In the wake of the labor market and housing market crises of the late 2000s, media reports and research studies have documented increased “doubling-up” of families as ever greater numbers of young adults returned to their parents’ homes or were slower to exit than in years past (Mykta and Macartney, 2011, among others).  A sign of the times, a 2009 USA Today article began: “Love isn’t all that’s keeping family together today. The bruising housing market is too.”  Other reports have pointed to rising rates of cohabitation resulting from the economic crisis in addition to the secular rise that was already underway (Kreider, 2010).  While aggregate trends are valuable, less is known about relative impacts on individuals and subgroups – for instance, those in their late 20s may not have yet purchased a home as compared to those in their 30s and 40s who may have already been homeowners.  Further, to what extent was it events in the MSA-area labor market or in the MSA-area housing market, or both, that affected individual decisions to double-up, own a home versus rent, or cohabit rather than marry? 

In this study, we append MSA-level data on housing and labor market conditions for 353 MSAs to individual-level data on doubling-up, homeownership, and cohabitation from the American Community Survey (ACS) for the period 2005 (pre-crisis) through 2011.  We build on our earlier research (Rogers & Winkler, IZA Paper No. 7263, March 2013) which identified considerable subnational variation in the relative timing of the housing and labor market crises.  We found that MSAs could be categorized into one of three types: "housing crisis first, labor market crisis second" (akin to the national picture), "labor market crisis first, housing crisis second" (as it turns out, the experience for the majority of MSAs), and those where the crises were effectively concurrent.  We make use of this categorization scheme in the present study.  This is an important advancement over earlier studies which have either looked at the effect of the housing crisis or the labor market crisis on decisions, but have not adequately accounted for the effects of both (e.g. Mykta and Macartney, 2011; Molloy and Shan, 2011).  In the empirical work, we focus on the role of negative equity, local-area foreclosures (using proprietary data obtained from CoreLogic), housing price expectations (data from FHFA), and depth of the employment decline.  By exploiting variation in time, place, and individuals, we are able to disentangle secular trends from cyclical patterns, which is especially important in drawing conclusions about the impact of the recent crisis and making policy recommendations.