A Quantitative Theory of Information, Worker Flows, and Wage Dispersion

Abstract

Employer learning provides a link between wage and employment dynamics. Workers who are selectively terminated when their low productivity is revealed subsequently earn lower wages. If learning is asymmetric across employers, randomly separated high-productivity workers are treated similarly when hired from unemployment, but recover as their next employer learns their type. I provide empirical evidence supporting this link, then study whether employer learning is an empirically important factor in wage and employment dynamics. In a calibrated structural model, learning accounts for 78 percent of wage losses after unemployment, 24 percent of life-cycle wage growth, and 13 percent of cross-sectional dispersion observed in data.