This paper analyzes the impact of anticipated old age support, provided by children to parents, on intra-family transfers and education. We highlight an education motive for remittances, according to which migrants have an incentive to invest in their siblings' education via transfers to parents, in order to better share the burden of old age support. Our theory shows that in rich families, selfish parents invest optimally in children education, while in poor families, liquidity constraints are binding and education is fostered by migrant remittances. We test these hypotheses on Indian panel data. Identification is based on within variation in household composition. We find that remittances received from migrants significantly increase with the number of school age children in the household. Retrieving the effects of household characteristics shows that more remittances tend to be sent to poorer and older household heads, confirming the old age support hypothesis.