That Esther Duflo and Abhijit Banerjee would receive the Sveriges Riksbank Prize for the Economic Sciences in memory of Alfred Nobel in their lifetime was never in doubt. That Michael Kremer would join them to receive this honour is unsurprising. That it should have come so soon is a matter of considerable surprise, since it is rare for winners to receive this award with so much of their academic career ahead of them.
The award feels special for several reasons. Few winners in the past have been awarded for work that seems so accessible to so wide an audience, for studying problems that are so familiar and proximate to our lives, ranging from remedial teaching in schools, deworming school children, microfinance, immunization, to mention a few. That India is a site for a lot of their research, especially for Banerjee and Duflo, makes us all feel that our affable neighbour just won the Nobel, although it would seem that the three laureates who have been lauded for their humility and mentorship are bound to inspire such sentiments among many across the world. The outpouring of joy on social media befits their status as rock-stars of development economics; and drowns the disappointed voices of the many critics of their work.
What is so compelling about their work? The announcement explains helpfully that prize was awarded to them “for their experimental approach to alleviating global poverty”, for focussing on small and precise questions and answering them with a great deal of care and rigour. They urge us to eschew seeking solutions to poverty at a global scale and to focus instead on the lived experience of the poor, to test different solutions to specific problems to identify what works best. These questions might pertain to improving learning outcomes in schools, improving immunization rates, getting more children under bed nets to protect them from malaria and so on. At the meso-level, questions relate to how information can influence voting behaviour, on tackling corruption, studying diffusion of technologies within networks, the relationship between microfinance and poverty, etc. This approach, according to them, is much more likely to offer concrete solutions to the problem of poverty, in ways that can feed directly in the policy-making process and shape interventions. They are of the conviction that global poverty can be tackled one small step at a time. In these, they offer a counterpoint to those who believe in tackling the deeper structural, social and institutional conditions that keep poor people poor. Banerjee, Duflo and Kremer have less to say and do about these big-picture issues.
The methodological approach they advocate is experimental, via what is called Randomized Control Trials (RCTs). Inspired by biomedical studies, this is accomplished, in its simplest form, by randomly selecting units of observation (schools, classes, pupils, mothers, children, etc.) to receive treatment, while treatment is withheld from those not selected. Before treatment, the two groups are, or assumed to be, comparable in terms of the intended outcome of interest and other observable attributes. The difference in outcomes after the intervention can therefore be credibly identified and attributed to the intervention.
This approach ensures that the relationship between an intervention and the outcomes is causal and not mere association. Banerjee, Duflo and Kremer are masters at what they do – they conduct well-designed RCTs with a strong theoretical anchor, implement these carefully and their analysis of these interventions has a rigour that can be awe-inspiring.
This approach of combining experimental methods to answering small well-defined questions that are policy actionable have found natural partners among policy makers within governments and civil society organizations worldwide. While most new ideas take time to diffuse, the approach they proposed has caught the imagination of a large number of young scholars in their universities and beyond very quickly, including in the countries they research. In this, their apparently stellar qualities as mentors, teachers and institution builders have helped a great deal. Much of their work, for example, is enabled by the organization Banerjee, Duflo and their colleague Senthil Mullainathan founded in 2003. Called the Abdul Latif Jameel Poverty Action Lab (J-PAL), which cringe-worthily implies that the poor are objects of experimental study in a laboratory, even if run by well-meaning researchers, JPAL has become a large and vibrant network implementing a large number of interventions as part of their wide-ranging research in developed and developing countries. Their presence in India, for example, is strong, and employs a large number of young aspiring researchers, who hone their skills and beef up their CVs, while looking forward a recommendation letter or two to doctoral programmes.
An idea such as theirs is bound to have critics (here, here, here, for example). Many of the critics express concern that the growing influence of RCTs could crowd out big and important questions that might not be amenable to the experimental approach, but might have catastrophic consequences for human life. Climate change and conflict are two examples. A second concern is that the inferences from RCTs can vary so much across contexts that transferring learning from one context to another can be difficult at best and dangerous at worst. Critics mostly agree that RCTs have a useful role, but assert that this role should be circumscribed and not overstated, if we are to ensure that economists collectively work towards eradicating poverty and deprivation.
While the body of works by Banerjee, Duflo and Kremer represents outstanding intellectual achievement, I am less likely to agree with assessments that the approach they advocate constitutes a radical rethinking. Nor do I ascribe to the widely-held view that they brought economics down from the ivory tower in a way that had never been done before, using field-based data to generate evidence for policy making. A rich tradition of field surveys have long preceded their work, for example, going back to the early debates on tenancy and credit markets in developing countries. The idea of evidence feeding into policy too is by no means new. For example, when the Integrated Child Development Services (ICDS) was established in 1975 in India, the contributions of the National Institute of Nutrition, in informing the design were crucial. Further, to suggest that they transformed our thinking about the poor is offensive to many empirical development economists, with nuanced understanding of the challenges, for whom doing economics was never about anything other than the lives of the poor. It is also unclear whether this experimental approach to poverty alleviation policy has germinated substantively innovative ideas. It appears that much of the RCT tradition involves testing ideas and innovations that are already being implemented or proposed, by those who tend to rely on experential learning. At best, economists have turned plumbers to fill in the gaps. There have been reports, for example, that the current research by Banerjee and Duflo led the Tamil Nadu government to employ an additional worker in the ICDS to focus exclusively on pre-school education and to introduce milk, is somewhat misleading. Both these have been part of demands of both activists and researchers for well over a decade. It seems that researchers have been far more successful in co-opting ideas that are `out’ there, restricting their contributions to measuring the consequences of the innovations well.
To me, the greatest challenge facing the `revolution’ is a failure that might be precipitated by its own success. Running experiments has become an aspiration for many young researchers. These experiments are not cheap; many of them do not run the course; in JPAL’s work in India, so far only a fraction have yielded results that have credibly fed into (re)shaping policy, while offering an insight that could not have been obtained via experential learning by practitioners. It is not clear what the benefits of this approach are relative to the costs. With more and more young scholars in India choosing to tread the `revolutionary’ path, there are several risks that make for a dubious legacy: a risk that bigger, more pressing questions and other critical forms of knowledge would be marginalized; a risk that the choice of what to research be driven overwhelmingly by what can be randomized, and worse to randomize things that have in ways that are ethically questionable ethics – for example, randomizing wages, interventions that churn up caste identities, that incorporate bribing as part of the experiment, that undermine solidarity within groups, are divisive and so on. There are already several disturbing examples of experiments that likely would never get ethical clearance at their home universities if they were to be implemented in the countries where they are located. The fact that Banerjee, Duflo and Kremer have got the Nobel Prize so young hopefully allows them the time and freedom to shape their legacies so that the revolution they have triggered flourishes in the best possible way.
(Sudha Narayanan is Associate Professor, Indira Gandhi Institute of Development Research and a Ph.D. in Agricultural Economics, Cornell University.)
(The Federal seeks to present views and opinions from all sides of the spectrum. The information, ideas or opinions in the articles are of the author and do not necessarily reflect the views of The Federal.)