Disparities in mortgage lending, bank performance, economic influence and regulatory oversight

Large racial disparities in mortgage loan denial rates and low minority representation among applicants for mortgage credit have created concerns that mortgage lenders discriminate against minorities. This dissertation investigates factors that influence changes in these disparities in 25 MSAs in the U.S. during 1991-1997. During the latter portion of the sample period racial disparities in mortgage denial rates declined significantly, while minority representation in the mortgage applicant pool increased. Some industry observers attribute these changes to improvements in regulatory enforcement of the fair lending laws and Community Reinvestment Act (CRA). I hypothesize that improvements in economic conditions during the period that reduce default loss probabilities on mortgage loans contribute to reductions in denial rate disparities, since minorities represent a disproportionate percentage of the marginal loan applicants in this sample. I anticipate that improvements in the financial condition and performance of financial institutions may have had a similar effect. In Chapter 4 of the Dissertation, an empirical model is developed to formally test the joint affect of market forces, including changes in the economy and lender financial condition and performance, as well as regulatory influence on mortgage lending outcomes. The model incorporates the Home Mortgage Disclosure Act (HMDA) data on characteristics of applicants for conventional home purchase mortgage loans in 25 metropolitan statistical areas (MSAs). These data are matched to census tract data from the 1990 U.S. census, bank-specific financial data from the bank reports of condition and income, and state and MSA-level macroeconomic series. A partial adjustment framework is employed to model lending institutions' response to established 'targets' for racial disparities in denial rates and minority representation. Since these targets are jointly determined, the two-stage least squares (2SLS) method for estimating simultaneous equations is used to estimate the 'disparity' and 'representation' models. The results of the 2SLS estimation overall are not consistent with a regulation-induced shift in racial disparities in denial rates and minority representation, after controlling for market forces. While there is some evidence to support a regulatory influence on changes in minority representation, this effect is restricted to institutions that already have satisfactory performance ratings under CRA. The results suggest that changes in the mortgage market during the period disproportionally benefited the minority applicant group. Specifically, changes in the quality of the applicant pools, economic conditions in the MSAs under study, and the financial condition and performance of the lending institutions are found to have a statistically significant influence on mortgage lending outcomes.