Experiments in Economics

THIRTY YEARS AGO, a group of Princeton economists kicked off the day as they often did, gathering for coffee around a counter before classes and meetings began. The newspaper that day had an article about an impending hike in New Jersey’s minimum wage, and the economists’ talk turned to what that change would mean. Conventional economic theory predicted that an increase in the minimum wage would lead employers to cut jobs, but was that true? Soon a research idea had formed — one that would help carry the economics profession toward a new way of conducting empirical research, and eventually lead to a Nobel Prize.

Taking part in the discussion that day were David Card *83, then a professor at Princeton and now at Berkeley, and Alan Krueger, who had joined the Princeton faculty about five years before. A study of the minimum wage resonated personally with Card, whose parents were dairy farmers in Canada and who himself had held minimum-wage jobs as a youth. Krueger, meanwhile, subscribed to The New England Journal of Medicine; he sometimes noted how the journal laid out each paper’s research design and lamented that in economics, it was more difficult to run real-world experiments. But with the minimum-wage hike, the economists saw an opportunity. 

Card and Krueger evaluated the impact of the law by surveying 410 fast-food restaurants in New Jersey and eastern Pennsylvania, where the minimum wage did not change. They compared changes in employment. And they found “no indication that the rise in the minimum wage reduced employment.” 

The study, which stood out for its simplicity, was among Princeton’s most notable “natural experiments” in economics, and it had a huge impact on the profession. In October, three practitioners of natural experiments — Card, Joshua Angrist *89, and Guido Imbens — were named Nobel laureates in economics; the Nobel Committee said they showed how this approach can answer central questions in economics. Many observers believe Krueger, who died in 2019 at 58, would have shared the prize had he lived. 

“I’m not sure where the profession would be now without Card and Angrist, but the American Economics Association publishes an entire journal, AEJ: Applied Economics, that is filled almost exclusively with the sort of empirical work that they pioneered,” says Michael R. Ransom *83, an economist at Brigham Young University. Perhaps even more important, their work and work by other Princeton economists have attracted widespread attention outside economic journals and fueled policy changes at both the local and national levels.

Orley Ashenfelter *70

S. Bola Okoya/Primo Supremo Photography

The latest Nobel Prizes for economics were forged in an out-of-the-way nook in Firestone Library, an enclave that was home to a tight-knit group of Princeton economists known as the Industrial Relations Section. Collectively, members of the section have moved smoothly between the worlds of academia and government, providing guidance on what policies should be implemented to improve outcomes for the economy and for individuals. 

The IR Section “really transformed the way we think of empirical evidence,” says Lawrence Katz, a Harvard University economist who worked closely with both Card and Krueger. It made the study of economics “much more transparent and clear, which made it more policy-relevant and more rigorous,” Katz says.

The section was created in 1922 with financial support from John D. Rockefeller Jr. Labor unions had grown stronger during World War I, but when the war ended, the labor-management relationship grew tumultuous, and unions quickly lost ground. During the Great Depression, Princeton’s young IR Section saw itself as serving both company executives and union leaders (though companies used its work far more frequently than unions, PAW reported), disseminating reports on such subjects as “labor capitalism,” pensions, the five-day work week, and “age limitations” of workers. Later, under such figures as Albert Rees and future University president William G. Bowen *58, the section began to experiment with new approaches, most notably rigorous data techniques.

Card recalls that the hothouse atmosphere of the IR Section — in Firestone, physically separated from the rest of the economics department — enabled innovative approaches to flourish. Graduate students worked side by side with senior faculty — an unusual arrangement at the time, says Card. (Card even met his wife at the section’s home in Firestone: Cynthia Gessele *89 had a part-time job typing for the economists. Her secret to getting the job? “Not many people could type math notations,” he says.) 

David S. Lee *99

Denise Applewhite/Princeton University

The section’s physical setup was so important that when it moved to the Louis A. Simpson International Building in 2017, officials included a long counter with a coffee maker where faculty and students could bat around ideas. In a tour of the space, economics and public affairs professor David S. Lee *99 points out the setting, noting that it continues to host offhanded “counter talk” among the members.

“Almost any topic was fair game — thoughts about a question one of us was attempting to address, an issue in a paper, just ‘blue-skying it’ as we thought about the latest news, or what we were going to have for dinner,” says economist Cecilia Rouse, the former dean of Princeton’s School of Public and International Affairs, who is on leave from Princeton to chair the White House Council of Economic Advisers (she spoke to PAW in her personal capacity). “Those unstructured moments often led to new ideas, and the section had the resources to allow us to pursue many of them.” Sometimes so much time would pass in discussion that IR Section members would get stuck in the library when Firestone locked its doors for the night.

“It was an intimate setting,” recalls Lee. “The offices of faculty and students were right next to each other, with a common area, like a dorm. As a student, many of the things I learned came from just hanging out around the coffee machine and listening to David, Alan, and Orley talk.” 

“Orley” is Princeton professor Orley Ashenfelter *70. Now 79, Ashenfelter has been a presence in the IR Section off and on since the 1960s, helping to lead its transition from focusing on old-fashioned company-union mediation into cutting-edge econometrics. Ashenfelter was “one of the very first economists to try to make estimates of economic effects more credible and defensible” by moving beyond theory and using quantifiable data, says Gary Burtless, an economist with the Brookings Institution. 

James Heckman *71 recalls that the first time he met Ashenfelter — when Ashenfelter was a graduate student and Heckman was considering working toward a Princeton graduate degree himself — they spent the whole night arguing, ending only after the sun was up. 

Ashenfelter loves direct intellectual combat, says Heckman, a University of Chicago economist and a Nobel laureate. “I still think of it today,” he says. “A lot of my thinking came from those interactions.” 

The late 1960s and 1970s were an exciting time for labor economists, as federal officials tried to deal with rising inflation and other economic tribulations by implementing innovative policies. Economists were needed to test the real-life effects of these policies — but few academic departments were well-prepared to do that. As late as the 1970s, academic industrial-relations departments “were strongly opposed to measurement, even of the simplest kinds,” Heckman says. “The study of labor supply was the province of sociologists, they believed.” Theories prevailed over quantitative studies. The younger economists coming to Princeton rejected this approach as outdated, however, and, led by Ashenfelter, they became active in launching a quantitative economics that turned hard data into models. 

The gold standard for academic experiments is the “randomized controlled trial,” in which researchers randomly assign test subjects into one of two groups — one group that receives the “intervention” they’re trying to study, and a control group that does not. But in economics, a randomized controlled trial is not always feasible. “Natural experiments” were the next best thing, allowing researchers to mimic a random controlled trial using data that came from a fortunate accident of history.

Joshua Angrist *89

Nobel Prize Outreach/Risdon Photography

One day, when Angrist was a graduate student, his professor, Ashenfelter, mentioned the long-term health effects of being drafted into the military. Epidemiologists “had done this very clever thing where they used the fact that draft lottery numbers were randomly assigned,” Angrist recalled in an interview with the Federal Reserve Bank of Richmond in 2020, “and they compared people who had high and low numbers to test the causal effects.” Ashenfelter suggested that someone could use lottery data to study the earnings of men who had been drafted compared to those who had not. When the class ended, Angrist went to the library to get started on the project. It became his Princeton dissertation. 

Today, natural experiments are standard in economics, says Lee — in fact, generations of scholars take this for granted. The impact of the early Princeton researchers has been multiplied through their students and their students’ students. Card, for example, was one of Angrist’s dissertation advisers and also advised other prominent Princeton economists. Economics professor and Nobel laureate Angus Deaton recalls introducing Card to an audience of perhaps 1,000 economists and asking how many had been one of his students. “There was a forest of arms,” he says.

The studies by these economists deal with questions affecting the lives of ordinary Americans, like wages and education, and so frequently wind up in news reports and policy discussions, not just professional economics journals. For example, Card and Krueger’s conclusion that raising the minimum wage did not harm employment echoed what had been found in several previous papers, but the simplicity of its findings helped attract widespread attention among policymakers and the general public. (“The paper is four numbers — before and after employment in New Jersey, and before and after employment in Pennsylvania,” Ashenfelter notes.) As expected, that study attracted some criticism, says Janet Currie *88, who now chairs Princeton’s economics department, but “over time, more and more people accepted the finding.” And while the paper did not lead Congress to raise the federal minimum wage, it helped convince states and localities to increase theirs. 

Other natural experiments flowed from the IR Section. A 1991 paper by Angrist and Krueger explored how much of an economic return comes from spending additional time in school. To study this, Angrist and Krueger focused on the quarter of the year in which someone was born. People born in the first quarter would start school when they were older, so they would reach the minimum age to legally drop out sooner than slightly younger peers. In other words, through the accident of birth, students born earlier in the year would be able to leave school earlier, with less education. The researchers concluded that extra schooling did, in fact, lead to extra earnings later. Yale economist Joseph Altonji *81 says the work has been “a key part of the case for investment in education.”

Card and Krueger, meanwhile, looked at affirmative-action policies in higher education, using data from California and Texas both before and after those states eliminated preferences in college applications. Minority-student acceptance rates at selective state universities fell, they concluded, but application rates did not, suggesting that highly qualified minority students were not dissuaded from applying to elite public schools either because of reduced campus diversity or because of uncertainty about their admission prospects. 

Princeton’s IR researchers have long had ties to government. Heckman recalls that when he was a graduate student in the group, one of the professors mused about having advised President Herbert Hoover. The link grew tighter in the 1960s and 1970s. Scholars in the group undertook several efforts to study a proposal for a version of negative income tax — a guaranteed income — by President Richard Nixon. (Their findings were mixed.) In time, the researchers would explore many other public-policy questions, including the effectiveness of government-subsidized worker-training programs, whether expanding Medicaid reduced labor supply, and whether the 1980 Mariel boatlift, which brought Cuban migrants to Miami, affected wages in that city (the study found almost no effect on wages or unemployment rates). 

One seemingly humdrum yet crucial development was the emergence under presidents Lyndon Johnson and Nixon of new streams of federal statistical data. Data collected from programs that were part of Johnson’s War on Poverty greatly increased the granular material available to researchers who were interested in conducting natural experiments. And government officials welcomed academics to play a role. “There was an optimism that informed scientific research could actually help alleviate poverty or racial or gender wage differences,” Card recalls. “It was an optimistic time, where you felt you could potentially have an impact on things that were relevant for people’s lives.” Some of the IR economists have served in the White House, including Rouse, who worked at the National Economic Council during the Clinton administration, served on the Council of Economic Advisers for President Barack Obama, and has been Biden’s top economic adviser since the start of his presidency. 

Not all economists have the stomach for working inside the government, however. Winning a Nobel Prize is one thing. But policymaking? “To get things changed is a huge process of advocacy and bending wills,” says Card. “It’s not something I feel I have much ability to do.” 

Louis Jacobson ’92 is a senior correspondent for PolitiFact, where he has written extensively about economics.