A semi-nonparametric copula model for earnings mobility

In this paper we develop a novel semi-nonparametric panel copula model with external covariates for the study of wage rank dynamics. We focus on nonlinear dependence between the current and lagged worker's ranks in the wage residuals distribution, conditionally on individual characteristics. We show the asymptotic normality of the Sieve estimator for our preferred mobility measure, which is an irregular functional of both the finite- and infinite-dimensional parameters, in the double asymptotics with N,T Ç É. We derive an analytical bias correction for the incidental parameters bias induced by the individual fixed-effects. We apply our model to US data and we find that relative mobility at the bottom of the distribution is high for workers with a college degree and some experience. On the contrary, less-educated individuals are likely to remain stuck at the bottom of the wage rank distribution year after year.