The Effect of Migrating on Young People's Education: A Cross-National Analysis

Despite longstanding scholarly interest in migrant academic achievement, there has been little work to isolate an effect of migrating on student outcomes. This paper develops a causal inference and dissimilation framework for studying migrant outcomes and implements it to evaluate the effect of migration on young people's learning outcomes on a global scale. Using test score data from the 2012, 2015, and 2018 Programme for Student Assessment, it compares 8,726 migrants in 42 destination countries to 603,280 counterparts in 55 countries of origin. After adjustment for individual characteristics using Generalized Random Forests and Augmented Inverse Probability Weighting, multilevel techniques developed for meta-analysis aggregate effect estimates for each country pair and test the effects of country-level mediators. Generally, migrant students obtain slightly lower scores on these exams than similar stay-at-homes in their countries of origin, but results vary by countries. Children who migrate to wealthier and more democratic countries tend to outperform stay-at-homes, and there is partial evidence that between-school tracking is associated with lower scores relative to stay-at-homes. In addition, children who migrate at older ages and who have two rather than one migrant parent tend to score lower. Despite their somewhat weaker performance than stay-at-homes, these migrants' dissimilation results paint a rosier picture than comparisons to local non-migrants, where estimates are much more negative.