How do you make decisions and manage resources in the face of complexity? It’s a tough nut to crack. Owen Barder recently wrote several blog posts on the implications of complexity theory for development. The whole series is highly recommended reading. I could write several more posts just highlighting all the great insights there, but instead I’m going to focus on my one gripe. (Cause I’m a blogger, and that’s just how we roll.)

First, a short and totally inadequate summary

The overall analytical framework of Barder’s posts resonates deeply. In short: Development is an emergent property from a complex adaptive system encompassing political, economic and social factors. Complexity theory implies that development policy can’t actually create development. Policymakers should instead take a more humble approach, emphasizing dynamic properties like experimentation and feedback. Barder laid out the basis for a “complexity-aware approach” to the results agenda that recognizes the need for adaptive management of complex interventions.

(Got it? Seriously, just go read his whole series — starting with What Is Development?)

But something is amiss…

Despite all the parts that I agree with, something bugs me about the third post in the series (co-authored with Ben Ramalingam). When it came time to lay out the practical implications of complexity theory, Barder and Ramalingam pointed to two ideas that the Center for Global Development (CGD) is supporting: Cash-on-Delivery (COD) Aid and Development Impact Bonds. It felt like a bait-and-switch.

I’ve written previously about my discomfort with COD aid. I think the approach treats a developing country government like a contractor. It increases the government’s accountability to the donor in a way that could greatly undermine whatever democratic accountability may exist between the government and its citizens. In response to my previous post, CGD’s Bill Savedoff argued that the transparency components of COD aid (e.g. publicly sharing the contract) will at least be better than current practice. That’s true. But I remain unconvinced that those components are inherent to cash-on-delivery. If donors and recipient governments were willing to agree to a real level of transparency, those components could be implemented as part of current practice.

One metric to rule them all

In my mind, the defining feature of COD aid is the use of a single metric to trigger disbursement. Before the contract is signed, the various actors involved must agree on a narrow, clearly defined, and measurable metric of success. Then they stake large amounts of money on that metric. This approach locks in the result being sought. The target can’t be changed later, or the whole process would lose legitimacy.

Development Impact Bonds (DIBs) take the COD aid concept a step further by bringing in another set of actors: private investors. Under COD aid, the developing country government fronts the resources needed, while expecting to get cash down the line. Under DIBs, the upfront resources come from private investors, who would then receive payout at the end. The accountability chains are a bit different in each and it will be interesting to see what role the private investors actually play, but the importance of the chosen metric remains the same in both approaches.

Embracing complexity… from a distance…

This is why I called it a bait-and-switch: pegging large sums of money to a single metric is the opposite of engaging with complexity. Rather than embracing complexity, these approaches seem to be avoiding it.

Barder and Ramalingam gave a few reasons why they think COD aid and DIBs are good complexity-aware approaches, including the following:

[COD aid and DIBs] avoid the need for an ex ante top-down plan, log-frame, budget or activities prescribed by donors. Because payment is linked only to results when they are achieved, developing countries are free to experiment, learn and adapt.

There is no attempt to follow money through the system to show which particular inputs and activities have been financed; it is important for governments to learn about whether certain activities are working, but it is futile for donors to speculate about the extent to which those changes would happen without them.

Reading these reasons, it seems like the benefit of these approaches is to absolve the donor of dealing with the complexity. Unfortunately, these approaches offer no guidance to the developing country government and their partners who are implementing under these frameworks. The work itself becomes no easier. Sure, the donor avoids a top-down plan, log-frame, etc. and no longer has to monitor activity as closely, but someone still has to plan and manage the efforts. Someone still has to budget, to put financial controls in place, to hire and evaluate staff, to ensure the effort achieves the pre-defined result — all within a context of complexity.

Adaptive leadership

Barder and Ramalingam offered general principles for what they call “results-enabled adaptive leadership” drawn from evolution theory: variation, an appropriate fitness function, and effective selection. In other contexts, they’ve written more about adaptive leadership (e.g. see this one from Ramalingam). I agree with most of what they have to say on the topic. Where I part ways with them is on the idea that COD aid and DIBs create the necessary environment for this type of leadership. I don’t see it. The agencies operating under one of those agreements would be free from certain donor constraints and reporting requirements, but they would also be locked into the One Big Metric with no flexibility in the goal. I’m not sure that’s a net-positive in terms of complexity-awareness.

My critiques aside, I appreciate the effort to marry complexity thinking with the results agenda. I share Barder’s conviction that the two can co-exist harmoniously. However, I think the correct approach is to inject complexity into our understanding of results, rather than to become rigid on the results and kick the complexity down the line.


Interested in complexity in aid? Then you should definitely be reading Ben Ramalingam’s blog, Aid on the Edge of Chaos. You can also check out other complexity posts on Find What Works.

  1. Dave

    Many thanks for this thoughtful post.

    As I read it, you are making two related points. First: cash on delivery aid sets a single objective, and this uni-dimensional approach ignores the complex nature of what we are trying to achieve. Second: the donor does not provide guidance and support to the partner country on how they can manage the complexity they face.

    On the first, for me the key point is that cash on delivery should as far as possible be linked to an outcome rather than an input or process. While there may be a great deal of uncertainty about how outcomes are achieved, there is much more consensus about what those desired outcomes should be. So this approach does indeed step over the marsh of complexity onto the solid ground of agreed goals. That’s a feature, not a bug. (I would nuance this by saying that in an ideal world I would like to see donors offer a menu of ‘cash on delivery’ offers from which partner governments can choose the relative priority of different outcomes they will work towards).

    On the second, let us distinguish between, on the one hand, offering guidance and support to a partner who wants it to navigate complexity, and on the other hand building the details of that support into the aid agreement itself. Certainly developing country partners may want to draw on help from others to implement the policies and programmes that enable them to deliver the envisaged services. But that should be on the basis of getting the help they actually need, and being able to adjust as they discover successful approaches or encounter obstacles. It is a disadvantage of conventional aid relationships that the entire plan – inputs, activities, milestones, outputs etc are all designed from the outset and written into the aid agreement. So I don’t accept that Cash on Delivery aid means abandoning developing countries to solve their own problems: instead I think it means solving those problems together as partners rather than as implementers of a predetermined master plan. So this again is a feature not a bug.

    I think we agree that it is possible to marry complexity and the results agenda – indeed, I would say that a focus on results is indispensable in a complex world. The ideas of Cash on Delivery and Development Impact Bonds are nothing more than hypotheses that deserve to be tested, both as tests of the ideas themselves, and as ways to explore the more general principles of focusing aid relationships more clearly on independently audited outcomes and creating space for experimentation.

    Thanks again for a very interesting post.



    1. Owen, thanks for your reply. Your first point, about the single objective, makes me realize that I didn’t state the case as clearly as I could have. I certainly agree with your emphasis on outcomes rather than inputs/process. However, I disagree with your claim that there is consensus on the desired outcomes, especially when you consider the relative priorities among outcomes.

      Political discourse in any country reveals sharp disagreements over the outcomes people want on dozens of issues. Environmental protections, increased health care access, educational attainment, economic growth, etc. We constantly argue over how much we want of each of these and whether they are worth the trade-offs. The consensus you cite may exist among donors, development policy experts, and even recipient government officials — but if the people in the country don’t agree, then that’s potentially a problem. We know that the interests of government officials are not always the same as the interests of their citizens. To give a small hypothetical: A government that relies on the support of teachers’ unions might be keen to expand educational access (more schools = more teachers = more supporters) in exchange for donor money, but it would be less interested in committing to higher levels of educational achievement (which would require actual accountability for teachers). What if the country’s citizens prioritize increasing achievement over expanding access?

      That brings me to the issue of incentives. I think you underestimate the potential dangers of getting incentives wrong. There are countless examples of what can go wrong when you incentivize one outcome over others without room for flexibility. Keeping with the education theme, the No Child Left Behind Act in the US punishes or rewards schools based on test scores, to the exclusion of other outcomes. Not only did Americans discover that assessing students is harder than it sounds, but we also learned that an obsessive focus on test scores can undermine real learning, squeeze valuable extracurricular activities out of the budget, and create incentives for otherwise honest teachers and administrators to cheat. Get the incentives wrong in a complex system, and you never know what might happen. (Sorry that I don’t have any examples from British politics!)

      If you combine these two elements — lack of consensus on outcomes and the effect of incorrect incentives — then you have to be concerned that a COD/DIB contract could create large incentives for the government to do something other than what the citizens want it to do. That’s the point I’ve been trying to articulate on democratic accountability. Having multiple goals and changing priorities constantly is a feature (not a bug, as you say) of complex adaptive political systems. Locking in on one outcome that can’t be adjusted democratically works counter to fostering the kind of accountable institutions that will ultimately be necessary for enabling the emergence of development, as articulated in the “golden thread”. Transparency mechanisms in COD aid might actually reinforce this problem, as they only enable citizens to hold their government accountable for one specific thing: fulfilling the contract obligations.

      On your second point, about offering guidance on managing complexity: I’m not actually as worried about this issue. This help could come from a partnership with the donor in parallel with the COD contract, from the private investors who funded the DIBs, from an independent capacity building organization, or from an entirely separate source. There’s no reason to bundle this with the COD aid. My stance is simply that the challenge of managing complexity is 80% of the problem; even if COD/DIB work as intended, they only address the 20% of creating an enabling environment.

      As I’ve said before, I’m happy to see pilots moving forward. I have my doubts, but I definitely agree that these are ideas worth testing. If they succeed in spite of my concerns (or even better, if my concerns can contribute in a small way to making the pilots more successful), then the first round is on me.


      1. Owen, Dave,

        I think the issue is complex, and it is dangerous to focus on only one way to do go for it. For the reasons Dave is explaining, it might be a bad idea to go for the Big Goals, which are difficult to measure and often there is no consensus, nor on the evidence, indicators, nor goals itself.

        However, if we would choose less ambitious goals in sectors where there is evidence, it might work as Owen intends it to work. Diminishing child mortality from pneumonia and measles. We know what it takes, in general we know how to do it. COD could work. By taking lower level, less political goals, a lot of problems mentioned by Dave just go away. And what with the higher level goals? Perhaps it is too complex to have a single long term way forward yet?

        An aspect of complexity approach which is probably more important than defining the long term goals is how you work.

        I notice that projects and programs, national or donor funded, follow in general a rigid log-frame approach. So if we would go for COD, probably the country itself would approach it in a static way. This means there would be no advantages addressing complexity from this approach. I notice that there are modern project management systems out there ( I love the scrum approach), but they are not too often used in development, as donors ask for static log-frames.

  2. […] Complexity theory, adaptive leadership, and cash-on-delivery aid: one of these things is not like th… ( […]


  3. […] Complexity theory, adaptive leadership, and cash-on-delivery aid: one of these things is not like th… […]


  4. […] Barder and Ben Ramalingam describe cash-on-delivery aid as a complexity-aware approach; though I disagree with them on whether COD aid qualifies, the point here is that complexity thinking can change the way we address problems. Another […]


  5. […] making the case for doing development differently, must involve engaging with the results agenda, injecting complexity into our understanding of results as Dave Algoso puts it, and shifting the focus to results that are useful for learning and downward […]


  6. […] making the case for doing development differently, must involve engaging with the results agenda, injecting complexity into our understanding of results as Dave Algoso puts it, and shifting the focus to results that are useful for learning and downward […]


Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: