In a recent and very interesting article, William Easterly tried to put an end to his debate with Jeffrey Sachs’ claim that aid would end poverty as exemplified in the Millennium Villages project in Africa, where the hypothesis is that all problems of poverty have discrete technological fixes (for example, bed nets to deter malaria-spreading mosquitoes).
“… Sachs’ promise that aid would deliver an end to poverty wound up being convicted to a lesser charge: shoddy evaluation policies. To declare that experiments had been a success, any positive trends in the Millennium Villages should have been measured against Africa-wide trends … But because of the way Sachs set up the project, the comparison could not be done reliably”.
We now know two things: (1) In spite of technical fixes to specific problems, poverty persists in the Millennium Villages and; (2) Notwithstanding, we cannot judge the success or failure of the Millennium Villages without a proper evaluation. Since these multi-faceted interventions were not set-up with rigorous evaluations, there is no estimation of the counterfactual or how the fixes can be measured against what would have happened in the intervention’s absence.
But the heart of Easterly’s article, and this is a message that all development practitioners, myself included, should heed; is
“… If today’s development economists talk only about what can be tested with a small randomized experiment, they confine themselves to the small aid conversation and leave the big development discussion to others, too often the types of advocates who appeal to anecdotes, prejudice, and partisanship”.
Development economists must not exit the big development discussion given their microscopic focus on specific interventions. Instead, they should seek to evaluate where there are binding constraints to development.
In conducting experimental or quasi-experimental evaluations, we usually study very specific problems and link specific solutions to see how our proposed intervention fared in improving them.
Sachs’ interventions were multifaceted, attempting to address poverty comprehensively at the village level.
Poverty does not exist in a vacuum, it is complex and multidimensional. As evaluators, we must recognize this, and realize that any single evaluation of a given problem is only a small piece of the puzzle. We still need to know what works so that we may best use our development resources and be accountable for them – evaluation will always be a necessary part of aid-driven development.
The way forward then, is to prioritize what we evaluate. Often times, evaluation is demand-driven and is conducted where stakeholders, for varied reasons, happen to have an appetite for it. Instead, we should aim to evaluate where we know little, taking a strategic view in forwarding solutions to those issues that are most pressing and less understood on the grander development agenda.