Lee Crawfurd over at The Roving Bandit recently wrote a compelling post about why the question “Does aid work?” is fundamentally flawed.
The question “does policy work” is¬†jarring, because we immediately realise that it makes little sense. Governments have about 20-30 different Ministries, which immediately implies at least 20-30 different areas of policy. Does which one work? We have health and education policy, infrastructure policy (roads, water, energy), trade policy, monetary policy, public financial management, employment policy, disaster response, financial sector policy, climate and environment policy, to name just a few. It makes very little sense to ask if they all collectively “work” or are “effective”. Foreign aid is similar. Aid supports all of these different areas of policy. My colleagues and I at¬†OPM¬†work on aid-financed projects that support most of these different policy areas in different developing countries.
Lee, as he admits himself, is taking his cue from the combined work of Esther Duflo, Abhijeet Banerjee and Dean Karlan. Some policy questions are becoming more and more answerable: does X work in a given context is something that can be tested and applied. Lee is asserting that since aid is just used to fund policies, thus the question of whether aid works just boils down to the complex task of determining whether or not individual policies work.
Yet, when you examine this assertion a little further, it starts to fall apart. Let’s imagine we lived in a world where we, development economists from aid-giving countries, figured out all the good government policies. We knew exactly which actions developing country governments needed to take to save children’s lives, promote income and job creation, reduce hunger and conflict and so on. In the world which Lee has presented to us, policy works, because we knew what all the best policies were. Would aid work?
It certainly wouldn’t be guaranteed to. Aid isn’t just policy – it’s the transfer of financial resources and technical expertise from one country or entity to another. That transfer is inherently non-trivial: it can create huge differences in power, has the potential to distort the¬†recipient’s¬†decision-making, creates opportunities for rent-seeking and often is used for completely political purposes. We cannot, for instance, only judge a US-funded conditional cash-transfer programme in Afghanistan solely on its microeconomic impact – it has to be viewed within the context of the US’s ongoing military intervention in that country, and its likelihood of long-term success. The recent scrap between Rwanda and its donors over the security situation across the border in the DRC again shows that aid is vastly more complicated than simply choosing effective programmes.
Lee¬†acknowledges some of these differences and potential problems, but then dismisses them as things which are hard to gather robust evidence on. This preference to stick to what we know is somewhat admirable and tempting, but ultimately dangerous: it is incredibly difficult to gather widespread, robust evidence on the effects of aid on the macroeconomy or on local political economy. This should make us more, rather than less, wary of possible deleterious¬†effects.
I’m as equally horrified as Lee by the recent attention that right-wing attacks on aid have been getting in the UK, and I completely agree with him that aid cannot always be ¬†judged as a whole. The question, “does aid work?” doesn’t get us very far in life, especially since we have little concept as to what metric we should be using. That said, we have to tread carefully around the argument that small, neat questions are sufficient for success.¬†I agree that aid should be considered more carefully as a bundle of heterogeneous flows and relationships, but I also believe that “aid” is unique in several key ways, and it is only healthy if we continuously question whether the the things that make aid unique also undermine its effectiveness.