HMDA Data: What 10 Million Loan Records Reveal About Mortgage Pricing

What Is HMDA Data?

The Home Mortgage Disclosure Act (HMDA) is a federal law that requires most mortgage lenders to collect and publicly disclose data about their lending activity. Administered by the Consumer Financial Protection Bureau (CFPB), the HMDA dataset covers virtually all residential mortgage originations in the United States — approximately 10–15 million records per year. Each record includes details about the loan amount, interest rate, property location, borrower demographics, lender information, loan type, and outcome (approved, denied, withdrawn). This makes HMDA the most comprehensive public dataset on mortgage pricing and lending patterns in the country. At lenddy.io, we use HMDA data as a foundation for benchmarking Loan Estimates against real market data.

The Dataset: 10–15 Million Records Per Year

The HMDA dataset includes data from over 5,000 reporting institutions, covering banks, credit unions, non-bank lenders, and mortgage brokers. Each year's dataset contains detailed information on 10–15 million loan applications, making it one of the largest financial datasets publicly available. Fields include loan amount, interest rate, rate spread (the difference between the loan's APR and a benchmark rate), total loan costs, total points and fees, origination charges, discount points, lender credits, property value, combined loan-to-value ratio, debt-to-income ratio, borrower credit score range, and geographic data down to the census tract level. This granularity allows for meaningful analysis of pricing patterns across lender types, geographies, loan products, and borrower profiles.

Key Finding: Rate Distribution by Credit Tier

Our analysis of HMDA data reveals significant rate dispersion even within the same credit score range, suggesting that many borrowers pay more than they should. Among conventional 30-year fixed-rate purchase mortgages with credit scores above 740, the median interest rate in the most recent data year was approximately 6.75%. However, the 25th percentile (better-than-average pricing) was 6.50%, while the 75th percentile was 7.00%. This means a borrower with the same excellent credit could pay rates ranging from 6.50% to 7.00% depending solely on which lender they chose and how well they negotiated. The spread widens at lower credit tiers: borrowers in the 680–719 range saw a 25th-to-75th percentile spread of nearly 0.75%, representing thousands of dollars in potential savings over the life of the loan.

Key Finding: Fee Patterns by Lender Type

HMDA data reveals distinct fee patterns across lender types. Traditional banks tend to charge lower origination fees (median: 0.5–0.7% of loan amount) but often have slightly higher interest rates. Non-bank lenders (independent mortgage companies) typically charge higher origination fees (median: 0.7–1.0%) but may offer more competitive rates. Credit unions generally offer the lowest combination of rates and fees, but their product selection and geographic reach are limited. The data also shows a concerning pattern: total loan costs as a percentage of loan amount are, on average, 15–20% higher for borrowers with credit scores below 680 compared to those above 740 — even after controlling for the expected rate difference. This suggests that lower-credit borrowers face disproportionately higher fee loads in addition to higher rates.

Key Finding: Geographic Pricing Variation

Geography plays a larger role in mortgage pricing than most borrowers realize. HMDA data shows that median interest rates vary by 0.25–0.50% across metropolitan statistical areas (MSAs) for otherwise identical borrower profiles. States with more competitive lending markets (like California, Texas, and Florida, where dozens of lenders compete for business) tend to have slightly lower rates than states with less competition. Closing costs vary even more dramatically — total loan costs in high-cost metros like New York City and San Francisco can be 40–60% higher than in mid-market cities like Denver, Phoenix, or Charlotte. This geographic variation is driven by differences in state regulations, transfer taxes, title insurance pricing, and local market competition. The key takeaway: a 'fair' price depends on where you're buying.

What This Means for You

HMDA data proves three things that every mortgage borrower should know. First, there's significant price dispersion — borrowers with identical profiles can pay very different rates and fees depending on their lender, making comparison shopping essential. Second, fee patterns vary by lender type — the cheapest rate doesn't always mean the cheapest loan when you factor in origination charges. Third, geography matters — a 'good deal' in one city might be an 'average deal' in another. At lenddy.io, we use HMDA data as the foundation for our AI Loan Estimate Analyzer. When you upload your Loan Estimate, we compare your rate, fees, and total costs against real market data from millions of actual loan originations — not theoretical benchmarks or lender marketing materials. This gives you an objective, data-driven assessment of whether your deal is truly fair.

Frequently Asked Questions