Analysis of the public-private gap in educational attainment is important for the formulation of policies for improving education quality. This article presents an analysis of this differential in Brazil, using a decomposition method that reveals valuable information for each quantile of the conditional distribution of scores. Results indicate that private schools perform better at all quantiles of the distribution, and the gap is as large as 1.5 standard errors around the 60th percentile. Furthermore, the counterfactual analysis suggests that private schools perform relatively worse in the lower end of the score distribution; counterfactual scores for the weakest students would be below those observed in public schools.
quantile regression; counterfactual decomposition; educational performance; quantile treatment effects
