The Union

The individual worker, the opening plate argued, bargains from weakness: he must eat this week, while the employer can wait out the year. The union is the elementary response to that weakness, and it is nearly as old as the wage itself. Its principle is a single idea — that workers should meet the employer not as a crowd of isolated individuals, each replaceable and each in a hurry, but as one body, a single seller of labor that cannot be picked off one at a time.

The device that gives the combination its bite is the strike: the collective, coordinated withdrawal of labor. For the lone worker the only sanction is to quit — a threat the employer absorbs without noticing, since one departure among many costs little. For the organized workforce the threat is that all of them leave at once, which the employer cannot absorb, because the firm cannot run. The strike is, in effect, the workers' version of the employer's power to wait: it manufactures, on the workers' side, the staying power the individual lacked.¹ It converts a market of many weak sellers into a bargain between two parties of more even strength — which is exactly why employers have always resisted it, and why the law spent more than a century deciding whether to permit it at all.

That history is worth recalling, for it shows the asymmetry was no abstraction. Smith had observed that the masters' combinations were constant and quietly lawful while the workmen's were prosecuted; for a long time this was the literal state of the law. Britain's Combination Acts made workers' associations criminal outright; American courts treated them as conspiracies until, in 1842, Commonwealth v. Hunt held that a union was not in itself unlawful. Only in 1935, with the Wagner Act, did American law affirmatively protect the right to organize and bargain collectively — and only twelve years later, with Taft-Hartley, did it begin taking some of that protection back.² The union, in short, is the workers' combination finally made legal: the mirror of the masters' combination Smith had simply taken for granted.

What is the union for, in the economist's terms? The most useful framing was Galbraith's. The classical remedy for private economic power is competition: if no buyer and no seller is large enough to move the price, none can exploit the rest. But competition is often simply unavailable — markets concentrate, the employers in a town or a trade are few — and where it cannot be had, Galbraith argued, power on one side of a market tends to call up a countervailing power on the other. The union is the labor market's instance of the principle: a concentration of workers raised to offset a concentration of employers. Not the restoration of competition, but a deliberate counterweight to its absence.³

The union does not restore competition. It is a counterweight to its absence — power assembled on one side to meet power on the other.

This recasts the union's effect entirely. Recall the bargaining range: a band between the floor, set by the worker's options, and the ceiling, set by the value of the work, within which the wage is decided by power. The individual, bargaining alone against an employer who can wait, is pressed toward the floor of that band. The union's whole office is to push the settlement back up — to turn the lopsided bargain into something nearer a contest of equals. When a single organized workforce faces a single dominant employer, the case is the one economists call bilateral monopoly, and its outcome is, once again, indeterminate: a range, not a point, with the wage coming to rest wherever the balance of the two powers leaves it.⁴ The union does not abolish the range. It moves the wage within it — upward, against the gravity of the asymmetry.

Whether this upward push is good or bad — for workers as a whole, and for the economy — is the question over which the economics of unions has been fought, and the honest answer has two sides. It depends, exactly as the minimum wage did in an earlier plate, on the market the union is acting within.

Begin with the case the textbook begins with: a competitive labor market, where the wage already sits at the value of the marginal product and the employer has no power to exploit. Into this market a union introduces a wage above the going rate. The effect is the standard and unflattering one. At the higher wage the employer wants fewer workers — one slides up the demand curve of the plate on the margin — so employment falls. The union has raised the pay of those who keep their jobs by thinning the number of jobs; it has acted as a labor monopoly, and like any monopoly it leaves a wedge and a loss. This is the union of the textbook, and where the market truly is competitive, the textbook has it right.

Now place the same union in the market of the eighth plate: one where a dominant employer holds monopsony power, hiring fewer workers than competition would and paying them less than their work is worth. Here the union's wage floor does the very thing the textbook case forbids. Because the monopsonist was deliberately holding employment down to keep the wage from rising, a floor that fixes the wage strips that motive away: the employer can no longer depress pay by hiring less, and so it hires more. Wages rise and employment rises together, both moving back toward the competitive levels the monopsonist had suppressed. The union, in this market, is not a distortion but a correction — the same surprising result the minimum wage produced under monopsony, and for exactly the same reason.⁵

The lesson is not that unions are good, nor that unions are bad. It is that the sign of their effect is fixed by the market beneath them. In a competitive market a union trades jobs for wages; in a monopsonized one it can win both at once. And since real labor markets, this volume has argued throughout, are shot through with the frictions that hand employers monopsony power, the textbook's confident verdict — that unions simply price workers out of their jobs — is not the general truth it is usually taken to be. It is the special case.

There is a second reason the textbook verdict misleads, and it is the great contribution of Freeman and Medoff's study of the question, the one that gave the modern subject its title: a union has two faces, and economics had long trained its eye on only one.

The face it had seen is the monopoly face — the union lifting its members' wages by restricting the supply of labor, at the cost just described. The face it had missed is the union as voice. Borrowing a distinction of Hirschman's, Freeman and Medoff observed that a worker unhappy in a job has two ways to register the fact: exit, by quitting, or voice, by speaking up and pressing for change. Without a union, voice is dangerous and usually futile, so the dissatisfied simply leave; with a union there is a protected, collective channel for it — a grievance procedure, a seat across the table, a means of fixing what is wrong without walking out.⁶

The consequences of voice are not those the monopoly story predicts. A workforce that can speak quits less, so turnover falls, and the costly churn of hiring and training falls with it. Grievances are aired and settled rather than left to fester. Workers with secure tenure are worth training, and get trained. The upshot, Freeman and Medoff found across a wide body of evidence, is that the voice face can actually raise a firm's productivity — enough, in many settings, to offset the monopoly face's cost. The union the textbook had filed under pure deadweight turns out, seen whole, to be doing something the textbook had no category for: making the employment relationship work better, and not merely dearer.

The third face has the longest reach beyond the single firm, and it carries the plate back to the volume's spine. Unions raise their members' pay — the union wage premium, the amount by which a union worker out-earns an otherwise identical non-union one, has been put at around fifteen percent, and the raw gap is plain to see: in 2024 the median unionized worker in America earned about $1,340 a week against roughly $1,140 for the non-unionized.⁷ But the more consequential effect is on the shape of the distribution, not its average. Unions compress wages. They raise the bottom more than the top, narrow the gaps between occupations and between firms, and impose standard rates that flatten the arbitrary scatter of individually negotiated pay. A union, in the old phrase, is a sword of justice quite as much as a vested interest.

This compression scales up into one of the central facts of recent economic history. As American unions rose, the wage distribution narrowed; as they collapsed, it widened again. The timing is too close to be accidental, and careful work bears it out: the decline of unions is estimated to account for as much as a third of the great rise in wage inequality among American men over the last half-century — an effect on the same order as the far more discussed premium on a college education.⁸ The union, that is, did much of the quiet work of holding the wage distribution together. Its retreat is written into the widening gap between the floor and the ceiling — and, as the final plate of the volume will argue, into the shrinking share of the whole product that reaches labor at all.

And it has retreated very far. At its mid-century height, around a third of American wage-earners carried a union card; by 2024 the figure was 9.9 percent, and in the private sector 5.9 percent — the lowest on record. The collapse is the great structural fact of the American labor market across two generations, and its causes are braided together.

Some of the causes are economic: the long migration of employment out of the heavily unionized factory and into the harder-to-organize services, and the discipline that globalization and the standing threat of relocation imposed on union wage demands. But much of it is legal and political. Taft-Hartley had licensed the states to pass “right-to-work” laws barring union-security agreements, and more than half eventually did; employer resistance to organizing grew steadily more sophisticated and more aggressive; and enforcement of the rights the Wagner Act had promised weakened to the point where the penalty for illegally firing an organizer became, for many firms, a tolerable cost of doing business.⁹¹⁰

This matters for reading the bare number, because density understates how many workers would choose a union if choosing one were easy and safe; the gap between the share who say they want representation and the share who have it is very wide. Nor is the number destiny. In much of continental Europe, where an agreement struck in one firm or sector is extended by law to cover the rest, the coverage of collective bargaining has stayed high even as formal membership fell — in France barely a tenth of workers are union members, yet almost all are covered by a collective agreement.¹¹ The American collapse of density toward irrelevance is in part a feature of the American model itself, in which a union must be built and defended firm by firm, against determined opposition, with little help from the law.

Two honest complications remain before any verdict, and the plate would be propaganda without them.

The first is that the union, in pushing up the wage of its members, may do so partly at the expense of others — the workers it shuts out. If a union lifts wages above the market-clearing level, the jobs that vanish fall not on its members, who are shielded by seniority, but on outsiders: the young, the unemployed, those queued for the jobs the higher wage has eliminated. Economists call this the insider–outsider problem, and it has been pressed hardest as an account of stubbornly high European unemployment — the charge that protected insiders bargained for wages that priced the outsiders out and kept them out.¹² The compression that is a sword of justice within the union can be a wall against those beyond it. How large the effect is remains disputed; that the possibility is real is not.

The second complication is the strangest fact in the whole subject, and it sits at the center of the present moment. Public approval of unions in America stands near seventy percent — about as high as it has been since the early 1960s — while union membership sits below ten. Two-thirds of the country approves of an institution fewer than one worker in ten belongs to. The recent years have sharpened the paradox rather than dissolved it: a visible revival of organizing energy — a doubling of election petitions, the auto workers' successful strike and record contracts, the campaigns at the coffee chains and the warehouses — set against a density that has not stirred from its lows, drives that have organized a few hundred stores out of many thousands and have mostly failed to reach a first contract, employers facing record numbers of unfair-labor-practice charges, and a political turn that has set the law against the organizers once again.¹³ The country appears to want the thing in the abstract and to be unable, or not permitted, to have it in the concrete.

Which returns the plate to the question it began with. The union is the institutional answer to the asymmetry of the opening plate — power assembled on the weaker side to meet the power of the stronger. Whether that answer is a good one cannot be settled in the abstract, because, like the legislated floor and like the single buyer it is built to oppose, its effect depends entirely on what it is set against. In a competitive market the union is a monopoly, trading jobs for wages and walling out the excluded. In a market where employers hold power — which is to say, in most real markets — it is a corrective, winning higher wages and more jobs at once, pulling the wage back up from the floor toward which the lone worker is forever pressed. The right question is never whether unions are good. It is the question this whole volume keeps putting to every force that touches the wage: what, exactly, is it countering?