Fair Lending in Crisis: Racial Disparities in FHA Mortgage Approvals Before and During COVID-19

The Promise and the Reality

The Federal Housing Administration was created in 1934 to make homeownership accessible to Americans who couldn't qualify for conventional loans. FHA loans require only a 3.5% down payment, accept credit scores as low as 580, and insure up to 96.5% of a home's value — explicitly designed for low-income and minority households who are shut out of the conventional mortgage market.

So why, in 2022, were Black and Hispanic borrowers still being approved at significantly lower rates than White borrowers — even for the loan product built for them?

This study uses five years of Home Mortgage Disclosure Act (HMDA) data from Massachusetts (2018–2022) to measure the racial gap in FHA loan approvals, track how COVID-19 affected it, and explore whether the disparity is worse in higher-cost housing markets.

Methodology

The study analyzed 93,819 FHA loan applications from Massachusetts HMDA data (2018–2022). Three primary modeling approaches were used:

• Generalized Linear Models (GLM) to estimate racial disparities in approval probability, controlling for income, loan amount, property value, neighborhood characteristics, and pandemic timing

• Regression Discontinuity Design (RDD) to isolate the causal effect of COVID-19 onset on minority approval rates

• T-tests and ANOVA to test whether approval rate disparities vary systematically across property value tiers and counties

What the Data Shows

After controlling for income, loan amount, property value, and neighborhood characteristics, the results are stark:

• Black FHA applicants are 45.7% less likely to be approved than White applicants (log-odds coefficient: -0.4571, p < 0.001)

• Hispanic applicants are 20.2% less likely to receive FHA approval than White applicants

• Every other racial group — Asian, Pacific Islander, multiracial — is also less likely to be approved than White applicants, suggesting systemic structural patterns rather than isolated bias

These gaps remain significant even after accounting for credit proxies the data allows — meaning the disparity cannot be fully explained by the observable financial characteristics in the model.

COVID-19 Made It Worse

The pandemic appears to have reversed years of gradual progress on fair lending. Using a Regression Discontinuity Design with 2020 as the cutoff:

• The likelihood of a Black or Hispanic borrower receiving FHA approval fell 36% during the pandemic compared to the pre-COVID period

• For every additional year since 2020, the negative effect compounds by an additional 47%

The pattern in the data suggests that during times of uncertainty and financial stress, banks became more conservative — and the borrowers perceived as higher risk (often correlated with race) paid the price first. White borrowers, historically seen as more creditworthy, held a larger share of approvals as the overall approval pool contracted.

Higher Property Values, Lower Minority Approval Rates

The disparity is not uniform across Massachusetts. In higher-priced counties, minority applicants face a steeper climb:

• Counties with higher median property values have significantly lower minority approval rates (t-stat = -38.18, p < 0.05)

• The effect is most pronounced in high-cost suburban counties — the markets where lenders may perceive the greatest risk

• Notably, Suffolk County (Boston and Cambridge) is an exception: urban density, diverse lending competition, and community reinvestment activity appear to partially counteract the disparity in city cores

What This Means for Fair Housing Policy

The FHA is not working as designed. The loan program that was supposed to be a safety net for minority homebuyers is delivering unequal outcomes — and the gap widens precisely when economic uncertainty makes access to credit most critical.

Several mechanisms may explain the pattern:

• Lender risk aversion during downturns: When housing markets become uncertain, lenders tighten discretionary underwriting in ways that disproportionately affect already-marginal applications

• Regional property value effects: In high-cost markets, lenders may perceive minority applicants as less able to absorb value declines, even when controlling for income and loan size

• Structural credit score gaps: The credit scoring system itself reflects historical inequities — a point the FHFA and CFPB have acknowledged but not resolved

The findings call for stronger monitoring by FHFA, Fannie Mae, and Freddie Mac — particularly during economic shocks — and suggest that fair lending enforcement needs to be counter-cyclical, not just steady-state.