A “credibility revolution” has changed the economy since the 1990s. Before that, theory prevailed and empirical work was a bad second cousin. “Hardly anyone takes data analysis seriously,” said Edward Leamer of the University of California at Los Angeles in a 1983 article. Yet within a decade, new and innovative work had changed the career path so that the lion’s share of noteworthy new research is now empirical. To facilitate this transition, David Card from the University of California at Berkeley is sharing this year’s Nobel Prize in Economics, which was awarded on October 11, with Joshua Angrist from the Massachusetts Institute of Technology and Guido Imbens from Stanford University.
The messy real world can often defy attempts by economists to prove causality. Calculating how, for example, an increase in the minimum wage will affect employment is made difficult by the fact that some other influence (such as a chronically weak labor market) may have contributed to political and employment policy changes. In other areas, researchers establish causality by designing experiments that randomly assign subjects to different groups, only one of whom receives a particular treatment, so that the effect of the treatment can be clearly seen. More and more economists are also using randomized controlled trials – the 2019 Nobel Prize even rewarded such efforts. But many issues cannot be explored in this way for political, logistical, or ethical reasons.
This year’s award winners have overcome such hurdles with “nature experiments”, in which a peculiarity of the story has the same effect as a deliberate attempt. In a landmark paper published in 1994, Mr. Card and Alan Krueger examined the effects of a minimum wage increase in New Jersey by comparing the change in employment there with that in neighboring Pennsylvania, where the wage floor remained unchanged. Strikingly, although theory predicted that a rise in the minimum wage would be followed by a sharp fall in employment, such an effect did not appear to hold up in practice. The paper inspired further empirical work and brought new energy to thinking about labor markets. Krueger, who died in 2019, would likely have shared the award if he had lived.
The use of nature experiments spread quickly. Mr. Card analyzed another unique circumstance – Fidel Castro’s decision in 1980 to allow emigration from Cuba – to examine the impact of immigration on local labor markets. About half of the 125,000 Cubans who fled to America settled in Miami. Comparing the city’s experiences with those in four other locations, which were similar in many ways but saw no influx of migrants, Mr. Card found that neither the wages nor employment of local workers had suffered as a result of migration.
Mr. Angrist worked with Krueger to use a similar method to study the impact of education on labor market outcomes. Since students with a more academic disposition are likely to both spend more time in school and earn more work, what looks like a return to education may indeed reflect a natural talent. To determine causation, researchers took advantage of strange features of the American education system. Although the law usually allowed students to drop out of school at the age of 16, all students born in the same year started school on the same day, regardless of their birthday. Those born in January therefore received more schooling on average than those born in December – and, as the researchers found, also tended to earn more. Since the month of birth of a student can be assumed to be random, they concluded that the additional education caused the higher income.
The school education study found that an additional year of education increased later income by 9%. Such an effect appeared implausible to many economists. However, this reflects a difference in definition, concluded Mr. Angrist when working with Mr. Imbens. The two scientists found that the effects of a “treatment” in a natural experiment would not be the same for everyone. For example, if the drop-out age were raised from 16 to 17, some would have to get an additional school year; others who had always wanted to stay longer in school were not affected.
Together, the researchers developed statistical methods to make the conclusions from nature experiments more meaningful. Economists refer to the bizarre factor in nature experiments (such as the month of birth of a student) as an “instrument”. Messrs. Angrist and Imbens explained the assumptions that must apply in order for the use of an instrument to be valid: For example, it may only influence the examined result (in this case, earnings) through its effect on treatment (school years) and not through others Channels. With these assumptions, the researchers enabled a more differentiated analysis: The increase in earnings in the above case, for example, only applies to students who were born at the beginning of the year and have to stay in school longer than usual. In addition, the methodology described by the researchers improved the transparency of the economists’ results. The reader of a paper can judge for himself how well an instrument fulfills the necessary assumptions and devalue the result accordingly.
It’s only natural
The credibility revolution, like any great upheaval, had its excesses. Critics point to careless work and the digging of data in search of results that appear meaningful. Scholars are occasionally too eager to extrapolate knowledge from a particular natural experiment in ways that may not be warranted, given the uniqueness of the circumstances. Yet the innovations developed by this year’s award winners have undoubtedly changed the field for the better, illuminating questions that were once shrouded in obscurity, forcing economists to drive theory in directions that better describe real-world experiences – a reason to celebrate. ■