Who Are the Foreclosed? A Statistical Portrait of America in Crisis

Abstract

Data from the National Suburban Survey from September 2010 permit the first statistical portrait of Americans displaced by the mortgage foreclosure crisis. The average person who has experienced home mortgage foreclosure since September 2007 resembles the average American but is somewhat likely to be younger, Latino, and a parent. The foreclosed are also more likely to report various other measures of financial distress, including recent job loss. The experience of foreclosure is associated with more problems in the neighborhoods where respondents currently reside, including such problems as crime, unemployment, and a lack of affordable housing. Respondents who have not personally lost a home, but who know the foreclosed, are also experiencing more economic distress and more neighborhood problems than those who have not. These descriptive findings suggest the human costs of the foreclosure crisis and the limits of informal social safety nets for addressing those costs.

Keywords:

• foreclosure
• survey data
• race and ethnicity
• subprime crisis
• income and unemployment

Notes

1. These benchmarks are not the same as the data that were used to calculate the post-stratification weights. PSRAI calculated the sampling weights to adjust for differential sampling probabilities and to match known demographic parameters from the 2009 Annual Social and Economic Supplement of the Current Population Survey, cell phone usage data from the 2009 National Health Interview Survey, and geographic parameters from the telephone exchange database that provided the sampling frame (see PSRAI, 2010).

2. For an overview of HMDA's uses and limitations for identifying subprime loans and borrowers, see Avery, Brevoort, and Canner (2007) and Mumm and Wyly (2011).

3. This conclusion might be tempered if we had data on those who experienced foreclosure prior to the housing crash. The NSS asks about the experience of foreclosure from 2007 to 2010, when subprime loans made at the height of the boom (between 2005 and 2007) reset and became unaffordable. Although this period represents the peak of the crisis as measured by national delinquency rates, the crisis unfolded unevenly. Certain regions and populations experienced foreclosure earlier than others. Many African American urban communities experienced an influx of subprime credit that began as early as the late 1990s (Immergluck & Smith, 2005; Newman, 2009). Studies during this period also showed that neighborhood education levels were correlated with rates of subprime lending and foreclosure (Calem, Hershaff, & Wachter, 2004; Grover, Smith, & Todd, 2008). These pre-2007 foreclosures are not captured in the NSS. As house prices have fallen and unemployment has risen, subprime loans and foreclosure has become a growing problem among a broader range of households, including those who would have once qualified for “prime” mortgages.

4. The NSS income question was categorical. We recoded income in each category to the mean of the category bounds, and imputed a mean income of $335,000 to the top-coded category on the assumption that income is Pareto-distributed.

5. This model has the form γ = e^βx, where γ is the expected number of community problems, β is a column vector of parameters, and x is a row vector including a constant term, a dichotomous indicator for respondents who said they personally had lost a home to foreclosure, and selected control variables. Coefficients can therefore be read as unit contributions of each variable to the natural logarithm of the expected count. The model also includes an overdispersion parameter α to characterize the variance. We also estimated a computationally simpler linear regression with ordinary least squares and found qualitatively equivalent results.