Today I want to talk to you about the Nobel Prize in economics and “the Swiss army knife of the natural experiment” – Opinion –

It is not easy to explain what causes the results we observe. For example, we know that private school children have better adult jobs, but to what extent is that a consequence of their education, and to what extent is both work and school a consequence of their parents being richer? In the real world everything is connected; It is difficult to observe a cause and its effects without crossing variables.

By Kiko Llaneras, for Diario El País

Unfortunately, our intuition ignores that. People jump to conclusions with amazing ease. If we are told that among children who use the mobile phone there are twice as many cases of depression, we run to think that the phone depresses them … although it is perfectly possible that it is the other way around (that the depression of some children pushes them to abuse the mobile ). We have known this human failure for centuries, that is why it is a logical fallacy with a name: “If it happens after this, then it happens because of this” (post hoc ergo propter hoc).

The pandemic has left us other examples, but my favorite is traditional advice that has aged poorly. Do you remember that until 2018 people told you to bundle up so as not to get congested? In the popular imagination, the cold caused infections. Now it has become clear that it should not be that, because to avoid covid nobody tells you to wear a sweater, but to avoid indoors, that you do not cough on top of people and wash your hands. The congested ones circulate more in our winter and the temperature has some effect here, but if we associated being cold and getting congested, it was partly because we confused correlation and chance.

Economists knew that correlation did not imply causation, but they did little about it. At least this is how Justin Wolfers, a professor at the University of Michigan, explains what the work of the three winners involved: “We looked at the data and said ‘correlation is not causality’, but we quickly forgot it and made a lot of pseudo-causal judgments based on data that did not really support those claims. But david [Card], Josh [Angrish]and Guido [Imbens]They said, ‘Wait.’ His response was not the usual destructive vibe of ‘we can’t make causal claims’, but something entirely constructive – here’s a toolbox that can help you make credible causal claims. “

It is not a small achievement. The awardees invented new methods to tackle the deepest problem in social science.

The Swiss Army Knife of the Natural Experiment

The essential idea of ​​the winners was the natural experiment. To understand it, it helps to take a step back and think about how all the causal questions would be resolved: the solution would be to have counterfactual universes. Do you want to find out the effect of vaccinating a person? It would be enough to generate two identical universes, one vaccinated and the other not. As it is impossible to do that, the scientists use a plan B, which is the control group experiments.

These are trials that divide thousands of volunteers into two groups at random, so that they were equal in everything, except in the intervention of receiving (or not) the vaccine. It’s a great trick because it avoids interference. With a control group, it doesn’t matter whether the virus mutates or the incidence low: since these changes will affect the two groups, we can always compare them and attribute the differences to the effect of the vaccines.

But what happens when there are no experiments, because they are unethical, expensive, or simply not done? That’s where Card, Angrist and Imbens come into play. They showed that it was possible to search for them in the wild. In the real world there are a lot of experiments, or near experiments, that we can exploit to answer questions.

One of the earliest examples was David Card’s 1990 work in which he used the massive influx of Cubans to Miami to study the effects of immigration on employment. In just a few months, unskilled workers had increased by 20% in the region. Did that cause a drop in wages? Card compared the evolution of salaries in Miami with those of other cities and found that they did not. It was just a study confined to one place and time, but it pushed economists to rethink their theoretical models.

Another classic was Angrist’s work on serving in the military. What effect does enlisting have on a person’s life? It is not something easy to observe without more: if we see that veterans earn little money, we cannot know if it is a consequence of the war or if it would have been the case in any case (because correlation does not imply causality!). But Angrist found a natural experiment: He compared young men who were drafted into the ranks during the Vietnam War with other young men who were left on the verge of being elected. What he saw is that, ten years after the end of the war, veterans had 15% lower incomes than their peers.

Searching for these shortcuts was a methodological revolution. In the words of economist Alex Tabarrok: “The last thirty years of empirical economics have been the result of economists opening their eyes to the natural experiments that were all around them, everywhere.”

The case of the minimum wage

Of the first natural experiments, the most talked about these days is the one published by Card and Alan Krueger in 1993, on a still current question: Does raising the minimum wage destroy jobs?

To find out, the two economists took advantage of a 1992 law that raised it in New Jersey from $ 4.25 to $ 5.05. What they did was collect data from 400 fast food restaurants in that state and in neighboring Pennsylvania, where the minimum wage remained constant, reasoning that any other factor affecting one region would also influence the other. Did New Jersey employment grow less than Pennsylvania when the law passed? They found that no, contradicting many economists.

It was no accident that this result was controversial. It was normal for a new method to yield results in striking contrast to the beliefs of great economists in the older, more theoretical tradition. That showed the virtue of the new approach. Inventing credible techniques for answering questions with data should take the discussion to the ground of fact. If employment did not evolve worse in New Jersey than in Pennsylvania, that is a fact that theorists should incorporate.

After 30 years of study, there are empirical economists who still think that the minimum wage destroys jobs, and others – perhaps the majority – who think that its effects are minimal. But the important thing is to look at what one and the other have agreed: they agree that it is an empirical question that is answered by looking at reality.

That transformation is the story of this Nobel.

Card, Krueger, Angrist, and Imbens made economics a “more scientific” discipline, as Noah Smith puts it. “If even the most consensual, basic, and beloved theory of the discipline could be refuted by empirical data, that means that economics consists of a set of falsifiable claims about the world in which we live.” In a field dominated by almost philosophical approaches to big questions, the winners decided to investigate reality little by little, asking small questions, but obtaining firm answers. They embraced the curious gaze of scientific routine.

It was a revolution that was felt beyond economics. Since the 1990s, natural experiments have changed political science, sociology, demography, and public health. The winners, and the generation of scientists that followed, have pushed those disciplines to be less theoretical and more empirical. Social science is no longer a science without experiments. Denying it, today, is making excuses.

We would love to thank the author of this write-up for this remarkable web content

Today I want to talk to you about the Nobel Prize in economics and “the Swiss army knife of the natural experiment” – Opinion –

Hank Gilbert