Revisiting the contribution of schools to cognitive gaps: Evidence from Peru

Previous studies have found that differences in school characteristics can explain a significant share of the difference in cognitive skill between advantaged and disadvantaged children. These estimates, however, can be biased if the effect of schools depends on the skill attained by children before they enter school. I use longitudinal information on test scores for a large sample of Peruvian children to reassess the importance of schools for the urban/rural cognitive skill gap observed at age 8. I propose a decomposition strategy that allows the skill attained during early childhood to influence the effect of schools. I find a negative relation between early childhood skill and the effect of schools in the rural domain and confirm that schools account for a significant share (around 34%) of the urban/rural cognitive skill gap. This reveals that it is possible to exploit the equalizing potential of improvement in rural schools despite the large cognitive skill deficits of rural children. It is not "too late" for rural children currently at school.