Do Job Destruction Shocks Matter in the Theory of Unemployment?

The current DMP approach to labor markets presumes job destruction shocks are small. We relax that assumption and also allow un lled jobs, like unemployment, to evolve as a state variable. Calibrating an otherwise standard DMP framework, we identify a remarkable, (almost) perfect, fit of the empirical facts as reported in Shimer (2005, 2012). The results, how- ever, are also consistent with the insights of Davis and Haltiwanger (1992): that unemployment volatility is driven by large but infrequent job separation shocks. The approach not only provides an important synthesis of two litera- tures which, in other contexts, have appeared contradictory, it also identfies a more traditional view of the timing and progression of recessions.