We examine the effects of residential foreclosure on family outcomes, particularly migration patterns. Our data is the result of a merger between the 2008 Survey of Income and Program Participation (SIPP) and foreclosure data from RealtyTrac, We are able to extend the work on Molloy and Shan (2011) on this topic in two important ways. First, we are better able to control for the pre- foreclosure status (and changes in that status) of the sample households. The well- known “double trigger” hypothesis (e.g. Foote, Gerardi and Willem, 2008) suggests that default and foreclosure are not simply the result of negative equity in housing assets, but are also generated by coincident negative household events such as labor market or health shocks. Failure to control for this pre-foreclosure heterogeneity may result in biased estimates of post-foreclosure outcomes, particularly in the demographic outcomes that are the focus of this paper. We are also able to distinguish the effect of different stages of the foreclosure process on the moving decisions of households.

 In order to assess the effect of foreclosure on family outcomes we estimate a Cox regression model predicting the hazard of moving for families at risk of losing their homes to foreclosure. Our key independent variable is stage of foreclosure (notice of default, notice of trustee sale, real estate or bank owned).  The models also control for household characteristics, individual characteristics of the household head, changes in employment and health insurance coverage status, neighborhood characteristics, length of residence and state of residence. Further, we estimate instrumental variable regression models for the subsample of movers to examine changes in neighborhood characteristics at the block group level. Preliminary findings suggest that families entering the foreclosure process increase their risk of moving out of their home, and that the risk increases at each stage of the foreclosure process.  However, models predicting change in neighborhood characteristics at the tract level found no significant difference analysis by stage of foreclosure.