Decline and Fall: Union Election Trends and the Failing NLRB Strategy

Eric Dirnbach
13 min readJan 24, 2018

Labor movement data since the 1950s shows the near impossibility of growing union membership through running more and better NLRB elections. What should we do?

The new Bureau of Labor Statistics data on 2017 union membership has come out, and the news could be worse. The union membership rate, also called union density, which is the percentage of workers who are union members, has held steady at 10.7%. The private sector rate rose slightly to 6.5% from 6.4% and the public sector rate held steady at 34.4%. Union density is a quick and simple metric to gauge the overall strength of the labor movement and its ability to raise standards and defend the interests of workers. The reason why the news could be worse is that these rates usually decline slightly every year. Holding steady, and even slightly increasing the private sector rate, is what passes for good labor movement news these days. A good review of labor movement activity last year is here.

Here is a chart of union density over time, where we see the peak in the 1950s and then a slow decrease over time in the private sector until the present day. The public sector saw a burst of union organizing later on and has a much higher union density today.

From LBO News from Doug Henwood

This decline isn’t happening because workers are uninterested in joining unions. Polls over time have found that a majority of workers approve of unions and want to join one. In a previous post I discussed a number of reasons for this declining membership trend which include structural changes to the economy over time, including globalization, outsourcing and automation, as well as a fairly hostile set of labor laws. However, most salient for this discussion is the increased anti-union campaigning by employers over the decades which have made it much harder to organize.

I also discussed that, coinciding with this membership decline, the number of large strikes has plummeted. For example there were over 300 large strikes in most years during the 1947–1976 period, while in the last 10 years, the average number annually has been 14. Another previous post looked at the recent trend of National Labor Relations Board (NLRB) elections, which is the primary way new workers are organized.

After reviewing those numbers, I became interested in looking back much further to see what was happening decades ago as the union density decline was starting. So buckle up, because this is a fascinating roller coaster ride of labor data.

Elections Back in the Day

I want to look at union organizing activity over a long period of time. Below is a chart showing NLRB elections and win rates every 5 years from 1957 to 2017. Before we begin the discussion, a few notes first. Why go back to 1957? Because that’s around the time of maximum union density, so we will be looking at what happened during the long decline. Moreover, this is only looking at what are called “RC” elections where the union files for an election to get official certification as the bargaining representative of a group of workers. There are a few other kinds of elections that are much less common and not considered here.

Furthermore I should clarify that this data is just a proxy for overall union organizing activity. There are also union elections in the state and federal public sector which are handled outside of the NLRB and don’t show up here. That’s fine because, as the chart shows, the decline in union density has really been a private sector crisis. We can see that public sector density, for now, has held fairly steady, though this may change with the upcoming Janus decision.

Also, this data does not capture any private sector organizing which occurs outside of NLRB elections, which includes card-check campaigns and building trade sector organizing that directly signs up contractors to be union. That activity is potentially significant and would need to be included for a fuller picture. Moreover, this does not take into account any non-traditional, “alt-labor” campaigns among workers’ centers outside of the formal labor movement, which at this point don’t appear in official labor statistics.

Note: 1957 and 1962 includes the win rate for both RC and RM (management instigated) elections because I did not see the NLRB break these out separately. RM elections accounted for less than 5% and 6% of the total for those years, so this not a huge deal.

Let’s now turn to the NLRB election data, and this is fascinating in a number of ways. First, we see that the number of elections rose dramatically from nearly 4,500 in 1957 to about 8,300 in 1977, before plummeting to about 1,200 today. Moreover, a huge drop occurred in the 1980s, coinciding of course with the conservative Reagan administration. Years ago I reviewed a book about the disastrous PATCO strike of 1981 which set the anti-union tone for that whole era. So the labor movement is now running many fewer elections than it used to.

But what’s also interesting is the change in the win rate. From around 60% in the 50’s and 60’s, dropping to around 50% in the 70’s through 90’s, it is now over 70%. That’s a pretty decent win rate given the anti-union campaigns that are standard these days, which we’ll discuss.

Eligible Voters and Organized Workers

Here’s a look at some related data, the number of eligible union election voters and organized workers at the workplaces where a majority voted yes for the union, the union became certified, and the workers became official union members.

Note: 1957 and 1962 include Organized Workers for both RC and RM (management instigated) elections because I did not see the NLRB break these out separately. RM elections accounted for less than 5% and 6% of the total for these years, so this is not a huge deal. In 2012 and 2017 I estimated Organized Workers at about 47,000 and 49,000 based on election wins and estimated average unit size, as I did not find these numbers cited by the NLRB for those years.

We see a similar increase, peak, and then long decrease. 1967 had almost 600,000 eligible voters and about 340,000 workers organized through these elections. Today it’s pretty depressing, with less than 80,000 voters and about 49,000 organized workers. This gives a sense of just how few workers are currently organized in this way. To put this in context, the economy grew by 1.8 million jobs last year, and to increase union density by just 1%, we would need to organize an additional 1.4 million workers.

But this raises a few questions. Why are there so few elections, and if the win rate is so high, why don’t unions just run more elections and organize more workers?

Unfair Labor Practices

But first, let’s get back to the issue of the anti-union campaigns waged by employers during union organizing drives. Many folks have analyzed this over the years and a great overview is Kate Bronfenbrenner’s No Holds Barred report from 2009. The data is a bit old at this point, but still completely relevant. She discusses the myriad ways employers fight union campaigns, using both legal and illegal methods:

In the NLRB election process in which it is standard practice for workers to be subjected to threats, interrogation, harassment, surveillance, and retaliation for union activity. According to our updated findings, employers threatened to close the plant in 57% of elections, discharged workers in 34%, and threatened to cut wages and benefits in 47% of elections. Workers were forced to attend anti-union one-on-one sessions with a supervisor at least weekly in two-thirds of elections. In 63% of elections employers used supervisor one-on-one meetings to interrogate workers about who they or other workers supported, and in 54% used such sessions to threaten workers.

This gives a sense of the employer reign of terror that is so common in these campaigns. Furthermore, 75% of employers even hire union-busting consultants to help them do all this. Legal tactics include the mandatory meetings with workers (called captive audience meetings), where workers are subjected to anti-union lectures. Illegal actions include threatening or firing union supporters. These illegal tactics are called Unfair Labor Practices (ULP).

Here are the number of Employer ULP Charges over time, which is where the union formally accuses the employer of a ULP. Also shown is the number of ULP Charges per Election. A ULP Charge does not mean the employer is guilty, but the union believes they are. The charge could result in a finding that a ULP occurred (the NLRB finds that the charge has “merit”), or it could be dismissed for lack of evidence, or withdrawn by the union. The reason I included charges is that I think they give a more accurate picture of what happens in these anti-union campaigns. To rely on official ULP findings of merit would under-count the intense fear and intimidation that often pervades the workplace. After all, many ULP charges take years to process, could be lost because of a lack of evidence or workers decide to give up and move on. Moreover, as the No Holds Barred report says, cases are often settled before a merit determination is made, and unions sometimes don’t file potential ULPs because that will delay the election or for other tactical reasons.

This data is somewhat shocking. A huge rise in ULP Charges occurs as the number of campaigns grows, and then declines later on as the number of elections goes down. But looking at ULP Charges/Election, there is an incredible rise over time, which continues even as the number of elections decreases in recent decades. This really shows the increasing onslaught of anti-union activity that workers are subjected to when they try to organize.

But let’s also remember that the union win rate has climbed as well. So unions have over time figured out how to overcome this increased union-busting, which is great news. Moreover, the win rate dropped to around 50% for several decades that coincided with the huge rise of ULP Charges. It was only in more recent decades that unions have figured out how to deal with these ULPs and bring the win rate back up.

Declining Union Resources

I think a final piece of the puzzle to explain why there are fewer elections is to remember the declining union membership. Fewer members means less dues money to support new organizing campaigns. The private sector union membership has decreased by almost half (47%) since 1977, from 14.3 million workers to about 7.6 million workers. That’s a huge loss of available funds for organizing. Of course, we would need to estimate the actual money spent on organizing over time to see if that has really dropped and by how much. This is a complicated task since organizing funds are spent at the national, regional and local level by many unions, but I think a large decrease is a reasonable assumption.

Quality over Quantity

The decline in the number of elections over time coincided with the rise of anti-union campaigning by employers. So union elections have gotten tougher over the years, as many folks have observed over time. But what’s fascinating is that the union win rate has been increasing, along with the smaller number of elections and reduced union resources. I think there’s really only one conclusion that makes sense here - unions are focusing their resources and efforts on a smaller number of elections, and they are winning them. Unions are choosing quality over quantity.

Given the massive expected employer opposition, unions are now understandably cautious in choosing which elections to run. You need to file signed authorization cards from at least 30% of the workers to have an election, but it’s generally considered malpractice to file with less than 60%, given the potential loss of support once the anti-union campaign starts. My guess is that in the 1970s unions filed with much less and accepted the lower win rate. In that sense, the labor movement of the time was bolder and took greater risks than unions do today.

Leaving aside ethical considerations for a moment, should we accept a lower win rate and spread resources toward running many more campaigns? If we could multiply the number of elections and accept a lower 50% win rate, what would that mean for the number of organized workers?

Based on 2017 data, we ran 1,193 elections with a 71% win rate, which is 847 election wins and an estimated 49,000 workers organized. Increasing the number of elections and decreasing the win rate, and assuming the same average number of organized workers/election, we get this:

In this scenario, a herculean effort of running 10,000 elections with a 50% win rate, yields about 289,000 organized workers. Recall that we need over 1 million new workers to increase union density by 1%. That’s pretty depressing.

Also remember I said that in this scenario we would need to leave ethical considerations aside. Why did I say that? Because a union election campaign is a serious fight for justice in the workplace which will alter the balance of power. As discussed, the boss may subject the workplace to a massive campaign of threats, intimidation and firings. The workers who go through this process need absolute full support from the union they are working with. Anything less leaves them at the mercy of their employer. That’s why it would be wrong to make the trade-off between more elections (quantity) and a lower win rate (quality).

However, we could avoid this trade-off and run many more elections while keeping a high win rate if we dramatically increase the resources devoted for organizing. If the 10,000 union elections had the same high win rate of 70%, this would lead to about 404,000 organized workers, enough to slightly increase union density by about 0.3%. This is better but still not enough. And of course 10,000 elections is about 8 times the number we ran last year. Increasing organizing resources is possible, but an increase of 8 times is a fantasy.

Then What Do We Do?

We’re in a tough position here. Through NLRB elections, we’re not organizing enough workers now and increasing the number of elections, even under the best scenarios, leads to very small increases in union density. This is not enough to build real power in the labor movement on any satisfying time scale. It must be asked whether it’s even worth running NLRB elections if this is the depressing trap we’re in. For any individual workplace where a union win and decent contract is a life changing event for workers, of course it is.

But on a macro scale, I’m certainly not the first to raise this question, and many folks have criticized the NLRB election system for decades. Add to this the difficulty newly unionized workers have in negotiating a first contract. The No Holds Barred report states that 1/3 of workers still don’t have a contract after 2 years.

To continue to use the NLRB system, we need a dramatic change in labor law that makes it easier for workers to organize. Of course politically, that’s very difficult. Unions pushed for labor law reform in the Carter years and lobbied for the Employee Free Choice Act (EFCA) during the Obama administration. EFCA would have allowed card-check as an official organizing process and increased penalties for ULPs. Unfortunately it failed to pass.

Alternatively, we can increase organizing outside the NLRB election system. This has been a big discussion in the labor movement for well over a decade. In fact, one could make the argument that the low number of elections means that unions are already abandoning the NLRB in favor of these alternatives. As discussed earlier, there has been an effort to run card-check campaigns and the growth of the worker center/alt-labor movement is promising. High profile examples include the SEIU-backed Fast Food/Fight for $15 and the UFCW-backed OUR Walmart campaigns which are agitating for better conditions but are deliberately not running store by store elections. These efforts have yielded results like wage increases, but no official union membership. The Coalition of Immokalee Workers is an example of a workers’ center successfully improving working conditions without formal union membership. Another model is the direct action campaign of the IWW’s Stardust Family United, where they are union members fighting for improvements without seeking an election or a formal union contract.

Many of these workers are invisible in the government union membership statistics. It’s possible that the BLS, in its annual union survey, could be persuaded to change its data collection to include workers involved in these kinds of campaigns, to give a more complete picture of labor movement membership. This would increase the numbers but it raises the question of how meaningful these results are. If workers are fighting in some way for better working conditions, but don’t have official bargaining status, then what are we really measuring? What if their main fight is for a local minimum wage increase in their city or state? Trying to measure this kind of worker intervention in the political system may take us too far away from meaningful labor membership data.

And in any case, official statistics are really beside the point. What matters is that the working class needs to organize, build power, and improve wages and working conditions, however it does it. Government certified unions with signed contracts are the traditional way since the 1930s, but not the only way. They are just one means to an end. However we do our organizing, what’s clear is that the resources devoted to organizing need to be seriously increased, and many more organizing campaigns, of any kind, need to happen, while maintaining the kind of quality that leads to high win rates. And there are plenty of interesting policy ideas that could help including fighting for just-cause job security, making union organizing a civil right, or having industry-wide sectoral bargaining.

Let’s Remember Our History

But we need to remember that the great growth spurt in union membership that started during the 1930’s was not primarily because unions staffed up and ran excellent election campaigns or because we had great policy proposals. More importantly was the tremendous desire during the Great Depression of millions of workers to improve their working conditions and their willingness to strike and otherwise be disruptive. Something similar happened with public sector workers in the 1970s, resulting in that surge of organizing. That culture of militancy and solidarity has to be renewed on a large scale. Joe Burns in recent years has made the case for reviving the strike as a crucial part of rebuilding the labor movement. This is not easy but necessary. And Labor Notes has made the case for more democratic, militant unionism since the late 1970s.

Disruption of business as usual creates working class power which opens space for improved conditions and labor laws to codify the new status quo. We can’t skip that step with just Beltway lobbying and political donations to Democrats. And the data presented here shows that we are unlikely to slowly grow our way back to power by running more and better NLRB elections. The numbers just don’t support that. This conclusion is not new, but looking at this data makes the case all that much clearer.


2017 and 2012 elections:, ULPs:

Earlier years, elections and ULPs:



Eric Dirnbach

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