Families, Neighbourhoods and Children's Educational Outcomes

Understanding whether differences in the outcomes of children who grow up in different locations represent location effects or residential sorting is an important question in economic research and policy. We estimate location effects by controlling for differences in observed family characteristics across locations using machine learning and rich Swedish administrative data. We focus on university enrolment, and find that observed family characteristics explain 70-80 percent of the differences between children who grow up in different locations. The remaining unexplained gap is an upper bound for the size of location effects, as it also includes the effects of unobserved family characteristics. We systematically analyse heterogeneity in the size of the unexplained gap for children from different types of families, finding that it is larger for children of low-educated parents. The unexplained rural-urban gap is larger for boys and second-generation immigrants, while the unexplained gap between rich and poor neighbourhoods in cities is larger for girls and those with native-born parents. Overall, the results suggest that differences in university enrolment across locations are mostly due to residential sorting of families rather than location effects.

Researchers

Yaroslav Yakymovych, Majken Stenberg, Matz Dahlberg, Torsten Santavirta