In this paper we investigate the effect of Head Start on long term education and labor market outcomes using data from the NLSY79. The contributions to the existing literature on the effectiveness of Head Start are threefold: (1) we perform a nonparametric bounds analysis that relies on relatively weak stochastic dominance assumptions, (2) we consider outcomes
for individuals in their early 30s, and (3) we bound entire potential cumulative distributions of education and labor market outcomes. The results show that Head Start has a statistically significant positive effect on years of schooling, in particular for girls, blacks and hispanics.
For income we also find evidence that Head Start has beneficial impacts, with effects located
at the lower end of the distribution.