The one anomaly in the regression is the impact of per-capita GDP in the source country. This variable has a strong positive effect on the log entry wage, but a strong negative effect on the rate of wage growth. The negative correlation reported in the first column of Table 7, however, turns out to be very sensitive to model specification. Consider, for example, the regressions reported in columns 3 and 4, which add a vector of country-of-origin fixed effects to the specification. Most of the source country variables have the same impact as in the simpler regression, so that a decrease in income inequality within the country raises the rate of wage growth of immigrants in the United States. The coefficient of per-capita GDP, however, becomes insignificant.
Finally, it is worth noting that the source-country characteristics—which, at best, are rough measures of the effective human capital stock of a particular cohort of immigrants— explain about 60 percent of the dispersion in wage growth among the various cohorts. In other words, source country characteristics matter a great deal in determining the rate of immigrants in the Unite States.
The regression specifications reported in columns 3 and 4 of Table 7 include the average educational attainment of the immigrants at the time of entry. Column 3 implies that a one-year increase in educational attainment increases the rate of wage growth by 1.2 percentage points. Note, however, that the independent impact of educational attainment disappears when the regression includes a vector of country-of-origin fixed effects.
It would be of interest to include measures of English language proficiency in the wage growth regressions.