Dismal science

Finally finished reading what I found to be a fascinating and somewhat maddening article in the May 17th issue of The New Yorker - a profile of Esther Duflo, a development economist who directs MIT's Poverty Action Lab. Finally b/c I read it in spurts of a few paragraphs in the mornings, here and there between making ham and cheese for Beanie's lunch and pouring more Cheerios for Bubbie and brushing my own teeth. Such is my life.

The article describes Duflo's work - the economics of poverty, which necessarily means looking not only at poor "people" but poor "nations" (and rich people and rich nations) - and her particular approach to it - running randomized clinical trials of economic interventions in order to be able to compare their effects.

It was interesting to me to read that "the dominant economic model of the poor was that they were 'poor but efficient' - that is, they acted with the same freedom and self-interest as the wealthy" - that simply recognizing this might not be true represented a paradigm shift in development economics. For me, this illustrates not that the pointy heads miss what seems like a big duh, but that science works in a controlled and exacting manner. Even radical transformations require talking in the terms of the discipline. For better and worse.

In the article, Duflo contends that too much development policy has been implemented (and too much development aid invested) without really knowing whether or not the interventions work. "We have no idea. We're not better than the medieval doctors and their leeches," she is quoted during a presentation that she gives.

Really? I mean, do we really not know what is good or bad for poor people and poor nations?

This is where I wish to return to the idea that talking in the terms of the discipline works for better and worse. When Duflo claims that "we" have no idea, it seems to me worth asking exactly who is this "we." At first, I found the idea of randomized trials intriguing, even promising, but the further along I read, the more I started to feel outside Duflo's "we."

Silly rabbit. Such tricks are for the stakeholders of development - not poor people or poor nations, but the rich people and rich nations that implement and invest development policy and aid. That is the "we" interested in efficacy and efficiency. Those are the kind of "we" (not me) that might be expected to read The New Yorker. I imagine that others like me might become absorbed into the "we" as they become convinced by reading the article?

I was not, or I am not. I admit that I side with bleeding heart Friend of Bono economist Jeffrey Sachs, who commented that "careful measurement and comparisons are, of course , vital" - that is what StraightMan and other development anthropologists also observe and study - but that, "if I go into a village without bed nets, it pains me." I think this gets to the point that poverty is like disease (with which it is shown to be correlated) in that withholding a therapy known to work (which I believe bed nets have been) runs counter to research ethics even or especially in randomized clinical trials. Not to mention human ethics. In this sense, I feel like Duflo, strangely enough, underestimates the significance of her work, as when she defends her methodology by pointing to its versatile applications: "I once sat through a presentation where randomized trials had been used to choose the best packaging for yogurt." Did I miss the point, or did she? I see development interventions as more like medical therapies than like the packaging of Yoplait.

Also, as a teacher and scholar from the so-called softer side of the sciences - anthropology - I maintain that not everything actually can be learned from randomized trials. It is only one method among others, some of which might be themselves more effective and efficient for learning and understanding what we seek to know. The New Yorker quotes Angus Deaton, who critiques Duflo's work on the ground that (1) "experiments are frequently subject to practical problems that undermine any claims to statistical or epistemic superiority" and (2) "even if your data are perfect, how can you generalize from the information? Does the policy that works in India work in Brazil?" Spoken like someone who might appreciate a bit of anthropology and ethnography.