Wage Discrimination: Reduced Form and Structural Estimates

Abstract

Regressions explaining the wage rates of white males, black males, and white females are used to analyze the white-black wage differential among men and the male-female wage differential among whites. A distinction is drawn between reduced form and structural wage equations, and both are estimated. They are shown to have very different implications for analyzing the white-black and male-female wage differentials. When the two sets of estimates are synthesized, they jointly imply that 70 percent of the overall race differential and 100 percent of the overall sex differential are ultimately attributable to discrimination of various sorts.