Despite Worker Militancy, Union Rates Are Still Down

Despite an apparent upsurge in labor militancy, unions made no gains in their share of the workforce last year. Something needs to change — and fast.

SAG-AFTRA members march on Labor Day in Manhattan, New York. (Barry Williams / New York Daily News via Getty Images)

Despite an apparent upsurge in labor militancy, unions made no gains in their share of the workforce last year. According to newly released figures from the Bureau of Labor Statistics (BLS), in 2023, 10.1% of the workforce belonged to a union, unchanged from the previous year and down 0.2% from 2019. There was a spike in union density in 2020, as more nonunion workers left the workforce than their organized counterparts in the early COVID days, but that was quickly reversed in 2021.

Since 1965, union density has risen only four times from one year to the next; it’s fallen in forty-five. Although long-term comparisons are dangerous — definitions, coverage, and methods can change radically over time — it looks like the private sector union share today is lower than it was in 1900.

The public sector, once the brighter spot in the union picture, has lately been leading the way down. Since 2011, the decline in public sector density has been four times that of the private sector. The Right’s war on public sector unions has taken a serious toll. The grim story is graphed below.


Much of the material difficulties experienced by the working class over the last few decades can be traced the shrinkage of unions — lower wages, stingier benefits, less job security. Pay is the most obvious measure. As the graph below shows, union members have weekly paychecks 16% larger than those of nonunion workers. The union premium is larger for groups that are the victims of discrimination — women and non-whites. Women enjoy a 19% union premium overall; black women, 20%; and Hispanic (as the government calls them) women, 32%. Non-white men also enjoy bigger union premiums than white men. The only exception is workers of Asian origin; it’s likely that the high share employed in tech (a sector where unions are rare) is responsible.


Union density varies widely by state, from Hawaii’s 24.1% to South Carolina’s 2.3%. The map below shows some clear regional patterns, with the Northeast, upper Midwest, and West showing the highest densities and the South the lowest. The ten states that gave Donald Trump the highest share of their 2020 vote have an average density of 6.3%; the ten lowest, 15.2%. The states of the old Confederacy have an average density of 4.8%; the rest, 10.7%. Curiously, the state with the lowest density, South Carolina, had the largest share of its population enslaved before the Civil War.

Shrinking Premium

That union premium has been shrinking over time, as unions have weakened more generally. Here’s a history showing the ratio of the weekly pay of all union workers to nonunion since 1994. (Figures for 2000 are unavailable. There’s no simple source of the historical data — you have to enter it by hand from annual releases.) The premium has declined steadily for three decades, from 37% in 1994 to 16% in 2023. While 16% isn’t nothing, the premium is less than half what it was back in the early Bill Clinton era.

This graph is based on weekly wages, which are a half-decent proxy for workers’ “normal” wage (though people do cycle in and out of work, and sometimes hours can be hard to come by). But since workweeks can vary by time, industry, region, and worker characteristics (as economists say) it’s also worth looking at hourly pay.

That too is in decline, as the chart below (from UnionStats data) shows. The “unadjusted” line has fallen from a high of 31% in 1977 to 6% in 2023, a decline of 80%. But the unadjusted average, like the BLS averages above, mixes together workers of varying demographics (like age, sex, region, and race) from a variety of industrial sectors, with wide differences in pay and unionization rates. As those mixes change over time, the averages change, which obscures the pure union effect.

The line marked “adjusted” is an attempt to control for those compositional changes and compare workers who are similar except for their union status. The premium has declined there too, though not as much — from 21% in 1977 to 10% last year, or just over 50%. That’s a lot, but it’s less than the 80% decline in the unadjusted numbers.


Of course, any premium greater than 0% will inspire bosses to hate unions and want to destroy them — and not just for the higher pay. Unions are a constraint on boss behavior; they can force them to be less racist, less abusive, and less able to fire. But clearly they’ve lost a lot of their bite over time.

As I say every time I do these reviews:

There are a lot of things wrong with American unions. Most organize poorly, if at all. Politically they function mainly as ATMs and free labor pools for the Democratic party without getting much in return. But there’s no way to end the forty-year war on the US working class without getting union membership up.

It’s still true, though I should probably add a few years to that “forty” timespan.