The Neighborhood Distribution of Subprime Mortgage Lending

Abstract

Subprime lending has grown over the past decade into a highly visible part of the United States' mortgage market. Previous studies have shown that a relatively large share of subprime lending occurs in census tracts with high concentrations of low-income and minority households, generating the concern that minority households may not be provided equal opportunity in the prime mortgage market. This study expands the literature by utilizing data from the Census, HMDA, and proprietary data on neighborhood risk factors, to perform a spatial analysis of subprime lending across Census tracts, supplemented with a logit regression analysis at the borrower level. Results indicate that lending patterns across neighborhoods are strongly influenced by neighborhood risk composition. Across census tracts, increased credit risk is associated with a larger subprime share of loans in the tract or increased odds of subprime borrowing relative to prime. We also find a substantial concentration of subprime lending in neighborhoods where homeowners are predominantly African American, and we find that African-American borrower is strongly correlated with being a subprime borrower in the logit regressions, with or without the control variables.