Unobserved Worker Ability, Firm Heterogeneity, and the Returns to Schooling and Training

We offer in this paper an alternative way of controlling for worker and firm heterogeneity. Our strategy assumes that the gap between the individual wage and the firm average wage, unexplained by differences in observable characteristics, gives the extent to which the individual unobserved ability deviates from the unobserved average ability in the firm at which she/he works. Based on an extended set of longitudinally observed attributes, including participation on workplace training, our results indicate that the typical human capital function covariates are highly correlated with unobserved ability which of course leads to the presence of a large bias in standard OLS regressions. We also found that high ability workers are more likely to switch jobs, while at the same time the quality of job matching is expected to increase. In turn, after controlling for worker and firm effects, the gender gap virtually vanishes. Given the visible impact of unobserved ability on wage determination, it follows, in particular, that standard state subsidies to firm training do entail the risk of greater wage inequality.