This seminar will empirically examine the effects of exposure to ethnic diversity during school age on students’ human capital outcomes, as measured by the years of schooling completed. Combining three data sets from Indonesia, we link children’s neighbourhood-ethnic diversity and their average educational outcome ten years later. Diversity is measured by the Herfindahl-based index constructed at the sub-district level from the population census of 2000. We control for neighbourhood characteristics that may be correlated with population mix and density. The ordinary least squares estimates tend to support the idea that diversity is harmful, with more diverse communities being associated with lower mean years of schooling completion. However, when we employ instrumental variables to mitigate potential endogeneity issues, the diversity effects become positive and significant: more diverse neighbourhoods tend to have better school completion outcomes. The mechanism behind this is likely through exposure to greater pools of talent.