Estimating the Effects of Unions on Wage Inequality in a Panel Data Model with Comparative Advantage and Nonrandom Selection

This article considers the estimation of the structure of wages in union and nonunion sectors. It proposes an estimator that extends standard panel data techniques to the case in which the return to the permanent component of the error term is differently rewarded in the two sectors. The econometric model is used to estimate the effect of unions on both the level and the variance of wages in Canada. The findings indicate that unions increase the average wage of workers and compress the returns to observable measures of skill and to a time‐invariant unobservable measure of skill.