Skilled migrants typically contribute to the welfare state more than they draw in benefits from it. The opposite holds for unskilled migrants. This suggests that a host country is likely to boost (respectively, curtail) its welfare system when absorbing high-skill (respectively, low-skill) migration. In this paper we first examine this hypothesis in a politico-economic setup. We then confront the prediction of the theory with evidence. In doing so, we reckon with an endogeneity problem that arise because the skill composition of migration is itself affected by the generosity of the welfare state.