Post-Migration Earnings Profiles: An Application of Human Capital and Job Search Models

Several studies have noted that post-migration earnings of migrants rise more rapidly than earnings of non-migrants 13, 42; 9. Human capital models would attempt to explain this by differences in the investment profiles of migrants and non-migrants. These differences could be due to migration selectivity, whereby migrants have different investment incentives due to ability or motivation; or to the loss of human capital by migrating and the recoupment of this capital after migration. Search models would explain this phenomenon by imperfect information. Migrants may have less information about wage distributions at the destination than non-migrants, enter the destination job market at a low wage and engage in subsequent post-migration search for better earnings. Or, as Yezer and Thurston 15 have suggested, search may be selective and migrants whose reservation wages are not met after some time at the destination leave; i.e., we observe migrants having steeper earnings profiles since those who do not obtain high earnings leave. As one might expect, neither of these pure models is alone sufficient to explain the factors that create the different earnings streams. Instead, they can be viewed as complementary explanations.' The failure to recognize that migration involves both job search and human capital changes could lead to biased tests of hypotheses originating from only one model. A good example of this, and the one addressed in this paper, is testing hypotheses regarding the source of differences between experience-earnings profiles of migrants and non-migrants. For example, Yezer and Thurston 15 establish and test the hypothesis that the distance migrated should be positively associated with income gains after migration. They find the expected relation using regressions with a distance dummy.2 However, it is well known that more educated individuals travel greater distances 11 and generally have steeper earnings profiles than less educated persons. The correlation between education and distance migrated makes any test of the Yezer and Thurston hypothesis biased toward accepting their search model when education is not controlled.