Testing for Racial Discrimination in Mortgage Lending and Assessing the Community Reinvestment Act via Mortgage Default Equations

Abstract

This paper studies racial discrimination in the mortgage market at the point of loan underwriting decisions and evaluates the Community Reinvestment Act (CRA), a piece of legislation focused on preventing discrimination in mortgage lending. I study the period from 2006-2016, which allows the consideration of both pre- and post-crisis trends in the mortgage market. This analysis relies on novel data from Fannie Mae and the Federal Financial Institutions Examination Council. The baseline assumption states that in the absence of discrimination in loan underwriting, the odds of default for all racial groups should be statistically indistinguishable. Higher odds of loan default for a minority community relative to a White community suggests overly permissive lending; lower odds of default indicates discriminatory under-lending. Controlling for borrower, loan and property characteristics, I find that a one percentage point increase in the Hawaiian/Pacific Islander population in a county decreased the log odds of loan default in that county by 104.5. This result points in the direction of discrimination against this population. I find no evidence of discrimination against other minority populations in the sample. The evidence of the efficacy of the CRA is fairly mixed. In pre-crisis, not CRA-eligible county-years, a one percentage point increase in Asian population in a county raised the log odds of loan default by 10.21; CRA-eligibility lowered this value to approximately zero. In post-crisis, not CRA-eligible county-years, a one-unit increase in American Indian/Alaska Native population increased the log odds of default by 45.50; CRA-eligibility raised this value to 47.12. Post-crisis origination tended to bring the log odds of default of minority communities relative to White communities closer to zero.