The missing economist at this year’s Nobel Prize
On a flight in 1987, Alan Krueger sat next to a woman called Ginna Ashenfelter. Krueger was just finishing his Ph.D. in Economics from Harvard University and was still looking for a job. Ginna, as it turned out, was the wife of Orley Ashenfelter, the kind of man who could offer Krueger a job. Orley led a group of professors and students at Princeton University and, recognizing Krueger’s potential, invited him to join them.
One of the students didn’t think much of Krueger at first, but was soon persuaded by his charm and intellect. This was Joshua Angrist, one of the recipients of this year’s Nobel Prize in Economics. Angrist would end up getting his Ph.D. in 1989 and following the opposite path to Krueger’s, going from Princeton to Harvard.
However, the two were already close by then. In 1988, they had started working together to answer an important question: How much does schooling affect earnings? The first thing you could do to answer this is to look at data and check whether people with more years of schooling have higher wages. You would find, to no one's surprise, that, on average, they do. But this is not a very good answer, because it only shows you that there is a correlation between the two variables. It does not tell anything about causality, i.e., what is actually causing what, since there are a multitude of things that could explain this correlation even if education did not affect income at all.
For instance, it could be the case that people with more years of schooling usually come from wealthier families, which affects both education and future income. As measuring family income is pretty straightforward, economists could easily check whether this is the case. But some things are much harder to measure, such as one’s inner motivation and work ethic. If some people are more hard-working than others, they would tend to stay in school for longer and to earn more, which would also cause the data to show a correlation between earnings and years of schooling even if there was no causal path between them.
In the natural sciences, there is an easy way out of this mess: you conduct an experiment. But when you realize what that would mean in this case, it is easy to see why it couldn’t be done. You would have to get a bunch of kids and divide them randomly into groups, then give each group a different amount of years of schooling. And then you would have to follow the kids throughout their lives to measure how much these differences in education affect their earnings. For obvious ethical and practical reasons, this could never happen.
Luckily, Krueger and Angrist had a clever solution to deal with questions like this. In fact, a solution that they had thought of even before they posed this particular question. They realized that if they had something that affected a person’s years of schooling that was not related in any other way to their earnings in the future, they could isolate the effect of education on income. This is, if they had a variable that caused some people to study less and some people to study a little more, but that had nothing else to do with wages, it would essentially look like the mad experiment described earlier, even if no experiment was done.
And they had such a variable: the day people were born.
Date of birth is a reasonably random variable that has, at first glance, no reason to affect how much money one makes. However, Krueger and Angrist reasoned that given that everyone starts school at the same time, people born in January are older when they first start school. For instance, American students start school in the calendar year they attain age six. Dropping out of school, on the other hand, can be done as soon as a student reaches a certain age, which varies from sixteen to eighteen depending on the state. This means that a student born in January will be legally allowed to drop out of school almost a year earlier than a student born in December. This causes people that were born at the beginning of the year to have, on average, fewer years of schooling than people born at the end, a fact the data corroborated.
In sum, Krueger and Angrist had found a variable that caused some people to study more than others but that had nothing to do with earnings in some other way. The law of compulsory schooling and the natural randomness of date of birth had created the sort of experimental results they needed without anyone having to perform any experiments whatsoever. Combining this idea with some fancy statistical techniques, they were able to show that an extra year of schooling significantly impacts future earnings, even when you control for the effects of family income, personal traits, and other unobservable variables. The duo had found an elegant and powerful solution to finding causality in complicated data.
Krueger knew that you could use natural experiments — these kinds of situations in which a random variable accidentally ends up creating different conditions for different people — to study much more than just schooling. Led by his “expansive curiosity”, to use Angrist’s words, he started collaborating with other members of the Princeton bunch to study things like inequality and labor. The group would meet at the basement of one of Princeton’s libraries to discuss projects and research methods. Some, like Orley Ashenfelter, would spend only the evening there. Others, like Angrist, would leave to have dinner with their families. But there was one member in particular that never left the basement, and it was him that Krueger chose to partner with to produce what would become his most important paper. His name was David Card, and he was another one of the recipients of this year’s Nobel Prize.
Krueger and Card took advantage of a policy change to study a crucial question in economics: Does raising the minimum wage increase unemployment? The traditional answer is that it does, and it is easy to see why. As the minimum wage increases, it becomes costlier for companies to employ low-skilled workers, causing layoffs or hiring freezes to contain costs.
A natural experiment emerged in 1992 when the state of New Jersey raised its minimum wage from $4.25 to $5.05. As the neighboring state of Pennsylvania kept their minimum wage at $4.25, Krueger and Card could compare what happened in both states after the policy change. It was crucial that the states were neighbors so you could guarantee that they had similar conditions and that no other relevant changes had happened at the same time — that there wasn’t anything other than the minimum wage increase differentiating the two markets. Surprisingly, they did not find any significant changes in unemployment in New Jersey compared to Pennsylvania. These results considerably shook economists’ understanding of the minimum wage, and are controversial to this day.
In 1993, Krueger and Angrist ran into some trouble when one of Krueger’s former classmates from Harvard, John Bound, together with David Jaeger and Regina Baker, found some issues in the fancy statistics they had used in the schooling and earnings’ paper. By then living in Israel, Angrist felt defeated. Krueger didn’t. He started coming up with some clever ways to show how their findings still held. This led to a whole new research project on the statistical methods used to analyze natural experiments. To help them in this endeavor, Angrist resorted to a friend he had made during his time at Harvard. Together, the three researchers came up with some creative tools to study natural experiments and find causality in messy data. The friend was Guido Imbens, the final recipient of this year’s Nobel Prize.
Krueger continued exploring and trying to find new methods to answer important questions. Collaborating with Nobel laureate James Heckman, he studied education and inequality. With also Nobel laureate Daniel Kahneman, he explored the economics and the psychology of well-being. His last collaboration with Angrist was in 2001, in a paper about natural experiments. With David Card was in 2005, in a paper on affirmative action. From 2009 to 2013, Krueger worked under President Barack Obama, who described him as having “a perpetual smile and a gentle spirit — even when he was correcting you.” More recently, Krueger expanded his research scope even further, looking into things like the effect of Uber on the economy and of opioid addiction on labor markets. In 2019, he finished writing a book called Rockonomics, on the music industry and what it teaches us about economics.
A few months before the book was published, on March 16, 2019, Alan Krueger took his own life. When the Nobel committee announced Joshua Angrist, David Card, and Guido Imbens as this year’s laureates in Economics, there was one economist missing.
Thanks to Arthur Alberti, Helena Werneck, François Boris, João Nobre, and William Radaic for reading drafts of this.