Panel estimates of the gender earnings gap. Individual-specific intercept and individual-specific slope models

This paper develops individual-specific slope (as opposed to individual-specific intercept) models to account for unobserved heterogeneity in estimating male-female wage differentials. Estimates from these models are compared to traditional OLS, as well as to fixed- and random-effects individual- specific intercept approaches. We find unambiguously that unobserved heterogeneity accounts for about 50% of the male-female wage gap, with individual-specific slope models explaining a slightly smaller proportion of the wage gap. Based on inferences from life-cycle earnings models, we conclude that most individual-specific differences manifest themselves early in one's work career, probably even in the type of schooling received.