Employer Learning, Productivity, and the Earnings Distribution: Evidence from Performance Measures
利用一家大型企业20年的薪酬与绩效数据，研究雇主学习和动态生产率异质性如何共同导致收入差距随工龄扩大，发现两者均重要，且不完全学习会扭曲人力资本投资激励。
Pay distributions fan out with experience. The leading explanations for this pattern are that over time, either employers learn about worker productivity but productivity remains fixed or workers' productivities themselves evolve heterogeneously. We propose a dynamic specification that nests both employer learning and dynamic productivity heterogeneity. We estimate this model on a 20-year panel of pay and performance measures from a single, large firm. The advantage of these data is that they provide us with repeat measures of productivity, some of which have not yet been observed by the firm when it sets wages. We use our estimates to investigate how learning and dynamic productivity heterogeneity jointly contribute to the increase in pay dispersion with age. We find that both mechanisms are important for understanding wage dynamics. The dispersion of pay increases with experience primarily because productivity differences increase. Imperfect learning, however, means that wages differ significantly from individual productivity all along the life cycle because firms continuously struggle to learn about a moving target in worker productivity. Our estimates allow us to calculate the degree to which imperfect learning introduces a wedge between the private and social incentives to invest in human capital. We find that these disincentives exist throughout the life cycle but increase rapidly after about 15 years of experience. Thus, in contrast to the existing literature on employer learning, we find that imperfect learning might have especially large effects on investments among older workers.