The 2019 Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel – colloquially known as the Nobel Prize in Economics – has gone to Abhijit Banerjee, Esther Duflo and Michael Kremer “for their experimental approach to alleviating global poverty.” Banerjee, Duflo, and Kremer are pioneers in the use of field experiments, or randomized-controlled trials (RCTs), to study economic phenomena. An RCT in economics is analogous to its counterpart in medicine. The economist wishes to study a particular rule or policy (an “intervention,” in the jargon) – say, giving teachers a financial reward to show up for classes. Rather than work out a theoretical model and then “test” it against historical data on incentives and attendance, as most contemporary economists would do, the economist conducts an experiment in real time. One group of teachers in an actual school is given the treatment, another the control (i.e., no change in incentives), and the differences between the groups are compared. The results are then generalized to suggest that this incentive is effective at reducing teacher absenteeism.
Reaction to the prize has been almost uniformly positive. Alex Tabarrok provides a useful summary of the field with links to the Laureates’ key papers. Other summaries and evaluations can be found in the usual places. Reactions on #EconTwitter have been glowing, with testimonies from Banerjee’s, Duflo’s, and Kremer’s colleagues, current and former students, and many admirers. Besides their purely academic work, which has had an enormous impact on the field of development economics, Banerjee, Duflo, and Kremer have also been active in advising, consulting, teaching, and otherwise influencing development policy.
While some commentators have been surprised by the timing (Duflo, at 46, is the youngest ever recipient of the award), there was no doubt that these RCT specialists – “randomistas,” to their critics – would eventually be recognized with a Nobel prize. Their work is an important part of the empirical turn I wrote about earlier this year. Mainstream economics is moving away from grand theory, big questions, general models of interdependence and coordination, and classic problems and toward small-scale, incremental, experimental, and largely atheoretical measurement exercises. Schumpeter and Samuelson, Friedman and Becker (and of course the Austrians) are out, Raj Chetty is in. Among younger economists, the most heated debates are not about Keynesian versus New Classical or Austrian business cycle theories, partial versus general equilibrium approaches, or even socialism versus capitalism, but Stata versus R . In previous centuries, economists needed to know logic, rhetoric, and history. Then it became math and statistics. Now the most important skills are how to code – and how to conduct a RCT.
Within the mainstream literature on economic growth, the randomistas have their critics, such as William Easterly or fellow Nobel laureate Angus Deaton . The abstract from Deaton’s 2018 article with Nancy Cartwright is worth quoting in full:
Randomized Controlled Trials (RCTs) are increasingly popular in the social sciences, not only in medicine. We argue that the lay public, and sometimes researchers, put too much trust in RCTs over other methods of investigation. Contrary to frequent claims in the applied literature, randomization does not equalize everything other than the treatment in the treatment and control groups, it does not automatically deliver a precise estimate of the average treatment effect (ATE), and it does not relieve us of the need to think about (observed or unobserved) covariates. Finding out whether an estimate was generated by chance is more difficult than commonly believed. At best, an RCT yields an unbiased estimate, but this property is of limited practical value. Even then, estimates apply only to the sample selected for the trial, often no more than a convenience sample, and justification is required to extend the results to other groups, including any population to which the trial sample belongs, or to any individual, including an individual in the trial. Demanding ‘external validity’ is unhelpful because it expects too much of an RCT while undervaluing its potential contribution. RCTs do indeed require minimal assumptions and can operate with little prior knowledge. This is an advantage when persuading distrustful audiences, but it is a disadvantage for cumulative scientific progress, where prior knowledge should be built upon, not discarded. RCTs can play a role in building scientific knowledge and useful predictions but they can only do so as part of a cumulative program, combining with other methods, including conceptual and theoretical development, to discover not ‘what works’, but ‘why things work’.
The last point is important: the economist’s task is not to demonstrate (empirically) that x is associated with y, but to provide a causal explanation of how x affects y. For Austrians, that causal explanation is impossible without a priori theorizing (though the application to specific cases requires historical understanding). While RCTs are usually described as a method for “causal inference,” they don’t deal with causality in the way that economists have traditionally understood it.
Along these lines, Carmen Dorobăț and I have both criticized Duflo’s idea that economists are best understood as “plumbers,” not philosophers. As I wrote two years ago:
[E]xperiments, and RCTs in particular, are not substitutes for economic theory and more conventional forms of applied economics, because they deal with very small problems. By “small” I don’t mean socially unimportant — Banerjee and Duflo became famous for their Poverty Action Lab, an attempt to alleviate poverty in the world’s least developed areas — but rather, problems that don’t involve much economics beyond something like, “incentives matter.” RCTs have been used to study how to get students to study harder for tests, how to write fundraising letters to get more money, how to get people to eat healthier food (maybe), and other social issues. It’s unclear that they can provide any insight into the core questions addressed by economic theory and policy, both Austrian and neoclassical. What is the basis of social cooperation? How does an economy grow? What causes business cycles? Should we adopt the gold standard? Does regulation protect private interests? There is nothing wrong with providing a little extra understanding, on the margins. But RCTs don’t easily handle the big questions.
To echo David Henderson’s remarks on the prize, we know a lot about what causes poverty – the lack of free markets. Many small-scale experiments on teacher pay or operating practices of health clinics or similar studies may be interesting, but they don’t give us much big-picture knowledge we didn’t already have.
More evidence that the economics has gotten small.
