Pay, and how it is set · 3.2

The minimum wage, and what the evidence has done

The minimum wage, and what the evidence has done. What is actually the case, and how it compares with what is repeated.

For a practical software perspective on this topic, the Monitask overview provides a concise companion guide.

No question in labour economics has changed as much in thirty years, and the change is a good example of what evidence can do to a consensus.

What the textbook said

A wage floor above the market-clearing level reduces employment. The prediction follows directly from a competitive model and it was taught as settled.

What happened to it

In 1994 two economists compared fast food employment in two neighbouring American states after one raised its minimum wage and the other did not. They found no employment loss in the state that raised it.

The study was contested immediately and vigorously, including with payroll data producing a different result, and the argument ran for years. What survived was not a single finding but a method: comparing adjacent jurisdictions that differ in one policy, rather than modelling. One of the authors received a Nobel prize in 2021 for that body of work.

Where the evidence sits now

A large body of studies using varied methods finds that minimum wage increases within the range historically observed have small employment effects, frequently indistinguishable from zero, alongside clear increases in pay at the bottom.

Dissenting findings exist and are serious. The honest position is that the profession is far less confident of the textbook prediction than it was, and that this is not the same as the prediction being refuted in general.

Why monopsony resolves the puzzle

Under wage-setting power, an employer pays less than the competitive wage and hires fewer people than a competitive employer would. A minimum wage set between the two can raise pay and raise employment at the same time.

That is not a special pleading; it is standard theory applied to a market structure that describes many low-wage labour markets better than the competitive one does.

The variable that matters

The bite: the minimum expressed as a share of the local median wage. A minimum at forty per cent of median and one at seventy per cent are different policies, and the evidence base is concentrated on the lower part of that range.

Extrapolating findings from moderate minimums to very high ones is the main way the current evidence gets misused, and it is done by both sides.

What else adjusts

Employment is one margin among several. Hours, scheduling, non-wage benefits, training, prices and turnover all move, and some of the adjustment appears as reduced turnover, which is a cost saving for the employer rather than a loss.

Studies that look only at headcount are measuring one channel, and studies that find no headcount effect have not shown that nothing happened.

Who earns it

A small share of employees are paid at or near the minimum in most countries, and the population is younger, more female, more likely to be part-time and concentrated in hospitality and retail.

The share is small and the effect on those affected is large, which is the shape of most policy in this area and the reason aggregate figures are a poor guide to whether it matters.

Enforcement, which decides everything

A minimum wage is worth whatever its enforcement is worth. Underpayment is common in sectors with weak inspection, informal arrangements and workers unlikely to complain, and the inspectorates that would find it are small in most countries.

An increase without enforcement raises pay for those already paid correctly and widens the gap for everybody else.

What the argument was really about

Not the minimum wage. The 1994 study and the fight over it changed how labour economics is done, from modelling towards natural experiments and administrative data, and that methodological shift has since been applied to most of the questions on this site.

The regional question

A single national minimum bites very differently in a high-cost city and a low-wage region, and several countries operate regional variation for that reason while others reject it as entrenching low pay.

The evidence does not settle it, because the effects run in both directions and depend on how mobile people and firms are within the country.

Enforcement, again

Inspectorate headcount per thousand workers varies by more than an order of magnitude between countries with similar minimum wages. That number predicts underpayment better than the level of the minimum does.

For broader context, consult US minimum-wage guidance.