If there is one thing that most of the donor community can believe in, it is this: aid gets better results in countries with better governance. The data linking aid effectiveness and governance go back to the work of David Dollar and Craig Burnside around 15 years ago. And though there have been many challenges to the original findings, more recent studies by Aart Kray and colleagues on the World Bank’s (WB) own portfolio confirm that over the past 25 years, WB projects have performed better in countries with better governance. But for most people, it’s not the data that convinces them of this simple, but powerful maxim; it’s just a matter of plain common sense. Or is it?
Something strange and unexpected has happened to the WB’s portfolio in the last few years. Since 2009, projects in fragile and conflict affected states (FCS) have out-performed projects in the rest of the portfolio as judged by both internal and independent evaluations. The share of “satisfactory or better” projects has been 5-10 percentage points higher in FCS versus non-FCS over a three year moving average. Of course, a few years of volatile project performance data are not enough to challenge one of our most deeply held assumptions about aid performance. But what is going on here?
Spoiler alert: I don’t have the answer, just a lot of potential hypotheses. Here are some that come to my mind. You’ll undoubtedly have others.
FCS projects are getting more staff time and resources. That’s true. Over the past few years, the rate of increase of WB staff on the ground in FCS and budgets for FCS projects is much higher than in non-FCS. We are now spending three times more of the WB’s own budget in FCS versus non-FCS for every dollar of IDA assistance provided. This translates into much higher growth in budgets for project design and supervision. But this also raises difficult issues of potential trade-offs, since in the flat budget environment of recent years, we are taking money out of the non-FCS portfolio to cover the higher implementation costs in FCS, potentially “robbing Peter to pay Paul.”
FCS projects are getting simpler in design and less ambitious in results. For years, we have been telling teams to stop over-designing projects. Keep it simple. Define results that can be achieved during the project’s lifespan. These lessons may have sunk in better in FCS, where they are so patently obvious. In the better governed countries, our ambitions may still get the better of us. Certainly, projects in FCS are smaller. We don’t have a simple measure of their design complexity or ambitiousness of their results. We should. This might also explain why better performing FCS projects with their less ambitious outcomes are not necessarily leading to better performing FCS countries in terms of poverty reduction.
Projects in FCS are implemented differently than in non-FCS. “Using country systems” is all the rage in the aid community. Clearly, it makes sense that if we want to build up government capacity for sustainable results, we need to route aid projects through government systems and stop using special project implementation teams (PIUs) and special fiduciary rules. Of course, WB projects go through government systems in all countries, including FCS. But we are clearly going the extra mile in those places with better governed country systems. Perhaps the more traditional approach to project implementation is getter better immediate results in FCS than in those countries with greater reliance on country systems. If this is true, it has important implications for the country systems debate.
Projects are evaluated differently in FCS than non-FCS. A more mundane, but equally plausible, hypothesis is that we are either less rigorous or perhaps more generous in evaluating projects in FCS. With weaker statistical systems in FCS, we have less data to monitor results. This could either lead to expected outcomes that are more difficult to falsify. Or given the higher risks in FCS, it might lead to greater willingness among project staff and evaluators to “look the other way” in FCS projects and to accept higher variation between the results expected at the start of the project and those ultimately achieved in light of the risks.
We need to get the bottom of this, since there are important implications of each hypothesis. And we have no idea if these recent trends will be maintained over time, so we need to understand what is driving them. There is also the question that though we may increasingly be doing projects right in FCS, we might not be doing the right projects to address the fundamental drivers of conflict and fragility, and hence poverty reduction, in FCS.
But regardless of what is driving these results, they raise questions about some of our most deeply rooted assumptions about aid effectiveness. Many important aid funds are allocated according to formulas heavily weighted on governance indicators. Many still wonder whether we can get results in FCS. Many still say that increasing aid to FCS is “throwing good money after bad.” Three years of volatile project performance don’t constitute a trend and we’ll need to know much more before we draw any conclusions for practice. But for now, let’s at least change the way we talk about these FCS countries. Let’s not assume that there is an inherent trade-off between poverty and performance in aid allocation. Let’s not assume that more aid to FCS will necessarily lead to poorer results. Let’s recognize that if we want to end extreme poverty, we need to be ahead of the curve to address the increasing concentration of the extreme poor in FCS. By testing the hypotheses above, we might even be able to draw lessons from FCS to improve the performance of all aid projects. Now that would be a switch.