In this study the concept of the marginal migrant is used to develop a theoretical model of the skill composition of US immigrants. Two skill classes are distinguished (high-skill - - professional, technical, and kindred, as well as managers, officials, and proprietors - - and less-skill, which includes all other occupations). Because of the relationship between age and skills, three age classes also are distinguished (20-34, 35-49, and 50-64). Since most high-skill immigrants are males who enter the US as numerically restricted, we study only numerically restricted male immigrants. Annual data (1972-1991) for 86 source countries are pooled, and the Hausman-Taylor instrumental variable technique is used to estimate regressions that satisfy adding-up restrictions.
Along with measures of differential economic opportunities, migration costs, and US institutional controls, the empirical model incorporates a unique set of variables relating to social programs in source countries. It is expected that social programs may provide very important forces in shaping the composition of migration flows, especially since many countries have social program systems that are more extensive and more generous than that of the United States. Four broadly defined types of social programs are studied: (1) old-age pension programs, (2) sickness and maternity programs, (3) unemployment insurance, and (4) family allowance systems. The social program vector includes several variables relating to whether a specific type of program exists, the population covered by a particular program, and controls relating to the financing of such programs. The basic idea that underlies the inclusion of the these social program variables in the model is that the presence or absence of different types of social programs (as well as coverage and financing features) will influence differential economic advantage among members of the source country population, since the programs can entail income transfers among indigenous population groups. Other things being equal, those individuals who receive net social program benefits (costs) should have a lower (higher) likelihood of becoming international migrants.
The preliminary empirical analysis indicates that the extent to which social programs influence migrant skill composition is related to the age of potential migrants. Whereas the skill composition of the youngest migrants appears to be influenced only by the presence of old-age pension programs, the composition of the oldest group (50-64) seems to be affected by all social programs except family allowance systems. Skill composition also appears to be influenced by the specific social program's coverage as well as the method used to finance the program.