What HMDA Data Reveals About Lending Inequality

Every year, financial institutions report data on millions of mortgage applications under the Home Mortgage Disclosure Act (HMDA). This data—covering applicant race, income, loan amount, property location, and outcome—provides the most comprehensive look at lending equity in America. The picture it paints is deeply troubling.

The Overall Picture

In the most recent HMDA data (2023 filings), mortgage denial rates by race tell a consistent story:

Controlling for Income

The most important finding in HMDA data is that racial disparities persist even after controlling for income. A Black applicant earning $100,000-$150,000 per year has a denial rate of 17.4%, compared to 8.9% for a white applicant in the same income bracket. That's a gap of 8.5 percentage points—nearly double the denial rate.

At every income level we examined, the same pattern holds:

Denial Reasons

Lenders are required to report the primary reason for denial. The most common reasons differ significantly by race:

• Debt-to-income ratio: Cited for 38% of Black denials vs. 29% of white denials
• Credit history: 31% of Black denials vs. 22% of white denials
• Collateral: 14% of Black denials vs. 18% of white denials
• Insufficient cash: 12% of Black denials vs. 8% of white denials

These denial reasons themselves reflect systemic inequalities. The racial wealth gap means Black families have, on average, one-sixth the wealth of white families ($44,900 vs. $285,000 median net worth). This affects down payment savings, debt loads from education financing, and credit history depth.

Geographic Hotspots

Some metro areas show particularly stark racial disparities in lending:

Notably, the worst disparities are concentrated in Midwestern cities with histories of aggressive redlining and residential segregation. Milwaukee's 3.8x ratio is the highest in the nation—a Black applicant there is nearly four times more likely to be denied than a white applicant.

The Algorithmic Question

As more lenders adopt algorithmic underwriting, questions arise about whether technology perpetuates or alleviates discrimination. Studies by the Brookings Institution and UC Berkeley have found that fintech lenders show smaller but still significant disparities compared to traditional lenders—roughly 40% lower racial gaps in denial rates, but still systematically different outcomes.

The challenge is that algorithms trained on historical data will reflect historical discrimination. Credit scoring models, for instance, don't explicitly consider race—but they consider factors (credit length, credit mix, debt ratios) that are themselves shaped by generations of differential access to banking, wealth-building, and employment.

Loan Pricing Disparities

Even when approved, HMDA data reveals that Black and Hispanic borrowers pay more. The average interest rate spread (rate above a benchmark) for conventional purchase mortgages:

• White borrowers: 0.42% above benchmark
• Black borrowers: 0.71% above benchmark
• Hispanic borrowers: 0.58% above benchmark

On a $300,000, 30-year mortgage, a 0.29% higher rate costs an additional $18,700 over the life of the loan. This pricing gap extracts wealth from communities of color at an enormous scale.

What Would Equitable Lending Look Like?

1. Special purpose credit programs: The ECOA allows lenders to create programs specifically designed to increase access for disadvantaged groups
2. Alternative credit data: Using rent payments, utility bills, and other regular obligations to build credit profiles for "thin file" borrowers
3. Down payment assistance: Targeted programs to address the wealth gap's impact on homebuying ability
4. Fair lending enforcement: Robust DOJ and CFPB enforcement using HMDA data to identify and prosecute discriminatory patterns
5. Algorithmic auditing: Required disparate impact testing for automated underwriting systems