Poverty has declined dramatically in Nepal since the end of the conflict. The sharp increase in
revenues primarily comes from remittances. From a household's perspective, choosing the appropriate
number of migrants is crucial: they must trade-off the direct cost and loss of local income earners
versus the potential remittances. We build a standard game theory model of remittances to emphasize
two effects of the number of migrants on the total remittances received that go in opposite directions.
On one hand, if each migrants earns more than when he stays home, then there is more to remit, and
we expect higher total remittances. On the other hand, when there are more migrants, the incentive to
free ride also increases: each of them remits less and the total remittances could decrease. We use the
last nationally representative survey to test our theoretical predictions. We find that the total
remittances received (per household member) in fact substantially decreases with the number of
migrants, in line with the free-riding argument. We use past conflict intensity in the district to predict
current number of migrants and clear our estimates of endogeneity biases. We also discuss the
plausibility of alternative explanations.