The phenomena of Donald Trump and Brexit reflect a deep disillusionment with politics as usual throughout the developed world. How can responsible leaders get things back on track and make progress toward a more prosperous and less acrimonious world?
One piece of advice from two Nobel-winning economists: Stop treating the poor and left-behind like losers.
The economists, Massachusetts Institute of Technology professors Abhijit Banerjee and Esther Duflo, have spent their careers seeking ways to alleviate global poverty, specifically by conducting carefully designed field experiments in the developing world. This is the subject of their first book, “Poor Economics,” and the work that won them and collaborator Michael Kremer the 2019 Nobel prize.
In a new book, “Good Economics for Hard Times,” Banerjee and Duflo turn their attention to the big issues facing the developed world, including immigration, trade, growth, inequality and climate change. They marshal evidence from economic research, including their own, to separate myth from reality and identify promising policies – such as showing more compassion, and sharply increasing assistance, for people who lose their jobs to automation or import competition.
I spoke with Banerjee and Duflo about their book, their recommendations and the role that economists and their methods can play in crafting policy. Here’s an edited version of our conversation.
Mark Whitehouse: You say you felt the need to write a second book because the public conversation about core economic issues – immigration, trade, growth, inequality and the environment – had gone so awry. Do you think we’re in a teachable moment? By bringing fundamental issues to the fore so starkly, could the era of Brexit and Donald Trump create a unique opportunity to explain and address them?
Abhijit Banerjee: That’s a great way to look at it. We feel very much that many people hate being placed in a position where it’s always us versus them. That’s our target audience: people who feel they want a reasoned resolution to the conflicts of today but do not see how to get there.
Esther Duflo: The only danger is that people are so obsessed with the day-to-day politics, and so angry with each other, that they are not ready to pause and think about how we got where we are now, and how to solve these core issues. We wrote this book because we really wanted to hold on to the hope that we could still have this conversation.
Also read: Off The Cuff with Abhijit Banerjee
MW: You describe how the policies of recent decades have aggravated inequality, and you advocate finding better ways to help the poor, the disadvantaged and the left-behind. You also offer a masterful overview of the relevant economic research. But amid all the potential solutions you survey, I struggled to discern priorities. If you had the power right now, what would you do first?
AB: We do not suggest a silver bullet in part because there is probably no single one. In fact, the fact that we should look at many silver pellets and not one silver bullet is one of our recurrent themes. But we are quite insistent what must be done: 1) change the conversation about welfare, acknowledging that it is mostly not a favor that we do for “losers” but a just recompense for bad luck; 2) restore faith in government, as a critical agent for building a more humane society; 3) raise tax revenues to finance a meaningful redistribution.
ED: Indeed! There are many concrete policy proposals in the book that could be done at scale, such as a G.I. bill for those displaced by trade or automation (which would include extended unemployment insurance and a generous tuition allowance for education). What we might do first is raise top tax rates on income, which is popular and technically feasible to do relatively swiftly, then use the extra resources to fund an ambitious rollout of this program and build support.
MW: The predicament of the poor often stems not from market forces or misplaced priorities, but from systematic bias and even outright exploitation. In the U.S., for example, a deep racial divide confounds efforts to reduce inequality. What, if anything, does economics say about how to address such entrenched injustice?
AB: Economics says a lot, for example, about the circumstances in which affirmative action works or can backfire. In our book, we talk about the much-abused term “merit.” It’s a fallacy that we can allocate jobs or university slots based on merit rather than ethnicity — as if we have objective measures of merit that are independent of how people have been treated historically.
MW: You suggest that it’s surprisingly easy to get people to set aside certain prejudices, noting how Abhijit and colleagues used street theater and puppets to get Indian voters to focus on development issues rather than caste. Suppose you were doing an experiment on how to counter the ethnic and racial tensions that Trump has amplified in the U.S. How would you design it?
AB: We haven’t done any of this kind of work in the U.S. yet, so what I’ll say is pure speculation. Maybe we could start by organizing some debates where people have to defend positions that are anathema to them, and see if they become more open to alternative views? There are groups in the U.S. and the U.K. that are trying to organize such conversations.
ED: The Democrats ran a successful experiment in the 2018 election by focusing clearly on economic issues that had a lot of common support. And even in predominantly Republican states, people voted for the expansion of Obamacare, which represented a sharp departure from the increased polarization we have seen in general. People revert to their basic ethnic and racial prejudices in part because they have lost hope that they can express their voice on anything else. The first step toward gaining the attention (and votes) of people who feel abandoned by traditional politicians is to offer them an economic package that they find credible and attractive.
MW: Economics differs from a hard science such as physics in a crucial way: It involves the study of human behavior, which is subject to change across time and populations. The effect of policies can disappear as people learn and adapt. If a program helps address violence in Chicago in 2013, it might not do so in other places and over time. Given this, how can you know that your understanding is deepening, rather than merely changing?
AB: We feel that the world is more stable than you imply. Our experience from working in the developing world is that for a wide range of interventions, we get similar responses in very different places at different points of time (admittedly the time variation is not that important, given that our work is mostly new). That does not mean that everything works the same everywhere, but there is a fair amount of stability. And if programs don’t have the same effect everywhere, what is the alternative? Sit on our hands? It seems better to do a number of experiments, see what can be replicated and what can’t, and try to understand the common principle that explains different behavior across contexts. This can help guide the next round of policy design.
MW: Finally, a question for the wonks out there. Your primary tool of inquiry, the randomized controlled trial, has drawn some criticisms. One is that when researchers choose a trial sample (say, villages near the capital city), they’re typically not selecting randomly, so the results don’t necessarily apply to the entire population of interest (all the country’s villages). Another is that the average effect doesn’t tell us how the program will affect any given individual — the average can be positive even if the effect on most people is negative. A third is that RCTs don’t offer insight into why A causes B: It could be due to contextual factors unique to the group being studied – say, trust in local government (or lack thereof). Do you consider these issues important, and if so how do you address them in your work? What are the implications for designing policy using the results of RCTs?
AB: We have written extensively about these issues. We definitely take them seriously, but I think in each case the claim is a bit unfair. First, this idea that we work in three villages near the capital is mostly made up. Many of our experiments span millions of people in thousands of villages. Most of Michael Kremer’s experiments in Kenya are located around Busia, which is hardly next to Nairobi — in fact, it’s pretty much as far as you can go before ending up in Lake Victoria, and it’s by no means one of Kenya’s major cities or tourist destinations.
On the second question of average treatment effects, I have two points. First, we do try to estimate quantile treatment effects, which show how the entire distribution of outcomes has shifted, and this is increasingly how papers report their results. Second and more damningly, show me a method that you can apply at scale, that gives reliable answers and that doesn’t have the problem that you can’t say what happens to individuals. It’s the nature of empirical research, not just RCTs.
On your third point, again, two responses. First, in many cases where people claim to have an “understanding” of causal channels, it’s a product of a model that they assume represents the truth. That’s often as easy to do with RCT data as with anything else, as long as you are willing to swallow the assumptions. There’s more and more of this work, but I am not always sure I believe the stories we end up telling. Second, many experiments are designed to pick up causal mechanisms, more and more as time goes on.