Risk or Race: An Assessment of Subprime Lending Patterns in Nine Metropolitan Areas

Over the past decade, numerous studies have raised concerns that subprime lenders were inappropriately targeting economically vulnerable neighborhoods and that some borrowers, particularly African-Americans, were paying more than necessary for mortgage financing. Indeed, there is a substantial body of literature consistent with the finding that predominantly African-American neighborhoods had much higher shares of loans originated by subprime lenders than areas where whites predominated, while even controlling for a range of factors that are likely to influence the allocation of mortgage credit. Yet, despite the importance of the topic, efforts to explain the underlying causes of the observed patterns have been limited by the lack of data on mortgage credit risk.

Using newly available data on neighborhood-level measures of credit scores for each of nine metropolitan areas selected to reflect a range of racial and ethnic population characteristics, this paper presents a detailed analysis of what has come to be known as the “risk or race” question. In particular, this report summarizes the results obtained from a series of weighted least squares regressions for models that seek to explain census tract-level variation in the share of higher-priced subprime mortgages for home purchase and refinance. In addition, the paper also reports the findings obtained from a series of logit regression models of the impact that tract-level mortgage credit scores have on the probability that an individual borrower will obtain a higher-priced as opposed to a lower-priced prime mortgage. Following the path-breaking work of Susan Wachter and her colleagues, these models included various measures expected to influence the allocation of mortgage credit, including income and other demographic factors, mortgage credit and other measures of credit risk, and various neighborhood-level housing market variables. For both sets of equations, the key focus is on whether and how the inclusion of tract-level credit score information influences the relationship between racial factors and access to prime mortgages.