Unobserved Ability, Efficiency Wages, and Interindustry Wage Differentials

One competitive-market explanation of interindustry wage differentials that is not challenged by the persistence of these differentials is that they are due to differences across workers in unobserved ability or quality. In contrast, this paper explores the unobserved ability hypothesis by using test scores as error-ridden indicators of ability, and family background variables as instruments.This approach avoids two potential problems with using first-difference methods to remove omitted-ability bias from wage equation estimates of industry effects the exacerbation of measurement error from misclassification of industry, and selectivity with respect to industry changes. However, this approach may introduce other problems, for two reasons. First, it depends on using test scores that are correlated with the type of ability that is rewarded in labor markets. Second, because the test scores are undoubtedly error-ridden measures of ability, identifying assumptions are needed to correct for measurement error. Thus, the approach taken in this paper should be viewed as complementary to first-difference methods. Results indicate that ability can account for only a small portion of interindustry wage differentials in cross-section wage regressions.