This paper exploits longitudinal employer-employee matched data from the U.S. Census Bureau to investigate the contribution of worker and firm reallocation to changes in earnings inequality within and across industries between 1992 and 2003. We find that factors that cannot be measured using...