How Structural Is Unemployment in the United States?

Abstract

In this paper, the role of matching efficiency (equivalently, mismatch) at the aggregate level in driving unemployment fluctuations is estimated using a TVP-SVAR model. Modelling mismatch at the aggregate level sidesteps the problematic implicit assumption of orthogonality of sources of mismatch at disaggregated levels (industrial, occupational, geographical, etc.) and is not sensitive to the level of disaggregation by construction. Observing that estimated aggregate matching efficiency lags business cycles, I identify a structural shock to aggregate matching efficiency using standard timing restrictions. Based on impulse response analysis and forecast error variance decompositions, I conclude that the matching efficiency shock explains no more than 20% of the variation in unemployment in the United States between 1967-2013, whereas aggregate shocks explain well above 70% of unemployment fluctuations. Related, the rise in the unemployment rate during the Great Recession is dominated by a slump in aggregate demand rather than driven by structural factors.