Estimating the Causal Effects of Education on Wage Inequality Using IV Methods and Sample Selection Models

We develop semiparametric and instrumental-variables approaches to self-selection problems in comparisons of wage inequality between educational groups. We propose new estimators to identify the causal effect of college education on scale parameters of wage distributions, using symmetry conditions on the joint distribution of outcome and selection errors, along with kernel weighting schemes. A simulation study indicates that the proposed estimators perform well in finite samples. We illustrate these methods with a well-known study by Card (1995), using college proximity as an instrument for schooling. Estimation results suggest that college education significantly increased the degree of wage inequality in 1976. JEL Classification: C24, C14, C13.