We analyze an international migration episode for which we are able to gather individual-level data covering all relevant countries, namely the exodus of Ecuadorians to Spain and the US in the aftermath of the economic collapse of 1999. Specifically, we produce selection-corrected predictions of counterfactual individual earnings and use them to estimate a discrete-choice migration equation that allows for correlated errors across destinations and a rich structure of migration costs. We find that earnings significantly shape individual migration decisions, even in an episode in which Ecuadorians mostly chose Spain where earnings were lower than in the US, and they contribute to explaining the observed composition of migration flows. Moreover, our estimates show that changes in earnings at a particular destination have a larger effect on destination choice conditional on migration than on the scale of migration.