The Monopsony

The plates so far have read the wage as a measure of the worker. The marginal principle made it the worth of his last hour; human capital made it the yield on his skill. This plate turns the whole picture over. Under a single change to the model — a change not exotic but, on inspection, ordinary — the wage ceases to measure what the worker is worth and begins to measure what the employer can get away with. The number does not change; its meaning inverts.

The change concerns the supply of labor as the firm sees it. The competitive theory assumes a firm can hire as many workers as it wishes at the going wage, which it cannot itself move — its labor supply is flat, the market wage a fact handed to it. Suppose instead that to hire more workers a firm must offer a higher wage: that labor does not flood in at the posted rate, and the firm faces an upward-sloping supply. This single buyer, or buyer with such power, is a monopsonist — the mirror of a monopolist, a dominant buyer rather than a dominant seller — and the consequence is precise. To hire one more worker the firm must raise the wage, and, unless it can pay the newcomer more than the rest, it must raise the wage for everyone already on the payroll. The cost of the marginal worker is therefore not his wage alone but his wage plus the raise conceded to all the others — so the marginal cost of labor stands above the wage. The firm, maximizing profit, hires only up to the point where that marginal cost meets the value of the marginal product; and because the marginal cost lies above the wage, the wage is left below the marginal product. The firm holds employment down precisely in order to hold the wage down.¹

That gap has a name and a size. Pigou called the proportional shortfall of the wage below the marginal product the rate of exploitation, and its magnitude is governed by one quantity: the elasticity of labor supply to the firm — how readily workers come and go in response to its wage. Where workers are abundant and mobile, ready to leave for a penny's difference, supply is highly elastic, the marginal cost of labor barely exceeds the wage, and the gap is small. Where workers are scarce, tied down, slow to move, supply is inelastic, and the gap yawns. In the limit of perfect mobility the elasticity is infinite, the gap closes, and the competitive result returns; monopsony is simply what stands between the actual labor market and that limit.²

For a long time this was treated as a textbook curiosity, important in logic but rare in life. The image attached to it was the company town: the single coal mine or cotton mill that is the only employer within a day's travel, its workers with nowhere else to take their labor. Such cases were real — and a few labor markets, nursing among them, were studied as genuine instances, the hospitals in a region being too few to bid against one another for nurses. But the cases seemed special, the exception that proved the competitive rule.³

The modern revival of the idea, owed above all to Alan Manning, was to see that monopsony power requires no single employer at all — only that workers have limited choice among employers, which is the normal condition of the labor market rather than a rare one. Finding a job takes time and effort; jobs differ in location, hours, conditions, and culture, so that workers do not treat them as interchangeable; moving is costly; information is poor. Each of these frictions, on its own, gives every employer a sliver of the monopsonist's power, because a firm that quietly cut its wage would not lose all its workers in an instant, as perfect competition demands, but only some of them, and only gradually. The flat labor-supply curve of the textbook turns out to be the special case; the upward slope is the general one. By this light, monopsony is not the exception but very nearly the rule, and the competitive labor market the curiosity.⁴

Once monopsony is admitted as the general case, three things that puzzle the competitive theory fall into place. The first is the minimum wage. In the competitive model a wage floor set above the market rate must reduce employment, pricing the lowest-paid out of work — the standard verdict, and the subject of an earlier plate. Under monopsony the verdict reverses over a range: a minimum wage set between the monopsony wage and the competitive wage actually raises employment, because it strips the firm of its reason to hold hiring down — the firm can no longer suppress the wage by hiring fewer workers, so it hires more. This is the mechanism behind the celebrated empirical finding that increases in the minimum wage did not produce the job losses the competitive model predicted.⁵

The second is the long, much-discussed divergence between productivity and pay — output per worker rising steadily while the typical wage lags behind it. The competitive theory has no easy room for such a gap, since it ties the wage to the product by definition. Monopsony has room to spare: if the wage sits below the marginal product by a margin set by employer power, then a widening gap between what is produced and what is paid is just what one would expect as that power grows, and the explanation need not run through skill or technology at all. The third is the renewed attention of antitrust: where the gap is set by how little choice workers have, employer concentration and the devices that limit mobility — agreements among firms not to poach one another's workers, clauses forbidding workers to leave for a competitor — become, quite literally, instruments for manufacturing and entrenching monopsony power.⁶

Here is the inversion stated plainly. Marginal productivity made the wage a mirror of the worker: what you are paid is what you are worth. Monopsony makes the wage, in its larger part, a mirror of the employer: what you are paid is what you can be made to accept. The very same number — the wage — that the orthodox theory read as a measure of contribution, the monopsony theory reads as a measure of power, and the two readings coincide only in the competitive limit that real labor markets, frictional and concentrated and slow, may never reach. The figure and the ground change places: the market structure, which the orthodoxy treated as a frictionless backdrop against which marginal products were simply revealed, turns out to be doing the decisive work — fixing how much of his own product the worker keeps and how much the employer takes.

It is important to be exact about what survives, because the inversion is often overstated. Monopsony does not abolish marginal productivity. The marginal product still sets a ceiling — no firm will long pay a worker more than he adds — and it still governs the relative demand for one kind of labor against another. What monopsony abolishes is the equals sign: the identification of the wage with the marginal product. The wage is loosed from the ceiling and falls somewhere between two figures — the worker's marginal product above, his next-best alternative below — and where in that interval it comes to rest is not read off the technology but settled by power. Which is to say that Smith's unequal bargain, banished by the competitive idealization a century ago, has returned as the general case, and returned now with a formal apparatus to measure it.⁷

One special case sharpens the moral. A monopsonist able to pay each worker a different wage — to offer every recruit just enough to secure him and not a penny more — would actually hire the efficient, competitive number of workers, since for such a firm the marginal cost of an extra worker is only that worker's own wage. Employment would be efficient; but every worker would be paid only the least he would accept, and the entire surplus of the labor over its reservation price would go to the firm. Efficiency and exploitation, that case shows, are not opposites: one can have the right number of workers and the most lopsided possible division of what they produce. The ordinary single-wage monopsonist is in a sense kinder — it cannot squeeze each worker individually, so it squeezes employment instead — but the lesson stands, that where the wage measures power, fairness and efficiency are separate questions, and the wage answers neither on its own.

The wage, then, is not a fact of nature read off the worker's productivity. It is a division of a surplus between two parties, and the division reflects, in the end, who can hold out the longer — exactly what the first plate said, before the competitive century talked itself out of believing it. What the orthodoxy called simply the wage, the rest of this volume is obliged to call something less tidy: the worker's share of a contested product, the size of the share set less by what he makes than by what, at the table, he has the power to keep.