The Floor

A minimum wage places a public condition on the private bargain: covered work must command at least the payment the law specifies. It adds a legislated floor to the worker's own reservation wage, the least the worker would accept while weighing other possibilities. The two can differ. Someone may be willing to work below the legal minimum and still be entitled to it; someone whose alternatives are good may refuse a wage well above it. The policy therefore changes the terms on which a job can be offered, and the question is what employers and workers do with the changed terms. The simplest competitive model gives a precise answer: a binding floor raises the wage while reducing the quantity of labor demanded. The argument deserves to be followed carefully, because the scope of its prediction is also the scope of the test it faces. A rule about the minimum payment becomes a question about the market that must make that payment.

In the competitive model, the market wage settles where the quantity of labor demanded meets the quantity supplied. Each employer takes that wage as given and hires until the contribution of another unit of labor just covers its cost. Set a legal minimum above the clearing wage and the firm moves along its demand curve: at the higher cost, fewer units of labor are worth hiring. Workers move along the supply curve in the opposite direction, offering more labor at the better wage. The legal rate thus produces two quantities where the clearing rate had produced one: the amount firms will hire and the larger amount workers would like to supply. Employment is limited by the former. The gap between them measures excess supply at the legal wage; the decline from the old employment level measures the jobs or hours removed by the demand response. Keeping those two magnitudes distinct matters. Some of the people unable to find work at the higher wage may have entered the market because of the higher wage, while others may have lost work they previously held.¹

The size of the employment response determines how the wage gain is distributed between workers who keep the work and workers who lose it. A less elastic demand curve permits a larger pay increase with a smaller reduction in labor demanded; an elastic demand curve permits a larger quantity adjustment. Even the total wage bill depends on both movements, since it is the wage multiplied by the labor paid for. A gain in that total would still leave the worker who lost a job in a different position from the worker who kept one. The same care applies to the unit being counted. A firm can preserve the number of people on its payroll while changing their hours, and a higher hourly wage does not by itself establish higher weekly earnings. The model supplies a disciplined way to ask these questions, provided its competitive setting and its chosen measure of labor are kept in view. Its force comes from the mechanism it specifies. Turning the diagram into a prediction for a particular industry requires evidence about the demand response and the scope employers have to adjust.²

A policy change can test the prediction when it provides a credible comparison with what would otherwise have happened. New Jersey raised its minimum wage in April 1992 while neighboring Pennsylvania left its rate unchanged. David Card and Alan Krueger surveyed fast-food restaurants on both sides of the boundary before and after the increase, reporting their findings in 1994. The comparison asks whether employment changed differently in the affected stores; it does not attribute every movement in New Jersey employment to the law. Its causal reading depends on the neighboring stores providing a useful account of the change that would have occurred without the increase. In their sample, the researchers found no indication that the higher minimum reduced employment. The result challenged the competitive prediction in a setting where many workers were directly affected, and made the choice of comparison central to the argument. A policy conclusion now had to survive questions about the establishments observed, the measure of employment, and the counterfactual path. The before-and-after experience of the treated employers alone could not settle any of them.³

Looking across the wage distribution provides a further way to ask whether a new floor raises the price of existing work or removes the work itself. Jobs previously paid below the new minimum must disappear from that part of the distribution under compliance. The consequential question is whether an offsetting number appears at or above the new floor. Doruk Cengiz, Arindrajit Dube, Attila Lindner, and Ben Zipperer used this approach in their 2019 study of 138 prominent state minimum-wage increases between 1979 and 2016. They found that the missing jobs below the new minimum were largely offset by excess jobs at or slightly above it, leaving overall low-wage job counts essentially unchanged over the following five years. This compares the number of jobs in parts of the wage distribution; it does not follow each displaced worker into a new position. The distinction matters to the interpretation, just as the distinction between headcount and hours does. An employment total can stay steady while particular workers change jobs, and a positive wage result for a group need not describe every member's experience.⁴

A small measured employment response leaves a substantive question about how the higher wage was absorbed. Employers may have room to adjust prices, profits, hours, benefits, or the organization of work; a change in pay can also change recruitment and turnover. The next plate gives one particular mechanism a full derivation. An employer with wage-setting power may have been restricting employment to keep the wage down, and a floor within the appropriate range can remove that incentive, raising both pay and employment. That is a coherent explanation to investigate. An unchanged job count alone cannot tell us how much of the adjustment came through that channel, how much through other margins, or how large a wage increase the same market could sustain. The evidence must therefore do two jobs that are often run together: estimate what changed after a policy, and distinguish the mechanisms that made it change. Finding room for higher pay is consequential in itself. Assigning that room a precise size, cause, and endpoint requires more than the absence of a detected job loss.⁵

The legislated floor joins the union and the safety net as an institution that can alter the worker's position in the bargain. Each works through a different means: a collective demand, support while searching, or a public rule governing the permitted offer. The effects depend on the labor market those institutions enter, and the minimum wage makes that dependence unusually visible. The competitive diagram identifies the quantities that would move under its assumptions; empirical comparisons ask how they moved in particular markets; the account of employer power explains why another structure can produce another response. Keeping those tasks distinct makes the policy argument more exact. The question is how far a specified floor can raise earnings, how its costs and gains are distributed, and what happens to access to work and to the conditions attached to it. The eighth plate now returns to the wage before the law intervenes, and changes one assumption: the employer must offer more to hire more. From that change comes the possibility the simple competitive diagram leaves out.