The Dark Corners of the Labor Market

What can happen to unemployment after a severe disruption of the labor market? Standard models predict a reversion to a long-run steady state. By contrast, this paper shows that a large shock may set the economy on a path towards a different steady state with possibly extreme unemployment. This result follows from the empirical behavior of the U.S. job finding rate over the last 25 years. First, I estimate a reduced-form model for the labor market and show that --once allowing for nonlinearities-- it implies a stable steady state around 5 percent unemployment and an unstable one around 10 percent unemployment. Second, I consider an extension of a basic Diamond-Mortensen-Pissarides (DMP) model in which multiple steady states arise due to skill losses upon unemployment, following Pissarides (1992). Based on only observed rates of job loss, this model endogenously explains most of the observed fluctuations in the job finding rate and the unemployment rate, thereby dramatically improving over a basic DMP model with a single steady state.