The minimum wage that did not cost jobs

Every introductory economics course teaches the same diagram. Put a floor under a price and you get a surplus; put a floor under wages and the surplus is unemployed people. In 1994 David Card and Alan Krueger published a study of fast food restaurants on either side of the Delaware River showing the opposite sign, and the profession spent a decade trying to make it go away. It did not go away. Card got a Nobel for it in 2021.

Two states, one river

Full-time-equivalent employees per fast food store, February 1992. Pennsylvania stores are slightly larger to begin with.

FTE employees per store, Feb 1992 and Nov 1992.

A natural experiment on a state line

In 1992 Card and Krueger phoned 410 fast food restaurants on both sides of the Delaware. Pennsylvania was the control group: same labour market, same burgers, different legislature.

The result that started the fight

New Jersey raised its minimum by 80 cents. Employment in New Jersey stores went up slightly while Pennsylvania stores shed staff. The textbook prediction had the sign backwards.

Then somebody pulled the payroll

Neumark and Wascher obtained actual payroll records, partly through the industry, and got the opposite answer. Phone surveys of shift managers, it turns out, are a noisy way to count employees.

Both sides went to the administrative data

Card and Krueger came back with government payroll data for the same period and found essentially no effect. The honest summary of round three is small and uncertain, not clearly negative.

A better question

Thirty years on, the field stopped arguing about two states and started counting jobs by wage across 138 state minimum wage increases. If a rise destroys jobs, the jobs below the new floor should disappear without reappearing above it.

They reappear

The missing jobs below the new minimum are matched almost one for one by extra jobs just above it. Total employment for low-wage workers barely moves; the wage bill goes up. That is the modern finding, and it has held up.

The claim

On 1 April 1992 New Jersey raised its minimum wage from $4.25 to $5.05 an hour. Pennsylvania, next door, did not. Card and Krueger surveyed 410 fast food restaurants in both states before and after the change and compared how employment moved. New Jersey stores, the ones hit by the increase, ended up with slightly more full-time-equivalent staff than Pennsylvania stores, not fewer.

The difference-in-differences estimate was positive, on the order of two to three full-time equivalents per store. Not “the minimum wage creates jobs.” Something more useful and more disturbing: a moderate increase did not visibly destroy them.

410

restaurants surveyed

New Jersey and eastern Pennsylvania, 1992

$4.25 → $5.05

New Jersey minimum wage

138

state minimum wage increases in the 2019 re-test

Cengiz et al., 1979–2016

How they found it

The design is the whole contribution. Before Card and Krueger, minimum wage studies mostly regressed national teen employment on the national minimum wage over time, which asks the data to separate a policy from every other thing that happened that decade.

The state line does the work instead. Fast food restaurants in Camden and in Philadelphia sit in one labour market, sell the same product, and face the same customers, weather and burger prices. The only sharp difference between them in April 1992 is which legislature set their wage floor. Compare the change in one group to the change in the other and the common shocks cancel.

That structure, now called difference-in-differences, is the reason this paper matters beyond the minimum wage. It moved labour economics from arguing about national trends to hunting for borders, cutoffs and policy discontinuities.

The mechanism

If the textbook diagram fails, something must be absorbing the cost. The candidates, roughly in order of how much support they have:

Monopsony. In a labour market with few employers and real search frictions, firms set wages rather than taking them. A wage floor below the competitive level can raise both wages and employment. This is now the standard explanation and it has its own body of evidence.

Prices. Fast food is mostly sold to local customers who have no alternative supplier of a nearby lunch. Some of the increase goes straight into the price of the sandwich.

Turnover. Low-wage jobs churn. Higher pay means fewer people quit, which means less hiring and training cost, which offsets part of the wage bill without any change in headcount.

Reduced profit and slower growth. Sometimes the store just makes less, or opens the next location a year later.

How much to believe it

The first round was messy and the critics were right about that. David Neumark and William Wascher (2000, AER) re-ran the comparison with actual payroll records rather than telephone interviews and found New Jersey employment falling about four percent relative to Pennsylvania. Their objection to the survey was reasonable: asking a shift manager over the phone how many part-timers work there is a noisy measurement, and noise that correlates with treatment is fatal.

The second round moved to administrative data and the effect went to roughly zero. Card and Krueger replied in 2000 using government payroll data covering the same period and found essentially no relative change. Two teams, the same underlying event, better data: small and uncertain, not clearly negative. That is where the original episode honestly ends.

What settles it is what came after. The New Jersey study is now the least important evidence for its own conclusion. Dube, Lester and Reich (2010) generalised the border trick to every pair of adjacent US counties straddling a state line, hundreds of them over sixteen years, and found employment effects close to zero with clearly positive wage effects. Cengiz, Dube, Lindner and Zipperer (2019, QJE) took a different angle entirely: across 138 state minimum wage increases they counted jobs by wage level, and found the jobs that disappeared from below the new minimum reappeared just above it, in almost equal number. That is the scene in the middle of this piece, and it is the single most convincing picture in the literature.

The limits are real and the good researchers state them. All of this concerns moderate increases in places where the new floor lands well under the local median wage. The Seattle studies (Jardim et al.) looked at a very large increase in a high-wage city and found reduced hours for some workers, and that finding has its own disputes. Nobody serious claims the demand curve never slopes down; the claim is that within the range the US has actually tested, disemployment effects are small.

What it means for you

Two things, and the second is the one that applies to your work.

First, notice how the field made progress: not by arguing harder about the same national time series, but by finding a comparison where the confounders cancel. When someone hands you a before-and-after number, your first question should be what the Pennsylvania is.

Second, the bunching estimator is a template worth stealing. Instead of asking “did the total go down,” Cengiz and colleagues asked where in the distribution the units moved. Totals hide compositional change. A flat headline with a redistributed distribution underneath is the single most common thing an aggregate metric conceals, and the fix is always to look at the histogram before and after, not the mean.

Suggested chart

The bunching picture, done properly with real data: the count of jobs by hourly wage bin before and after a minimum wage increase, with the new minimum marked. Shade the missing mass below the line in one colour and the excess mass above it in another, and label both areas with their totals. Public use CPS microdata supports this directly and it is the clearest thing in the whole literature.

Sources

Card, D. and Krueger, A. B., 1994, “Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania,” American Economic Review, 84(4), 772-793. nber.org.
Neumark, D. and Wascher, W., 2000, “Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania: Comment,” American Economic Review, 90(5), 1362-1396. aeaweb.org.
Card, D. and Krueger, A. B., 2000, “Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania: Reply,” American Economic Review, 90(5), 1397-1420. aeaweb.org.
Dube, A., Lester, T. W. and Reich, M., 2010, “Minimum Wage Effects Across State Borders: Estimates Using Contiguous Counties,” Review of Economics and Statistics, 92(4), 945-964. mitpressjournals.org.
Cengiz, D., Dube, A., Lindner, A. and Zipperer, B., 2019, “The Effect of Minimum Wages on Low-Wage Jobs,” Quarterly Journal of Economics, 134(3), 1405-1454. academic.oup.com.
Jardim, E. et al., 2022, “Minimum Wage Increases, Wages, and Low-Wage Employment: Evidence from Seattle,” American Economic Journal: Economic Policy. nber.org.