Class composition effects and school welfare: evidence from Portugal using panel data

Using student-level longitudinal data of 6th graders I estimate class composition effects impacting on individual academic achievement. The richness of the dataset allows to tackle endogeneity stemming from between and within-school non-random sorting of students and of teachers and other confounding factors through the inclusion of many control covariates that characterize the students’ cumulative process of learning and several fixed effects, namely school, teacher and cohort fixed effects. I find that increasing the percentage of high achievers, in a 6th grade class, has a negative effect on student performance. Larger shares of low-income classmates improve performance for non-low-income students. The shares of male and foreign students yield non or faintly significant results. Using the setup of a particular school, representative of the sample, I also compute improving classrooms’ allocation of students by rearranging the existing students through the existing classes using the estimates of the education production function and different social welfare functions. This way I assess how the actual distribution of students across classes of a given school-grade deviate from what can be considered an improving distribution of classmates. Pareto improving allocations were not found, nevertheless utilitarian welfare functions yield marginally improving allocations