The spatial distribution of subprime/higher-priced mortgages and its relationship with housing price variations within the Philadelphia metropolitan area: global model vs. local model

Abstract

Over the last decade, the United States had experienced a boom and bust in the subprime mortgage market. The ups and downs of the subprime mortgage market became a primary factor triggering the most severe global economic recession since the Great Depression. The dissertation contributes to the literature by inquiring whether the subprime lending has exacerbated social inequity between subprime neighborhoods and other neighborhoods, through analyzing the subprime mortgage market in the Philadelphia MSA from 2000 through 2010, and focusing on two research questions: (1) the spatial distribution of subprime mortgages across census tracts; (2) the relationship between subprime intensities and housing price variations across zip-code areas. As the dissertation's study area expands from an urban to a MSA, spatial heterogeneity merits attention in this relative huge area. As a result, this dissertation not only employs a global, Ordinary Least Squares (OLS) model, but also a local, Geographically Weighted Regression (GWR) model to examine spatial variations across different neighborhoods. For the first research question, the dissertation finds: (1) a higher concentration of higher-priced mortgage for purchase and refinance in tracts with higher proportion of African-American and Hispanic residents, lower median household incomes, higher-unemployment rates, lower self-employment rates, and higher capitalization rates, after controlling for other variables; (2) the association between higher-priced mortgages and explanatory variables varies across census tracts. Because the dynamics of neighborhood subprime originations are heterogamous, the association between subprime mortgage origination and socioeconomic characteristics may be stronger in some neighborhoods than other neighborhoods. For the second research question, the dissertation finds: (1) subprime mortgage shares have a significant negative association with housing price appreciations during the housing boom period (2001-2006); (2) subprime mortgage shares have a significant positive association with housing price depreciations during the housing bust period (2006-2010); and (3) the association between housing price variations and explanatory variables differs across geographic submarkets within the Philadelphia region. The result confirms that areas where more residents obtained subprime mortgages have suffered more severely than other from the housing market's ups and downs over the last decade. The empirical results can draw broad policy implications. The primary implication is that it is time for the federal government to rethink its homeownership policy. Increased homeownership levels arising from the expansion of subprime mortgages are not sustainable, and subprime lending has exacerbated social inequity between subprime neighborhoods and other neighborhoods. The second implication is that the government needs to enforce the fair lending laws, because the cluster of subprime mortgage origination reflects the unequal opportunities of prime mortgage accessibility across different neighborhoods. The third implication is that the government needs to promote place-based policy making. As the GWR demonstrates, the dynamics of the mortgage market and housing market are uneven across different neighborhoods. Therefore, place-based making can increase the efficiency of public policy. These implications based on the dissertation's empirical results are helpful for designing more efficient, effective, and sustainable housing policies of the United States.