Pre-print, Working paper: I build a simple theory of self-selection into migration and immigration policy formation. I show that any immigration policy affects immigrants skill composition, and this effect may drive the policy outcome in the receiving country. For example, restricting immigration when it is low skilled may worsen immigrants' self-selection and thus the receiving country skill distribution. Hence, understanding the migration decision becomes crucial for analyzing the political economy of immigration. By this composition effect, some natives may support further restrictions even though current immigrants are not harmful for them, and immigration restrictions may be optimal even in a purely utilitarian world.