Esther Duflo: Experiments against poverty

Author: Ethan Zuckerman
Posted: 10/24/09 18:10 UTC
Info: Bio
Tags: Development, Economics, esther duflo, Poverty, experimentation, rigor

Professor Esther Duflo is a new Macarthur fellow, recognized for her pioneering work in studying a wide set of issues in development economics. Her reseach has looked at the effectiveness of financial remittances (it’s better to send money to grandmothers than grandfathers) and the health issues associated with charcoal stoves. Her weapon of choice is the randomized trial, used extensively in drug discovery research, but rarely applied to the field of development economics.

Her Pop!Tech talk is titled “Creative experimentation and the fight against povery”. She tells us that her goal is to help us think through the impacts of innovation, specifically for the people living on less than a dollar a day, the 25 million children who go unimmunized, the millions who die of preventable diseases. These facts are well known, but they’re so disturbing we generally don’t think about them.

Esther Duflo, photo by Kris Krüg

The tempation is to look for a single, silver bullet to make these problems go away. For some people, it’s foreign aid – Sachs, Bono, Angelina Jolie. Bill Easterly and Dambisa Moyo argue that aid is the problem and that free markets work. These views compete on the moral high ground, and also on scientific grounds, looking for support for these positions.

It’s not surprising that the scientific facts are in dispute – these problems are extremely complex. There’s a demand for discourse that’s scientifically legitimate and leads to policy prescriptions.

If you want to argue that malaria is important, you can show that countries without malaria are richer. You can also argue that free markets are important, showing that countries with free markets are richer. Is Sri Lanka richer than Bangladesh because of free markets? Or because of the elimination of malaria? Too many things move at the same time.

We cannot identify the secret to ending poverty by comparing the historical experience of hundreds of countries. We need to decompose the problem into lots of small problems. Poverty has many facets – lack of health, lack of education, lack of choice for self-realization. When we focus on these facets, we can look at questions like, “How do we get children to school? How do I get them to learn something?”

You can then identify significant ways to improve the lives of the poor. And then we can run experiments. She quotes Franklin Delano Roosevelt: “The country needs, and unless I mistake its temper, the country demands bold, persistent exprimentation…”

We, as a community, need to guide policy towards creative experimentation. This means thinking outside the box, but it also means taking rigorous, scientific methods of evaluation and applying them to problems in development.

Economists have a positive tradition – the idea that economic agents are billiard players and econmists are physicists – they can infer the law of physics from looking at the balls, but not interfering. This isn’t a very accurate understanding of economics. We need to understand that people communicate, learn, help each other… and experiment.

Farmers in poor countries know a lot about agriculture, because their lives depend on it. But this makes them very reluctant to experiment, because if you fail, your family dies. You can look at what your neighbors do… but if everyone’s looking, no one is innovating. You need people who are in the business of experimenting and sharing the results of those experiments. Those innovations will never be generated by the market – they will be generated by community effort.

We cannot invent something new without modeling reality. When we do that, we often ignore critical variables. It’s not a reason to do nothing. Everyone gets it wrong – it’s not a reason to avoid acting, but it is a reason for humility and careful evaluation.

It’s not done very often because it’s not particularly easy. If you design a car and no one buys it, you realize your car wasn’t very good. But designing social policy is different. Perhaps we subsidize something, because we believe people need it. People will obtain it because of the subsidy… but that won’t tell us whether the intervention was a good or bad idea. You’re trying to contrast the outcome of people who experienced a policy to what people would otherwise have experienced – you don’t see the counterfactual. Beneficiaries of the program are generally not comparable to others.

What’s the solution? Randomized evaluation. We can introduce new policies by randomly assigning populations into a treatment and comparison group. We can then compare the effectiveness of policies by comparing those two groups.

This isn’t a tool for accountability – it’s not designed to punish people. It’s a tool that helps us figure out what does and doesn’t work. It helps us determine that policies like deworming – which aren’t especially sexy – are extremely effective and worth making policy around. It’s also a tool that tells us how people actually behave so we can design future interventions.

Here are three things which might make a huge difference for the poor. Giving a kilogram of lentils with an immunization raises the rate from 5% to 37% in a poor Indian district. It’s actually cheaper to give out the lentils because it keeps the health workers fully occupied. Informing girls of the relative risks of HIV amount older and younger men reduces risky sexual behavior by 67%, letting girls avoid sex with older men. This compares to no effect in normal HIV education, which simply tells girls that all sexual behavior is risky.

If we’re talking about 1 billion poor people, are we really making progress by distributing lentils and warning girls about sugar daddies? Yes. These aren’t silver bullets. But they’re a strategy for transforming the ways we do development and encouraging innovation. Evaluating smaller ideas may be the way to unlock the “bigger machine”, the complex set of factors that govern developing world economies.

She closes, telling us: I would like to practice a true human science – rigorous, impartial, a science of humans in its imperfections and complexities, humble and humane and generous.