Migration of workers from developing to developed countries and the resulting remittance flows are important development policies. World Bank calculations show that restrictions on international migration have larger welfare costs than the more widely studied restrictions on international trade. But estimated gains from migration may be affected by selection bias, with differences in outcomes for migrants and non-migrants reflecting unobserved differences in ability, skills, and motivation, rather than the act of moving itself. This poster illustrates this selection bias in commonly used statistical corrections for nonrandom selection. A unique survey conducted by the authors of Tongan migrants in New Zealand, and of non-migrants in Tonga is used. New Zealand allows a quota of Tongans to immigrate each year with a lottery used to choose amongst the excess number of applicants. Experimental estimates of the income gains from migration are obtained by comparing the incomes of migrants who were successful in the lottery to the incomes of the unsuccessful applicants who stayed in Tonga. We also conducted a survey of individuals who did not apply to migrate. Comparing this non-applicant group to the migrants allows us to use non-experimental methods to obtain alternate estimates of the gains from migration. Comparison of the two sets of estimates finds that non-experiment methods overstate the income gains to migration by 11 to 82 percent. Thus, assessments of global gains from increased international migration are likely to be sensitive to the modelling of selectivity bias.