Discrimination: Minority Mortgage Market Experiences Leading Up to and During the Financial Crisis

Dave here. It took this many studies to get to “Discrimination exists”? Ah well. Important to have the data underpinning the reality. But it’s pretty obvious: minority borrowers have a larger majority of their wealth tied up in their homes, so of course they’re more vulnerable to downturns that have wide-ranging effects on their economic well-being. If anything that’s a legacy of centuries of discrimination.

This is another reason, incidentally, why industry rebuttals to the foreclosure crisis and its associated frauds always fell back on the “deadbeat” trope. Quite simply, it’s playing to a crude stereotype, one created by racially discriminatory lending.

By Stephen L. Ross, Professor in the Department of Economics, University of Connecticut. Originally posted at VoxEU

The foreclosure crisis and the growth of subprime lending that preceded the crisis have disproportionately affected low-income and minority neighborhoods, having dire consequences for minority homeownership (Geradi and Willen 2009). Some have suggested that subprime lending played a substantial and general role in the foreclosure crisis (Mian and Sufi 2009); others argue that high cost lending in minority neighborhoods contributed to the high foreclosure rates among black and Hispanic borrowers (Reid and Laderman 2008). Bhutta and Canner (2013) find a lower incidence of high cost loans within banks’ Community Reinvestment Act (CRA) assessment areas, consistent with mortgage brokers playing a role in high cost lending. Further, discrimination in the underwriting of prime mortgages may cause minority borrowers to select into the subprime market (Munnell et al. 1996, Ross and Yinger 2000). In fact, the U.S. Department of Justice has recently filed several cases against major lenders for steering minority borrowers into subprime loans.

At the same time, the purpose of subprime lending was to provide credit to borrowers who faced credit constraints in the primary mortgage market; such borrowers should be expected to face a higher price of credit and have worse credit market outcomes. African-Americans and Hispanics have lower levels of wealth, and as a result are likely to face significant downpayment constraints that discourage homeownership (Deng et al. 2003, Gyourko et al.1999, Duca and Rosenthal 1994). Minority mortgage applicants tend to have lower credit quality and incomes – as well as higher loan-to-value ratios – than white borrowers (Ross and Yinger 2000, Bhutta and Canner 2013). In fact, many studies of loan pricing at the individual lender level find at most modest and often zero racial and ethnic differences in prices when controlling for observable characteristics (Black et al. 2003, Courshane 2007, Courshane and Nickerson 1997).

However, few studies assess market wide racial and ethnic differences in either the price of credit or credit market outcomes, such as mortgage delinquency or foreclosure. Several studies use proprietary samples to examine the outcomes of minority borrowers. On the price of credit, Reid and Laderman (2009) find substantial racial and ethnic differences in the likelihood of receiving a high cost or rate spread loan, and find that the wholesale origination channel plays a major role in explaining the incidence of high cost loans for all demographic groups. Haughwout et al. (2009) find no racial differences in the price of credit for 2/28 mortgages in 2005, while Ghent et al. (2013) find racial and ethnic differences for longer-term adjustable rate mortgages that are concentrated in the purchase market and among non-depository lenders. Further, Reid and Laderman (2009) find substantial racial differences in the likelihood of default among loans to African-Americans and Hispanics. One key limitation of these papers is that the samples tend to be dominated by privately securitized loans, and so are composed primarily of non-conforming loans originated in what is thought of as the subprime mortgage market.

A recent paper by Bayer et al. (2014) examines racial and ethnic differences in the incidence of high costs loans in a sample that provides broad coverage of the mortgage market during the crisis and the period leading into it. They draw samples of home purchase and refinance loans in seven major metropolitan areas between 2004 and 2008 from the Home Mortgage Disclosure Act data. They merge this data with detailed assessor, transaction, and lien data to obtain information on the house price, the combined loan-to-value ratio, the type of mortgage (e.g. adjustable or fixed rate), and the address – then providing these addresses to a major credit repository in order to obtain anonymous loan files with detailed credit history information. Using the resulting sample, they estimate models on the likelihood of obtaining a high cost or rate spread loan using detail controls for mortgage risk factors and find that blacks and Hispanics are more likely to have high cost loans for both home purchase and refinance loans across all metropolitan markets. Furthermore, a majority of the observed differences are associated with the lender, rather than resulting from differences across races at the same lender. For blacks in the home purchase market, the racial differences are spread across the entire market, rather than being concentrated among loans with credit risk factors that are typically associated with subprime lending. For Hispanic homebuyers, differences are concentrated among borrowers with high loan-to-value ratios and subprime credit scores. Finally, they find that racial differences in high cost loans are associated with locations where most black borrowers have relatively low levels of education, suggesting a role for financial sophistication in explaining the pervasiveness of high cost loans among black borrowers.

In a companion paper using the same sample, Bayer et al. (2012) examine racial and ethnic differences in loan performance. They examine the likelihood of mortgage delinquency and foreclosure from 2005 to 2009 for mortgages underwritten between 2004 and 2008. They find substantial racial and ethnic differences in the likelihood of delinquency and foreclosure, even after controlling for credit risk factors and for contemporaneous estimates of negative equity and county-wide risk of unemployment. Racial differences in unemployment risk can explain part of the difference for black homebuyers, and the remainder appears to be concentrated among borrowers who have very high debt-to-income ratios. For Hispanics, ethnic differences in county employment rates cannot explain the observed differences, but the delinquency and foreclosure differences are concentrated almost entirely in counties with the highest overall unemployment rates for Hispanics. Further, the unexplained racial and ethnic differences are concentrated among individuals who purchased their homes near the peak of the housing and mortgage market expansion, and these effects are largest among borrowers with high debt-to-income ratios and who have been exposed to low employment rates. Bayer et al. argue that this evidence is consistent with a process where borrowers sort into homeownership in part based on their expected risk of adverse future events. As credit becomes more easily available, the new entrants to the housing market will tend to be those who face the highest risks of negative outcomes during an economic downturn, and these effects are felt most dramatically among minority borrowers who tend to be more vulnerable to economic downturns (Hoynes et al. 2012).

Minority homebuyers – especially blacks – tend to face a higher cost of mortgage credit and had substantially worse credit market outcomes during the recent downturn than white homebuyers with equivalent mortgage risk factors. In terms of the price of credit, a majority of the unexplained differences are associated with the lender from which the homebuyer obtained credit. These effects are felt most among minority borrowers with the lowest levels of education, and are likely due in part to the concentrated activity of subprime lenders in minority neighborhoods and a lack of knowledge of financial markets among minority borrowers with low levels of education. On the other hand, most of the racial differences in loan performance that are unexplained by traditional credit risk factors cannot be captured by controlling for the lender or other aspects of subprime lending. African-Americans and Hispanics appear to be more vulnerable to an economic downturn and to the associated risks of unemployment and housing price declines than observationally similar white homeowners. This higher vulnerability is most pronounced for borrowers who purchased their homes right before the onset of the financial crisis, even after controlling for the increased risk of negative equity associated with buying at the peak of the market. While the expansion of the subprime sector may have contributed to a higher cost of credit for black homebuyers, their concentration in high cost loans (and in the subprime market more generally) can explain only a small portion of the racial differences in foreclosure. Rather, a broad spectrum of black and Hispanic borrowers appear to be especially vulnerable to the economic downturn and associated shocks to their ability to meet their mortgage commitments.