INITIAL WAGE, HUMAN CAPITAL AND POST WAGE DIFFERENTIALS

Insufficiency in information with which firms judge the productivity of a worker for the first time in the market creates more randomness in initial wages than in later wages. This paper examines whether the initial randomness in wages may have a persistent effect on post wages. We set up a human capital accumulation in which an individual may respond to the positive error in initial wage by adjusting hours worked thereafter in her career, and consequently may receive higher future wages than those who draw a negative error in initial wages but otherwise are equivalent. The model predicts that the initial wage, in particular, its random component, is a persistently important factor having positive effect on future wages. Using data from the National Longitudinal Survey of Youth 79, we find empirical evidence that this effect is indeed positive and persists even after 20 years since the initial entry to labor market. The decomposition of initial wages by both parametric and nonparametric IV methods further shows that this effect is derived by the random component, not the observable component, of the initial wage. It implies that the observed cross-sectional wage variation within group can be accounted for the initial randomness in wages.