Abstract
This article focuses primarily on determining the economic consequences of family migration for husbands and wives in matched married-couple families, using data from waves 1 and 2 of the National Survey of Families and Households. The analysis is designed to determine whether or not the return to migration for husbands and wives is similarly affected by their relative earning potential, as predicted by the human-capital model of migration. The study's secondary contributions include its estimation of the effect of moving on earnings for both husbands and wives within matched married-couple families and its avoidance of the problems of self-selection bias and unobserved variable bias associated with cross-sectional models by using panel-data methods. The results indicate—as predicted by the gender-role model of family migration—that the effect of family migration on individual earnings is largely a function of gender: family migration causes an increase in the husband's income and no change in the wife's income even if a wife has a greater earning potential than her husband. Thus, the study does not support the human-capital argument—that family migration decisions are egalitarian and symmetrical, such that each spouse's absolute and relative earning power is given equal weight in the migration decision. This research makes a strong statement that the gender-role model of family migration is of greater utility for understanding family migration behavior than the human-capital model of family migration.
Notes
*0.05 < α < 0.10
**0.01 < α < 0.05
***α< 0.01
a None of the estimates are statistically different from zero.
*0.05 < α < 0.10
**0.01 < α < 0.05
***α < 0.01
*0.05 < α < 0.10
**0.01 < α < 0.05
***α < 0.01
1. While this effectively eliminates individuals whose general human capital changed between wave 1 and wave 2 the possibility still exists that individuals may have differential changes in the acquisition of specific human capital that are not captured by the time change between wave 1 and wave 2.
2. See CitationClark and Withers (2002b) for a discussion of issues concerning the use of the CPI-U and some alternative approaches.
3. Even if a married-couple individual is unemployed or out of the labor force, the individual will be included in the analysis, with yearly income set to zero. While this may not be the traditional way of viewing labor-market activity, the focus here is on changes in earnings between wave 1 and wave 2 as a function of family migration behavior. Excluding individuals with no earnings in either the origin or the destination would exclude those individuals who may have the greatest losses and gains to family migration and would bias the results.
4. See CitationCooke (2001) for a discussion of the relative merits of panel data versus cross-sectional data models in migration studies.
5. Importantly, wave 1 family income is considered to be a proxy for both general and specific human capital and, as such, is assumed to reflect earning potential.
6. In the case of the estimates of differences between the parameters for husbands and wives, a model that pooled both groups together was estimated. Statistically, this joint model is identical to the separate models (CitationGujarati 1995), but it is not presented for ease of discussion. The author will provide the results upon request.