Race and Mortgage Denials: 21st Century Discrimination and the Need for Reform

Background and Purpose

Mortgage approvals and denials are, at least in theory, based principally on the risk of default. There are three levels of information around which mortgage application decisions are based: (1) borrower characteristics, (2) property characteristics, and (3) neighborhood characteristics. All three can affect the ability of the borrower to repay the loan. Consequently, the mortgage decision-making process is awash in information. After decades of discrimination, the information lenders use to make mortgage decisions is mandated to be race-neutral. Yet, race continues to matter to denial decisions in Detroit, Michigan. This paper aims to analyze the ways that race contribute to mortgage denials, and thus racial differentials in homeownership within Detroit.

Methods

To capture mortgage lending dynamics, we used mortgage application denials because they are in reaction to both supply-side and demand-side housing market forces. Extending beyond other studies about racial lending discrimination, we account for each of the three levels of information with measurable data from the Home Mortgage Disclosure Act (HMDA) public-use data file for 2012 to 2021; the 2010 U.S. Census; and five-year estimates from the American Community Survey (ACS) for the metro Detroit area.

The HMDA data were merged at the census tract level with the estimates from the ACS on a variety of neighborhood characteristics. The HMDA data capture the volume of mortgage loan applications in the region. Each observation represents a mortgage loan application and includes information on the loan’s purpose, application decision, denial reason, borrower demographics, and property geographic information.

With the HMDA data we construct information variables on the volume of lending activity in the neighborhood, assuming that the most recent information is the most important to the lending decisions in the current year. Consequently, we calculate the ratio of originations to applications in the neighborhood from the previous year. The ratio is necessary because the metro Detroit region sees considerable variation in mortgage application activity by neighborhood. We account for this curvilinear relationship by calculating the log of neighborhood-level applications in the current year.

Results

Controlling for individual, property, and neighborhood factors, race remained a significant predictor of mortgage denial each year that we modeled. Specifically, Black borrowers were significantly more likely to be denied than White borrowers in the Metro Detroit area. Black borrowers, for example, were 75% more likely to be denied in 2015 than white borrowers. This figure was never less than 50% between 2012 and 2017.

Conclusions and Implications

This paper concludes with a discussion of several other factors that contribute, alongside race, to mortgage denials, including collaterals and credit history. It is important that future research identify what mechanism accounts for the racial disparities in denials. Also included is a list of policy recommendations meant to ensure greater accountability for lending institutions, to address structural, racial inequalities that are unveiled in the lending process, and to provide support and resources to potential homebuyers in order to address the racial gap in mortgage approvals.