A stable capital inflow of low-risk loans is the foundation for a sustainable neighborhood: equal access to low-risk mortgages plays an important role for neighborhoods to recover from decline caused by foreclosures during and after the subprime crisis. However, inequalities in mortgage lending by neighborhood racial composition have been persistent in the U.S. housing market since the 1930s.
Beginning with redlining in the early 20th century, disparities in mortgage lending by race and neighborhood racial composition have been evident in the lack of inflows of residential mortgages into predominantly minority neighborhoods, compared to predominantly white neighborhoods. During the subprime boom, lenders engaged in reverse redlining, steering predominantly minority neighborhoods to high-cost subprime mortgages. The mortgage crisis that stemmed from the surge of subprime mortgages brought widespread foreclosures. Neighborhoods with a high concentration of subprime loans, particularly predominantly minority neighborhoods, suffered from high rates of foreclosures, which caused neighborhood decline. In the wake of the crisis, major conventional mortgage lenders seriously tightened their underwriting practices. The extent to which households can access mortgages still depends on their race and ethnicity (Goodman, Zhu, & George, 2014, 2015; Immergluck, 2011; Richardson, Mitchell, & West, 2016; Stein & Nguyen, 2010). This was despite the fact that capital inflows were required for communities to recover from decline (Ding, 2014). Furthermore, the number of higher-cost residential mortgages, which is defined as mortgages with annual percentage point of 150 basis points higher than the market rate, has gradually increased since 2010, remaining around twice as high for among minority borrowers compared to white borrowers during the crisis period (Board of Governors of the Federal Reserve System, 2015).
Given these trends, this study aims to examine spatial mortgage lending patterns by asking the following question: Were inflows and costs of mortgage capital different by neighborhood racial composition during and after the crisis? This study hypothesizes a polarization of mortgage inflows and costs whereby predominantly African American neighborhoods are (1) less inflows of mortgages capital and (2) higher costs of mortgage capital than predominantly white neighborhoods. To address the research question and test the hypothesis, this study investigates spatial lending patterns at the zipcode level in the five metropolitan statistical areas (MSAs) in Ohio including Cleveland, Cincinnati, Columbus, Dayton, and Toledo during three periods spanning from 2004 to 2015—the housing mortgage boom between 2004 and 2007; the bust of foreclosures between 2008 and 2011; and the recovery from 2012 to 2015.
Using cumulative data from the U.S. Census, the American Community Survey, the Home Mortgage Disclosure Act, and CoreLogic©’s mortgage origination data, this dissertation begins with a description of (1) the number of conventional mortgages originated in a neighborhood, (2) the market share of FHA insured mortgages in a neighborhood, (3) the market share of subprime mortgages in a neighborhood, and (4) the median annual percentage rate (APR) spread of conventional mortgages in a neighborhood during the boom, crisis, and recovery periods. In addition, this descriptive analysis incorporates a geographic information system (GIS). This analysis is followed by a multivariate analysis employing a cross-sectional ordinary least squares (OLS) regression, cross-sectional binary logistic regression, and longitudinal random-effect regression models to evaluate the association between these four lending patterns mentioned above and neighborhood racial composition, controlling for neighborhood-level information such as median credit score in a neighborhood and median neighborhood income. In addition, while accounting for loan-level information such as borrower’s credit score and neighborhood information such as median neighborhood income simultaneously, this study conducts two more multivariate analyses. The first investigates lending patterns of FHA insured versus conventional loans in relation to neighborhood racial composition, employing binary logistic regression and hierarchical regression models. The second investigates APR spread of conventional mortgages in relation to neighborhood racial composition, employing OLS regression and hierarchical regression models.
The primary results confirm spatial differences in flows of mortgages by neighborhood racial composition during and after the foreclosure crisis. Crucially, findings indicate that higher proportions of African American residents are significantly associated with lower inflows of conventional loans in neighborhoods in and across the five MSAs (except Cincinnati) during this study period. Furthermore, during the crisis and recovery periods, across the five MSAs, predominantly African American neighborhoods tended to experience greater inflows of FHA insured loans, which are potentially costlier than conventional loans. On the other hand, Cleveland and Dayton, in which inflows of conventional mortgage capital were lacking as compared to the other MSAs, also indicated the lack of FHA mortgage inflows in the predominantly African American neighborhoods. These findings indicate that overall inflows of mortgage capital (including both conventional and FHA) were seriously limited in predominantly African American neighborhoods in the two MSAs during the crisis and thereafter.
With respect to costs of mortgage capital, the results varied by MSA and the local housing market condition. Spatial differences in costs of mortgage capital by neighborhood racial composition during and after the foreclosure crisis are not uniformly confirmed across the five MSAs; rather, it was only confirmed in Cincinnati and Columbus. Predominantly African American neighborhoods had higher costs of conventional mortgages in Cincinnati during the crisis period, and in Columbus during both the crisis and recovery periods. Most importantly, the findings indicate that inflows and costs appear to be related in some MSAs. In Columbus, which experienced relatively large inflows of mortgage capital, predominantly African American neighborhoods within the MSA had higher costs of mortgage capital. By contrast, in Cleveland, which experienced lower inflows of mortgage capital as compared to other MSAs, predominantly African American neighborhoods within the MSA had lower costs of mortgage capital. An expected reason for those anomalies (higher costs mortgages went into African American neighborhoods in Columbus; lower costs went into African American neighborhoods in Cleveland) is that only those who have higher financial credibility had access to conventional loans and thus, the cost of mortgage tends to be lower in Cleveland, and vice versa in Columbus.