Recent research in the Quarterly Journal of Economics offers previously unseen levels of detail unraveling the relationship between labor unions and income inequality in the U.S.
The study, “Unions and Inequality over the Twentieth Century: New Evidence from Survey Data,” suggests rising union membership from the 1930s to the 1960s strongly contributed to closing the income gap between the richest and poorest Americans during those decades, with particular gains for racial and ethnic minorities. The results are largely based on responses to more than 500 Gallup surveys from 1936 to 1986.
Further, a “premium” of 10% to 20% higher income for union households compared with non-union households from the 1940s through the mid-2010s remains “relatively consistent over our long sample period, despite the large swings in density and composition of union members that we document,” the authors write.
Union density generally refers to the share of workers in an industry who are in a union. The Gallup surveys did not consistently ask about health insurance coverage and paid vacation time, so the authors do not explore benefits other than higher wages that can come with union jobs.
New measures of union density
Before 1973, economists estimated union membership at the national level. For example, Rutgers University labor economist Leo Troy used union dues revenue and other data to estimate national union membership from 1897 to 1962 for a book the National Bureau of Economic Research published in 1965.
Eight years later, the U.S. Census Bureau and the Bureau of Labor Statistics began asking about union membership as part of their Current Population Survey, a monthly survey of 60,000 households. Labor economists could, for the first time, analyze individual union membership.
But by then, union density was trending downward.
The new analysis of Gallup data shows the share of U.S. households with a union member had fallen under 30% by that time, down from a high of nearly 35% in the mid-1950s.
“Almost everything economists knew had come from this post-1970 period when unions were in decline,” says Suresh Naidu, an economics professor at Columbia University and co-author of the recent paper with Henry Farber, Daniel Herbst and Ilyana Kuziemko. “So we were like, ‘Let’s see what this looks like when you look at the period when unions were increasing.’”
The Gallup data come from a trove of surveys, available through the Roper Center for Public Opinion at Cornell University, that starting in 1937 asked respondents whether anyone in their household was a union member. The authors go back an additional year using a survey about household spending on union dues in 1936 conducted by the Bureau of Labor Statistics and the now-defunct Bureau of Home Economics.
Using those surveys and the Current Population Survey since 1973, along with other data sources, the authors find the share of union households skyrocketed from just over 10% in 1936 to roughly one-third in 1955. Union density held steady around 30% through the 1960s before dropping below 25% by 1985.
The rise of unions from 1936 to 1968 explains about 25% of the decline during that period in the Gini coefficient, a common measure of income inequality, according to the paper. The lower the Gini coefficient for a nation, the narrower the gap between its highest and lowest income earners.
After 1968, falling union membership explains roughly 10% of increasing income inequality over the next five decades, the authors find.
In 2020, about 11% of wage and salary earners were union members — 35% of them in the public sector and 6% in the private sector, according to the Current Population Survey. Union workers earned a median of $1,144 per week in 2020, compared with $958 for non-union workers. Put another way, union workers earn $1 for every 84 cents a non-union worker earns.
Although dozens of local union shops remained racially segregated in the South during the years after World War II, unions generally also “drew in disadvantaged groups such as the less educated and nonwhite households,” find Naidu and co-authors Henry Farber, Daniel Herbst and Ilyana Kuziemko.
“That suggests a reason why unions mid-century were a powerful force for equality,” Naidu says. “They brought in the people worst off in the labor market and raised their wages a lot.”
Black workers in particular continue to be represented by unions at a relatively high rate. In 2020, nearly 14% of Black workers could claim union representation, compared with 12% of white workers, 11% of Hispanic or Latino workers and 10% of Asian workers, according to the Bureau of Labor Statistics.
The rise and fall of manufacturing in the U.S.
The U.S. economy from the Great Depression through the postwar years looked very different than it does today. Before, during and after the war, unions organized some of the biggest firms engaged in domestic production for domestic buyers — think Ford, General Motors and U.S. Steel.
The 1940s specifically were “a decade of extraordinary wage compression,” as economists Claudia Goldin and Robert Margo explain in a February 1992 paper in the Quarterly Journal of Economics. The difference between the highest and lowest wage earners narrowed so significantly that Goldin and Margo dubbed those years the “Great Compression.”
As Farber, Herbst, Kuziemko and Naidu find, the Great Compression was partly spurred by unionization.
Likewise, an August 2018 paper in The Economic History Review finds that parts of the U.S. where unions grew briskly during the 1940s also show greater reductions in wage inequality than places where unions expanded less quickly.
“In 1950 we lived in an economy where Americans bought American-made goods,” explains Farber, an economics professor at Princeton University. “It’s pre-globalization. There was this overarching agreement between unions and management that management would accept the unions, unions would allow management to manage, and there would be labor peace without the opportunity to strike during a contract. Roughly speaking, it was sharing the gains that came from having a protected product market, because Americans didn’t want to buy cars made elsewhere.”
National markets are also more interconnected than they were midcentury. The U.S. is now a net importer of goods. The U.S. imported $2.3 trillion worth of goods in 2020, $911 billion more than it exported. Census data on trade balances go back to 1960, when the nation imported $15 billion worth of goods — nearly $5 billion less than it exported.
Finally, U.S. jobs are no longer dominated by manufacturing. In 1950, manufacturing firms produced $225 billion worth of goods, roughly 40% of the nation’s entire industrial output, according to the Bureau of Economic Analysis. By late 2019, before COVID-19, manufacturing represented 16% of the nation’s industrial output.
Today, the biggest firms in the U.S. — Apple, Microsoft, Amazon, Alphabet and Facebook — are data companies. If they do build physical products, like cell phones, that largely happens overseas. Income inequality has grown, with the highest income earners in the U.S. having gradually taken home a larger share of the national income from 1980 onward.
Here’s how the manufacturing decline has played out in the labor market: Nearly 30% of full- and part-time U.S. workers were employed in manufacturing in 1950. By 2019, 8% worked in manufacturing while 18% worked in retail, hospitality or food services.
Citing federal labor statistics, Vanderbilt University sociologist Daniel Cornfield writes in a 1986 paper in the American Journal of Sociology that in “manufacturing — the traditional source of union membership — the percentage of unionized workers declined from 51.3% to 39.9% between 1956 and 1978.”
The decline in manufacturing as a share of unionized workers has continued over the last 20 years. In 2000, 15% of union jobs were in manufacturing. By 2020, that figure had dipped to 8.5%, according to the Bureau of Labor Statistics.
Unionization rates for government jobs have held relatively steady over the last two decades, with about 37% of the public sector unionized today. Private sector unionization has fallen from 2000 to 2020, from 9% to 6%. The educational services sector is a notable exception, growing from 12% unionization in 2000 to 14% in 2020.
Utilities, transportation, warehousing and telecommunications remain the most unionized industries, though union density in each has declined by over 5% since 2000 — and by almost 10% in telecommunications alone.
How unions grew
The “Unions and Inequality” authors identify two primary ways unions expanded before and during World War II and the Great Compression.
The first was the Wagner Act, which President Franklin Roosevelt signed in 1935 and which the Supreme Court upheld in 1937. The act, sponsored by New York Sen. Robert Wagner, established the National Labor Relations Board. It brought legal protection to private sector unionization and collective bargaining activities, with some unions turning their focus toward organizing unskilled workers after it became law.
Before the Wagner Act, most workers involved in industrial manufacturing were not in unions. Large companies sometimes used physical violence to break organizing campaigns. “Henry Ford, whose brutal private army was well known by his workers, set the tone for how to crush unions,” recounts University of Rhode Island labor historian Erik Loomis in his 2018 book, “A History of America in Ten Strikes.”
Governments at all levels also had historically sided with employers, “with military deployments and judicial repression commonplace,” the authors of the recent paper write in the appendix.
The second event that led to union expansion was Roosevelt establishing the National War Labor Board in January 1942, weeks after the U.S. entered World War II. The board seated 12 representatives drawn from private firms, unions and the public sector. It was tasked with settling labor disputes before they affected wartime production.
“The government was letting very large defense contracts for armaments and so on and requiring, as a condition of the contract, by executive order, that companies be open to unionization,” Farber says. “In places that had a lot of government contracts, there was a growth in unionization and inequality in those places at those times declined.”
The Wagner Act also protected recognition strikes, says Naidu. Over the five years after the act passed, strikes were successful in gaining union recognition 40% of the time, compared with about 20% during the seven years before the act, the authors find.
The Korean War provides an important check on their findings. A smaller conflict than World War II, the Korean War from 1950 to 1953 still required massive production of tanks, planes, ammunition and guns.
Difference was, firms that got government contracts during the Korean War didn’t have to allow unionization efforts. The authors find “no correlation between Korean War Defense spending and changes in state union density or inequality measures.”
Expert Commentary