State of the U.S. Unions 2023

Eric Dirnbach
11 min readFeb 24, 2023


This is my annual roundup of labor data on union membership, elections and strikes. How is the U.S. labor movement doing?

At the beginning of each year a bunch of annual U.S. union data is released, and this is a good time to see what it tells us about how the labor movement is doing. We’ll examine data on union membership, NLRB elections and strike data, and then discuss what it all means. My summary from last year can be found here.

Union Membership

In January, the U.S. Bureau of Labor Statistics (BLS) released its annual Union Members summary. The labor movement membership rate went down, a depressing annual reminder that we’re not doing well enough. As the BLS states,

The union membership rate — the percent of wage and salary workers who were members of unions — was 10.1 percent in 2022, down from 10.3 percent in 2021, the U.S. Bureau of Labor Statistics reported today. The number of wage and salary workers belonging to unions, at 14.3 million in 2022, increased by 273,000, or 1.9 percent, from 2021. However, the total number of wage and salary workers grew by 5.3 million (mostly among nonunion workers), or 3.9 percent. This disproportionately large increase in the number of total wage and salary employment compared with the increase in the number of union members led to a decrease in the union membership rate. The 2022 unionization rate (10.1 percent) is the lowest on record.

Union membership is up slightly to 14.3 million workers, but the overall workforce grew much more, so the union membership rate dropped to 10.1%. It’s very likely that it will go below 10% in the next year. The rate for the private sector is even lower, falling to 6.0%. The public sector rate also fell a bit, but remains much higher at 33.1%. The numbers for all represented workers, which also include workers who choose to not be union members, but are covered by the union contract, are slightly higher for all these.

Particularly alarming is that union membership has such a limited geographic distribution. Almost half of all union members are in just six states (CA, NY, IL, PA, OH, NJ), and 11 states have union membership rates less than 5% (AR, FL, GA, ID, LA, NC, SC, SD, TX, UT, VA). The labor movement is not spread out enough and has too small a presence in too many states. One consequence is that elected officials in many states face no consequences when voting against worker and union interests.

Overall, these are really disappointing numbers, especially because the last year saw an exciting upsurge in organizing and strike activity, which we’ll discuss more. Here is the union density trend for the last century, and we see this latest decrease just continues a long term tendency.

From LBO News with Doug Henwood

Union Elections

The U.S. National Labor Relations Board (NLRB) released its annual summary of union elections for fiscal year 2022, which covers the year from October 2021 through September 2022. And there is some great news to report. The number of union RC elections (the main type of election) increased 49% from last year and the percentage of union wins increased to a very good 72%.

It has to be acknowledged that a huge reason for this increase is the extraordinary Starbucks Workers United campaign, which ran 276 store elections during that fiscal year, winning 231 of them, which is an amazing 84% win rate. But even aside from Starbucks, the number of all the other elections increased 16% over the year before, and the number of wins increased 30%.

The chart below shows this data for the last few decades. We can see the downward trend in elections over this time, which makes this last year all the more remarkable. However, we can also see that although unions have been running fewer elections over time, they have been getting better at winning, with a roughly 67% win rate for the last three years.

These are the RC petitions that end up in elections. But there are more petitions that are filed, with some that are withdrawn before elections. That withdrawal rate is interesting and is a measure of a few possible things. Withdrawn petitions likely indicate that workers had insufficient preparation for the election when encountering the employer’s union busting. So a lower withdrawal rate indicates that unions are starting campaigns that are better prepared. The trend shown below, using NLRB data, shows that is what seems to be happening, with the petition withdrawal rate declining over the last decade from the low 30% range to the low 20% range. So unions are starting better prepared campaigns, bringing more of them to election, and then winning more of them, all good trends.

But how many workers are these elections organizing? Below we have the total workers in elections, and the workers who won their elections. We can also see a long term decline, but with a sharp jump in the last year, with 62,000 workers voting and about 47,000 workers winning their union. To see the net gain of organized workers, we would need to subtract a relatively small number of workers lost in RD union decertification elections, and add an even smaller number gained through management instigated RM elections, both not shown here. These numbers are also assuming that the relatively small number of multi-union elections (either two or three unions) are adding newly organized workers, rather than the results of raids on an already organized unit.

This last year is an improvement, but by historical standards is very low. From the 1950s through the 1970s, an average of about 500,000 private sector workers participated in NLRB elections every year, about 1% of the workforce annually. But in the last three years, only an average of about 54,000 workers have participated in these elections, with less than 0.1% of all workers participating.

We should also look at voluntary recognitions, which may be increasing. A majority of workers in a workplace can petition their employer to voluntarily recognize their union without an election, and it does happen sometimes. This data is not readily available from the NLRB, but efforts by Labor Data to collect it through FOIA requests have found that in 2022 there were 209 voluntary recognitions involving 11,182 workers.


There have traditionally been two publicly available sets of strike data provided by the federal government. The one that gets the most attention is the BLS annual report on Work Stoppages. These are strikes or lockouts involving at least 1,000 workers for at least one shift, but they are mostly strikes. For 2022, BLS reports that there were 23 large strikes that started that year.

In 2022, there were 23 major work stoppages beginning in the year, the U.S. Bureau of Labor Statistics reported today…There were 120,600 workers involved in major work stoppages that began in 2022.

As with most labor data, the number of annual strikes has been falling for many decades. There was some excitement in 2018 and 2019 when large strikes went up to 20 and 25, led by the many education strikes. There was some loose talk of a new “strike wave.” Perhaps I may have made that claim. But then the number dropped back to eight in 2020, and rose again in 2021 to 16 amidst the commentary about #Striketober, which I also participated in, because we got excited about the John Deere, Kellogg’s and other major strikes. So 23 strikes is an encouraging number for recent years, but is very small by historical standards.

Here’s the trend below of the average number of annual large work stoppages and workers in each decade (with the 2020s = 2020, 2021, 2022), and we can see the incredible fall.

The 1950s had an average of 352 large work stoppages per year involving 1.6 million workers. Incredibly, the entire time period from the 1950s-1970s had over 300 large work stoppages per year on average (nearly one per day!) involving over 1.4 million workers annually. However, the 2010s only had an average of about 15 per year involving 150,000 workers. The huge drop in the 1980s happened in the era of the failed PATCO strike, leading to the term “PATCO Syndrome,” describing unions’ fear of striking and losing.

Other strike data has traditionally been collected by the Federal Mediation and Conciliation Service (FMCS), which is an agency that was formed out of the anti-union Taft-Hartley bill of 1947 to settle labor disputes and prevent strikes. It used to have monthly Work Stoppages reports which cover all the work stoppages that ended each month arising from private sector contract negotiations, and again these are mostly strikes. It also has an occasional public sector strike where it was asked to mediate. I have more information about this data here. So these aren’t all the strikes that happen, but looking at this data over time is another interesting indicator of strike activity. I say the FMCS “used to have” this data, because it mysteriously disappeared from the website around the time President Biden took office, and I have never heard why that happened. Here are the number of work stoppages available from the last few decades, and we can see the clear decline.

Strike data is also kept by Bloomberg Law, and is not fully available, but they do write articles based on it. It tracks strikes differently than BLS and FMCS. Their recent analysis puts the number of strikes last year at 314, involving 222,000 workers. Here is their latest strike chart, which shows the long decline and recent upsurge.

And now we also have the Cornell-ILR Labor Action Tracker, which has two full years of strike data. This much needed public resource tracks strikes and labor protests, with rigorous documentation and verification procedures. Major indicators show an increase in strike activity from 2021 to 2022. The 2022 annual report states that the number of work stoppages increased 52% from 279 to 424. The number of workers involved increased 60% from 140,000 to 224,000. The number of strike days (one strike day is one worker out for one day) increased 36%, from 3,269,186 to 4,447,588. In both years, about one-third of all work stoppages were by workers not officially unionized, which is really interesting. These are clear increases in activity, but again the historical context is important, as total work stoppages were about 10 times higher in the 1970s (see Table 1 of this BLS report — the BLS used to collect much more data but scaled back in the Reagan era). This also shows the importance of collecting data on all strikes, as the BLS data now leaves out a lot.

There’s also the factor of credible strike threats that impact all these strike numbers. There are many strikes that could have happened but didn’t because employers were moved to settle. So in an era these days with relatively low unemployment and many employers scrambling to find workers, the working class is in a relatively strong position to win improvements at work without striking. So a fuller analysis of organizing activity would include strike authorization votes that lead to collective bargaining settlements for unionized workplaces. For formally unorganized workplaces, we would have to look at direct actions that lead to improvements.

What Does It All Mean?

First we always have to remember that unions are extremely popular, with a 71% approval rating. Moreover, half of all workers, that’s tens of millions of workers, want to be in a union, and workers googling “how do I form a union” is at a 10 year high! But it’s very difficult to organize because of intense employer union busting. An EPI analysis found that employers were charged with violating the law in over 40% of union elections. As another EPI analysis says, “The fact that tens of millions of workers want to join a union and can’t is a glaring testament to how broken U.S. labor law is.”

It’s worth mentioning that President Biden is implementing dozens of pro-union recommendations made in the report from the White House Task Force on Worker Organizing and Empowerment, although the results remain to be seen. But we will not get the PRO Act anytime soon, so we can’t have a primary strategy that depends on getting labor law reform. In any case, the government will not save the labor movement.

Primarily this is a matter of the amount of resources the labor movement is committing to organizing and how we’re doing it. The number of union members increased by 273,000 workers last year, and 193,000 in the private sector alone. I think it’s critical to realize that of that private sector total, only about 60,000 joined a union through NLRB elections and voluntary recognitions, or 31%. There are also some number of workers organized through building trades campaigns and Railway Labor Act elections where the data is not as readily available. But it’s likely that the majority of new union workers came into unions through the normal churn in the economy where unionized employers hired more workers, rather than organizing. We can’t depend on that to grow the movement.

Of course, this data doesn’t capture the entirety of the labor movement. There are workers who are “organized” in other ways, for example through worker associations and worker centers, who are not necessarily counted in this data. That is harder to track, but it would be useful.

But here’s the scale of our challenge. The size of the non-union workforce is about 125 million workers, so to meaningfully increase the union membership rate again, we need to organize over one million workers every year, and we are still very far from that. At the very least, we would need to increase our amount of organizing at least 10 fold to match the effort of decades ago, and ideally 20 fold to organize enough workers. Do unions have the money for this? Well a new report on labor movement finances shows that there are billions of dollars that could be used to hire tens of thousands of additional organizers. This current level of organizing is a cautious and risk averse approach.

We need a huge increase in organizing, in whatever form it takes. This also should include training many more non-union workers in organizing and helping them launch their own campaigns. As an example, I have written about the Emergency Workplace Organizing Committee, which has a large network of volunteers that has had success helping workers organize during the pandemic. This kind of project needs to scale up. I wrote a review of the great recent book Class Struggle Unionism, which discusses labor movement problems and urges that unions embrace a bolder and more militant approach.

There has been a lot of talk about a labor upsurge, and I am here for it. We may be in the relatively early days of a sustained increased in organizing, as measured by union campaigns and strikes. If true, this should appear in the union membership data at some point. What’s clear is that the labor movement needs to find ways to significantly amplify these recent trends.



Eric Dirnbach

Labor Movement Researcher, Activist, Campaigner, Organizer, Educator, Writer & Socialist, based in New York City. @EricDirnbach