What Actually Causes Mortgage Defaults, Redefaults, and Modifications

I use new household level data to simultaneously test three competing hypothesis about the primary cause of mortgage default. I consider the potential interactions of individual unemployment spells, negative equity, and unsecured debt. Unlike past studies that were only able to control for regional unemployment rates, I find that the double trigger event of job loss and negative equity is crucial in determining default as has been widely hypothesized in the literature. In every specification considered, the interaction term between unemployment and negative equity is significant indicating that double trigger events are the culprit in generating persistent defaults as hypothesized by Foote, Gerardi, and Willen (2009). I find that a person who has negative equity and who is unemployed is 10% more likely to default than a person who has just negative equity. The negative equity alone only makes a person 1.45% more likely to default, a factor of 6 smaller than the interaction probability. Unsecured debt plays a tertiary role compared to negative equity and unemployment, and past modifications make someone 20% more likely to have defaulted at the survey date. Redefaults, which are defaults after modification, are also well predicted by an interaction between unemployment and negative equity, and modifications are best predicted by past negative equity which makes someone 8% more likely to receive a modification.