Sources of Wage Inequality: Decomposing the Conditional Gini Coefficient

I develop an econometric method to determine the characteristics that affect the disparity in the distribution of a positive random variable, e.g., income or wages. I relate the conditional Lorentz curve to the conditional quantile function to decompose the conditional Gini index. The proposed procedure uses polynomial approximations of the estimates of the conditional quantile regression coefficients to decompose the factors and characteristics that contribute most to the inequality of the distribution of the random variable. Moreover, this paper proposes a technique to disentangle the temporal changes in the distribution. I demonstrate the use of the method using U.S. hourly wages for the years 1986 and 2015. I find that increases in the proportion of workers with high school and associate degrees are significantly correlated to reductions in inequality of the wage distribution in the U.S. during the analysis period. JEL Classification Numbers: C13, C21, C43, D63