Abstract

We study scar formation and persistence after a house price bubble has burst using data on 3,089 US counties and county equivalents over the period 1980q1–2019q4. We date house price booms and busts for each county, and identify periods with explosive house price developments. Applying a sharp bubble definition to the data, we observe the regularities that the probability of a housing bubble increases when housing supply is inelastic and when access to credit is easy. We differentiate between non-bubble price accelerations and bubble price accelerations, and demonstrate that there is scar formation after the latter. Conditioning on a set of factors, including county-fixed effects, our results show that house price reductions are larger and macro aggregate responses are stronger in areas in which there was a house price bubble. In particular, areas that experience a housing bubble burst are areas in which, subsequently, there are stronger and longer increases in unemployment and decreases in household income.