If you work in governance, or for that matter in any area of aid in where success depends on a functioning government, and if you only have time to read one academic paper this year, make “Capability Traps? The Mechanisms of Persistent Implementation Failure” that paper.
The CGD working paper — written by Lant Pritchett, Michael Woolcock and Matt Andrews — isn’t perfect but nevertheless makes two important contributions to thinking about governance and development; contributions to which practitioners should pay heed to.
The first of these is that quality of governance very rarely improves rapidly. Contrary to the predictions of the Modernisation theorists of old, most of the world’s poorest countries aren’t set out on an inevitable journey towards well-run states. In fact, most of them are stuck: Pritchett and his co-authors (henceforth PWA) provide empirical evidence suggesting that in almost all of the world’s poorest developing countries governance is improving very slowly if at all.
PWA also contend that external agents such as aid agencies have done little or nothing to help developing country governance become unstuck. We may have tried but, contrary to the assumptions of much aid work, our efforts in this area rarely achieve dramatic results. Donor efforts in conjunction with the actions of domestic actors do not, PWA argue, tend to lead to substantive improvements in quality of governance. Instead they often end up reproducing the formal structures of governance found in developed countries, but with little of the function. PWA call this Isomorphic Mimicry and this is the second key insight of the paper. It may be easy enough to create something that looks like a Ministry of Health, but it is much harder to create an entity that interacts with on the ground realities (social structures, norms, capacity etc.) in a way that actually sees hospitals and clinics run successfully.
There is more to the paper but these are the two standout points; the ones of critical importance to aid. And for emphasising and elaborating on these two points alone the paper is a must read. Yet, as I noted earlier the paper’s not without shortcomings.
One problem is that by using performance on a range of governance indicators as its metric of success, the paper arguably paints too grim a picture of development progress. Governance in many of the world’s poorest countries may not be improving rapidly, but at the same time, as Charles Kenny ably documents in “Getting Better”, in terms of human development at least, most of the world’s least developed countries are seeing improvements. Isomorphic Mimicry may not bring good governance but maybe it still helps states govern well enough to allow for some progress in areas such as health and education. And to eschew it altogether might be to send the baby sailing out along with the bathwater.
PWA’s paper is least convincing when it discusses potential solutions to the challenges of governance and alternatives to Isomorphic Mimicry. PWA’s preferred solution is Cash on Delivery Aid (CODA), the brainchild of other CGD researchers. CODA is a proposed approach where donors cease to prescribe to recipient governments how development goals are to be met, instead simply agreeing the goals and then giving aid; providing more aid to those governments that perform well and less aid to those that perform poorly. This system, rewarding good performance and punishing poor would, it is argued, incentivise success amongst recipient country elites while at the same time providing them space to be innovative in developing systems which succeed.
To me CODA seems simple, elegant, and almost certainly destined to fail.
Not only because it assumes that incentives and money are all that are required to elicit positive change, but also because it assumes that elites in developing countries actually have sufficient agency to respond to the CODA incentives. I think a much more likely outcome of CODA in the world’s worst governed countries is that, constrained by complicated problems of political economy, elites, even with incentives in place, wouldn’t be able to improve either governance or development outcomes at a pace more rapid than that already occurring. For their failure they would then be punished by decreased aid flows and we would likely see a system where the most needy countries receive the least aid. (For a detailed critique of CODA see here [pdf], for a sophisticated defence see here.)
In my opinion, a better way forward, bearing in mind the difficulties of good governance and the limitations of isomorphic mimicry, would be as follows:
- First accept that aid is a long run phenomenon. Forget about myths of exit strategies and handovers. We’re going to be giving aid for quite some time yet.
- Keep it simple. Where governance is weak, accept the fact that government can’t do all the things that it does in our own countries. Focus on those activities that really matter. And don’t try and do too much. If you do you run the risk of overwhelming systems of government (PWA make this point too). Also, focus on systems that achieve results with minimal coordination and as few steps as possible. If at all possible, avoid putting money into byzantine systems where success is many transactions away (like the PNG health service for example).
- Study governance at the micro-level. PWA’s macro study is instructive, but within countries some government departments work much better than others and some aspects of governance come more easily. We need to learn a lot more about why this is the case. Study micro-successes (and micro failures) and try and figure out why some things work and others don’t.
As far as approaches go, this has none of the tidiness of CODA but in the messy world of development I think it’s far more likely to help.
Despite these limitations the paper is still excellent though. In the imperfect world of global development work, this imperfect paper is probably the most important one you’ll read this year.
Terence Wood is a PhD student at ANU. Prior to commencing study he worked for the New Zealand government aid programme.