The Floor

This volume has now found the floor of the wage twice. The plate on subsistence located the classical floor in the body — the cost of keeping a worker alive. The plate on insurance located the modern floor in the reservation wage — the lowest offer a worker will accept, set by their options and the cushion a society provides while they search. Both are floors the market arrives at. This plate is about a third floor, of a different kind: not one the market finds, but one the law imposes. The minimum wage is a number below which the wage may not legally fall, whatever the market would otherwise do.

It is the most direct intervention a society can make in the wage, and the most bitterly argued. In the United States the federal minimum has stood at $7.25 an hour since 2009 — the longest it has ever gone without a raise, its real value steadily eroded by inflation. Around that frozen federal floor the states have split: some twenty still sit at $7.25, while around eighteen states and the District of Columbia have legislated minimums of $15 or more, the fruit of the “Fight for $15” that began in 2012. For the first time, more American workers now live under a floor of $15 or higher than under the federal one.¹ The question this plate asks is the one beneath all the politics: when the law lifts the floor, what happens to the workers it is meant to help?

The answer that dominated economics for most of the twentieth century was discouraging, and it followed from the simplest model of the labour market. Suppose the market is competitive — many small employers, many workers, no one with power over the wage. Then, as the early plates described, the wage settles where the supply of labour meets the demand for it: at that wage, everyone who wants to work at the going rate can, and every employer who wants a worker at that rate can find one. The market clears.

Now impose a minimum wage above this market-clearing level. The logic of a price floor is merciless. At the higher wage, two things happen at once. Employers, facing a higher price for labour, want less of it — they slide up the demand curve, and hire fewer workers. And more workers, drawn by the higher wage, want the jobs — they slide up the supply curve. The result is a gap: more workers wanting work than there are jobs on offer, which is to say unemployment. And the workers who lose out are exactly those the policy meant to help — the least skilled, the least experienced, the young, whose product is worth least and who are therefore first to be priced out. On this account the minimum wage is a cruel illusion: it raises the pay of those who keep their jobs by destroying the jobs of those who do not.² For decades this was the profession's settled view, taught in every introductory class, and the case against the minimum wage rested on it almost entirely.³

The case held only as long as its premise did — that the labour market is competitive. The eighth plate of this volume argued that it often is not; that search frictions, switching costs, and concentration give employers a measure of wage-setting power, making the typical labour market not a perfect competition but a monopsony of degree. And in a monopsonised market, the minimum wage behaves in a way that would have seemed, to the textbook, impossible.

Recall the monopsonist's predicament. To hire one more worker, the employer must offer a higher wage — and must pay that higher wage not only to the new worker but to everyone already employed. The cost of an additional worker therefore exceeds that worker's wage, sometimes greatly, and the employer, to keep the wage down, deliberately hires fewer workers than they otherwise would. Monopsony means a wage below the value of the work and employment below what competition would deliver: the worker is doubly shortchanged, in pay and in numbers.

Now set a minimum wage above this monopsony wage. Something remarkable happens. The employer's reason for restricting employment was that hiring more drove up the whole wage bill; but with the wage fixed by law at the minimum, hiring more no longer raises the wage at all — the firm can take on additional workers at the minimum without paying everyone more. The brake on hiring is released. The firm hires up to the point where the minimum wage meets the value of the marginal worker: more workers than before, at a higher wage than before. The minimum wage raises both pay and employment, undoing at a stroke the distortion the monopsonist imposed.⁴ It is the mirror image of the union of the ninth plate, achieving by law what the union achieved by organisation — and within a range, it is pure gain. The catch is the range. The result holds for a minimum set above the monopsony wage but below what a competitive market would pay; push it higher than that, above the true value of the work, and the textbook case reasserts itself, and jobs begin to vanish after all.⁵

For most of the century the two cases — the textbook's job losses, monopsony's job gains — remained a theoretical standoff, because the question they disagreed about is empirical, and the empirics were hard. You cannot run the experiment cleanly: raise the minimum wage and watch, and you can never be sure whether the jobs that did or did not disappear would have done so anyway, for reasons having nothing to do with the wage.

Then, in 1994, David Card and Alan Krueger found a way to ask the question the world itself had set up. In 1992 New Jersey raised its minimum wage while neighbouring Pennsylvania did not. Card and Krueger surveyed fast-food restaurants — the archetypal employer of minimum-wage labour — on both sides of the state line, before and after the increase, and compared what happened. If the textbook were right, employment should have fallen in New Jersey relative to Pennsylvania. It did not. If anything, employment in New Jersey rose slightly. The increase had not destroyed the jobs the theory insisted it must.⁶

The finding was a bombshell, and it was fiercely fought. Critics re-examined it with different data — payroll records rather than survey responses — and reported smaller or negative effects; the dispute ran for years and is not entirely settled even now. But the deeper consequence of Card and Krueger's work outlasted the particular quarrel. They had shown how to use a natural experiment — an accident of policy that treats one group and not an otherwise-similar other — to pull the causal effect of a policy out of the tangle of everything else going on. This “credibility revolution” transformed empirical economics, and Card was awarded a share of the 2021 Nobel for it. The minimum wage was the occasion; the method was the legacy.⁷

Three decades of research have followed Card and Krueger — hundreds of studies across many countries and many minimum-wage changes — and a rough verdict has emerged, not the old certainty in either direction but something more careful. For moderate minimum wages, the disemployment effect the textbook predicts is small, often too small to detect: raise the minimum within a reasonable range and the wages of low-paid workers rise while their employment barely moves. The most careful recent work, tracking the whole distribution of wages around a minimum-wage increase, finds that the jobs lost just below the new minimum are largely offset by jobs and pay gained just above it — the floor lifts the bottom of the distribution without emptying it.⁸

But — and the evenhandedness matters here — this is a statement about moderate minimums, and the qualifier carries weight. A minimum wage is not a magic number that is always benign; it is benign within a range and harmful beyond it, exactly as the monopsony analysis predicts. Push the minimum high enough relative to the local wage — and a figure that is moderate in an expensive city may be extreme in a poor rural county — and the disemployment effects do appear. The studies of the steepest increases, such as Seattle's rapid climb toward $15, have reached genuinely conflicting conclusions, some finding lost hours and jobs, others finding none.⁹ The honest summary is neither the old slogan that the minimum wage always kills jobs nor the newer one that it never does. It is that the effect depends on how high the floor is set relative to what the market would otherwise pay.

Step back from the policy quarrel and ask what the evidence has been telling us about the wage itself, for that is this volume's concern. The single most important fact to come out of thirty years of minimum-wage research is this: a moderate minimum wage can raise pay without destroying employment. And that fact is not neutral between the theories of the wage — it is a near-decisive piece of evidence for one of them.

If the wage were simply the value of the work — if the competitive model held, and every worker were paid their marginal product — then a minimum wage above that value could only do harm, exactly as the textbook says. The very fact that it does not, that there is room to raise the wage without pricing the worker out, means the wage was not equal to the value of the work to begin with. There was a gap — a wedge between what the worker produced and what the worker was paid — and the minimum wage moved into that gap and closed some of it. But a gap between the wage and the value of the work is precisely what the eighth plate called the monopsony markdown, the signature of employer power. So the benign employment effect of moderate minimum wages is not merely consistent with employer power; it is a measurement of it. Wherever a minimum wage raises pay without cutting jobs, it is telling us that employers there had been paying less than the work was worth — that there was power in the market, and the law has clawed some of its proceeds back for the worker.¹⁰

This does not make the minimum wage a free lunch, and honesty requires the other side. Even where employment holds, costs can surface elsewhere — in reduced hours, in slower hiring, in higher prices passed to consumers, in a quickened push to replace workers with machines, a theme the next plate takes up. And the minimum wage is a blunt instrument against poverty, since many who earn it are not poor and many of the poor do not work; more targeted tools, like wage subsidies paid through the tax system, reach the needy more precisely. A thoughtful person can favour a higher minimum and still grant all of this.¹¹ But the central lesson stands, and it is the volume's: the room the minimum wage has to do good is the exact measure of how far the wage had fallen below the worth of the work — which is to say, the measure of power.

Place the minimum wage, at last, in the bargaining range that has framed this whole volume. The wage, we have said throughout, is settled somewhere between a floor and a ceiling — the ceiling the value of the work, the floor the worker's walk-away point — and where exactly it lands is decided by power. The minimum wage is the law reaching directly into this range and raising the floor by fiat.

Seen so, the three plates on the floor come together. Subsistence was the floor nature imposed; the reservation wage is the floor the market and the safety net arrive at; the minimum wage is the floor the law decrees. And the minimum wage is, like the union and like the safety net, a power-shifting institution — a way of moving the floor up, and the worker's share of the surplus with it, against the downward pull of employer power. It belongs to the same family as the ninth plate's union and the tenth's insurance: three different mechanisms — collective organisation, social provision, legal command — for one end, which is to give the worker, in a bargain they would otherwise enter on their knees, something more like an equal footing.¹²

And the minimum wage carries a lesson the others only imply. Because it intervenes so directly, its consequences are unusually legible, and what they reveal is the central claim of this volume written in policy. A minimum wage that raised pay only by destroying jobs would be proof that the wage had been a measure of worth all along — that there was no slack, no power, nothing to reclaim. That moderate minimum wages do not work this way is the proof, as close to direct as economics offers, that the wage was never only worth: that beneath it lay a gap cut by power, into which the law could step. Set the floor inside that gap, and it lifts the worker at the employer's expense, reclaiming what power had taken. Set it above the true ceiling, and it reaches past worth itself, and does the harm the textbook warned of. The minimum wage is, in the end, a wager on a single quantity — how much of the wage was power, and how much was worth. This whole volume has been an argument that the answer is: more power than we are usually told.¹³