Sunday, January 24, 2010

RCTs, Micro and Macro tussle in development

"There is no magic bullet for sources of growth; growth regressions have not produced anything substantial to identify the sources of growth."

Thats from Bill Easterly an event in Brookings last week. Based on the title of the event, I thought it would focus on what works on development -- either a smaller grassroots development approach (micro approach) or a traditional top-down development approach (macro approach) or a combination of both. Unfortunately, the whole discussion focused on Randomized Controlled Trial (RCTs). Is RCT a grassroots development approach? Jessica Cohen and Easterly debated discussed the pros and cons of RCTs. To be very fair, Cohen's talk was extremely boring (may be thats the drawback of powerpoint-less presentation). She kept on forgetting questions asked by Raj Kumar, the moderator of the event, and the audience.

Meanwhile, Easterly was to the point narrating his dislike for top-down, expert-led approach in development. [After the event, one development economist asked me: “Ain't Easterly behaving like an expert himself despite his dislike for expert’s advise in development?”]. Basically, he favors (in fact, most of the economists do but they differ in the working modalities) bottom-up approach to development that is more attuned to market principles. There is (potentially) more accountability and transparency. However, micros do not add up to macro. So, we might need both approaches-- in fact, there are some complementariness between the micro and macro approaches.

Easterly repeatedly emphasized that the extensive use of RCTs to assess pretty much everything is turning into a social engineering project (he joked: you can do RCTs in pretty much everything but the ones who do it!). He cautioned against conflict of interest among donors who fund RCTs projects and the academicians who evaluate results. His warning: RCTs is going to fail if it gets captured by the aid agencies.

Cohen tried to defend RCTs by arguing that the way it is conducted and the results derived should be noncontroversial; it is a step in the right direction. It is the best way to test any theory. People can learn what works and what does not, paving a way to incrementally do better in implementing projects and increasing effectiveness. She argued that the strong feedback mechanism derived from RCTs could potentially help in aid effectiveness.

I was more interested in Easterly thoughts on HRV's Growth Diagnostics approach, a policy-oriented approach to growth studies that looks at a set of strategies to identify the binding constraints on economic activity and tries to figure out relevant policy to relax the constraints so that the resulting change in the objective function (growth) is the highest.

More precisely, "the strategy is aimed at identifying the most binding constraints on economic activity, and hence the set of policies that, once targeted on these constraints at any point in time, is likely to provide the biggest bang for the reform buck." This approach takes into account the fact that different countries have different binding constraints on growth and that the same policy used to relax a constraint in one country might not work in another country; it is time, context and country specific, in general. This approach to growth studies differs from other approaches like cross-country panel growth regression, growth accounting, and international rankings/benchmarking. It is heavily policy-oriented. It seeks answer to the question: "what is constraining growth?" instead of "what causes growth?" I find this approach very neat, easy to follow and reality-based.

I wanted to ask Easterly what he thinks of growth diagnostics approach as it basically addresses almost all the concerns he has with previous growth studies. Before I could raise my hand, someone from the Woodrow Wilson Center asked him the same question I had in my mind. Easterly's reply: "It is a good approach but if your whole point is that policy effect differs in countries, it does not lead too far." First, if it does not lead too far, then it also means that it leads somewhere, usually in the positive direction; it might not lead way too far on the expected long run growth curve but it does lead in that direction in the short run, which is what policymakers are concerned with. I was expecting a stronger and weighty response from Easterly. Got disappointed :((

Here is Rodrik:

I know from my own experience that there is still a lot that we need to learn about how to do this right. To those who say that the framework is difficult to implement in practice, my answer is "right, it is indeed hard to determine policy priorities, but this approach at least forces you to confront those difficulties in a systematic way."

Instead, Easterly said what he always says: "outside experts cannot figure out what drives growth... rather than experts from HKS and NY, local people know it better"--its a classic Easterly bash. I like his work but feel uncomfortable with his strong inclination to Hayekian principles. It reminds me of this paper where Samuelson and Hayek debate (not through emails but snail mails and publications) on the 'inevitability thesis' (a summary of the paper is here).

The event was organized to launch a book "What Works in Development?: Thinking Big an Thinking Small". The book has contributions from a range of eminent development economists. You can find papers presented at the Brooking Development Conference here.