Default When House Prices are Uncertain

During the 2000-2011 housing boom and bust, changes to self-assessed house prices in 20 different metro areas are highly correlated with changes to the Case-Shiller-Weiss (CSW) house price indexes, but do not change percent-for-percent with the CSW indexes during the boom or bust. We argue this evidence is consistent with an environment in which homeowners imperfectly observe the value of their home and update their guess of home value using a Kalman filter. Using data on self-assessed house prices and the CSW indexes, we estimate the parameters of a simple model of imperfectly-observed house prices. In an out-of-sample forecast exercise, we show the model is nearly perfectly able to replicate the sequence of self-assessed house prices from 2006-2011 in many MSAs, from places experiencing a modest boom like Atlanta and Detroit to the "bubble" areas like Miami and Phoenix. We then specify an economic model of the optimal default decisions of homeowners when homeowners imperfectly observe the value of their home. We show that, relative to a standard model of default, homeowners delay their decision to default after a bad house price shock. We argue this phenomenon helps to explain default patterns in the United States during the 2007-2011 housing bust.