The New Jersey Fast Food Study That Rewrote the Minimum Wage Debate

In the late winter of 1992, telephones started ringing at Burger King, KFC, Wendy's and Roy Rogers restaurants across New Jersey and the eastern edge of Pennsylvania. The callers worked for two young economists at Princeton, David Card and Alan Krueger, and they asked the managers a set of plain questions. How many full-time workers do you have? How many part-time? What is the starting wage? How much is a meal? In November and December, after New Jersey had raised its minimum wage, the phones rang again at the same restaurants with the same questions. Nobody on the other end could have guessed that their answers about headcount at the fry station would end up in the American Economic Review, in a Nobel Prize citation, and in nearly every argument about wages for the next thirty years.

In short: When New Jersey raised its minimum wage to $5.05 in April 1992, Card and Krueger compared fast-food restaurants there with similar ones across the river in Pennsylvania, where the wage stayed put. They found no drop in employment in New Jersey relative to Pennsylvania. The result defied the standard textbook prediction, started a fierce argument, and helped launch a way of doing economics, using real-world comparisons, that won Card a Nobel Prize in 2021.

The prediction everyone learned

To see why a phone survey of burger joints caused so much trouble, you have to remember what students were taught. In the simplest model of a labour market, wages are set by supply and demand. If the government forces the price of labour above the market rate, employers buy less of it. A minimum wage, on this view, must cost some jobs, especially for the least experienced workers, who are the ones it is meant to help. The only question was how many.

By the 1980s, many economists treated this as close to settled. Studies using national data over time tended to find modest job losses for teenagers when the federal minimum rose. Surveys of economists found large majorities agreeing that minimum wages increase unemployment among the young and unskilled. Some prominent economists wrote that anyone who claimed otherwise did not understand basic economics.

The problem with the national studies was that they were trying to untangle the effect of the minimum wage from everything else happening in the economy at the same time: recessions, booms, changes in schooling, shifts in industry. When the federal minimum went up, it went up everywhere at once, and there was no clean way to know what would have happened without it.

A border as a laboratory

Card and Krueger saw an opening in a simple fact of geography. On April 1, 1992, New Jersey's minimum wage rose from $4.25 to $5.05, then the highest state minimum in the country. Pennsylvania's stayed at the federal $4.25. The two states share a long border along the Delaware River, and the restaurants on either side of it face similar customers, similar weather, similar regional economic conditions and the same national chains with the same menus and the same ways of doing things.

That made eastern Pennsylvania a natural comparison group. Whatever happened to fast-food employment in Pennsylvania between early 1992 and late 1992 was a reasonable guess for what would have happened in New Jersey without the raise. The difference between what actually happened in New Jersey and what happened in Pennsylvania could then be read as the effect of the wage increase.

Economists call this approach a natural experiment, and the specific comparison a difference in differences. Neither idea was entirely new. Card had used something similar a couple of years earlier to study what happened to Miami's labour market after the Mariel boatlift brought tens of thousands of Cuban immigrants to the city in 1980. But the New Jersey study made the method famous, partly because the result was so unexpected and partly because the question was so politically charged.

What the phone calls found

The team surveyed about 410 restaurants in two waves, one in February and March 1992, before the increase, and one in November and December, after it. They made a careful effort to reach the same restaurants both times, including going in person to some that did not answer the second time.

Starting wages in New Jersey rose as expected; many restaurants that had paid near $4.25 moved to $5.05. In Pennsylvania they did not. The surprise was employment. Measured in full-time-equivalent workers, employment at New Jersey's fast-food restaurants did not fall relative to Pennsylvania's. If anything, it rose slightly. Restaurants in New Jersey did raise meal prices a little, which is one way businesses absorb higher costs.

Card and Krueger also compared New Jersey restaurants that had been paying well above the old minimum with those that had been paying right at it, which gave them a second comparison inside the state. The restaurants most affected by the increase did not cut jobs more than the ones barely affected.

The paper was published in the American Economic Review in September 1994. A year later the authors expanded it, along with their studies of other increases, into a book called Myth and Measurement.

The counterattack

The response was intense, and some of it was personal. Prominent economists wrote that the result had to be wrong because it violated well-established theory. Business groups funded efforts to discredit it. The minimum wage was a live political issue; Congress raised the federal minimum in 1996, and supporters cited the New Jersey study as evidence that a raise need not cost jobs.

The most substantive challenge came from economists David Neumark and William Wascher. They argued that telephone surveys of managers were too noisy to measure employment changes precisely, and they analysed payroll records from a set of New Jersey and Pennsylvania restaurants instead. Those records had been gathered with help from the Employment Policies Institute, a group backed by the restaurant industry, a fact critics pointed to and the authors addressed by checking the data in several ways. Their analysis found that employment in New Jersey had fallen relative to Pennsylvania, contradicting the original result.

Card and Krueger responded by turning to a source that nobody could accuse of bias: the Bureau of Labor Statistics' own records of employment at fast-food restaurants, drawn from unemployment insurance filings. Those data, covering many more restaurants, showed little or no relative decline in New Jersey. The exchange was published together in the American Economic Review in December 2000.

Here it is worth being fair to the critics. Neumark and Wascher raised legitimate concerns about measurement, and the original survey did have noise in it. Their broader research has continued to find job losses from minimum wages in some settings, particularly for teenagers. Reasonable economists still weigh the evidence differently. What the New Jersey fight did was not end the debate but shift the burden of proof. After 1994, it was no longer enough to say that theory predicted job losses. You had to show them.

Why it changed economics

The biggest legacy of the study may not be about the minimum wage at all. It is about method.

In the decades that followed, economists increasingly looked for situations where some quirk of geography, timing, rules or luck created something close to a controlled experiment: a policy that applied on one side of a line and not the other, a lottery that assigned some people to a programme and not others, an arbitrary birthday cutoff that decided who started school a year earlier. By comparing groups that differed mainly in exposure to a policy, researchers could estimate cause and effect with far more confidence than by fitting models to national time series. Some economists began calling this shift a credibility revolution.

In October 2021, the Royal Swedish Academy of Sciences awarded half of that year's economics prize to David Card for his empirical contributions to labour economics, and the other half jointly to Joshua Angrist and Guido Imbens for their work on the methods behind natural experiments. The prize citation and the committee's explanation discussed the New Jersey study directly.

Alan Krueger was not there to share it. He had gone on to serve as chief economist at the Labor Department under Bill Clinton and as chairman of the Council of Economic Advisers under Barack Obama, and he died in March 2019. Card spoke of him when the prize was announced, and many economists noted that the work had been a partnership.

What the theory missed

If the textbook model predicted job losses and careful studies kept finding small or no effects, something about the model had to be off. The most common explanation is monopsony, a term for a market in which employers have power to set wages rather than simply accept the market rate. That happens when workers cannot easily switch jobs, because there are few employers nearby, because changing jobs is costly, or because workers lack information about other options.

When an employer has that kind of power, it can pay workers less than the value they produce, and it may keep a position unfilled rather than raise pay for everyone. A minimum wage that pushes pay up can, in that case, actually make it easier to hire and keep workers, and employment need not fall. Fast-food turnover is famously high, and a higher wage may reduce the costs of constantly hiring and training new people.

Monopsony is not a fringe idea any more. Research on labour market concentration, on non-compete agreements for low-wage workers and on how rarely many people switch employers has made it a central topic in labour economics, and it has implications well beyond the minimum wage.

What it does and does not prove

It is worth being careful not to overclaim on the study's behalf, because overclaiming is how good evidence gets discredited.

The New Jersey study looked at one industry, in one region, after one increase of about 19 percent, in a particular year. It does not prove that any minimum wage, at any level, costs no jobs. Card and Krueger never said it did. What it showed, along with the many studies that followed it, is that moderate increases relative to local wages need not cause the job losses the simple model predicts, and that the right way to find out is to look carefully at what actually happens.

For progressives, that is both an encouragement and a discipline. The encouragement is that the evidence has been friendlier to minimum wage increases than the textbook once suggested. The discipline is that the same method that overturned a confident prediction can be turned on our own claims too, and we should welcome it.

From one border to every border

The New Jersey study was one comparison across one state line. Its most important descendants took the same idea and multiplied it. In 2010, Arindrajit Dube, T. William Lester and Michael Reich published a study that compared pairs of neighbouring counties on opposite sides of state borders with different minimum wages, over many years. Restaurants in adjacent counties share a local economy, so differences between them are a cleaner test than differences between distant states. They found that minimum wage increases raised restaurant pay and did not reduce restaurant employment in a detectable way.

The approach spread far beyond wages. In 2008, Oregon expanded its Medicaid programme by lottery because it could not afford to enrol everyone who applied, and researchers used the lottery to study what health insurance actually does for people, comparing winners with otherwise similar applicants who lost. Economists have used school admission lotteries to study charter schools, draft lotteries to study military service, and sharp cutoffs in eligibility rules to study everything from unemployment insurance to financial aid. Each of these studies owes something to the idea that a policy boundary can serve as a laboratory.

The method has its critics, and they make a fair point: a natural experiment tells you what happened in one place under one set of conditions, and it can be hard to know how far the answer travels. A minimum wage increase in a booming metro may behave differently from the same increase in a struggling rural county. Good researchers now spend much of their effort on exactly that question, testing whether results hold in different settings before treating them as general rules.

There was also a direct political consequence. When Congress debated raising the federal minimum in 1996, supporters used the New Jersey evidence to argue that a modest increase would not cost jobs. The increase passed with votes from both parties, and Bill Clinton signed it in August of that year.

What you can do

• Look at the numbers where you live. The state minimum wage list shows what every state requires now. New Jersey's floor, once the highest at $5.05, has climbed far above that, and some neighbouring states have followed.
• Use the evidence in local debates. When a city or state proposes a raise, the most persuasive testimony usually comes from people who can explain what careful studies found, including their limits. Find your state legislators and their contact details through the representatives lookup.
• Pay attention to employer power. The monopsony question links the minimum wage to non-compete clauses, mergers and union rights. The corporate power topic page gathers coverage of how concentrated employers shape pay.
• Follow the wage data over time. The inequality page charts real wages and the minimum wage beside the wealth shares of the top and bottom, which puts any single study in a longer frame.