Rows 6 and 7 of the two panels in Table 2 show the mean log wages of union and nonunion workers and a set of adjusted union-nonunion wage gaps, estimated from simple OLS regression models with standard covariates.10 In 1973-4, the unadjusted gaps in mean log wages between union and nonunion men and women were 0.196 and 0.230, compared to the adjusted wage gaps of 0.178 and 0.220. The similarity of the unadjusted and adjusted gaps suggests that the observed characteristics of union and nonunion workers were similar — as would be expected given the distributions in Figure 1. In 1993, however, the unadjusted wage gaps — 0.254 for men and 0.313 for women — were higher than the corresponding adjusted gaps, suggesting that union workers had higher observed skills than nonunion workers. The large unadjusted gap for women in particular is a reflection of the rising union rate across the skill distribution illustrated in the lower panel of Figure 1.

Rows 8 and 9 of Table 2 present measures of the dispersion in wages within the union and nonunion sectors. The entry in row 8 is just the standard deviation of log wages, while the entry in row 9 is the residual standard deviation after adjusting for the effects of a standard set of covariates (allowing separate coefficients in the union and nonunion sectors). These simple measures of wage inequality illustrate three important facts. First, wages are less disperse in the union sector. Second, wages of women (in either union or nonunion jobs) have lower dispersion than wages of men. Third, wage dispersion in both the union and nonunion sectors rose between 1973-4 and 1993.

Tags: , ,