Voucher Program Incentives and Schooling Performance in Colombia: a Quantile Regression for Panel Data Approach∗

Abstract. Angrist et al. (2002) study is reconsidered using a relatively new quantile regression approach for panel data models. The evidence suggests that the effect of winning a Colombian voucher is largest in the lower tail of the educational attainment distribution, a possibility that was conjectured by the original authors, but could not be confirmed empirically using existing regression methods. This finding supports the hypothesis that lottery winners’ incentive to study harder could account for the impact of the vouchers, including lower repetition rate.