Mean and quantile regression Oaxaca-Blinder decompositions with an application to caste discrimination

Abstract

This paper extends the Oaxaca-Blinder decomposition method to the quantile regression random-coefficients framework. Mean-based decompositions are obtained as the integration of the quantile regression decomposition process. This method allows identifying if the observed differences between two groups differ across quantiles, and if so, what is the contribution to the mean-based Oaxaca-Blinder decomposition. The proposed methodology is applied to the analysis of caste discrimination in Nepal. The results indicate that much of the discrimination occurs at the lowest quantiles, which implies that disadvantaged groups are the ones who suffer the most caste discrimination.