The aggregate effects of long run sectoral reallocation
In this paper, I estimate a series of long run reallocative shocks to sectoral employment using a stochastic volatility model of sectoral employment growth for the United States from 1960 through 2011. Reallocative shocks (which primarily measure construction and technology busts) have little effect on the natural rate of unemployment or on long run productivity, but there is mild evidence that they are recessionary. A broad class of theoretical models suggests that the contractionary effect of a reallocative shock should come from the direct aggregate effect of the underlying shock and not from human capital mismatch. Looking at the period of the Great Recession, reallocation has had no detectable effect on the natural rate of unemployment and can count for a 0.5% rise in cyclical unemployment from 2007 through the end of 2009 and 0.3% through the beginning of 2011.