RISING WAGE INEQUALITY: Allowing for Differences Across Skill Groups

As noted earlier, there are several reasons to suspect that the naive calculations in Table 3 may overstate the role of unions in widening wage inequality. Using the framework developed in Section I it is possible to calculate union inequality effects within and across skill groups, allowing for differences in union coverage and union effects by group. In this paper I apply the framework to the “skill deciles” used in Figure 1. Table 4 shows the distributions of union densities, (unadjusted) union wage gaps, and (unadjusted) union variance gaps across skill deciles for men and women in 1973-4 and 1993. Perhaps the most interesting feature of these data is the pattern of union wage gaps across skill groups. For men, these range from 30-40 percent for the lowest skill group to -10 percent for the highest skill group. For women, the union wage gaps at the bottom of the skill distribution are comparable to those for men, but the decline in the gap for higher skilled women is less dramatic. As earlier studies (Card, 1996; Lemieux, 1992) have noted, unions seem to “flatten” the wage structure across skill groups by raising wages more for less-skilled workers.


Figure 2 illustrates the differences in the union and nonunion wage structures across skill groups and over time. To construct these figures I first fit a series of flexible wage models by sector, gender and year. For each gender I then constructed mean predicted wages in the union and nonunion sectors in 1973-4 and 1993 for each of the 10 skill groups (using the populations of 1973-4 workers).12 Finally, I plotted mean predicted wages in the union sector in 1973-4 and in the union and nonunion sectors in 1993 against mean predicted wages in the nonunion sector in 1973-4. For reference, the figures also show the 45 degree line, representing the counter-factual of equal wage structures in the union and nonunion sectors and over time.

Tags: , ,