Blinder-Oaxaca as a Reweighting Estimator

A large literature focuses on the use of propensity score methods as a semi-parametric alternative to regression for estimation of average treatments e¤ects. We show here that the classic regression based estimator of counterfactual means studied by Alan Blinder (1973) and Ronald Oaxaca (1973) constitutes a propensity score reweighting estimator based upon a linear model for the conditional odds of being treated – a functional form which emerges, for example, from an assignment model with a latent loglogistic error. As such it enjoys the status of a “doubly robust”estimator of counterfactuals as in Robins, Rotnitzky, and Zhao (1994) –estimation is consistent if either the propensity score assumption or the model for outcomes is correct. To illustrate the method, the Blinder-Oaxaca estimator is applied to LaLonde’s (1986) study of the National Supported Work program where it is found to compare favorably with competing approaches.