Blinder-Oaxaca Decomposition for Tobit Models

Abstract

In this article, a decomposition method for Tobit models is derived, which allows the differences in observed outcome variables between two groups to be decomposed into a part that is explained by differences in observed characteristics and a part attributable to differences in the estimated coefficients. Monte Carlo simulations demonstrate that in the case of censored dependent variables this decomposition method produces more reliable results than the conventional Blinder-Oaxaca decomposition for linear regression models. Finally, our method is applied to a decomposition of the gender wage gap using German data.