Can we end poverty?
Esther Duflo, a development economist at MIT and director of the Abdul Latif Jameel Poverty Action Lab, explains how our understanding of the economic lives of the poor has grown more complex in recent years. While Duflo doesn't see any silver bullet that will end poverty, she points to progress, in part from the use of randomized control trials to solve specific problems.
Q: Could you explain how you understand poverty?
The key to understanding poverty is the acknowledgment that it's not one thing. It's not just lack of income or any other single factor; it's a complicated set of issues.
The problems that poor people face aren't going to be solved with one philosophical framework. Some problems are embedded in the political historical structure, some are due to geography, some to human psychology, some to lack of markets.
Unless we acknowledge that we are not going to understand everything at once, we are never going to understand poverty. That's my view, but, at least among scholars, this has also become a relatively accepted point of view.
Q: What does the randomized trial approach add to our understanding of poverty?
Once you understand that you need to break up this very large question of why poor people are poor into smaller, manageable questions, you can start answering the concrete questions in a concrete way. For example, why is the quality of the schools the poor attend so low and what can be done about it?
You could say, "One reason why poor children don't learn is because the classrooms are too heterogeneous. Teachers are not able to manage so many different levels, so they're teaching only to the best students." If that hypothesis is correct, dividing classes into levels of preparation should help everybody.
That's where randomized control trials come in. It's quite late in the chain of thinking. Randomized control trials are a tool which allow you to determine, with a lot of confidence, the impact of an intervention. The intervention could be a program run by a NGO or government; it could be a price in a market. It is characterized by the fact that you are changing the environment in some way.
To continue the example, from 200 schools, you randomly pick half of them to implement this segregating of classroom by achievement level while the other half continue with heterogeneous classes. You would wait for a year to see whether the outcome has improved, and then you compare the results. If you get a very clear answer to the question, that in turn would shed light on your original hypothesis of what the main problem of education quality is.
What we found doing this experiment in Kenya was that grouping students by preparedness or prior achievement has positive effects for all levels at a very low cost.
Q: It sounds like, in addition to testing the outcome of programs, randomized control trials push researchers to ask better questions.
Randomized evaluation changed the way we do development because it forced us to ask better questions. Once you have the method to produce good answers it's important to have the right questions. The questions become inspired by what kind of intervention helps us better understand how the poor live.
Q: Your book Poor Economics with Abhijit Vinayak Banerjee touches on the daily experience of the economics of being poor. Could you explain some of those everyday realities?
So, when you wake up in the morning, if you want a glass of water, you turn the tap. The water's clean. It has enough fluoride for your teeth and your children's teeth. You don't have to think about it.
You wake up in the morning in a poor country, and for the vast majority of people, not even just the poorest people, there's no water. You have to go get it somewhere. And it isn't clean, so you have to remember to do something about it: adding chlorine or boiling it or otherwise treating it. Once you've treated it you have to keep it in a safe place; you have to make sure that your children don't put the scoop on the ground then back in the water—otherwise the water will get re-contaminated. This one example gives you some insight into their whole life.
If you spend all your effort and energy and time in getting small things right, it gives you very little time to think about the important things, like what you want to do with your life. Clean water is something that we never have to think about in rich countries, but it is filling up the day, particularly for women, in developing countries.
Sitting on our couches in the U.S., we think, "Well, why aren't the poor getting their act together and doing things right?" But we forget that even taking easy steps is difficult when there are so many of them and you have to take them all correctly every day, every time.
Q: You have started to look at hope and psychology more broadly as a component of development economics. What is that about?
The impact of psychology and the role of hope is something that we are just trying to discover, but it is potentially very important. Something that I am becoming increasingly convinced of is that part of the problem for the poor is their perspective on what's possible. Sometimes the rational response may be to hold back and not to try. For example, if you're thinking that there's no chance that your girls are ever going to get a job, then there is no reason to try to educate them. Giving people opportunities, for themselves or for their children, so they can aspire to something different can start a virtuous circle.
In a lot of cases the impact of your perspective—your expectations—on your current behavior is rational. If a subsistence farmer hears about a new technique but knows that if it fails no one is going to help feed the family, the farmer doesn't really want to try. Your view of the future has an impact on what you decide today in a completely rational way.
There are also things that may be less rational. If your future is really bleak, then you don't want to think about it. You might shut off thinking about very important things—for example, that you might become sick or you might die. If it is too sad to contemplate, you may fail to protect yourself from these things.
Q: Economist Sendhil Mullainathan and psychologist Eldar Shafir describe priming participants in an experiment with a scenario, "Your car is going to need $100 worth of work" and then having them take an IQ test. Then they change the scenario to "Your car is going to need $1,000 worth of work." They found having the more difficult problem consuming the mind led to lower IQ test results.
Exactly. It's very hard to put ourselves in this situation. When we don't have these endless small challenges, we can be a bit baffled by some of the choices the poor make or some of the choices they don't make. But in constrained circumstances, you may end up making the wrong choices.
The poor have too many things they have to do with their limited financial resources and with their limited time as well. When you have too many things to do with limited resources you have to sacrifice something.
Q: Are there key pieces of infrastructure, whether it's physical, institutional, or social infrastructure, to make sure people have a chance to get out of poverty?
I always look at things in a very pragmatic way. At the broadest level, one key goal of policy should be to make peoples' lives easier in very mundane ways. Bringing clean water to people is important because of disease, but it's important to free up this mental space as well.
The goal of infrastructure, physical or political or social, would be to have institutions for the poor that looks a little bit more like ours, where they will not have to make choices every time on things on which everybody pretty much agrees.
In rich countries, we are very much kept on the straight and narrow path without having to think about it. Our water is clean. Education is compulsory. Immunizations are compulsory. If you work at a firm that is large enough, someone is putting money in a retirement account. If not, at least you have the option of having a 401(k). Basically things are organized for a modicum of security and comfort. The first difference between the rich and the poor is that if the rich do nothing, mostly they'll end up in the right place; if the poor do nothing, mostly they end up in the wrong place.
Changing that can only work in a progressive way, slowly, problem by problem. There is no one broad answer.
Q: Can you give an example on addressing a specific problem?
I'll give you an example from my own work on the enforcement of environmental regulation in India. We're working in a very industrial state, Gujarat, which is growing very fast and which has some of the most polluted places on earth. It's not because they don't have the right laws. They have very stringent environmental regulations, but they aren't enforced well. Firms regularly flaunt the regulations.
We got an opportunity to work directly with the regulator, the Gujarat Pollution Control Board, to try to improve the effectiveness of their enforcement. We did that in two ways: one, by arranging more regular inspections and, two, by resolving a conflict of interest.
The firms must hire an auditor to perform environmental inspections every year. The problem is that the auditor is paid by the firm, so their main objective is to be hired by the firm again. Therefore they're absolutely trying to give the firms what they want, which is a clean bill of health for as little money as possible.
When we started our work, everybody—the firms, the auditors, the government—was convinced the new scheme would useless because most auditors were not even doing inspections; they were just filling out the form, pretending the firm was not polluting, when in fact some were very polluting.
This third-party audit system is very similar to what we have in the U.S. for financial auditing. After the financial crisis in 2008, there was a lot of discussion that the financial audit and credit rating system needed to be reformed. In India we had a chance to experiment with a potential reform, the structure of which is similar to what has been proposed in financial sectors of some rich countries.
In Gujarat, we randomly selected a set of firms and said, "For you the system must change. Now you won't be able to choose your auditor. We are going to assign your auditor to you randomly. The auditor is going to be paid from a central pool, and there will be a random back-check to verify their performance."
It cut the link between the firm and the auditor. Now to continue getting business, the auditor needs to provide an accurate rating, because if they don't the back-check will find it out and they'll be kicked out of the pool.
The result was phenomenal. Auditor accuracy improved dramatically. Instead of always hearing that the firms are just below the regulatory standards for highest pollutant, we had a range of results reported.
Interestingly, it also had an impact on very large polluters, where they reduced their pollution compared to firms that were not in the system. Now the government of Gujarat is scaling up the system [Read the paper.]. It is still underway, but our hope is that it's going to change the way policy is made in Gujarat, and then, I think, other states might be interested in following suit.
Q: With all the stakeholders involved, this was a complex experiment. Methodologically, how did the discipline get to this point?
The randomized controlled trials 10 or 15 years ago might have involved a few schools and an NGO. That simpler structure really allowed us to iron out the problems. Progressively we could become more and more ambitious in terms of the scope and the partners that we could manage. This project in Gujarat involved multiple government bureaucracies and hundreds of firms.
Others are also working directly with government. Frederico Finan at Berkeley has been able to randomize the wage for a civil servant in Mexico. People have been asking whether if you have higher wages it's going to attract better people in the civil service. It's hard to do, but somehow they've managed to convince the government to hire randomly for this program. And they showed that higher wages did result in more able applicants. [Read the paper].
Q: Where do you see development economics going in the next decade?
There may be even more work on whether administrative or political structures can be changed or to what extent they are stuck. I think it's one of the big debates that is just starting.
I'm hoping we're also going to see a better integration of microeconomic work into our understanding of the overall development of an economy.
For more on Duflo's work, visit the Poor Economics or Abdul Latif Jameel Poverty Action Lab websites.
Interview conducted and edited by Ted O'Callahan.