New Evidence on the Relationship Between Race and Mortgage Default

The traditional method for testing for lender discrimination has involved the estimation of a mortgage rejection equation to determine whether there is an independent race effect, after controlling for credit risk factors that lenders typically consider when underwriting a loan. In the early 1990s, the now widely cited Boston Federal Reserve study used this methodology to provide what many consider to be convincing evidence of discrimination against black and Hispanic mortgage applicants in the Boston metropolitan area. In the mid-1990s, a series of papers by Berkovec, Canner, Gabriel, and Hannan (BCGH) offered an alternative model to test for lender discrimination based on the performance of the mortgage following origination. In their model, lenders discriminate by holding minorities to higher credit standards, a practice that suggests minorities would have lower default probabilities than non-minorities for given values of other default-related factors. One problem with the BCGH work was that their estimated default equation did not control for the credit history of the FHA borrower. Recent studies have found credit history (e.g., past record of making monthly payments on time) to be an important determinant of mortgage default that is also highly correlated with race, with minorities, on average, exhibiting below-average past credit records. Thus, BCGH’s finding that black borrowers had higher FHA default probabilities could be traceable to their not controlling for the effects of credit history. The limited twofold purpose of this study is to report the findings obtained by including a measure of borrower credit history in a model of FHA defaults that is similar to the BCGH default model and to demonstrate the bias attributable to omitting such data.