“Randomista, C’est Moi”

Photo: Kris Krug, via Pop!Tech

Esther Duflo: she’s a short, plain-looking French woman who delivered her TED lecture in a thick French accent. As we reported a few weeks ago, she was also recently granted the prestigious John Bates Clark award, which goes to the best economist under 40 in the US.

Combined with the increasing prominence of development economics, then it seems almost natural that she landed a 12-page profile by Ian Parker in The New Yorker Innovator’s Issue.

The piece begins with a condensed—but still extensive—rundown of her career: hired as professor of economics at MIT—which strayed from departmental protocol that disapproves hiring its own graduate students—tenured by age 30, co-founded the Abdul Latif Jameel Poverty Action Lab (J-PAL), received the MacArthur “Genius” grant in 2009, named one of Foreign Policy’s survey of the Top 100 Global Thinkers and lectured at the highly regarded Collège de France, the youngest person ever to do so—one-upping Jean-Paul Sartre.

Her claim to fame is testing social policy ideas with randomized control trials (RCTs) in order to confirm whether a particular policy—say, microfinance—works or not, much like clinical drug trials. Described as “a new economics being born” by one colleague, this new approach to poverty alleviation, which requires a combination of “field-work patience, behavioral insight and mathematics,” is championed for its ability to connect cause and effect.

It aims to answer questions such as “Does increased education result in increased wages?” a question that was answered in Duflo’s Ph.D. thesis using data from an Indonesian school-expansion program in the seventies. RCTs can also answer more complex questions pertaining to the merits—or shortcomings—of female empowerment: “Do villages led by women exhibit less voter prejudice?” The answer turned out to be yes, and the RCT convinced Duflo that public policy can, in fact, influence voter prejudice for the better. She is now conducting a follow-up RCT, investigating whether people need to experience female leaders, or whether education, independent variable-ized in the form of a morality play extolling the capabilities of women, would suffice.

In the wide spectrum of development economists, Duflo places somewhere in between Jeff Sachs and Bill Easterly. She does not believe in the “proverbial grand soir” that is the Sachs-ian End of Poverty—she is, however, an advocate of aid—but she also shuns the pessimism of the Easterly camp, as her RCTs continue to make small but incremental gains in policy know-how.

Since its inception in 2003, J-PAL has carried out 206 RCTs, in both developed and developing countries. J-PAL is now entering its second phase, which aims to engage policymakers—developing country governments, the World Bank and NGOs, according to Duflo—as RCT proponents, ensuring that development policy is statistically supported by evidence of efficacy, in lieu of the “fad to fad” zigzagging that has long dominated development policymaking.

Naturally, this new-fangled branch of development economics, redolent of social engineering (and its failures), has been controversial. Among economists, Duflo and her colleagues are sometimes pejoratively referred to as randomistas.

Parker lists a few common arguments against RCTs. First, they are subject to practical shortcomings, seeing as it is impossible to section off a country into a treatment and control group due to attrition, contagion and slippage. Secondly, there are major ethical concerns, because half of the population, the control group, misses out on a possibly life-saving policy proposal. Duflo confidently stated at the 2010 TED Conference that she was taking “the guesswork out of policymaking”; these brash blanket statements have roiled several development economists, the two most prominent being Lant Pritchett at Harvard and Angus Deaton at Princeton, perhaps the most outspoken critic of randomization.

Pritchett points to the success of the civil rights movement or feminism, concluding that great social changes did not require RCTs. Deaton’s argument is twofold: first, there are too many practical obstacles to robust RCTs (mentioned above) and secondly, it is impossible to generalize from the conclusions of RCTs, effectively eliminating that virtue of international development, scalability. He calls the second argument “the problem of external validity”—a policy proven to work in India may very well not work in Brazil.

Bill Easterly commented upon the Esther-mania that has managed to spread from academia to the general public (or, rather, the New Yorker-reading educated elite). He enjoyed the personal anecdotes that have accumulated to form Duflo’s views on development—namely, her “protestant left-wing Sunday School” childhood—as well as her general optimism vis-à-vis global development using incremental step-by-step gains. However, he criticized Parker’s facile treatment of the arguments against randomization, noting that Parker spent more time analyzing the tone, rather than the contents, of the arguments.

Advertisements

One thought on ““Randomista, C’est Moi”

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s