In the recent housing crisis, housing prices fell an average of 34 percent across the nation. Current appreciation rates leave over 6 million families exposed to negative equity and the risk of foreclosure. This study employs a novel dataset and empirical strategy to estimate the causal effect of negative equity on children’s education outcomes in Florida. While some recent studies (e.g., Dastrup and Betts, 2011; Bradbury, Burke, and Triest, 2013) find a negative effect of foreclosure on children’s attendance and math and reading test scores, no studies to date examine the effects of negative equity on children’s education outcomes.

Examining the effects of negative equity on children’s education outcomes is important – not only because of the large number of families exposed to negative equity, but also because ironically, being underwater might benefit children academically by making their housing situations more stable.  Negative equity may reduce the family’s mobility because the family would have to resort to savings to pay the mortgage balance in the case of home sale; that spatial lock-in may reduce the number of times the children might otherwise move between houses or between schools. 

Our study also addresses a major empirical issue faced by any study that hopes to untangle the relationship between foreclosure and/or negative equity and children’s outcomes. Negative equity is endogenous, and negative equity and foreclosure are outcomes that are driven at least in part by family attributes that are also correlated with children’s education outcomes. We address this endogeneity bias using a unique, three-pronged empirical strategy. First, we develop a proxy for the risk of negative equity based on county level housing price changes, constructed so that it does not depend upon either the change in housing prices at the individual or neighborhood levels.  We then link schools in Florida to their address zip code as a proxy for school attendance zones. We use statewide, longitudinal student data to examine student test scores as a function of this proxy for negative equity, while controlling for a comprehensive set of fixed effects: student level fixed effects, purchase year by current year fixed effects, and zip code/school by current year fixed effects.  Finally, we focus on a sample of students who moved into the state during the run up to the crisis (2003-04 to 2005-06 school years). 

Our results show that children in Florida had much higher test score growth if their original zip code of residence in the state and year of moving into state implied a high likelihood of negative equity.  In others words, students who should be most risk of negative equity and foreclosure had the most improved test scores during the foreclosure crisis.  In fact, the test score gains are practically monotonic in the risk of negative equity.  This relationship is robust to a number of alternate specifications, and present in both simple difference and difference-in-difference analysis.