Does union activism increase workers’ wages?
by Nathan Wilmers
In October 2016, after a month-long strike, 750 dining hall workers at Harvard University won an increase in their minimum annual salary to $35,000. This wage boost seems to demonstrate the pay-offs to union activism.
But, for skeptics the link is not so clear. The wage increase happened while the labor market tightened for low-skilled workers, as the Massachusetts unemployment dropped to 3.6%. Maybe Harvard would have struggled to retain good dining hall workers at a lower salary.
And Harvard is rich. Absent union pressure, the University would probably still pay dining hall workers more than its less well-funded peer institutions. Maybe the observed 10% to 25% national union wage premium comes from unions organizing employers, like Harvard, that would have paid higher wages with or without a union.
For these reasons, recent research has been skeptical about unions’ capacity to increase workers’ wages. Since the early 1980s, labor union membership density has been cut in half, from around 1 in 5 workers to 1 in 10 today. If product markets are getting more and more competitive, employers’ margins should narrow, and there might be scant room for unions to bargain. Studies of close union representation elections show that when a union barely wins, wages do not increase more than when a union barely loses.
So do unions matter for workers’ wages? Or do they just get credit for pay increases that would have come along anyway?
To distinguish the effects of unions from those of changing market conditions, I examined the effect of union activism above and beyond the market position of workers and their employers. What would have happened to the Harvard dining hall workers if their union hadn’t done anything for them, but Harvard was still rich and the unemployment rate was still low?
To answer this question, I look directly at what happens when unions engage in more or less activism on behalf of their members. In extreme cases, like that of the Harvard dining hall workers, this activism could look like a strike. But unions also conduct everyday activism like printing a contract update newsletter about pay negotiations, or vigorously pursuing grievances on wage-related contract violations. If union activism pays off, then when unions do more, their members’ earnings should increase.
To sort out whether or not union activism raises wages, I tried to find situations where unions randomly get more active. First, I looked at returns from unions’ $23 billion financial investments. I isolated instances when a union’s investment portfolio performed randomly better or worse than overall market returns. In better years for a union’s investments, the union has more resources and can engage in more activism on behalf of its members.
Second, I considered the timing of union officer elections. Because union officials worry about losing internal union elections, they seem to spend more union money on activism in the lead-up to union elections.
Uncertain investment returns and pre-scheduled union officer elections are uncorrelated with tight labor markets and unaffected by the profitability of union employers. So, when idiosyncratic union investment returns or officer elections boost union activism, any ensuing wage increases can be attributed to union activism.
I find that in both of these cases, increased union activity increases a measure of union members’ earnings. A one percent increase in union spending increases a proxy for union members’ wages between 0.15 percent and 0.30 percent. These wage effects are larger in years of active collective bargaining, and when unions increase spending in ways that could pressure companies.
So, it looks like union activism actually does increase union members’ wages.
A key caveat to all of this: a big limitation of my data is that I cannot measure union members’ wages directly. Instead I use the union dues collected per member, which are indexed to union members’ earnings. This indirect measure could be misleading, so I carefully consider potential biases and test the validity of the measure. For instance, I find that only dues indexed to wages, and not unindexed dues, increase with investment shocks and elections, which is reassuring.
Overall, this research suggests that even in an era of union decline, unions still matter. It also suggests that unions might not need wall-to-wall dominance, like the Teamsters or Autoworkers enjoyed in the 1950s, to increase their members’ wages. Union activism per se, not just workers’ market position, increases pay for union members.
These findings have implications for how we might get U.S. workers a raise. From 1947 to 1979, earnings for the bottom 90% of U.S. workers increased by an average of 2% each year. Since 1979, annual earnings increases stagnated to an average of 0.5%. During the same period, the labor share of national income has inched down by around 5 percentage points.
The continued influence of union activism on wages suggests that further union decline will extend the period of stagnating median wages and declining labor share of income, while even partial and uneven union revitalization could begin to reverse it.
Nathan Wilmers is a doctoral candidate in sociology at Harvard University. This post summarizes findings from “Labor Unions as Activist Organizations: A Union Power Approach to Estimating Union Wage Effects” in Social Forces.
Image: International Ladies Garment Workers Union Photographs via Wikimedia Commons (CC by 2.0)