Three U.S. Economists Win Nobel Prize in Economics for Contributions to Estimation of Causal Relationships through Natural Experiments

Faculty Fellow, RIETI

It has been announced that the Nobel Prize in Economic Sciences in 2021 will be awarded to three U.S. economists.

Half of the prize money will be given to David Card of the University of California, Berkeley, for his contributions to empirical analysis in the field of labor economics, while the remaining half will be equally shared by Joshua Angrist of the Massachusetts Institute of Technology and Guido Imbens of Stanford University for their contributions to an econometrics methodology closely related to labor economics.

What is common among the three economists' contributions is that they have established methodologies for estimating causal relationships between variables based on observational data at the levels of individual persons and companies and have applied those methodologies to various labor economics issues.

It is difficult to identify the presence of causal relationships based on the observational data usually used by economists. That is because the matter that is examined as the cause of a certain outcome may not be independently determined from the outcome. Natural scientists identify a causal relationship by artificially manipulating the cause in an experiment in order to observe its effects on the outcome, but economists cannot do that in many cases.

Labor economists who used to be affiliated with Princeton University, including Card, Angrist, Orley Ashenfelter, who was the mentor of Card and Angrist, and Alan Krueger, who was a co-author with those two, dedicated themselves to conducting research under a methodology called "natural experiments" as a method of identifying causal relationships based on observational data. This methodology estimates a causal relationship by using a situation in which, by some historical chance or institutional quirk, the cause under examination can be conveniently observed as if it was artificially manipulated.

For example, in the United States, children born between October and December tend to receive a longer period of compulsory education than children born between January and March because of institutional factors. Using that situation as a natural experiment, Angrist and Krueger estimated causal effects of education on earnings and reached the conclusion that an increase in the number of years of education leads to a rise in earnings (see the figure below).

People Born Late in the Year Have More Years of Education and Higher Earnings
(An increase in the number of years of education received has positive effects on earnings)
People Born Late in the Year Have More Years of Education and Higher Earnings (An increase in the number of years of education received has positive effects on earnings)
Source: Prepared by Johan Jarnestad of the Royal Swedish Academy of Sciences based on data used by Angrist and Krueger.

♦  ♦  ♦

Card's main contribution to the field is a paper published in 1994 estimating the effects of minimum wages on employment which was co-authored with Krueger.

While raising the minimum wage is supported as a poverty mitigation measure, there are concerns that doing so may reduce employment. Therefore, labor economists were strongly interested in the relationship between minimum wages and employment. However, the minimum wage is not determined independently of the employment situation and it tends to be raised when the economic conditions are good. Therefore, estimating a causal relationship between them is considered to be difficult.

To resolve this problem, Card and Krueger focused their attention on the fact that the minimum wage was raised in 1992 in New Jersey, which is located in the same economic region as eastern Pennsylvania, where minimum wages remained unchanged. A survey of changes in the employment situation at fast food stores, which employ many low-wage workers, showed that employment in New Jersey increased slightly more than in eastern Pennsylvania, despite expectations to the contrary.

Meanwhile, what has been recognized as Angrist's major contribution is his study in 1990 that estimated the effects of military service experience in the Vietnam War on post-military earnings.

When considering what the appropriate compensation for war veterans is, it is important to estimate the effects of military service experience on earnings, and whether or not people have such experience is not independent of their earning capacity. For example, people who are physically strong or who find it difficult to find a private-sector job may be more likely to apply for military service. To overcome this problem, Angrist used the draft lottery during the Vietnam War as a natural experiment and showed that military service experience has negative effects on post-military earnings.

Angrist was aware that the findings of his own study represented the causal effects observed among draftees and that the causal effects among volunteers were not included. While Angrist made his argument on the premise that the causal effects of military service experience on earnings are different between draftees and volunteers, it was difficult to handle such heterogeneity under the framework of econometrics at that time. The problem was resolved through a joint study by Imbens and Angrist by introducing the "counterfactual framework," which was originally developed by Donald Rubin, a statistician.

♦  ♦  ♦

The papers of these three economists who have won the Nobel Prize have had a significant impact on subsequent studies in two respects.

First, their papers affected empirical economics in general. The methodology that they advocated, which uses natural experiments to estimate causal relationships, has upended approaches to empirical research in all fields of economics. Research papers in the field of empirical economics that are written without employing these methods of estimating a causal relationship are no longer adopted by major academic journals.

Their papers also affected the development of an econometrics approach that focuses attention on the heterogeneity in causal effects. Before the study by Angrist and Imbens, the standard approach was to assume that causal effects are constant. However, their study served as a catalyst for the development of an econometrics framework premised on heterogeneous causal effects.

Card's study created policy controversy. His finding that raising minimum wages does not negatively affect employment invited criticism from many quarters, as it overturned the conventional wisdom among economists who argued that higher wages reduce employment.

In particular, comments made by David Neumark of the University of California, Irvine, and William Wascher of the Federal Reserve Board in 2000 and counterarguments to their comments made it clear that the results may vary if different datasets or estimation methods are used, indicating the importance of using high-quality datasets and appropriate estimation methods.

From a theoretical perspective, attention turned to the hypothesis that the reason why raising minimum wages does not reduce employment is the existence of the state of monopsony in which companies have the power to determine wages in the labor market. Efforts to empirically verify that hypothesis are ongoing.

The conventional wisdom among labor economists that a labor supply increase holds down wages was countered by a paper in 1990 that reported, based on a natural experiment using the massive influx of Cuban immigrants into Miami, Florida, that an inflow of immigrants into a region does not work to restrain wages there. This paper came under strong criticism, for example from George Borjas of Harvard University, who published a paper that directly argued against its findings, titled "Labor Demand Curve is Downward Sloping."

Both minimum wages and immigration are politically sensitive issues. Apart from engendering ideological conflict, the controversy created by Card's study has promoted the deepening of labor economics as a science in terms of theoretical development, data quality improvement, and refining of the empirical approach. That owes much to Card's calm attitude as a scientist.

The awarding of the Nobel prize to Card, Angrist and Imbens is testimony to the importance of access to high-quality data in the field of empirical economics. Angrist's 1990 paper concerning military service experience during the Vietnam War used administrative records of social security. Meanwhile, Card, together with other researchers, including Patrick Kline of the University of California, Berkeley, has in recent years analyzed the role of companies in the labor market using administrative records from Germany and Portugal.

Establishing the access to the administrative records and microdata of government statistics for academic research while protecting privacy not only helps to improve policymaking but is also essential to contributing to the advance of economics and the social sciences in general. Promoting the use of administrative records and micro data of government statistics for academic research lays the foundation for Japan to contribute to academic development internationally as one of the world's advanced countries.

>> Original text in Japanese

* Translated by RIETI.

October 20, 2021 Nihon Keizai Shimbun

December 2, 2021