Quantile decomposition of performance differential among military schools in Brazil

Abstract

This paper aims to evaluate the performance differential of Brazilian military schools along the test score distribution of the 2017 National Basic Education Assessment System with unconditional quantile regressions and the Oaxaca-Blinder decomposition. The results show that among highest deciles of the scores most of the differential is due to explained factors, especially the better infrastructure of the military schools compared to other public schools. Conversely, among the lowest deciles most of the differential cannot be explained by observable factors, thus suggesting a positive effect of the military administration on these students, although a share of the impact might be attributable to self-selection bias.