Abstract

Abstract: Inequality of educational opportunity is frequently considered a primary driver of inequities in well-being. A common measure of limits to educational opportunity is the magnitude of sibling correlations in educational attainment. These capture a myriad of effects, including shared effects of schools as well as processes of sorting and matching.

In this study I propose a novel variance decomposition, which identifies the unique contributions of sibs, schools and sorting effects to variation in learning outcomes. I apply this technique to test score data for over one million children across East Africa. The results indicate that schools and households jointly account for over 50 percent of learning inequalities. However, the bulk of this effect is due to positive sorting between schools and households.