Labor Reallocation, Employment, and Earnings: Vector Autoregression Evidence

Abstract

An increasing number of data sources have measured the components of reallocation of jobs across employers and workers across jobs. Whether and how job reallocation across employers and excess worker “churn” affect other measures of the health of the U.S. economy remains an open question. In this paper, we present time series evidence for the U.S. 1993-2013 and consider the relationship between labor reallocation, employment, and earnings using a vector autoregression (VAR) framework. We find that labor market churn Granger-causes higher employment and lower unemployment, while job destruction does the opposite. We also find more limited evidence that churn and job destruction predict increased earnings, although this is not found for all earnings measures.