The Inside Agenda Blog

Solving Poverty with Science

by Allison Buchan-Terrell Wednesday November 16, 2011

In the foreword to "Poor Economics," authors Abhijit Banerjee and Esther Duflo bemoan the tired dichotomies in the debate about how best to alleviate poverty:

[The] urge to reduce the poor to a set of cliches has been with us for as long as there has been poverty. The poor appear in social theory, as much in literature, by turns lazy or enterprising, noble or thievish, angry or passive, helpless or self sufficient. It is no surprise that the policy stances that correspond to these views of the poor also tend to be captured in simple formulas: "Free markets for the poor," "Make human rights substantial," "Deal with conflict first," "Give more money to the poorest," "Foreign aid kills development," and the like. These ideas all have an important elements of truth, but they rarely have much space for average poor women or men, with their hopes and doubts, limitations and aspirations, beliefs and confusion.
And, the authors say the field of anti-poverty research, perhaps not unsurprisingly, is filled with lots of failed experiments based on these false premises. Banerjee and Duflo argue that to move forward, we must abandon our habit of reducing the poor to cartoon characters and instead make the time to really understand their lives and all the complexity that comes with doing so. How do you come to truly understand the lives of the poor?
The authors have dedicated their careers to on-the-ground research in search of an answer to that question. In particular, they've adapted a tool from medical research: the random control trial (known as RCTs), which is used to test new drugs, and applied it to development economics to find out how the poor make decisions and which interventions work and which don't. RCTs are something of a fad right now in the development world. In an RCT, individuals or communities are randomly assigned to different programs or different versions of the same program. Because the individuals assigned to different treatments are exactly comparable (since they were chosen at random) any difference between them is the effect of the treatment.
For example, Banerjee and Duflo helped set up a pilot project in India to see if offering two pounds of dal (dried beans, a local staple) for each immunization and a set of stainless steel plates for completing the entire course of immunization, would increase the number of people immunizing their children. They partnered with a local organization running immunization camps and offered the incentives at 30 camps. What they were testing was the claim that immunization had no place in the traditional belief system. People believed that children died because of catching the evil eye, and the way to catch the evil eye is to be displayed in public, so parents would not take their children outside for the first year of their life. And so, it was thought that it would be difficult to encourage parents to immunize their children without changing their beliefs.  
But, what they found was that in the villages where these camps were set up, the immunization rate increased sevenfold, to 38 percent. While this is certainly not at the level necessary to achieve what doctors refer to as "herd immunity" -- an immunization rate of 80-90 per cent, which is the rate at which the whole community is protected -- there is still a significant social benefit from increasing full immunization rates against basis diseases from six per cent to 38 per cent. 
More importantly, Banerjee and Duflo are interested in what this experiment tells us about how the poor make decisions -- in this case about health:
what the two-pounds-of-dal experiments demonstrate is that in Udaipur at least, the poor might appear to believe in all kinds of things, but there is not much conviction behind many of those beliefs. They do not fear the evil eye so much that they would pass up the dal. This must mean they actually know they are in no position to have a strong basis to evaluate the costs and benefits of vaccines. ... So, although some beliefs the poor have are undoubtedly strongly held, it is a mistake to consider that it is always the case. ... But, it is also wrong to assume, as both the left and right wing do, that action follows intention: that if people were convinced of the value of immunization, children would be immunized. This is not always true and the implications are far reaching.
And through such experiments, what is really being asked is: Is there anything special about the poor? Do they live as everyone else does, but with less money? Or, is there something fundamentally different about life in poverty?
RCTs have become the "gold standard" in program assessment and represent a vast improvement on how programs used to be assessed -- checking whether the money was spent, but there are many critics who argue it is being seen as a panacea. In fact, the March/April 2011 of the Boston Review was dedicated to the debate about the use of RCTs. In that issue, World Bank economists Shantayana Devarajan, Jishnu Das and Jeffrey S. Hammer cite two concerns about the use of RCTs in development: 
  1. RCTs only tell us what will happen if you intervene and not "Should you intervene?" The answer to the latter question lies in whether there is a market failure or need for redistribution. 
  2. Even if there is a good reason to intervene, RCTs, usually implemented by an NGO that manages a well-defined program, tell us little about what will happen if the program is implemented by real government officials, who face political pressures and may not be able to (or want to) run the program in the same way. 
Of course, Banerjee and Duflo are quick to point out that one experiment, like the two-pound-of-dal experiment, is not the final answer on whether a program will universally work. Rather, it is a first step:
"But we can conduct a series of experiments, differing in either the kind of location in which they are conducted or the exact intervention being tested (or both). Together, this allows us to test the robustness of our conclusions (Does what works in Kenya work in Madagascar?) and narrow the set of possible theories to explain the data. ... The new theory can help us design interventions and new experiments and help us make sense of previous results that make have been puzzling before. Progressively, we obtain a fuller picture of how the poor really live their lives, where they need help, and where they don't." 
The authors, and others who promote the use of RCTs in development, hope to persuade us that their patient, step-by-step approach is a more effective way to alleviate poverty. There is no one magic bullet solution, they say, so let's use an example from Poor Economics: "the millions of well-intentioned people across the world -- elected officials and bureaucrats, researchers and NGO workers, academics and entrepreneurs -- in the quest for the many ideas, big and small, that will eventually take us to the world where no one has to live on 99 cents a day." 
If you want to learn more about Abhijit Banerjee and Esther Duflo and their book, "Poor Economics," they have an excellent interactive website for their book that not only includes chapter summaries, but data tables and links to all of the studies cited. You find out more about the work they do at the Abdul Latif Jameel Poverty Action Lab (JPAL) at MIT here

Photo of Abhijit Banerjee and Esther Duflo by L. Barry Hetherington.