Analysis: Neighborhood income data understates mortgage access gaps

HMDA data shows lower origination rates, higher denial rates and wider loan-size gaps by applicant income vs. census tract income

New analysis from iEmergent, a forecasting and advisory services firm for the financial services, mortgage and real estate industries, finds that mortgage access gaps for low-income households are substantially wider than neighborhood-level data suggests. The analysis draws on the last five years of Home Mortgage Disclosure Act (HMDA) data and compares individual loan applicant income with the median income of the census tract where the subject property is located. 

The two measures produce materially different pictures of mortgage access, including who applies, how often applications result in originated loans and how much borrowers receive. That distinction matters because lenders and regulators use HMDA data to evaluate mortgage lending to LMI borrowers and in LMI geographies, including as part of fair lending and Community Reinvestment Act (CRA) examinations. But the implications extend beyond regulatory compliance: for any lender, understanding how its market performance changes when measured by borrower income versus geography can affect how it identifies access gaps, benchmarks its performance against peers and evaluates where its business opportunities lie.

Key findings include:

• Who applies: Applications tied to low-income census tracts, where median family income is less than 50% of area median family income, account for just 3% of all mortgage applications. But by applicant income, low-income borrowers account for 11% of all mortgage applications.
• Who gets a loan: Half or more (50% to 56%) of applications for properties in low-income census tracts convert to originated loans, whereas only about four in 10 applications (39% to 43%) from low-income applicants result in an originated loan.
• How much they borrow: The average purchase loan amount in low- and moderate-income (LMI) census tracts, where median family income is less than 80% of area median family income, is 31.6% lower than in non-LMI census tracts. But by applicant income, the average purchase loan amount for LMI borrowers is 50.7% lower than for non-LMI borrowers.

"A census tract's income level is a neighborhood average, and averages smooth over the households facing the steepest barriers," iEmergent CEO Laird Nossuli said. "If lenders rely only on tract-level data, they can conclude that access for low-income borrowers is better than it actually is and build strategies around the wrong problem."

Income affects mortgage access through two separate barriers. One is whether a household is in a financial position to apply for a mortgage at all. The other is whether an application becomes a loan. Application rates and origination rates do not necessarily move together, and analyzing borrower income can show lenders whether lower-income households are underrepresented among applicants or whether applications are falling out before they become loans.

"Growing the number of low-income households who apply and getting more existing applicants to closing are different challenges that call for different products, outreach and operational fixes," iEmergent COO Bernard Nossuli added. "Analyzing applicant income alongside census tract income gives lenders a more complete view of where access gaps exist, how those gaps vary by market and where there may be opportunities to respond.”

These findings are part of iEmergent's multipart HMDA data analysis:

• Part 1: Overview of 2025 HMDA Data Trends
• Part 2: 2025 HMDA Data By Race/Ethnicity
• Part 3: Mortgage Lending By Income Level