Income Maximization and the Sorting of Emigrants across Destinations

Recent data on emigrant schooling suggests that emigrants are almost universally more skilled than non-emigrants from the same country. For the case of poor source countries and rich destination countries, this runs counter to the negative selection predicted by the Borjas-Roy model of income maximization. We present a different income maximization model that admits positive selection while at the same time making sharp predictions that can be tested with data on immigrants in multiple destination countries. The model predicts that emigrants from a given source country should sort themselves across destination countries according to the rewards to skill among destinations: destinations with larger rewards to skill should receive a higher-skilled mix of immigrants on average than destinations with lower rewards to skill. We test this prediction using data on emigrants from 192 source countries to a dozen OECD destination countries. Our tests are generally consistent with positive sorting. Furthermore, our results show that destination-country immigration policy has important effects on the skill composition of migrants. Economic and policy factors can explain a substantial share of the immigrant skill differences that we observe among destination countries.