Twenty-one years ago this summer, I was fortunate enough to witness the start of a new movement of randomized evaluations that have transformed development economics. (This week, CEGA and IPA celebrated the anniversary in Nairobi.)
I had already met Michael’s family in Kansas but in 1994 he took me to see his other family, with whom he had lived on a small farm in rural Kenya for a year after college. Michael relates the story of how a chance meeting with an old friend who worked for International Child Support (ICS) led to the first RCTs on education in Busia, Kenya, a then very sleepy town on the border with Uganda which has become famous among development economists.
There is something very special about Busia and the ethos of inquiry and partnership that developed and flourished there. Here a craft was developed and honed. As the movement becomes much wider it's worth recording some of the principles that were so successful in Busia.
A deep commitment to craft. A thousand small decisions separate a good RCT from a poor one: Framing the question on the survey exactly right; designing the intervention so that it carefully reflects local needs; capturing data at the right time of day or year; meticulously monitoring enumerators; making it easy for the data to be entered correctly; capturing not just the outcome but all the steps along the way to better understand the outcome. Busia was where many researchers came to develop and learn the intensely practical craft of running a good RCT: Lorenzo Casaburi, Pascaline Dupas, Esther Duflo, David Evans, James Habyarimana, Ted Miguel, Owen Ozier, Jon Robinson, Simone Schaner, Frank Schilbach, Rebecca Thornton, Alix Zwane, and many others all worked there. And here they developed many of the protocols that have become standard in RCTs.
Long-term partnership. Many of the best RCTs have come from long-term partnerships between researchers and NGOs or other local partners. The trust that develops with collaborations over many years allows innovative ideas to be tested, often springing out of the lessons of past failures or partial successes. One RCT builds on the findings of the previous. The relationship between researchers and ICS was the first model for this long-term relationship. (Several years after going to Busia I also saw Michael and Abhijit Banerjee forge the start of a second important relationship with Seva Mandir. See Neelima Khetan, former chief executive of Seva Mandir, discuss this summer like no other here.)
Thorough understanding of context. An important ethos of the researchers working in Busia was the importance of understanding the local context. OK, not everyone could spend a year living in a local family, sleeping on cow dung in a thatched hut and teaching in the local school--but this example set the tone. Many PhD students and research assistants lived with families in town and learned Swahili during their long stays in Busia. Scott Guggenheim, an anthropologist who has worked closely with Ben Olken and others on RCTs in Indonesia criticizes economists' reluctance to talk about the detailed qualitative work they do in preparation for an RCT: its an integral part of how the best economists work, but if you don’t talk about it people will think you can fly in and fly out and do a good RCT.
Building high-quality research infrastructure. Why did so many PhD students and junior researchers come through this backwater town? Yes, for the collaboration with other researchers, but also to take advantage of the research infrastructure that was built--first as the evaluation group at ICS which then migrated to become IPA Kenya. It is hard enough for PhD students to do an RCT as their thesis, but if they had to find a way to hire Kenyan enumerators legally and find people to enter the data, etc., etc., it would be a lot harder. The lower entry costs were critical to the movement taking off. Again the model was copied in other places: first in India, then in IPA offices and J-PAL regional centers across the world. It has been wonderful to see some of the Kenyans who helped build this infrastructure be recognized. This summer, Michael and I (with other Busia alumni) were with Carol Nekesa (who was part of the team who moved from ICS to IPA), to celebrate her graduation from the midcareer master's at the Harvard Kennedy School.