Circular labor migration is a core feature of low-income labor markets. Yet, evidence on how
this migration affects education investments in sending communities is limited due to lack of high quality
data and challenging identification issues. This is especially true in Africa, where children can substitute
for migrant adult labor. In this paper, we estimate the net effect of international migration on human
capital accumulation of children by exploiting two large migrant labor shocks in sending communities in
Malawi. An international mine labor treaty signed in 1967 initiated a 300% increase in the flow of
Malawians to South Africa. Seven years later, a mining plane crash prompted the Malawian government
to halt and reverse this expansion until 1977. Our strategy compares differences in long run human capital
accumulation across high and low shock areas, among cohorts eligible and ineligible for primary school
during the shock years. We use historical locations of mining recruiting stations to construct measures of
district-level exposure to this rapid expansion and contraction of foreign employment and earnings. We
match this spatial variation in migration costs to cohort-specific education outcomes from newly digitized
1977 and 1998 Census data. Both shocks to migration had large, positive impacts on education. Age
eligible cohorts with the highest shock exposure attained 10 to 15% more schooling and the share with
any primary schooling rose by 5 to 8%. Neither school supply-side interventions nor internal migration
dynamics account for our results. These long run effects are only apparent in districts without agricultural
estates, where child labor is less substitutable for missing male labor.