In my previous blog I discussed some of the challenges of working with governments on RCTs. My aim was not (as some have suggested) to disparage this work but to give some tips to those who are taking it on. Indeed, it is because I know just how hard it is that I have enormous admiration for those who manage to pull it off. I was fortunate enough to work with the Government of Sierra Leone's Decentralization Secretariat in the aftermath of the civil war as they introduced a system of decentralization to the country (not randomized) and experimented with decentralizing all the way down to villages with the GoBifo program (randomized). Ten years later I am still there and still working with the government.
Despite all the challenges, working with governments can have substantial payoffs. Some of these benefits flow to researchers. As discussed in my previous post, working with governments opens up a whole range of questions that cannot be addressed when working with NGOs or companies. Nor is there anything like trying to run an RCT with a developing-country government to understand just how these governments work and what constraints they operate under. But many of the benefits are in the form of public goods. Researchers often come in with knowledge of what has worked elsewhere and help the government design better programs. They typically also provide technical assistance to governments about good process monitoring. Government partners also gain, through an improved capacity to distinguish different types of evidence, to determine which evidence is best at answering what question, and how best to integrate research evidence into their programs. If the relationship goes well, the government partner will be much keener to work with subsequent research teams and will be much more effective in working with them. Importantly, governments are often in a good position to scale up a program that has been tested if the results are positive. A great example of how researchers working on a question of intense interest to the government can lead to a quick and wide-reaching policy change is the Raskin rice program in Indonesia.
Because the costs to the individual researcher are high, but many of the benefits accrue to the world at large, J-PAL has put significant effort into building up long-term partnerships with governments to make it easier for our researchers to work with them. Typically these partnerships include: discussing general lessons from previous research findings that may be relevant to the local context; capacity building to help civil servants read and incorporate lessons from research; and a demand-driven agenda of research to answer the questions the government most wants answered, and that J-PAL researchers are able to design studies to answer.
Some good examples of these collaborations are EduLab in Peru, which is a joint effort of J-PAL Latin America & Caribbean and IPA Peru; the collaboration J-PAL South Asia has established with the Government of Tamil Nadu; and the partnerships being developed by J-PAL North America as part of their State and Local Innovation Initiative.
Another approach is to provide subsidies and support to researchers who are investing in building relationships with governments. The Government Partnership Initiative at J-PAL does exactly this.
Other organizations, including Innovations for Poverty Action and the International Growth Center (both of which I work with), have similarly invested in building long-term relationships with governments to foster researcher–government collaborations. When doing RCTs or other impact evaluations, researchers at the World Bank and other multilateral development banks also benefit from working within a long-run relationship with governments, which can help with--though they do not solve--the challenges I discussed.