This is a joint post with Owen Barder.
In a recent blog post in the Guardian, Jonathan Glennie welcomes the new focus on results in foreign aid and has some good things to say about Cash on Delivery Aid (COD Aid) in particular. (We especially like this: “COD Aid demonstrates a post-ideological humility regarding how to achieve development.”)
In his article, but even more so in readers’ reactions below it, there are some questions about the possible effects of linking aid to results. We think that the commenters are right to raise these worries, because they are concerns that we share. But we believe that these concerns are relevant for other forms of results based aid, and to an even greater extent for conventional forms of aid, and that COD Aid would in fact go a long way towards allaying them.
Here are the main worries raised, with some reflections on whether and how they apply to COD Aid:
- When you set targets you create the risk that the recipient will get nothing. This is a risk for other kinds of aid, but COD Aid does not involve setting a target so it is much less likely in this case that countries would get nothing at all. COD payments would be made annually for incremental units of progress (e.g. $100 for each additional child that finishes school or passes an exam above a baseline or for each additional case of sustained malaria treatment). That means in the worst case scenario—no progress at all—the recipient gets nothing. But over a five-year contract period, the likelihood is that there is some payment, small or large, depending on the actual progress made. The approach encourages developing country governments to try new ways to make gradual improvements in service delivery and requires that they take responsibility for the new policies or programs involved. In contrast, existing aid programs do involve setting targets, triggers or conditions which can lead to a country losing an aid program altogether if a country fails to meet them, even though the target may have little relevance in reality to whether progress is made.
- If you focus on what you can measure, you may not focus on what really matters. The concern here is that in focusing on what we can measure (e.g. school attendance) we may ignore another result which is harder to measure but is what we really care about (e.g. quality of education). This is a risk every time aid is linked to anything at all. Most aid today is linked to inputs and processes (e.g. ‘increase education budgets’ or ‘buy X thousand textbooks’) or sometimes to outputs (e.g. ‘build more schools’) which are the result of a negotiation between the donor and the recipient. But those inputs and outputs are far removed from the impact we care about (e.g. ‘are children getting a good education?’). When aid programs focus on the points earlier in the results chain, we are even farther from and more uncertain about the impact that we care about. Furthermore, if we link aid to inputs and outputs it is impossible for developing countries to experiment and learn about how to make progress in their context, because their aid is linked to a pre-conceived ‘theory of change’. Linking aid more strongly with real world impact – such as children completing school and taking a standardized test – is more likely to keep us focused on what matters, while giving countries flexibility to learn how to achieve those results, than linking aid to inputs, processes and outputs. And by linking payments to development outcomes as closely as possible, COD Aid creates the incentives for improving the measurement of these outcomes.
- We may reward countries for delivering services but do not help them to build sustainable and accountable institutions. Responding to Jonathan’s post, Annahthomas writes that there’s a possibility that results-based aid “could force short term measurable interventions to dominate over longer term institution building”, a concern also echoed by other commenters. Too often in the aid system we demand that particular results are achieved with little concern for whether this actually helps a country increase its capacity to achieve those results for itself – a problem that does not only apply to results-based aid. That’s how we end up with policies which are ventriloquized by powerful donors, separate project implementation units, and a raft of conditions, benchmarks and triggers against which implementing organizations have to report. We think COD Aid could be a solution to this: it allows and encourages countries to find their own way to achieve results, adapted to local circumstances, thereby strengthening the national government’s own capacity. In this way, COD Aid is more focused on long-term development than conventional aid is. For more on this problem see a recent presentation on COD Aid here.
- By determining in advance what they will pay for, donors risk distorting the priorities of developing countries. A few commenters alluded to this concern in one way or another (e.g., “[targets] can end up skewing resources or efforts away from doing what’s best”; “[we’re] only funding what we want”). It seems inescapable to us that donors will have a view about what kind of things they want to pay for, regardless of the aid modality, reflecting the values and priorities of their taxpayers. It seems far less intrusive for donors to fund the results that they want, but not also simultaneously tell countries what they have to do to achieve them. With COD Aid, the country is free to decide for itself how to achieve results. We have also suggested that for a COD Aid agreement donors offer an open contract to any and all countries (only those who agree on the outcome to be paid for sign up), and that outcomes be linked to the Millennium Development Goals, which at least have the legitimacy of being an international agreement about key development objectives. Another possibility is that a donor agrees to pay for a menu of outcomes, which reduces the risk of distorted priorities. If experiments with COD Aid for one or another outcome prove a success, we think donors will be more likely to offer a menu of outcomes.
We don’t want to pretend that COD Aid is perfect, or that we know that it will work before it has been tried—the risks that the commenters on Jonathan’s Guardian article have identified are real and important. But we believe that these risks are already present everywhere in the aid system and that they are reduced, not increased, by linking aid to results as envisaged by COD Aid (as Bill Savedoff has also argued). Jonathan is pessimistic that aid will change very much, because of a donor-recipient dynamic which he regards as inevitable. Perhaps he’s right: but we’ve been impressed by the willingness of some donors to consider COD Aid as a way to make the aid relationship less damaging. The humility inherent in the COD Aid approach that Jonathan points out is an important reason why we think – as Nicholas Kristof and others have said – that COD Aid is worth trying, and why it represents a potentially big step forward for aid.