Table 2 compares the characteristics of union and nonunion workers in the early 1970s and the early 1990s. Unionized men are typically older, less educated, and more likely to be married than their nonunion counterparts. Interestingly, the mean gap in education has narrowed over the past two decades (from 0.9 years in 1973-4 to 0.3 years in 1993). This is consistent with the data in Table 1 showing that union densities fell most rapidly for less educated workers. In 1973-4 unionized women were also older and less educated than nonunion women, but by 1993 the education differential reversed, again consistent with the decline in union membership of less-educated women and the rise in union membership of more-educated women.

One way to summarize the shifts in union membership across different age, education, and race groups and illustrate the resulting changes in the distribution of “skills” among union and nonunion members is to define skill groups based on percentiles of predicted wages in the nonunion sector, and then compare union densities in the resulting skill groups. This method is illustrated in Figure 1, which shows union membership rate for predicted wage deciles. The prediction equation is fit to nonunion workers only (by year and gender) using a very flexible functional form, and then used to assign union and nonunion workers into 10 equally-sized groups. Panel A of Figure 1 shows that in 1973-4 male union membership rates followed an “inverted-U” pattern, with highest rates for workers in the middle of the skill distribution. The 1993 data show a similar pattern, with lower membership rates for all but the highest skill group. Among women, union rates were fairly constant across skill groups in 1973-4 and have “tilted” toward more highly-skilled workers over the past 25 years.

Tags: , ,