The Institutions Workers Choose
The first Labor Day was organized by worker institutions that provided community as well as bargaining power. Reviving that spirit requires space for new forms of association.
Early in the morning of Tuesday, September 5, 1882, 400 members of Bricklayers Union No. 6 assembled at New York City’s City Hall Park wearing white aprons. A band followed, then the Manufacturing Jewelers union, marching four abreast in derby hats and dark suits, each man carrying his cane on his shoulder. Typographical Union No. 6 turned out 700 strong and marched with precision, having rehearsed.
The Friendly Society of Operative Masons brought its own band. The Clothing Cutters Benevolent and Protective Union sent 250 men. Somewhere between ten and twenty thousand people walked up Broadway, circled the reviewing stands at Union Square, and continued north to 42nd Street. Most of them went on afterward to a park at 92nd and 9th for a picnic with speeches, music, and a lot of beer. There was no government-sanctioned holiday. Nearly everyone in that parade gave up a day’s wages to be in it.
That was the first Labor Day, and it’s worth looking at the marching order again: A trade guild. A friendly society. A benevolent and protective union. The Central Labor Union that organized the day was itself a semi-secret assembly of the Knights of Labor. The men in the aprons and the derby hats belonged to organizations that buried their dead, taught trades, insured their sick, and sometimes even engaged in secret lodge rituals. Collective bargaining was just one of the things they might possibly do.
The holiday spread the way the organizations themselves had: gradually. The Central Labor Union chose the first Monday of September in 1884 and urged other cities to copy it. They did. Oregon made it law in 1887 and in June 1894 Grover Cleveland made it a federal holiday days before the once labor-supported president sent in troops to bust the Pullman Strike. Within a generation the parades thinned. Within two, the day had become a long weekend and the final day for outdoor swimming pools. Nobody now forfeits much to observe it and, ironically, the blue-collar workers who founded it are probably the most likely to have to work.
We shouldn’t shed tears for the loss of a parade. To say the least, modern labor unions wouldn’t do well to bring back funny hats and secret rituals. The parade, however, provides evidence of one important thing: labor unions were organizations valuable enough that individuals paid for the privilege of being seen belonging to them. They were important mediating institutions. The question worth asking on Labor Day is: What it would take for them to be seen that way again?
Arguments about why worker representation collapsed in America can include union corruption, lack of value provided by unions, economic change, legal barriers, employer tactics, and various combinations of these factors. They often leave out the most obvious explanation: every other part of civil society also collapsed. Houses of worship, fraternal orders, and famously bowling leagues all generally lost members over the same decades. That union membership fell while Americans were giving up joining anything is not surprising in the least. What is interesting is where the reversals have since come. Volunteering rebounded sharply after 2021 (much of it online). Religious participation among young men has turned up for the first time in a quarter century. Around 24 million Americans now play pickleball, mostly on courts that did not exist a decade ago. None of that happened by persuading anyone to rejoin the institutions their parents cast aside. It happened through invention. And that’s not permitted because current law makes mutual aid in the workplace exceptionally difficult.
This is ironic. Unions are a civil society creation. The history of the American labor movement from the 1794 establishment of the Federal Society of Journeymen Cordwainers to the 1935 National Labor Relations Act is largely a story of an effort to create a set of legal rules around something that arose out of collective self-help. That history gives us a test for all proposals: An organization belongs to civil society if people join it voluntarily, if they pay (or volunteer significantly) for something they value, and if they have a meaningful voice in how it is run. Good union locals have historically, more or less, met all three (agency fees in non-right-to-work states and cronyistic leadership structures to the contrary) which is why they could be sources of solidarity and community quite apart from their role as bargaining agents. An organization that fails all three may still do good, but it is simply a different thing.
The place where many of my fellow travelers on the center-right agree with CSLI’s founders that the version of the model we have right now is badly broken. Indeed, it’s instructive that nearly all of today’s private-sector labor union success stories operate largely through what are basically loopholes in labor law.
Hollywood guilds and professional sports unions organize by profession rather than enterprise. Hiring halls in the construction industry have kept building trade unions strong in many places. Las Vegas’ Culinary 226/165 runs a massive training and social welfare operation partly under the Taft-Hartley plan exceptions and provides enough value that nearly all workers in its bargaining units pay dues that Nevada’s right-to-work laws make optional. The pattern deserves more attention than it gets. True, every one of these organizations is built mostly around an occupation rather than a workplace. But it’s just as important that they also rely largely on voluntary membership and provide something a member would miss if it disappeared. And each one exists by exception: the Taft-Hartley carve-out for jointly administered funds, the hiring hall gets use in construction, maritime, and entertainment and very few other places. The best institutions in American labor are the ones that exist outside of the ordinary rules.
In this context, there are both directly pragmatic and structural questions to be raised about efforts like California’s AB 1228. Structure first: Run a wage and standards board through the three tests above and it fails all of them. Nobody joins it. Nobody pays directly for anything. Nobody other than the state itself governs it and then only indirectly. That is not a scandal, because it is not trying to be an association; it is a regulatory body. But it means that even if a council succeeds brilliantly in its intended goals, it’s not going to contribute to civil society one jot. The practical worries are the familiar ones: fewer entry-level positions (which is partly the point) along with shorter hours and thinner staffing (which threatens thin-margin business models and higher menu prices.) I would have voted against it were I a California legislator. But I am glad California did it anyway, because I hope researchers and state officials learn from what it does.
The best way to allow for experimentation is to repeal the provisions that ban it. Section 8(a)(2) of the National Labor Relations Act, section 302 of the Labor Management Relations Act, and the Board’s 1992 Electromation decision together make it all but impossible for a union to collaborate with the businesses it organizes, to operate as an enterprise in its own right, or to try a new model of any kind. These provisions aimed to kill the company union (which was, to be sure, subject to abuse) and they accomplished that. But they also outlawed any collaboration. Worker and management interests align in all sorts of ways all the time. However, it’s illegal for employers to do anything to encourage that. So forming a mutual aid society is incredibly difficult and, often, illegal. Moving in the opposite direction will take careful work to guard against abuses. Laws mandating secret ballot elections for officers and disallowing employer retaliation for lawful activities undertaken under new types of organizations are among the reforms we should consider.
All of this is federal law. The legislatures in Sacramento, Tallahassee or Richmond can’t touch it. For the time being, they’ll have to do other things.
Industry-based councils deserve experimentation, and that has to be a genuine allowance. Rather than efforts to “raise standards” in a way that organized labor groups tend to like—higher pay, more benefits, shorter hours—they should be able to innovate. For example, a degree of flex scheduling is a very important business practice in some industries but can cause genuine problems for people with family or other obligations. The solution may not be to ban the practice, but to allow for “win-win” arrangements that, for example, might allow workers to waive time-and-a-half pay or work for a wage below the state floor in return for guaranteed stable hours.
All this said, the greatest promise and return to the roots of the labor movement should come from state efforts to allow new structures.
Portable benefits are the first and most promising of them. Utah opened the field in 2023 and Georgia followed, where my own organization led the charge, along with Idaho, Kansas, Louisiana, West Virginia, Wyoming, New Hampshire, and North Carolina. The laws are modest and they all do the same basic thing: they let a job-providing platform contribute toward an independent worker’s benefits without that contribution counting as evidence that the worker is somehow an employee. That safe harbor removes a reason platforms had to offer nothing at all. It’s a benefit structure designed around the worker rather than the job, which allows for enormous amounts of invention.
Uptake, though, has been slow, and nobody should be surprised. The safe harbor is a safe harbor from state law only. The federal classification standard sits untouched.
And the whole structure is voluntary, which brings me to the part of this that, as president of a right-of-center, libertarian-oriented think tank, my own side may not embrace warmly. Predicting what will emerge here is difficult, but getting it right will take experiments in red and blue states alike. And if the councils ask too little of civil society, the portable benefits laws that exist may ask too much of it. Purely voluntary contribution is hard. In a voluntary system, a platform that contributes nothing undercuts one that contributes something, and thin margins combined with a price-sensitive public may make this hard. Without some requirement that the platforms where gig workers earn must contribute to their accounts, those accounts could stay small enough to be beside the point. And a mandate could easily go wrong by becoming a payroll tax that makes it hard to employ those most in need of jobs, or resulting in something far too close to employee classification to please anyone who now supports the concept. If we’re to experiment with some type of mandatory contribution for all gig-providing companies (one-industry mandates already exist), there are reasons to design such regimes carefully and to try it in a few states before all 50.
An account that follows a worker from job to job might help create an association, but it’s not an association by itself. Instead, an organization that runs one must pass the afore mentioned three tests: a worker must join voluntarily, he or she must value what the fund pays for, and they should have a reasonable, individual say in the way that it is administered. (This means, almost certainly, that administration decisions must be made at a relatively low level.) The states that go furthest here should let these funds be governed by boards the contributors elect. This is roughly how the Taft-Hartley funds in construction and entertainment have worked for 70 years, and it is the difference between providing benefits to workers and representing them in any real way.
Second, states should ban front-end noncompete agreements outright and let workers draw unemployment benefits while they attempt to start a business themselves. Contrary to the popular impression, self-employment declined over the past thirty years and has only lately begun to recover. Noncompetes are a state matter, governed by contract law rather than by the Wagner Act, which is why 14 states have moved on them in two years while the federal rule died in court. Nothing gives a worker more power than the ability to go into business against a former employer, and nothing makes an organization more attentive to its members than knowing they have somewhere else to go.
This isn’t a full program. The point is clearing ground to allow new shoots of innovation to emerge. The ongoing argument over worker representation has been an argument about who sets the terms of employment, conducted almost entirely among employers, unions, and the state. The older question, the one the parade was an answer to, is what workers can build themselves and truly want. The men who marched up Broadway in 1882 bought their own holiday at the price of a day’s pay. And they valued the day. The most useful work the states can do now is to make it both legal and worthwhile for their successors to build something they actually want.