The three new Laureates deserve the prize in economics in memory of Alfred Nobel. But the award shows how poor the modern economics discipline is in terms of gender equality, research methods, self-examination, and genuine insight into the lives of the poor.
Congratulations are due to the recipients of ‘The Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel’, and special congratulations to Esther Duflo who is the second ever woman (and first female economist) to receive the prize. Recognition of their achievements is much deserved – not only for their contribution to a major shift in economics methods, but also in how economists think about and aim to shape the world. However, we are concerned with the direction of this shift. The ‘randomista’ project is not just a methodological one, it also pursues a normative agenda of paternalism and behavioural engineering. Moreover, we have concerns about the scientific rigour in some Randomised Controlled Trial (RCTs).
RCTs have profoundly transformed economics and policymaking
The press release accompanying the prize highlights how the three prize winners have helped transform development economics by introducing and popularising experimental methods. Michael Kremer, Abhijit Banerjee and Esther Duflo borrowed RCT methodology from the medical world and applied it to assessing social policy interventions. They are part of a wider group of randomistas largely based at two organisations: Abdul Latif Jameel Poverty Action Lab (J-PAL) and Innovations for Poverty Action (IPA).
Banerjee and Duflo’s most notable publication so far has been Poor Economics, an unself-critical 2011 book that claims to reveal the realities of poor people’s lives through social experiments. The authors suggest that experimentally-informed small changes, covering a wide range of topics, rather than blue-sky thinking, will serve to steadily improve the plight of the poor.
Their RCT methodology now dominates development economics. J-PAL alone has led 978 RCTs since its founding in 2003. Martin Ravallion and others, however, question whether such experiments can generate appropriate evidence, particularly because the methods seem to determine the questions. The ‘search for something to randomize’ has led to a methodological fundamentalism in some social policy debates, and randomistas often present experiments (ideally, their own) as the only truly credible measure of evidence. Yet RCTs are very expensive (usually around half a million dollars), only enable slow learning, and often deliver inconclusive results – as, among others, Robert Chambers has noted.
The behavioural baggage of RCTs
RCTs have delivered intriguing insights into how poor people think and act, but also into how behavioural economists do. For example, when a slew of high-profile RCTs failed to deliver the evidence that researchers expected on the ‘miracle of microfinance’, the researchers paid little heed to the implications of their insignificant and sometimes even negative findings. Instead, they focused attention onto some small (but statistically) significant behavioural changes in their data. These included microfinance services encouraging slightly higher propensities to engage in entrepreneurship and reduced purchasing of ‘temptation goods’ (a category in which Banerjee and Duflo included, for Indian slum-dwellers, tea and food on the street).
The problem is that these insights, far from shifting economic paradigms in a progressive way, and enabling greater realism and pluralism in economic thinking, have led to thinly-veiled efforts at behaviourally re-engineering the poor, which have gained traction in global development. The new behavioural paradigm, canonised in the World Bank’s 2015 World Development Report Mind, Society and Behavior, invokes targeted social norm-shifting, subliminal marketing through entertainment, ‘choice architecture’ and ‘nudge’, social pressures, and punitive conditionalities, to change poor people’s behaviours. The idea is to ‘help’ poor people overcome supposedly irrational ‘risk aversion’ in order to be more entrepreneurial, or more ‘time-consistent’ and save for a rainy day.
But these behavioural interventions can be too small and overly simplistic, disempowering and paternalistic, and stray into victim-blaming. The behaviourist paradigm interprets low incomes and precarious lives as a function of individual misbehaviours and cognitive biases, rather than a product of larger structural injustices of the economic and political system. Yet poverty traps are not just a function of individual cognitive insufficiencies: rich people make mistakes too – more often with other people’s money – but for them the consequences are far smaller. As fifteen leading development economists said in an open letter last year, RCTs and behaviourist approaches are practically designed to miss the bigger picture.
Questions about the soundness of the methods
RCT methodologies and the science of replications are hotly debated. One such example is Kremer’s work on deworming for Kenyan school children which has been the subject of extensive controversy.
Some of us have sought to engage with the prize-winning randomistas and their work to help make sure their RCTs remain grounded in reality, but the experience has been disheartening. In 2010, a research team was hired (by the same donor) to work together on a flagship J-PAL-led RCT on the impact of microfinance. The research team provided additional qualitative data and extensive contextual knowledge to inform the design of the RCT. However, despite their best efforts to collaborate, their inputs were ignored and the J-PAL team implemented a generic RCT. When the research team subsequently conducted a replication of the RCT, they discovered numerous flaws in the collection and analysis of the data – enough to throw the validity of any findings into doubt. Their replication study was countered with a rebuttal from Duflo and the J-PAL team which the replicators responded to. This example raises questions about how an RCT so allegedly flawed could pass through peer review and be treated by the wider academic community as reliable.
Behavioural versus progressive economics
Earlier prizes have recognized broader contributions to poverty reduction and human development, notably Sir Angus Deaton (2015), who is also a critic of RCTs, and Professor Amartya Sen (1996); both of whom set poverty reduction in the context of inequality. But recent years have already heralded the behavioural turn, with Richard Thaler, exponent of ‘nudge’ theory and avowed ‘libertarian paternalist’, winning the prize in 2017.
We argue the most recent economics prizes reveal the poverty of the modern mainstream economics discipline. It remains poor not only in terms of gender equality. 81 men and two women have won the prize. It remains poor in terms of methodology and epistemology, having only one of each, because quantitative analysis always trumps qualitative or mixed methods, and debates outside the narrow orthodox frame are relegated to ‘heterodox’ economics. And it remains poor in terms of compassion and insight.
A progressive economics for the 21st century would ditch the behaviourist way of thinking and methodology, and open itself up to the political, social and cultural questions about what causes poverty and inequality. Real discoveries await the ‘poor’ economists who reject the narrow frame.
Philip Mader is a Research Fellow at IDS; Maren Duvendack is a Senior Lecturer at the University of East Anglia; Richard Jolly is an Honorary Professor and Research Associate at IDS; and Solene Morvant-Roux is Assistant Professor at the University of Geneva.
*The prize is not technically a Nobel Prize. It was created by the Swedish central bank, and has been disowned by Alfred Nobel’s descendants as ‘a PR coup by economists to improve their reputation’. After criticism, the prize in 1994 was redefined as for the social sciences but mostly it has still been awarded to economists.