If you have spent any time arguing about the minimum wage, someone has probably mentioned New Jersey. Usually it comes up as a trump card: economists once believed minimum wages killed jobs, then David Card and Alan Krueger went and checked, and the jobs were still there. Case closed.

The real story is more interesting than that, and a site that argues for abolishing the minimum wage has an obligation to tell it straight. The New Jersey study was a genuinely important piece of research. It changed how economists study labor markets, and it deserves to be taken seriously by anyone who thinks wage floors cause harm. It also has real limits, and it set off a dispute over data and interpretation that was never fully resolved. Both of those things are true at once.

The natural experiment

On April 1, 1992, New Jersey raised its state minimum wage from $4.25 to $5.05 an hour, an increase of a little under 19 percent. Pennsylvania, next door, stayed at the federal minimum of $4.25. For economists, this looked like a rare opportunity. Two neighboring states with broadly similar economies, one of which had just changed the price of low-wage labor while the other had not.

Card and Krueger, both at Princeton at the time, set out to measure what happened. Their team surveyed fast-food restaurants in New Jersey and eastern Pennsylvania by telephone, focusing on four chains: Burger King, KFC, Wendy's, and Roy Rogers. The first wave of interviews took place in February and March of 1992, shortly before the increase. The second wave came in November and December, several months after it. In total they gathered data from around 400 restaurants, and they went back to nearly all of them for the follow-up.

Fast food was a sensible choice. It is a large employer of minimum-wage workers, especially teenagers. Chain restaurants are fairly standardized, so a Burger King in Camden and a Burger King outside Philadelphia are doing roughly the same work with roughly the same technology. And because prices and menus are similar across locations, it's easier to compare like with like.

The method they used is now called difference-in-differences. You measure how employment changed in New Jersey between the two survey waves, measure how it changed in Pennsylvania over the same period, and compare the two changes. Pennsylvania serves as the stand-in for what would have happened in New Jersey without the wage increase. Anything that hit both states equally, like the regional economy or seasonal patterns, washes out.

What they found

Standard competitive theory predicts that a binding increase in the price of labor should lead employers to use less of it. Card and Krueger found no evidence of that. Full-time-equivalent employment at New Jersey restaurants did not fall relative to Pennsylvania. In their main estimates it actually rose a little, while employment at the Pennsylvania stores declined over the same months.

They also looked within New Jersey, comparing restaurants that had been paying close to $4.25 before the increase (and so were forced to raise wages the most) with restaurants that were already paying $5.05 or more. Again, the stores hit hardest by the law did not show the job losses the textbook model would predict.

One finding that gets less attention: prices for meals appear to have risen somewhat in New Jersey relative to Pennsylvania. That matters, because it tells you at least part of the cost was passed along to customers. The higher wage wasn't free; it just didn't show up as fewer jobs in this sample.

The paper, "Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania," appeared in the American Economic Review in September 1994. A year later, Card and Krueger expanded the argument into a book, Myth and Measurement: The New Economics of the Minimum Wage, which pulled together this and other studies to challenge the professional consensus.

Why economists took it seriously The study's influence came less from its conclusion than from its design. Instead of running regressions on national time series and hoping the controls were right, Card and Krueger found a policy change, picked a comparison group, and went out and collected their own data. That approach has since become a standard way of doing empirical economics, on every side of the minimum wage question.

The payroll-data challenge

The result was controversial almost immediately, and the most substantial challenge came from David Neumark and William Wascher.

Their concern was the data itself. Card and Krueger's employment numbers came from telephone interviews, in which a manager or assistant manager was asked how many full-time and part-time workers the restaurant employed. Phone surveys like that can be noisy. The person answering in February might not be the person answering in November, and "how many part-timers do you have" is a question with some wiggle room.

Neumark and Wascher assembled a different dataset: actual payroll records from fast-food restaurants in the same areas, covering the same chains. The initial collection of those records was organized by the Employment Policies Institute, an industry-funded group that opposed minimum wage increases, and Neumark and Wascher later supplemented it with records they gathered themselves. Critics pointed to the industry connection; Neumark and Wascher argued that payroll records were simply a more reliable measure of hours worked than recollections over the phone.

Using the payroll data, they reached the opposite conclusion. Employment in New Jersey fell relative to Pennsylvania after the increase, by an amount consistent with the conventional view that higher minimum wages reduce low-wage employment. Their comment was published in the American Economic Review in December 2000.

Card and Krueger's reply

Card and Krueger answered in the same issue. They raised questions about how the payroll sample had been collected and whether it was representative, and they brought in another source: administrative data from the Bureau of Labor Statistics, drawn from the unemployment insurance records that employers file (often called the ES-202 data). These cover essentially all employers rather than a sample of volunteers.

With the BLS data, they again found no meaningful decline in New Jersey fast-food employment relative to Pennsylvania. Their estimates were small and not statistically distinguishable from zero. They concluded that the increase had, at most, a small effect on employment, and that it was hard to make the case for a large negative one.

So the exchange ended with each side holding its own dataset and its own reading. Plenty of economists came away thinking the payroll critique had weakened the original "employment went up" headline without establishing a large job loss either. Others thought the critique was decisive. Few people changed sides.

Being fair about the fight It's tempting for minimum wage skeptics to treat the Neumark–Wascher comment as a clean refutation. That overstates it. The more defensible summary is that the original survey result was fragile to the choice of data, and that the best available evidence for this particular episode is consistent with a small effect, positive or negative, rather than a large one.

The Nobel, and what it rewarded

In 2021, David Card received the Nobel memorial prize in economic sciences "for his empirical contributions to labour economics." He shared the prize with Joshua Angrist and Guido Imbens, who were recognized for their methodological contributions to the analysis of causal relationships. Alan Krueger had died in 2019, and the prize is not awarded posthumously. The Nobel committee's materials made clear that the New Jersey work was part of what it had in mind, and many observers saw the award as recognizing Krueger's contribution too.

The prize is often cited as if the Swedish Academy had ruled on the minimum wage. It didn't. What it recognized was a way of answering questions: using natural experiments to get at cause and effect in settings where you can't run a controlled trial. Card's other well-known work, including his study of the 1980 Mariel boatlift and its effect on the Miami labor market, used the same approach. The award says the method was important. It doesn't say that every result produced with the method is final, and Card himself has generally described the minimum wage evidence in measured terms.

What the study can and can't tell us

Suppose you accept Card and Krueger's numbers entirely. What exactly have you learned? Less than the trump-card version suggests, for a few reasons.

  • The increase was modest. Eighty cents an hour, from a starting level that was already low relative to typical wages in a fairly prosperous state. A finding about a small, moderate increase doesn't automatically extend to a much larger one, or to a floor set close to the local median wage.
  • The window was short. The follow-up survey came within the same calendar year as the increase. Some of the ways businesses adjust to higher labor costs, such as redesigning workflows, investing in equipment, opening fewer new locations, or slowly letting headcount drift down through attrition, take years. A study that ends after a few months can't see them. On the other side of the ledger, the increase had been legislated well before it took effect, so some adjustment may also have happened before the first survey.
  • It's one industry in one region. Fast-food chains in New Jersey and eastern Pennsylvania are not the whole low-wage labor market. Independent restaurants, small retailers, and businesses in poorer regions may have different margins and different options.
  • Pennsylvania may not be a perfect control. The whole design rests on the assumption that, absent the law, New Jersey and Pennsylvania employment would have moved together. That's a reasonable assumption but not a guaranteed one, especially during the early-1990s slowdown.
  • Headcount isn't everything. Even where employment holds steady, employers can respond through hours, scheduling, training, non-wage benefits, or prices. The study found evidence of price increases, which is itself a reminder that someone pays.

None of these points shows that the study was wrong. They show that its domain is narrow. It is strong evidence that a moderate increase in a particular setting did not produce a large, immediate drop in fast-food jobs. It is weak evidence about what a high national floor would do over a decade.

Where the wider literature stands The New Jersey study opened a long line of research rather than closing one. Later work using many state-level increases has produced a range of estimates, from near-zero effects for moderate increases to clearer losses where floors bind tightly or for specific groups such as teenagers. Monopsony models, where employers have some power to set wages, offer one explanation for why small increases might not cost jobs. Serious economists disagree about how often those conditions hold.

The honest takeaway for the abolition argument

Here is how we think the New Jersey study should sit with anyone making the case against minimum wage laws.

First, concede what it shows. A crude version of the anti-minimum-wage argument says that any increase, of any size, anywhere, will visibly destroy jobs. The New Jersey evidence, and much of what followed, makes that version hard to defend. Labor markets have frictions, employers have some wage-setting power, and businesses can absorb modest cost increases in ways a simple supply-and-demand diagram leaves out.

Second, notice what it doesn't show. The case for abolition doesn't actually rest on the claim that every small increase causes immediate layoffs. It rests on the idea that a legal price floor is a blunt instrument whose costs grow as the floor rises relative to local productivity, and that those costs fall disproportionately on the least experienced workers. A short study of a moderate increase in one industry doesn't speak much to that, in either direction. Large floors in low-wage areas, long-run adjustment, and effects on people who never get hired are exactly the cases the study wasn't built to observe.

Third, take the method seriously. Card and Krueger's real legacy is the insistence that claims about policy be tested against carefully gathered data. That standard should apply to everyone. If wage floors cause harm, the way to show it is with credible natural experiments and better data, not by repeating a textbook diagram louder.

Our own view remains that the minimum wage is a poor tool for raising the incomes of low-wage households, and that targeted alternatives like the Earned Income Tax Credit do the job with fewer side effects. But the New Jersey study is part of the reason that view should be held with a measure of humility. The evidence is contested, the effects of moderate increases appear smaller than early models predicted, and anyone who tells you the question has been settled, by Card and Krueger or by their critics, is overselling.

For the theoretical side of the argument, see what Milton Friedman actually argued. For why the workers who never get hired matter so much, see the first job ladder.

Sources & further reading