Try, learn, iterate, adapt

Lant Pritchett’s alternative road to development effectiveness

by Gastón Gertner*

Is my work at IDB so far from this road?

A few weeks ago, I sat at an Impact Evaluation Seminar organized by the Office of Evaluation & Oversight at the Inter-American Development Bank. Lant Pritchett delivered a provocative presentation: a sassy cocktail mix equally balanced in wit, irony and insights about development effectiveness. From the outset he seized everyone’s attention in the room with a simple question: is impact evaluation just a fad or a useful tool for development? His skepticism on the use of randomized controlled trials as a tool for development effectiveness draws on an array of different arguments, some new, others from ideas I have already heard in the past here.

Lant claims there are four problems with RCTs in how they lead to development effectiveness:

1.  Policy improvements in the world come out from national development (having markets in place and increasing state capacity). Since national development is impervious to RCTs, the use of experiments by themselves will not deliver development on its own.
2. There’s no empirical model of how the world works that guarantees that reforms and policies are put in place according to evidence found through RCTs.
3. Even when we know the right intervention for a certain policy, limited capacity in developing countries to implement these polices transform RCTs into a futile exercise.
4. Randomistas are wrong about how people, governments and agencies learn and change their behavior. Just because an RCT shows an effective intervention working, it doesn’t mean that it translates into the immediate implementation of this new policy.

In any case, and whatever your opinions are on these critiques, Lant Lant did not stop there. As promised, he continued his presentation looking forward, introducing an approach that shows how organizations that aim for development effectiveness should look like. His Problem-Driven Iterative Adaptation (PDIA) model puts the identification of problems and the quest for solutions at the center of development effectiveness.

And as he is going through the slides I’m wondering: “Wait, isn’t this exactly what we are doing at the IDB?”

Let’s break down the PDIA model:

1. Local Solutions for Local Problems

– Lant says: Don’t pretend to know the solution, start out with the problem, for example: how can we reduce crime? Increase children learning at schools?

– me at IDB: …I’m involved in a study on how to increase households’ take-up rate in connecting their bathrooms to the public sewage network in Santa Cruz de la Sierra, Bolivia. We plan to help the Government in Bolivia in testing “possible solutions” without claiming we know the answer.

2. Pushing Problem Driven Positive Deviance

– Lant says: …Consider contextual challenges; take small steps that facilitate positive deviations through an incremental approach.

– me at IDB: …(in the same sanitation project evaluation mentioned before), we are taking step by step interventions – a promotion campaign, use of incentives – that are assessed progressively, one at a time, to see how these mechanisms operate in making households connect to the sewage.

3. Try, Learn, Iterate, Adapt

– Lant says: …Use monitoring and rigorous experiential learning to allow on-the ground reality to shape the content of the policy

– me at IDB: …I’m part of an evaluation for a 2^nd phase of a community based nutrition program in El Alto, Bolivia in which continuous monitoring of the intervention will inform program delivery performance on the ground, making adaptation possible.

4. Scale Learning through Diffusion

– Lant says: …Bridge agents with power with those of ideas (for effective policy design)

– me at IDB:…for the evaluations I’ve been involved in, we dedicate time to build partnerships, disseminate our results in seminars, learn from what works and track for innovative policy designs at low scale to test them, while bringing in top head decision makers.

I remember when the panel ended, I left the room puzzled, filled with questions while craving for immediate answers. Writing this blog has been a useful introspective exercise to look for some of those answers.

Finally, I’m finding out my work here is already venturing into Lant Pritchett’s roads to development effectiveness.

* Gastón Gertner is an economist at the Office of Strategic Planning and Development Effectiveness at the IADB