The 2019 Nobel Prize in Economics was awarded to Esther Duflo, Abhijit Banerjee, and Michael Kremer for their “experimental approach to alleviating global poverty”. This is an important turn of events partly because of the enormity of the problem that they and others are trying to address. In 2019, there are still more than 550 million people living in extreme poverty: that is, they make less than $1.90 a day.
What is the approach Duflo and her colleagues use so effectively to involve themselves in the poverty-alleviation movement? They take the position that data matters and by studying programs and projects meant to alleviate poverty, they can help to determine what works and what does not work.
Some authors and practitioners argue that to end poverty, we must spend more money on social programs. Others posit that more programs will just make people more dependent on them. Who is correct? It’s difficult to tell until we put each theory to the test. Without solid evidence, it is very hard to know.
Duflo, Banerjee, and Kremer proposed to break problems into smaller, more specific questions, instead of trying to answer a big question such as, can we eliminate poverty by spending more? Duflo’s broader set of colleagues asked things such as, to fight Malaria in Africa, do bed nets, which keep away malaria-carrying mosquitos, help contain the spread of the disease? If yes, then it might be worth knowing if it was better to give the nets to people for free or to charge for them? These are the types of topics they like to study. In the end, people who don’t get malaria are more likely to survive, get an education, be more productive and overcome poverty.
Duflo and others’ innovation was to push toward using more Randomized Controlled Trials (RCT) to study poverty alleviation.
Sometimes it is best to describe RCTs by first stepping away from poverty alleviation. Imagine that you have an experimental pill for diabetes, and you want to test its effectiveness. You could create at least two types of major problems for yourself in your experiments:
Problem # 1: What if you give the experimental pill to a group, and some people start getting better. Will you know it was because of the pill, while if, at the same time, the people who took the pill started eating properly and exercising? This is the correlation vs causation problem.
Problem # 2: What if you give the pill to those participants who have a better chance to show improvement? For instance, let’s say that you give the pill to young, fit individuals instead of overweight, elderly participants? Chances are that those who are fit will get better results. This is called the sample bias.
RCTs solve these issues by separating participants into two or more groups with similar characteristics. The pill is given to one group and not to the other. This second group may be known as the control. Then there may also be a third group that is given a placebo. Researchers measure the difference in wellbeing between groups. To avoid problem # 2, participants are assigned to the groups randomly; therefore, each group will probably have young and elderly people alike, reducing the risk of biased results.
Returning to the example of bed nets, the natural assumption might be that in an impoverished area, giving away free nets is the obvious choice. Yet, there is always the question of whether people will value the product if it is given for free, maybe they won’t use it. In addition, if nets are given for free there may not be enough resources to make the program far-reaching. Consequently, with the intention to know whether people would use the nets if they are given for free, researchers prepared an RCT meant to study pregnant women who went to 16 Kenyan clinics. In some cases, the bed nets were given for free while others were discounted through vouchers handed to participants. The result was that people obtained fewer bed nets when they were charged. Nevertheless, both groups, the ones who paid and the ones who didn’t, used the bed nets.
Poverty continues to be one of the most important problems facing humanity. However, fighting it with programs and policies that are well-intentioned but are not supported by data may not be enough and may even worsen people’s situations. Using evidence-based research techniques, such as those used by the newest Economics Nobel Laureates, we stand a better chance to fight poverty.
RCTs may not be all that we need. Perhaps good macro-economic theory and practice will make a difference as well. But RCTs are an essential tool in the tool box of poverty alleviation experts.
John Hoffmire is Chairman of the Center on Business and Poverty. He also holds the Carmen Porco Chair of Sustainable Business at the Center. Mario Alejandro Mercado Mendoza, Hoffmire’s colleague at the Center, did the research for this article.