David Baron

According to the World Bank, close to one billion people around the world live on less than a dollar a day. This despite decades of well-meaning development programs and charitable efforts. A team of economists at MIT says it's time for a new approach -- one that makes prescriptions for poverty as scientifically-based as prescriptions for disease. The World's David Baron has more. (photos by David Baron)
Listen to the radio story

Baron: If anyone was born to study poverty, it would have to be Abhijit Banerjee.

Banerjee "I mean, I grew up in a middle class household in Calcutta, on the edge of one of the largest slums in Calcutta. So I have really a ringside view of slum life."

Baron: His parents were economists. And he became one too. He now teaches at MIT. Banerjee says many charities and governments try to help the poor, but too often their efforts have little effect.

Banerjee: "This enormous amount of money's been spent in the world without doing any good. You know, good intentions only go so far."

Baron: Development programs, he says, are driven too much by fads and hunches about what works, not hard evidence. So a few years ago, he and some colleagues founded a new center at MIT they called the Poverty Action Lab. Its goal is to make development a science like medicine, and it aims to do so by copying the techniques of medical research. Rachel Glennerster is the lab's executive director.

Glennerster: "We use the basic concept of a randomized trial, which has been used in medicine for a long time to test the effectiveness of drugs."

Baron: The randomized trial. It's a simple idea. If you want to know if a drug works, you assemble a group of patients. You randomly assign half to get the drug and half to get a placebo, a sugar pill. And then you compare the results. If the group that got the drug does better than the group that didn't get the drug, then you know the drug is effective.

Glennerster says her team can use the same technique to study the effectiveness of school vouchers, AIDS prevention programs, outreach efforts to farmers, scholarships for girls.

Glennerster: "It's exactly like testing a drug for cholesterol, but we apply it to poverty programs."

Hyderabad, India

Baron: The Poverty Action Lab now has dozens of studies underway around the world, including in Abhijit Banerjee's home country of India. Here in the southern Indian city of Hyderabad, the researchers are examining the effectiveness of micro-credit. Micro-credit provides small loans to the poor -- mostly to women's groups, like this one -- to help them start businesses.

Women's microcredit group

It's one of the hottest trends in development today. Last year's Nobel Peace Prize went to the originator of the idea. Borrowers who take out micro-loans have a good record of paying them back, but Banerjee says it's not clear if these loans actually lift recipients out of poverty.

Banerjee: "No really convincing study of micro-credit exists."

Baron: That's the gap in the research he's hoping to fill. A couple of years ago, his team chose 100 slums in Hyderabad for the study. Half were randomly assigned to receive micro-credit, and in these slums, many women have taken advantage of the loans.

Pushpa -- a mother of two -- used her loan to expand the offerings at her tiny corner store. She sells sugar, chili powder, soap. Her neighbor, Swarna, bought a sewing machine. She charges 75 cents to sew a blouse.

Swarna with sewing machine

Down the road, Padma lives in a simple shack with her family of eight. She borrowed money to buy and sell saris for a profit. Her home is now crammed with electronics.

Voice of translator: "See the TV they hold."
Baron: "The TV. That's a big TV."
Padma: "Color TV."
Translator: "Color TV, cell phone, VCD player."
Baron: "Is that a VCR up there?"

She seems to be thriving. But is she? When patients take medicine they may appear to improve initially yet suffer unintentional harm in the long run. Consider the problems that have turned up with the arthritis drug Vioxx or hormone replacement therapy for post-menopausal women.

Abhijit Banerjee says micro-credit may also have hidden dangers.

Banerjee: "You could imagine people getting addicted to borrowing. They're buying consumer goods and mortgaging their future. I don't know that we have any reason to be confident that that's not what's happening."

Baron: This is what the randomized trial is designed to learn. The researchers are conducting extensive surveys in the slums where micro-credit has been offered. They're looking to see if people have more money to spend and how they spend it. And then the researchers will compare these people to their counterparts in the other slums where micro-credit is not being offered.

In this comparison group is a woman named Kavitha. She lives in a squat brick shanty behind a gas station. She says she'd like to borrow money to start a business. She wants to sell vegetables or maybe buy a sewing machine.

Kavitha and her daughter

But the micro-credit organization MIT is working with is specifically not providing loans in her slum until the study ends. The MIT scientists say for their research to be rigorous, it's critical that half the people in their studies not get micro-credit, or scholarships, or whatever anti-poverty program is being tested. And this aspect of the research makes some in the aid community uncomfortable.

Milanovic: "There are several what I call ethically troubling features of the programs."

Baron: World Bank economist Branko Milanovic says the MIT research turns the world's poor into guinea pigs. He argues these studies treat people in ways most Americans would never accept.

Milanovic: "How would people feel if, for instance, unemployment benefits would only be given to 50% of people at random, or even if salaries that we get would also be randomly changed because maybe our use of that money is inappropriate or maybe somebody would believe that we should use it to buy books rather than to buy drinks."

Baron: The MIT researchers say their studies don't really deprive anyone of help. They usually test programs that are gradually being rolled out. That means some people get the aid first while other have to wait. The experiments just randomly select which people go first. The researchers also point out that their studies undergo strict ethical reviews, just like drug trials do. But the very comparison with drug trials has also provoked criticism.

Slum family in Hyderabad

Michael Patton is a prominent consultant who studies the effectiveness of aid efforts around the world. He says designing and evaluating anti-poverty programs requires an understanding of politics and culture and social dynamics that vary from place to place. He says simply dividing poor people into random groups misses these nuances.

Patton: "Community interventions cannot be reduced to the control conditions of laboratory experiments, which is what the medical and pharmaceutical model is about. The complexity is so by a hundred-fold different from administering a pill that the comparison actually breaks down."

Baron: He says the method the MIT researchers are using to study unproven anti-poverty programs is itself unproven.

But randomized trials of this kind have turned up useful -- and sometimes surprising -- results. In Kenya, researchers randomly assigned schools to receive different forms of aid. They found that giving textbooks to children produced almost no benefit in test scores. Yet, in another study, they found that giving the children just 50 cents worth of medicine to treat intestinal worms produced a huge improvement in school attendance. That finding has already influenced government programs in Africa.

In India, the researchers tested ideas for boosting childhood immunization. They found that by offering mothers a small incentive -- a couple pounds of beans -- they could vastly increase vaccination rates.

The MIT researchers argue these sorts of studies should be required before new social programs are widely implemented, in the same way pharmaceutical companies must prove that their drugs help patients and don't harm them before marketing those drugs. Esther Duflo co-directs MIT's Poverty Action Lab.

Duflo: "The wrong social program might not directly kill someone, but it might do so by omission. Because if we don't find the best way to promote immunization, and if we pass over the opportunity to convince parents to immunize their children, then we are in some sense killing the children. So I do think there's a life and death dimension to that, even though it may be less obvious than in the case of medicine."

Abhijit Banerjee and Esther Duflo, co-directors of MIT's Poverty Action Lab, in Hyderabad

Baron: Duflo recently helped establish a second Poverty Action Lab in India, and she's setting up a third in Paris, to further spread the gospel of randomized trials for social programs. And while some people are uneasy about the very idea of experimenting on the poor, Abhijit Banerjee responds this way: The World Bank, Western governments, and charities he says have been experimenting on the poor for decades by implementing anti-poverty programs without knowing if they work. Banerjee says at least his lab is honest about the fact that it's conducting experiments, and it's trying to find answers.

For The World, I'm David Baron, Hyderabad, India.