Determinants of labor force migration.

This dissertation makes use of the Panel Study of Income Dynamics (PSID) to test the notion that job-history characteristics of the family head are an important determinant of whether migration will take place. An overlapping sample of families, whose heads did not change over a three year period, was constructed. This sample was then divided into six groups by whether the head had experienced a voluntary job change, an involuntary job change, or no job change, and by whether the family had experienced a 'move'. A strong advantage of using panel data is that a clear and unambiguous time sequence of events can be established. For example, using job tenure as an explanatory variable is meaningful only if it is possible to measure job tenure prior to the move. (Use of data which lack the panel feature could result in measurement of job tenure after the move so that the short observed tenure would be a result rather than a cause of migration.) This ability to place events in time means that the direction of causality can be established with more confidence. The results of the estimation support the assumption that a proper model of migration behavior can only be developed if account is taken of the different work histories of family heads. Although the coefficients of some variables were significantly different from zero across all of the work history groups (home-ownership, size of city of residence, age, sex, and race of the family head) other variables were significant in one or more of the job-change groups, but not necessarily for all of the groups or for the sample as a whole. The number of children, whether the wife worked, job tenure of the head, receipt of unemployment benefits, income from welfare, income from assets, and husb and 's and wife's real hourly earnings are examples of variables which do not affect job-change groups uniformly. Since the migration-income interaction plays as important role in the literature, the equations were estimated with and without the income interaction. Few of the coefficients on the interaction terms were significant.PhDEconomicsUniversity of Michiganhttp://deepblue.lib.umich.edu/bitstream/2027.42/162009/1/8906990.pd