Author Description

We explore the types of data used to characterize risky subprime lending and consider the geographic dispersion of subprime lending. First, we describe the strengths and weaknesses of three different datasets on subprime mortgages using information from LoanPerformance, HUD, and HMDA. These datasets embody different definitions of subprime mortgages. We show that estimates of the number of subprime originations are somewhat sensitive to which types of mortgages are categorized as subprime. Second, we describe what parts of the country and what sorts of neighborhoods had more subprime originations in 2005, and how these patterns differed for purchase and refinance mortgages. Subprime originations appear to be heavily concentrated in fast-growing parts of the country with considerable new construction, such as Florida, California, Nevada, and the Washington DC area. These locations saw house prices rise at faster-than-average rates relative to their own history and relative to the rest of the country. However, this link between construction, house prices, and subprime lending is not universal, as other markets with high house price growth such as the Northeast did not see especially high rates of subprime usage. Subprime loans were also heavily concentrated in Zip codes with more residents in the moderate credit score category and more black and Hispanic residents. Areas with lower income and higher unemployment had more subprime lending, but these associations are smaller in magnitude.