I blogged earlier expressing my frustration at the excessive hold of the evidencariat on development thinking.
One variant of evidence fanaticism takes scepticism about what works to its extremes and disowns all priors. There is just no credible enough production function for the problem. Replication of any best practice is isomorphic mimicry. But this leaves them without anything to say about what can be done about the problem.
In this backdrop, the iterative adaptation model of tackling public policy challenges that Lant Pritchett and Co have formulated, Problem Driven Iterative Adaptation (PDIA), comes handy. So their proposition - we can tell you how to go about trying to solve the problem, but we cannot tell you what to do. But, as any practitioner worth his salt would say, this won't suffice if the objective is scaled up implementation or if the challenge is to design a program or policy to address a complex development problem. The practitioner needs to start with something baked up enough. Let me explain.
I am a strong believer in the model of Pritchett et al, especially in the implementation of programs. At some level, even without knowing it, many successful practitioners deeply internalise this approach when they try out new initiatives. They consciously iterate based on a minimum viable product (MVP) (generally arrived at combining knowledge on what has worked with contextual adaptations) as the initiative gets rolled out. The need for an MVP necessitates delving into some form of the production function.
And when we are talking of scale in a system with very weak capacity and poor median leadership, the ability to carry out such iterative adaptation starts looking questionable. So the MVP has to be far more prescriptive than is desirable.
The challenge with policy making or program formulation is even greater. The implementer, by being close to the cutting-edge, has the luxury of being able to string short and closely monitored feedback loops and iterate on multiple strands of the implementation model. However, the policy maker's flexibility in this regard is far limited. He is distant, his span encompasses several contexts, his monitoring and guidance bandwidth limited, and his own institutional capacity weak. He just cannot stand back and let the policy or program elements largely evolve iteratively.
One option then is to delegate or devolve finances and functions to provincial or state government levels. Let them figure out policies and formulate programs to address learning outcomes, poor primary health care, and poor nutrition levels. But two problems with this.
One, they too suffer from the similar span, bandwidth, and institutional capacity problems. Two, given the reality of weak capacity at provincial levels and the limited likely ability to iterate adaptively, such delegation could easily become counter-productive.
At a fundamental level, the binding constraint is clear. It is very weak state capacity. And approaches like PDIA, by their very nature, require strong states.
So, we need a PDIA plus to meaningfully engage with practitioners. The next step in the evolution of iterative adaptation literature should be the incorporation of the MVP. Ironically, the iterative adaptation model should include a basic version of the production function. Fortunately, there is a lever already available. Iterative adaptation literature talks about positive deviances. The learnings from these positive deviances should help construct an MVP.
An illustration with respect to education is available here.