Intermediatepolicy
Minimum Wage Research: What Economists Actually Found
Learning objectives
- Explain the textbook prediction about minimum wages and employment
- Describe the Card and Krueger study and why it was revolutionary
- Summarize later evidence, including the Seattle minimum-wage study
- Explain monopsony and other theories for small employment effects
- Assess where the research consensus stands and what remains uncertain
The textbook prediction
The standard competitive model of labor markets makes a clean prediction: a minimum wage set above the market-clearing wage must reduce employment. Employers facing higher labor costs hire fewer workers; some workers gain higher pay, but others lose their jobs or never get hired. The only question, in the textbook, is how large the job loss is.
For decades this was the near-consensus among economists. Surveys in the 1970s and 1980s found broad agreement that minimum wages reduced teenage employment, and the policy debate was framed as a trade-off: higher wages for some workers at the cost of jobs for others, with the poorest and least skilled bearing the employment risk.
This framework still structures the debate. Every minimum-wage proposal is implicitly a bet about the size of the employment effect — and the last thirty years of research have been, in large part, an extended argument about whether the textbook model describes low-wage labor markets accurately.
Card and Krueger's revolution
In 1994, David Card and Alan Krueger published a study that upended the field. When New Jersey raised its minimum wage, they compared fast-food employment in New Jersey with neighboring Pennsylvania, which did not raise its wage — a natural experiment. Contrary to the textbook, they found no employment loss in New Jersey relative to Pennsylvania; if anything, employment rose slightly.
The study's method — difference-in-differences using a neighboring state as a control — became the template for a generation of research. Its substantive finding launched a thousand follow-ups: perhaps low-wage labor markets do not behave like the competitive textbook model.
The paper was fiercely contested, and later work using payroll data (rather than the original phone surveys) found more mixed results. But its central contribution survived: modest minimum-wage increases do not reliably produce the job losses the simple model predicts, and the burden of proof shifted to those claiming large disemployment effects.
Card and Krueger did not prove minimum wages never cost jobs — they showed the effect was far smaller than believed at moderate levels. Card later won the 2021 Nobel Prize in Economics, partly for this empirical revolution in labor economics.
The Seattle experiment
Seattle's decision to raise its minimum wage toward $15 — then the highest in the US — created the most scrutinized natural experiment in the literature. A University of Washington study using state administrative data initially found significant reductions in hours worked by low-wage workers, suggesting the textbook effect at high wage levels.
But the finding proved fragile. Critics noted the study excluded multi-location businesses (like chain restaurants), missing much of the affected workforce. A later revision incorporating those employers found much smaller effects, and other studies of Seattle found little employment impact alongside rising pay.
The Seattle saga illustrates a broader pattern: at moderate levels (around 40–50% of the median wage), employment effects are consistently small; as the minimum approaches 60% or more of the median, evidence of hours and employment effects grows. The policy question is not 'does it cost jobs?' but 'how high is too high?'
Why might the textbook be wrong? Monopsony
The leading theoretical explanation for small employment effects is monopsony — buyer power in labor markets. If employers have market power over wages (because workers face search costs, limited options, or non-compete constraints), they pay below workers' productivity and hire fewer than the competitive level. A moderate minimum wage then forces wages up toward competitive levels, potentially increasing employment.
Evidence for employer wage-setting power has grown: studies find that labor markets are often concentrated, that workers do not readily switch jobs for small pay differences, and that firms systematically pay less where they face less competition. In such markets, the minimum wage corrects a market failure rather than creating one.
Other mechanisms also dampen job losses: higher wages reduce costly turnover, modest price increases pass costs to consumers, and productivity rises as firms reorganize. None of this means any wage level is safe — monopsony models also predict job losses once the minimum exceeds productivity — but it explains why moderate increases have small effects.
Where the consensus stands
The research frontier has moved from 'do minimum wages cost jobs?' to finer questions. Meta-analyses find average employment effects close to zero for the moderate increases most studied, with meaningful uncertainty. Recent high-quality studies using border-discontinuity designs — comparing counties on either side of state lines — consistently find small effects.
The emerging consensus: moderate minimum wages (up to roughly half the median wage) raise pay for low-wage workers with little detectable employment loss; higher levels carry growing risks, particularly for teens and the least skilled; and effects vary by labor-market concentration, with monopsonistic markets tolerating higher minima.
What remains uncertain: long-run effects on automation and business models, which short studies miss; effects in developing countries with large informal sectors; and the interaction with other policies like the earned income tax credit. The minimum wage is now an evidence-based policy with known unknowns — a vast improvement over the certainties of the old textbook debate.
Popular summaries often claim research proves minimum wages never cost jobs. The actual finding is subtler: moderate increases show small, often statistically insignificant employment effects — but effects grow at higher levels, and absence of evidence in short-run studies is not proof of zero long-run impact.
Key points
- The textbook model predicts job losses; three decades of research find small effects at moderate levels
- Card and Krueger (1994) launched the empirical revolution with a NJ–PA comparison
- Seattle's $15 experiment showed effects grow as the minimum nears 60%+ of median wages
- Monopsony (employer wage-setting power) is the leading explanation for small effects
- Consensus: moderate minima raise pay with little job loss; higher levels carry more risk
- Long-run automation effects remain the key unknown
Common misconceptions
✓ Truth: That was the consensus before the 1990s. Modern empirical work consistently finds small employment effects from moderate increases. Disagreement now centers on how high is too high, not whether any minimum wage destroys jobs.
✓ Truth: Studies find small average effects with real uncertainty, and clearer negative effects at high levels relative to median wages. The honest summary is 'small effects at moderate levels,' not 'no effects ever.'
Frequently asked questions
What is monopsony in labor markets?
Monopsony means employers have wage-setting power because workers cannot easily switch jobs — due to few local employers, search costs, or limited information. A monopsonist pays below competitive wages, so a moderate minimum wage can raise both pay and employment.
Why do studies of the same wage hike disagree?
Different data sources (surveys vs. payroll records), different control groups, and different coverage (single-location vs. chain employers) all matter. The Seattle debate showed how excluding multi-location firms can flip the headline result.
Does the minimum wage help reduce poverty?
Modestly. Many minimum-wage workers are not in poor households (e.g., teenagers), and the poorest often do not work. Targeted policies like the earned income tax credit reach poor households more directly, though the minimum wage still raises earnings for many low-paid workers.