Monopsony: the strongest case for a minimum wage, and its limits

Most of the arguments on this site start from a simple picture: wages are set in a competitive market, and a legal floor above the market rate prices some workers out. That picture is useful. It is also, in one important respect, incomplete — and the best critics of our position know exactly where.

The gap has a name: monopsony. If employers have real power to set wages, rather than simply accepting whatever the market pays, then the textbook prediction can flip. A well-chosen minimum wage could raise pay and increase employment at the same time. That is not a fringe claim. It is standard microeconomics, and a large modern literature argues it describes real labor markets better than the competitive benchmark does.

If you want to argue against the minimum wage honestly, this is the argument you have to meet. So let's build it at full strength first, and then ask what it actually implies for policy.

Joan Robinson's buyer's market

A monopoly is a market with one seller. A monopsony is a market with one buyer. The Cambridge economist Joan Robinson introduced the term in The Economics of Imperfect Competition (1933), and her canonical example was the labor market: a single employer facing a pool of workers who have nowhere else to go.

The mining or mill town is the classic image, but the logic does not require literally one employer. What matters is that the firm faces an upward-sloping supply of labor. In a perfectly competitive market, a firm can hire as many workers as it likes at the going wage; if it offers a cent less, everyone quits. Under monopsony, a firm that wants more workers has to raise its wage to attract them, and a firm that pays a bit less loses only some of its workforce, not all of it.

That one change has a surprising consequence. Suppose the employer pays everyone the same wage. To hire one more worker, it must raise pay not only for the new hire but for every worker already on the payroll. So the true cost of the marginal worker — the marginal cost of labor — is higher than the wage itself. A profit-maximizing monopsonist responds by doing what a monopolist does with output: it restricts quantity. It hires fewer people than a competitive market would, and pays each of them less than the value they add. Robinson called the gap between a worker's marginal product and the wage "exploitation," in a technical rather than rhetorical sense.

Why a floor can add jobs

Now introduce a minimum wage. Under monopsony, a floor does something it cannot do in a competitive market: it changes the employer's incentive at the margin. If the law says the firm must pay at least W* to everyone, then hiring one more worker no longer requires raising pay for the existing staff — they are already at the floor. Up to the point where the floor meets the supply curve, the marginal cost of labor is simply the floor. The reason to hold back hiring disappears.

Read the diagram from left to right. Without regulation, the employer stops hiring at L_m and pays W_m. Set a floor at W*, where the supply curve crosses the value of workers' output, and the firm's best response is to hire L_c — more people, at higher pay. In this model the minimum wage is not a price control that destroys jobs. It is a correction that restores the competitive outcome.

But look at the red dotted line too. Push the floor well above W* and the employer is back to rationing jobs, now because workers cost more than they produce. Past a certain point, a floor leaves employment lower than the unregulated monopsony would have. Everything depends on where the number lands.

What the modern evidence says

For decades, many economists treated monopsony as a curiosity: relevant to company towns and perhaps to nurses in a one-hospital county, but not to the labor market at large. That view has changed substantially, for three reasons.

Firms face upward-sloping supply even in crowded markets

Alan Manning's Monopsony in Motion (2003) reframed the question. You do not need one employer for monopsony power to exist; you need frictions. Job search takes time. Workers have commutes, childcare arrangements, preferences about coworkers and schedules, and imperfect information about what else is out there. Building on search models such as Burdett and Mortensen (1998), Manning argued that these frictions mean a firm that cuts pay slightly does not lose its whole workforce — it just loses workers a bit faster and recruits a bit slower. That is an upward-sloping supply curve, even in a city with hundreds of employers.

Empirical work has backed this up in a range of settings. Staiger, Spetz, and Phibbs studied a legislated pay increase at Veterans Affairs hospitals and found nurse labor supply to individual hospitals responded very little to wages. Dube, Jacobs, Naidu, and Suri found that even on Amazon Mechanical Turk — a market that looks about as close to frictionless as labor gets — requesters face surprisingly inelastic labor supply. A meta-analysis by Sokolova and Sorensen pooled a large body of estimates from across the literature and concluded that firm-level labor supply is typically far from the perfectly elastic benchmark, though estimates vary widely with method.

Many local labor markets are concentrated

José Azar, Ioana Marinescu, and Marshall Steinbaum used online job postings to measure employer concentration by occupation and commuting zone. They found that a large share of local labor markets would count as highly concentrated under the thresholds antitrust agencies use for product-market mergers, and that higher concentration was associated with lower posted wages. Other researchers have debated how best to define a "market" — workers can switch occupations and commute farther than a vacancy database assumes — but the core finding that many employers face limited local competition has held up.

Minimum wage effects look different where concentration is high

If monopsony is the mechanism, the theory makes a testable prediction: minimum wage increases should do less harm to employment, or even help it, where employers have more power. A study by Azar, Huet-Vaughn, Marinescu, Taska, and von Wachter found exactly that pattern — employment responses to minimum wage increases were more positive in more concentrated labor markets. This also offers one coherent reading of the long run of small or null employment findings that followed Card and Krueger's 1994 New Jersey study.

None of this settles the empirical debate. Neumark, Wascher, and others continue to find disemployment effects in many settings, particularly for teenagers and when floors are high relative to local wages. But anyone who claims monopsony is purely theoretical is out of date.

The limits of a legislated number

Grant all of it. Employers have wage-setting power; many markets are concentrated; a floor at the right level can raise both pay and employment. What follows for policy? Less than it first appears, for four reasons.

The window is narrow, and nobody can see it. The diagram shows that the employment-raising floor sits in a band whose upper edge depends on workers' productivity and the slope of supply in that particular market. Above that band, the floor destroys jobs. Regulators do not observe marginal revenue product or firm-level supply elasticities. They observe, at best, a few years of noisy data after the fact. Stigler's 1946 point was precisely this: the theoretically correct minimum differs by firm, by occupation, and over time, and a legislature has no way to find it.

One number, thousands of markets. Monopsony power is local. It varies between a rural county with one distribution warehouse and a metro area with dozens, between home health aides and retail cashiers, between this year and next. A federal or state minimum wage applies one figure to all of them. Where employers have little power, the floor behaves like the competitive case and prices out the least-productive workers. Where they have a lot, the same figure may be too low to matter or too high to help. A single number can only be right by accident, and only in some places.

The floor treats the symptom, not the cause. Monopsony power comes from concentration, from search frictions, and from legal restrictions on workers' ability to leave. A minimum wage leaves every one of those in place. It also only reaches the bottom of the wage distribution. If employer power holds down the wages of a worker earning $22 an hour, a $15 floor does nothing for them.

Employers adjust on other margins. A firm that cannot pay less can still offer fewer hours, less predictable schedules, thinner training, or fewer benefits. Under monopsony, a floor changes the wage it can set; it does not stop the firm from exercising power through the rest of the job package.

Fixing the problem directly

If the problem is that employers face too little competition for workers, the most natural response is to give them more. Several tools do that without requiring anyone to guess the right wage.

• Antitrust enforcement in labor markets. The Department of Justice and the Federal Trade Commission issued guidance in 2016 stating that wage-fixing and no-poach agreements between employers can be prosecuted, and the 2023 Merger Guidelines explicitly direct the agencies to consider a merger's effect on competition for workers. Alan Krueger and Orley Ashenfelter documented no-poach clauses in the franchise contracts of many major chains, including well-known fast-food brands, which barred one franchise location from hiring workers from another in the same chain. Several chains agreed to drop those clauses after state attorneys general investigated.
• Non-compete reform. Non-compete agreements directly limit a worker's outside options, which is the definition of monopsony power. They are not confined to executives: Jimmy John's drew national attention for using them with sandwich makers. The FTC's 2024 rule banning most non-competes was blocked by a federal court, and the agency has since stepped back from it, but states can and do act on their own. Several, including California, already refuse to enforce most of them.
• Occupational licensing reform. Licensing requirements raise the cost of moving between states and switching occupations. Every barrier like that makes workers less mobile and employers' supply curves less elastic. Interstate license reciprocity and trimming requirements that have no clear safety rationale widen workers' choices.

These tools have their own costs and debates. But they share an advantage the minimum wage lacks: they push toward competition everywhere, with no risk of overshooting. A rule that stops employers from colluding on wages cannot price a teenager out of her first job. And they help workers at every wage level, not just those near the floor.

Meanwhile, if the goal is to raise the incomes of low-wage workers, as opposed to correcting employer power specifically, the Earned Income Tax Credit does that directly, without asking anyone to locate the edge of the window in the diagram above.

Where we land

Monopsony is the strongest argument for a minimum wage because it is the one that rests on mainstream theory and a growing body of evidence rather than on intuition about fairness. It is right that labor markets are not perfectly competitive. It is plausible that some modest wage floors in some places have raised pay without costing jobs, and possibly even added a few.

What the argument does not deliver is a case for a single, politically chosen number applied across a whole state or country. The theory itself says the beneficial range is bounded, that it differs from market to market, and that overshooting does real harm. That is a description of a policy that requires local, precise, constantly updated knowledge — exactly the kind of knowledge a legislature does not have. Friedman's broader critique was never that markets are perfect. It was that blunt instruments tend to miss.

If employer power is the disease, the cure is more competition for workers. We would rather spend political capital there than on a number that might help in one county and hurt in the next.