Statistically Significant

by Oscar Pocasangre:

To most people, words like “randomization,” “regression,” and “dependent variable” seem more related to a statistics class than to poverty alleviation. To Yale economist Dean Karlan, however, this could not be further from the truth. Determined to harness the power of empirical research to help make the world a better place, Karlan founded Innovations for Poverty Action (IPA) in 2002. Currently employing over 100 researchers and professors, including many Yalies, IPA draws on statistical concepts to determine what interventions are effective in international development and disseminates its findings to experts and policymakers across the globe.

Currently operating in 20 countries, IPA’s staffers are mostly recent university graduates, often aided by student interns. According to Christopher Blattman, assistant professor of political science and economics at Yale and an IPA research affiliate, the group is “a little bit like a think tank, an NGO, and an academic department.”

A woman pays back a loan she obtained through one of IPA's partner's in Huarez, Peru, while the credit officer calculates how much she still owes. (Pocasangre/TYG)

To determine what works and what doesn’t, IPA performs randomized control trials (RCTs), which essentially transfer the scientific method to economic analysis. By dividing participants randomly into treatment and control groups, researchers measure the direct effects of interventions or policies, comparing differences and controlling for variables.

“RCTs strip a situation of many other confounding factors and really let you look at a particular question in a rigorous, objective way,” said Tania Alfonso, MC ’01, the Peru country director for IPA. “They help us come up with specific tools that help policy makers decide what to fund.” Pia Gugnani, SM ’11, was involved in an RCT while working for IPA on a study of the impact of a new savings device in Ghana. “While designing the study, I found myself applying the same techniques I learned in class,” said Gugnani.

The results of IPA’s RCTs have already led Karlan and others to reject some grandiose claims about development policies and projects. Microcredit, for one, has captivated the world’s imagination with its almost miraculous promise to end poverty, drawing funding from donors and organizations from all over the world. Recent IPA studies in the Philippines compared the outcomes of microloan recipients with those of a control group and found that the loans yielded only modest benefits. “You are not going to cure poverty by giving people loans,” said Karlan, though “you can do some good.”

Beyond microcredit, IPA’s research has also helped identify inconspicuous solutions to problems faced by developing countries. A study by Harvard professor and IPA research affiliate Michael Kremer found that the most cost-efficient way to increase school attendance in Kenya is not through traditional interventions like the
provision of free uniforms or meals but rather through the provision of treatment for intestinal worms. Kremer’s study has motivated a global movement to improve access to education through deworming.

To make sure that research like this does not remain shelved in economics journals, IPA aims to disseminate its findings widely and scale up successful projects. “We try to write things up in ways that are not academic,” said Karlan. “We sometimes go directly to some organizations and tell them about our results. Sometimes, it takes going to the ministers of education and telling them about the benefits we found.”

But RCTs are not silver bullets. Blattman discussed some of the method’s drawbacks, explaining that RCTs tend to be very expensive and involved, as they have to be planned in advance and then replicated across different experimental groups. Other critics argue that IPA’s studies fail to accurately assess long-term effects like those of microcredit, which they say could take more than just a few years for their full impact to be measurable.

Alfonso agreed that RCTs are better suited for some interventions than for others, but she remains optimistic. “Not every single project will produce results that are that dramatic,” she said, “but we do get pieces of the puzzle.”

As they search for these pieces in countries from Ghana to the Philippines, IPA’s researchers and student affiliates keep finding better ways to alleviate poverty and create a wave of statistically significant change in developing countries. Though there is large margin for error, their confidence interval spans the world.

Oscar Pocasangre is a junior Economics and Psychology double major in Berkeley College. He interned with IPA in Peru over the summer.