Thoughts on Dr. Abhijit Banerjee receiving this year's Nobel Prize in Economics

As a Bengali and someone who now teaches Economics, I did receive some questions about Dr. Abhijit Banerjee, one of the three recipients of this year’s Nobel Prize[1] and his contributions to the field. I had the chance to share my thoughts with my high-school friends and below is a lightly edited version of my reflections on Dr. Banerjee.

It's understandable that when somebody with the last name Banerjee gets a Nobel prize, many Indians, and Bengalis in particular, are proud of it. I for one, don't fall in that camp. To me, racial, ethnic, or gender identity matter little when it comes to my views on a person and the ideas that they espouse. What matters is the extent to which my ideological views line up with the person concerned and the extent to which I believe the other person is sincere in their views, even when those views differ from my own. To make this point concrete, on any given day I would rather vote for a white transgender individual who shares my views regarding the free market rather than an Indian male who advocates for more government intervention in the economy. And therefore, it is safe to say that my initial reaction to a Bengali winning the Nobel Prize is a yawn and a sigh; in other words, one of genuine indifference.

Because some of you may nevertheless be interested in knowing my views on Dr. Banerjee, let me proceed further and note that whether I am enthusiastic about his winning the prize depends very much on what I make of his contributions to the field of Economics and my assessment of whether he has moved the sub-field of Development Economics in a helpful direction or not. Before I begin let me state the obvious: I don’t operate anywhere in the league in which Dr. Banerjee does and neither am I an expert in his field of Development Economics; my areas of research are public finance and political economy. However, I have read a number of his papers and have some familiarity with his work and that of his co-author, spouse, and co-recipient of this year’s Prize, Dr. Esther Duflo. And based on that passing familiarity with their body of work, my view is that the influence of Drs. Banerjee and Duflo has been to push Development Economics in a direction that isn’t very helpful.

The reason I say that is because the duo have spearheaded the use of experiments in the field of Development Economics. A typical example might be one in which a bank randomizes loan offers to farmers. Say, the first group of farmers gets an offer of a 5% interest rate, a second group gets an offer of a 10% interest rate, and a third group gets an interest rate offer where the rate is not fixed, but conditional on how the weather turns out to be. If the weather in that area turns out to be good and farmers earn a healthy level of profits, they can naturally afford a larger payment and perhaps the interest rate is set at 16% in that case. However if the weather is bad, farmers are not required to pay any interest and are instead only required to repay the loan. In the context of such a research design, the questions might be:

1. Among the three groups of farmers, which group of farmers sign up for loans most often?

2. Is default lower for the third group of farmers who have the state-contingent loan, i.e. the loan where the interest rate depends on the weather?

My problem with that strand of work is that it doesn't help us answer the big questions in Economics that the average public (or policymakers) care about. It focuses by design on questions that are small because experiments can be designed only for questions with relatively low stakes. This approach of conducting experiments is completely unhelpful for helping us figure out answers to questions like:

1. Did demonetization help or hurt the Indian economy?

2. Did the several rounds of quantitative easing by the US Federal Reserve help or hurt the US economy? Did it lead, for example, to higher levels of inequality as some of my research on the effects of an easy monetary policy seems to suggest (https://www.sciencedirect.com/science/article/pii/S0165176519301958)?

3. Many Democratic candidates running for president have proposed wealth taxes of various kinds. What would be the impact of such wealth taxes, if imposed, on investment and capital formation? Would it reduce wealth inequality as their proponents claim?

These are all kinds of questions that the average person would care about a lot and the fact of the matter is that the experimental approach pioneered by Banerjee and Duflo would have little to say on these kinds of questions. If that were the only drawback of the kind of scholarship advanced by Banerjee and Duflo, that would be bad, but that alone wouldn’t get me riled up to the point where I feel like writing this blog post. What does get me somewhat riled up though is that this group of scholars who have pioneered the use of experimental methods in Development Economics have also tried to get economists across the board to think that unless the evidence for a question of interest comes from an experiment, that evidence is less than credible. Now that IS unfortunate because ultimately for the big questions we care about as a society, we will never have the luxury of being able to conduct randomized control trials. And that is primarily why I am not a fan of Abhijit Banerjee or Esther Duflo and not too happy with this year’s selections by the Nobel Prize Committee.

Another side-note to why I haven’t been fans of Abhijit Banerjee or Esther Duflo and the Abdul Latif Jameel Poverty Action Lab (J-PAL) – the research lab set up to facilitate conducting their work, in particular, randomized control trials. In my view, the duo have contributed to making the field of Development Economics less democratic. It takes a lot of money (and I mean A LOT) to fly into a developing country and do experiments like the one I gave an example of. Few institutions outside of MIT and Harvard have the resources to run these kinds of experiments. But because the duo have held editorial responsibilities at some of the most influential journals in the field of Economics, for example with Esther Duflo having been the Editor at American Economic Journal: Applied Economics for 10 long years and now serving as an editor at the American Economic Review, in my view, they have downplayed work that  does not rely on randomized control trials. The net result has been that the field of Development Economics has been monopolized by them and affiliates of J-PAL and in the process, become much less democratic.

A final thought in response to the idea that governments could in principle rely on the kind of evidence generated by these randomized control trials in formulating policy. As it turns out, few developing countries have the skilled human capital or the political will to conduct experiments side-by-side researchers from MIT and then implement the findings that come out of these experiments. Much of the work conducted by J-PAL for example has been in Africa where a newly elected government often starts off anew and abandons all projects that were initiated by the outgoing government. The kind of continuity in policy that is necessary for these experiments to bear fruit is simply not there in many developing countries around the world, where political stability is far from given. But I think the biggest criticism I have of the strand of work pioneered by J-PAL is that it is simply quite unknown if their experiments would scale up well.

To use my previous example, perhaps the loans where farmers’ repayment is dependent on the weather is best for the farmers and leads to lowest default rates. And because the differences between the three groups are statistically significant, that study leads to a very nice publication for its authors – exactly the kind of scholarship which the Nobel Prize Committee would consider at as it makes its decisions of who to award the prize to. However, the fact that the study got published in a leading journal does not mean that we can scale up this experiment as-is or that the results from such a scaled up experiment would benefit society at large. For example, if in fact there is bad weather and all the banks who made these state-contingent loans get back less money than is necessary for them to operate profitably (remember that banks do not receive any interest in the case of bad weather), does that raise the likelihood that some of these banks (or financial institutions) will go belly up? If so, that would surely result in a myriad of problems for the local economy. If, on the other hand, the government steps in to offer subsidies to these banks (because they are now incurring losses), that “solution” comes with its own set of problems such as an increase in the government’s budget deficit. Regardless, my point simply is that the kinds of experiments pioneered by J-PAL are not only unhelpful for the big-picture kind of questions we as a society care about, they are harmful in that the evidence on the scalability of these experiments is pretty razor-thin. In the jargon of economics, through these studies, we only know the partial-equilibrium response to small questions; we don’t know the general-equilibrium response to those very questions, let alone the general-equilibrium response to questions that are grander in scale and scope that we as a profession have a fundamental responsibility to grapple with. So in the end, if you find me less than thrilled that a Bengali has been one of the three recipients of the Nobel Prize in Economics, hopefully I have carefully explained to you why that might be the case.

---

[1] The Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel for the finicky among you