption is that unaffordable markets breed the worst racial inequities. That’s not what the data show. Across the 50 largest U.S. metro areas, the Black-white mortgage denial gap is actually narrower in high-cost metros and wider in lower-cost ones.

Markets with higher typical home values, higher median household income and larger average loan sizes tend to have smaller denial disparities. That’s not a clean bill of health for expensive metros — it just means the racial gap in who gets denied is more compressed there than in relatively more affordable markets.

The real dividing line is income and debt

Why are denial gaps wider in more affordable markets? The strongest associations point to the financial profiles applicants bring to the table — specifically income gaps and debt loads.

In metros where Black applicants earned significantly less than white applicants, the denial gap was wider. The same pattern held for debt burdens: in markets where Black applicants were more likely to carry debt-to-income ratios above 50%, the denial disparity was higher. Both of these relationships remain after accounting for local market conditions

Loan-to-value differences also tracked in the expected direction, but the effect was weaker and less precisely estimated than income and debt.

This matters because mortgage underwriting is, by design, a financial fitness test. Lenders weigh monthly payment obligations against income. When one group of applicants systematically enters that process with lower incomes and heavier debt burdens, wider denial gaps are more likely to emerge. The results here are descriptive, but they suggest that differences in applicants’ financial positions are closely tied to how wide mortgage denial gaps are across markets.

The effects of discriminatory economic policies, including redlining, likely persist today through neighborhood disinvestment in Black communities that led to fewer labor market opportunities, lower incomes and reduced credit access.

Where markets fit in

Local market conditions aren’t irrelevant. Home values do help explain where denial gaps land. But they work more as context than cause. In lower-value markets, households tend to have less financial cushion overall, and racial differences in income and debt tend to be sharper — making those differences more consequential at the underwriting stage. In high-cost markets, the barrier to buying is higher, but the financial distance between Black and white applicants applying for mortgages appears smaller.

That distinction matters for how policymakers and housing advocates frame the problem. Mortgage denial gaps are not primarily a story about predatory lending in expensive metros. They’re also a story about who has been able to build financial stability — and who hasn’t — long before a mortgage application is ever filed.

Black–White mortgage denial gaps, 2025

Methodology

This analysis draws on Home Mortgage Disclosure Act application-level data from 2023 through 2025, covering conventional, first-lien, home-purchase, non-reverse mortgage applications for principal residences across the 50 largest metro areas. Market-level data come from Zillow’s measures of typical home values and local income data. The analysis sample comprises 148 metro-years: all 50 metros in 2023 and 49 in each of 2024 and 2025.

The primary outcome is the black-white mortgage denial rate gap within each metro-year. The main regression, a cross-sectional metro-year specification, relates the Black-white mortgage denial gap to local market conditions and black-white differences in applicant financial characteristics. Standard errors are clustered at the metro level to account for correlation across years within the same market.

Appendix / robustness

The main model is designed to explain variation across metro areas rather than changes within metro areas over time. That means the results should be interpreted as descriptive associations, not causal estimates.

Several robustness checks support the main conclusions:

• Clustered standard errors: Inference is based on standard errors clustered by metro area rather than simple heteroskedasticity-robust errors.
• Minimum sample threshold: Requiring at least 50 Black and 50 White applications per metro-year does not drop any observations, indicating that the results are not being driven by thin cells.
• Low collinearity: Variance inflation factors are low, suggesting that the main regressors are not so highly correlated that coefficient estimates become unstable.
• In a model that adds metro fixed effects and identifies only off within-metro changes over time, the strongest relationship that remains is the black-white gap in the share of applicants with DTIs above 50%. The cross-sectional relationships with home values and applicant income are much weaker in that tighter specification.