How Banerjee and Duflo's partners in India view their experiments

NEW DELHI: Abhijit Banerjee and Esther Duflo , this year’s economics Nobel laureates, refined their techniques in India, working on the field with organisations like Seva Mandir and Pratham.

Seva Mandir is an Udaipur-based NGO that works across areas like nutrition, health, education, gender and environment.

Banerjee, who was friends with its former president Ajay Mehta, began using his randomised controlled trials (RCTs) in the mid-1990s to help Seva Mandir answer some niggling problems that plagued their work in a complex tribal dominated region.

One of the most famous RCTs conducted for Seva Mandir was about how a combination of cameras and pay incentives could reduce teacher-absenteeism in their schools. Some disagreed with the experiment ideologically, saying it was more important to improve the teachers’ intrinsic motivation rather than use these external pressures. The teachers for their part were persuaded, even if reluctantly, into being part of the experiment.

But the experiment with cameras and incentives clearly worked to stem absenteeism and Seva Mandir gained greatly from this study specifically and, even more, from the process of collecting and analysing evidence.

“I have huge respect for the way they spend time collecting and cleaning the data, piloting their questionnaires themselves. They were closely connected to the ground,” says Priyanka Singh, former CEO of Seva Mandir and now CSR head at InterGlobe Enterprises. But Singh’s takeaway from their painstaking research is that “evidence can help you change things, if you are located in an institution open to change”.

MIT economists and J-PAL cofounders Abhijit Banerjee and Esther Duflo in Hyderabad (file photo)


Interestingly, the same methods, when applied to a large-scale government setting, did not work. The attempt to use the camera and incentives model
for nurses in the Rajasthan government’s health subcentres yielded no results — Duflo’s paper on the experiment likened it to applying a “Band-Aid on a
corpse”.

Randomisation takes care of many problems of context, so caste and other social dynamics do not muddy the results, says Singh, but admits that

RCTs have their uses and their limitations. They run for a period of 20- 22 months; the setting they operate in is not a blank slate but a place where NGOs
have worked.

For instance, she points out that in 2009, “the central government’s National Rural Health Mission recruited nurses on a large scale, leaving very few, under-trained nurses from private colleges for the NGO to work with, which cramped its immunisation camps. This supply problem suddenly shifted the
context”.

Such background issues are not always factored in when it comes to the RCT approach. All solutions need recalibration, for which you do need “longterm work

in a community and sustained leadership,” says Singh.

Another organisation that has worked closely with Banerjee, Duflo and J-PAL (Abdul Latif Jameel Poverty Action Lab), the poverty research centre they co-founded, is the education NGO Pratham.

“We have had a long association of 20 years, from when we were in an early stage ourselves,” says Rukmini Banerji, the Pratham CEO.

J-PAL evaluated the Balsakhi remedial teacher programme that Pratham ran for municipal schools in Mumbai and Vadodara. Over the years, they did six RCTs with Pratham, from asking whether a state-mandated participatory committee really improved educational awareness, to what could be done to

teach at the right level (TARL), so that children did not fall behind in primary school. “Methodologically, it was a very productive experience to work with J-PAL, they brought real rigour and helped share learnings across contexts,” says Banerji.

For instance, the stringent tests that Pratham put its programmes through, and its impact on learning increases as attested to by J-PAL’s reaserch, seemed useful for educators in Africa who wanted to collaborate with the organisation.