flattening, either in simple cross sectional comparisons or in more sophisticated longitudinal estimators. Lemieux (1998) also uses longitudinal data to examine the apparent effect of unions on the dispersion of wages controlling for observed and unobserved skill components. This analysis suggests that some of the apparent reduction in variance in the union sector may be due to selectivity, rather than to a within-sector effect. Unfortunately, this inference is confounded by the potential selectivity of the group of union- status changers, and the fact that the variability of wages may be temporarily high just before and just after a job change. Overall, Lemieux (1998) concluded incorporating unobserved heterogeneity effects leads to a small reduction in the apparent effect of unions on male wage inequality. A similar conclusion was reached in Card (1996).Based on these findings, we conclude that the estimates of the equalizing effect of unions on maleworkers in the U.S., Canada, and the U.K. in Tables 1-3 are likely to slightly overstatethe true equalizingeffects. For women, the estimated effects in Tables 1-3 are very small anyway, and the existinglongitudinal research suggests there is no important bias. During the same period, the effect of unions on the variance of wages computed using the simplified model declined from –0.047 to –0.021 (a rise of 0.026). If this effect had remained constant over time, overall wage inequality would have grown by 31 percent less (0.026/0.082) than it actually did. The contribution of unions to the growth of inequality remains important though only about one-half as big (14 percent) when we use the more sophisticated estimates of the union effect that control for observable skills. The results for the U.K. are qualitatively similar. Between 9 and 29 percent of the 0.087 growth in the variance of log wages between 1983 and 2001 can be accounted for by the decline in union compression effects. Furthermore, in both the U.S. and U.K. union wage compression effects remained relatively constant between 1993 and 2001. In particular, the effects from the analysis that controls for workers’ characteristics are essentially unchanged in the period from 1993 to 2001. This is consistent with the slowdown in the growth of inequality in both countries in the 1990s, relative to the 1980s. As in the U.S. and U.K., the union wage compression effect has been steadily declining for
23 Canadian men since 1984. Unlike the U.S. and U.K., however, overall inequality has remained very stable in Canada over time, so overall inequality would have actually declined if union wage impacts had remained at their 1984 levels. Several developments may have offset the pressures toward increased
You've reached the end of your free preview.
Want to read all 47 pages?
- Fall '13
- Economics, Trade union, Nonunion