Perspectives — A Forum for RAND Guest Speakers

Marshaling the Evidence

How Science Can Help Fight the War on Poverty

The images of devastation witnessed by the world in the aftermath of the catastrophic magnitude 7.0 earthquake in Haiti on January 12, 2010, put the problem of poverty in the developing world front and center once again. People and nations have since responded by contributing more than $2 billion to the Haitian relief effort.

But how do we know whether such aid is making a difference? “When it comes to dealing with world poverty, some feel that providing aid is the answer, while others feel that such aid is at best irrelevant and at worst a major contributor to dependency and corruption. Unfortunately, the answer is almost impossible to know,” said Esther Duflo, the Abdul Latif Jameel Professor of Poverty Alleviation and Development Economics at the Massachusetts Institute of Technology and a founder and director of the Jameel Poverty Action Lab.

“Nine million children under the age of five die every year in the world, which is the equivalent of a Haiti earthquake every eight days.”

In a talk at RAND in March as part of the RAND Corporation’s Distinguished Speaker Series, Duflo used the example of Africa, which has received billions of dollars in aid over the past 50 years but does not seem to be getting any richer, judging by the gross domestic product per capita of the people. “While such data can support those who feel aid is bad,” said Duflo, “we really don’t know whether Africa would have been far worse off without the aid or perhaps better off without it. There is no counterfactual here — no “what if’ — because there is only one Africa.”

In the absence of convincing evidence one way or the other about the value of antipoverty programs at the global level, Duflo argued on behalf of scientific experiments to gauge the value of antipoverty programs at the local level. She focused on three types of relief efforts — immunizing children, subsidizing the purchase of bed nets to control malaria, and ensuring that children attend school — all of which aim to alleviate poverty worldwide.

Immunizations, Bed Nets, Schools

Duflo explained that 25 million children a year do not receive immunizations. “Despite the fact that we have the technology, the political will, and money to do it, it still doesn’t get done. It’s what’s called a ‘last mile problem,’” she said, referring to the challenge of transporting children to the closest immunization sites.

Malaria kills nearly 900,000 people a year — with 91 percent of the deaths occurring in Africa and 85 percent of the deaths worldwide being under the age of five. In warding off the mosquitoes that cause malaria, bed nets are effective and inexpensive (about $10 to manufacture one, ship it to Africa, and teach people how to use it), and they benefit even those who do not use them. “But does how we subsidize bed nets make a difference in how or whether they are used?” Duflo asked.

As for getting kids into schools, Duflo noted that there are many ways to do this, such as paying families to keep their kids in school, hiring more teachers, providing meals, paying for uniforms, or educating parents about the benefits of keeping kids in school. “But which option makes the most sense? Which will provide the biggest bang for the buck?”

DIANE BALDWIN

Esther Duflo, director of the Jameel Poverty Action Lab at the Massachusetts Institute of Technology, addresses a RAND audience in Santa Monica, California, on March 4, 2010.

The Value of Experiments

For each of these questions, science can help provide an answer through the method of randomized controlled trials, which do provide counterfactuals, according to Duflo.

On the question of immunization, she discussed experiments in a village in India where only 1 percent of children are fully immunized. “The ‘last mile’ here involved persuading parents to overcome the natural human desire to procrastinate in getting their kids immunized, a problem compounded by the disincentive of having to walk to an outlying village subcenter, which may or may not be open or staffed.” The experiment tested whether removing the disincentive (by setting up highly localized vaccination camps every month, rain or shine) would make a difference and whether providing an incentive (in this case, a kilo of lentils at such camps) would work even better. “The results were striking: Compared to control villages where there were no camps, the percentage of kids getting vaccinated increased by a factor of three with the camps, and when those camps provided a small amount of lentils, vaccinations increased by a factor of six.”

Similar experiments, such as those done at the Jameel Poverty Action Lab by Duflo’s colleague, Pascaline Dupas, were conducted to address the question of bed net subsidies. In this case, the question was broken into pieces by asking (a) whether people would pay for the bed nets, (b) whether people would use the bed nets more if they paid for them than if they received them for free, and (c) whether people would buy the bed nets in the future if they had received them for free at first. “The experiments showed that cost is a barrier — the rate of purchase goes down as the cost charged for bed nets goes up — but that the usage rates are the same whether people pay or get them for free and that individuals are more likely to buy them in the future if they get them for free or pay only a little at first,” Duflo said.

“Once policymakers have decided where to invest resources, evidence-based experiments can help them choose the best intervention option to pursue.”

As for making sure that children attend school, experiments done by Edward Miguel and Michael Kremer, two other colleagues of Duflo, sought to identify the most cost-effective of the many possible options. “Providing extra teachers, school meals, uniforms, and scholarships yielded about one to three years of extra schooling for each $100 spent. But surprisingly, providing deworming medication — so kids were healthy enough to attend and perform well in school — and simply telling families about the benefits of schooling yielded far more substantial gains: nearly 29 and up to 40 extra years of schooling, respectively, for the same $100 investment.”

Taking the Guesswork Out of Relief

Duflo acknowledged that world poverty poses a daunting challenge. “Nine million children under the age of five die every year in the world, which is the equivalent of a Haiti earthquake every eight days.” She stressed that there is no silver bullet for addressing poverty. “The problem in Haiti was visible, circumscribed, and salient; world poverty is not like that.”

She also recognized that experiments such as the ones she described would not help policymakers decide how to set priorities among the many types of interventions needed. “All such areas — be they in education or health or elsewhere — are critical. It is a political decision about where to allocate resources,” she said. “But once policymakers have decided where to invest resources, evidence-based experiments can help them choose the best intervention option to pursue.” square

Marshaling the Evidence