Does complexity thinking have anything to offer the complicated world of aid?

It wasn’t the most pugilistic episode of Development Drums ever (that award goes to Episode 20, which saw Mushtaq Kahn growling at Daniel Kaufman that “India had markets when Europeans didn’t know what a bath was”) but Ben Ramalingam and Stefan Dercon’s debate on the usefulness of complexity science for aid was a tussle.

Ramalingam’s argument, from his book (which, mea culpa, I haven’t read), is that aid work occurs within a world of numerous variables which “interact in an intricate pattern”, so as to generate systems that are very complex, but not completely chaotic. In this sort of world it is still possible to understand social life through careful study (the science of complexity science, which Ramalingam espouses) and even to intervene for the better. But woe betide anyone who thinks the dynamics to be studied will be simple, or productively explained by simple models. And woe-especially-betide anyone who thinks the issues emerging from complex systems can be eased by simple solutions.

And that’s the problem, according to Ramalingam. The aid world currently spends too much time proposing simplistic, input-based solutions drawn from overly simple analysis of the problems we hope to ease. What  we need instead, he argues, is complexity science and aid practice which inserts the work we do amongst the complex social, political and economic dynamics of the places we work.

Dercon (who is DfID’s chief economist), however, was having none of this, offering an assertive defence of aid and the economic analysis that underpins much aid work. And he was an adroit combatant: delivering his defence from several directions at once. First, he argued that Ramalingam has constructed a straw-man world of aid practice, and that, in fact, given the constraints they work amongst, the typical aid worker (or project) does a better job of incorporating the complex world they work within than they are given credit for. Then he belted out a defence of economics: when economists need to understand complicated systems they already model in sufficient complexity, even if they don’t call the tool they use isn’t called complexity science. Then, having done that, he tacked back to argue that, actually, parsimony often produces better understanding than incorporating everything anyhow.

So, who won the debate? I don’t think either did.

Ramalingam did at least convince me to read his book. And he didn’t have to convince me that we need to situate the problems aid seeks to solve amongst broader political economies (I’m an aspiring political scientist, after all). But he failed to convince me that complexity science would be necessary for this (or any better than economics, political science, or political economy analysis). But at the same time Dercon wasn’t that convincing in getting the world of aid off the hook. Good aid workers do take context into account, and good aid programmes do fund research into context, and some of this (particularly the work DfID has funded) is very good. But a lot of aid work (definitely in rhetoric, and to a lesser, but still real, extent practice) does still have a strong, linear element of “input leads to output” type thinking associated with it.

The trouble for Ramalingam is, I don’t think the cause of this simple linear thinking is to be found in (so simple and obvious a place as) the beliefs of (most) aid workers. Rather, the culprit, aptly enough, is the context of the aid endeavour itself — specifically the domestic context of aid giving. Here the culprit is, sometimes, beliefs, but these are the beliefs of people who make donations (the ideas that evoke wallet-opening) and not those of aid workers. At other times the culprit isn’t ideas at all but rather is simple political economy: politicians hate scandals and the risk of aid work going visibly wrong discourages innovation while also suppressing learning.

We give aid in a complicated world, but before anyone can convince me we need to need a whole different type of social science to understand it or to make it work, the first thing they would need to show me is that we have fully exhausted study of aid and development using conventional tools, and in the case of aid in particular, that we had carefully studied why we give aid the way we do, and the impact this has on aid’s efficacy.

Terence Wood is a Research Fellow at the Development Policy Centre. His PhD focused on Solomon Islands electoral politics. Prior to study he worked for the New Zealand Government Aid Programme.

image_pdfDownload PDF

Terence Wood

Terence Wood is a research fellow at the Development Policy Centre. His research focuses on political governance in Western Melanesia, and Australian and New Zealand aid.


  • Hi David, I think your comment is a bit dismissive of the potential of complexity, and just wanted to come back to share my thoughts on your points:

    1. Some of the best complexity work in development has utilised models AND participatory processes – a number of which I reference in my book, and have been involved in developing. IDS is doing some especially interesting work in this area, led by Danny Burns, which has informed, among other things, the Participate programme:

    2. Some of the most powerful applications of complexity thinking have come from the private sector – use of agent-based modelling in strategy, system dynamics in supply chain management, network analysis in marketing and social media, etc. When I worked with DFID to apply some of these approaches in their programme on wealth creation, most of the people who came on board to support the work had long experience in the private sector.

    3. Complexity science is being utilised in the West – and there many examples of programmes and funding streams. It’s not mainstream, granted. But neither is it non-existent. To take just one example, the OECD Science Group published this paper in 2009, and has continued to develop this area:

    4. You are right that Western countries didn’t need complexity to help their development. But taking this to its logical conclusion leads one to a rather odd place (a) we should only be using things in development that worked in the West and (b) we shouldn’t have any methodological or theoretical innovation that is specific to the challenges we face in development. Moreover, if you are going to apply this particular empirical criteria to the entirety of development policy and practice, what do you think would be left?

  • The positives of the idea of complexity in development, I think, are largely therapeutic. We get so bound up in our Newtonian model of the universe, that we come to think of our billiard-ball “logic models” as actually real. We start to fetishise them, doing what they tell us, and ignore the real-world signals on the ground.

    The negative aspect of complexity is that it proffers yet another mechanistic model, to distract workers from the real interactions and information that’s available through deep engagement with the operating environment, of which 99% can be summed in this word: people. The image, of workers scrawling equations on a blackboard, encapsulates this negative.

    1. It suggests to aid workers that they respond to failure by building yet another layer of modelling, so that the workers interact yet more with models than with real people and with real situations. This is nirvana for researchers and for people in HQ, because models are their stock in trade, and if this approach to aid is furthered, the value of their stock goes up.

    2. At the same time, private enterprise is moving AWAY from these mathematical models, and towards deeper engagements with their customers, suppliers, and partners as the best way to win. Oddly, this is what aid has always been fairly good at: engagement. It would be sad at this time if we started to lose our strengths in favour of approaches that other social institutions are abandoning (and with good reason, in my view.)

    3. From a moral perspective, I think it’s dubious to sell to the world’s poor products (complexity, in this case) that the world’s rich are so far refusing to buy. If we’re going to see if this stuff works, let’s experiment with the OECD middle class first, not with the poor in foreign lands. When I see complexity being used to ACTUALLY solve some of the West’s manifest social problems, then I think it will be okay to export it.

    4. There’s also an empirical problem. If development means the social changes we have seen since the Enlightenment—public health, education, equality, democracy, wealth, increased life expectancy and so on—then we in the West have achieved these without “complexity thinking”. The claim that “complexity thinking” is suddenly necessary for Africa requires explaining away a lot of empirical evidence (the entire history of development in Western countries to date) to justify why it’s all of suddenly necessary for Mozambique.

  • Dear Terence, thank you for a great article – I really enjoyed it. I did want to clarify a couple of points: I don’t believe that complexity science provides all the answers to the afflictions of aid, but that these broad families of approaches / methods are proving useful and valuable, and are a worthy focus for further experimentation and learning.

    I also don’t claim that it is individual beliefs that sustain the damaging simplicity of aid, but rather that these stem from aid institutions, namely the ‘rules of the game’ in aid.

    Thanks again, and would love to hear what you think if you do get a chance to read the book!


    • Hi Ben,

      Thank you for your comment, and clarification. In the episode of DD you definitely convinced me that complexity science is a useful tool in a a development intellectual tool kit. So we’re in agreement on your first point. (Although I would only view it as one tool amongst many and still be reaching for other tools first in many instances.)

      On the second point, my response would be: if the problem is institutional rather than idea-based, why do we need a new intellectual tool, rather than – say – political economy analysis of world of aid donors?

      Thanks again for taking the time to comment.


  • Great post and summary of their debate. Even having listened to it, I think this crystallized key points in a more coherent way than what I took away of my own accord.

    When you got to the last two paragraphs, discussing the “cause of this simple linear thinking…” I was expecting you to go in a different direction. Or rather, an additional one. The points about donor preferences and politician risk aversion are well taken, but…

    When you mentioned “political economy,” I thought you were going to go to a more structural point, about the contractors, NGOs, and industry interests who influence aid allocation choices (via, for instance, the Council of International Development Companies). There are famous examples like the shipping industry/agricultural industry/food aid “iron triangle,” but the issue definitely extends beyond that.

    On balance, it strikes me that the influence of these actors tends to move the system towards more simplistic interventions, which take less account of context (and more of US commercial interests). And it seems to me that the influence here would be large, given Congress’ need to (internally) justify aid in terms of national security interests, commercial interests, and *then* humanitarian interests – in that order (although I don’t know how they do it in New Zealand…).

    I’m curious to what extent you think these structural political economy barriers are relevant? Or perhaps I am off the mark, in which case I would love to hear more.

    • Hi Nathan,

      Great comment thanks. I completely agree with you on the influence of domestic vested interests on aid giving, and New Zealand provides some examples from recent years of the power of domestic interests to win out over altruistic aid (although in our case it isn’t NGOs or contractors so much as our own domestic business interests, and generally I would see NGOs as a healthy countvailing force to the pull of, for example, our farmers on how our aid is given.) I also think this influence does play a role in driving the simplification of aid. Although the primary harm it does, I think, it how it makes aid about what we want to give, not what is needed. And I also think that risk aversion and the simple sell are probably the primary domestic causes of aid given without adequate consideration for the complicated world we live in.



  • My 20+ years in development suggests Ramalingam is right about the prevalence of linear logic or, as he says “Newtonian thinking” in the heads of “aid workers”.
    I think you are right though when you suggest that the characteristics of the way aid works is possibly the deeper problem. My guess is that many aid workers learned how to think linearly in order to succeed in aid agencies (which, by the way, Ramalingam writes about).

    Ramalingam suggests this problem can be usefully combated through more understanding and use of the tools and theories of complexity science. I think he is right, but overly optimistic about what those tools will get us.

    Take a moment and ask yourself: when was the last time I read a good analysis of the linkage between the way aid agencies work and the observable problems in the delivery of development assistance on the ground?
    [I will give you a moment]
    [Do you need more time?…]
    Time is up.
    The last, and only, analysis I have read (besides the superficial stuff you can find in various McKinsey and BCG assessments of agencies’ operation and activities) was in 2005 – in a book by Elinor Ostrom, Clark Gibson and others called “The samaritan’s dilemma”, Oxford University Press.

    The team used institutional economic analysis – many of the tools developed by Ostrom and others associated with the Workshop in Political Theory and Policy Analysis at Indiana University – to examine how aid agencies are structured and operate and how that links to problems observed in the delivery of aid.
    I highly recommend the book.
    If you know of another high quality study on this topic, please let me know.

    Upshot: Ben’s conclusion that we need some new social science methods and thinking is surely right. I believe that we should be looking to institutional economics for those methods – not instead of complexity science, but, in addition.

    • Thanks April,

      That’s a great comment. And I think you may be right that new aid workers are acculturated into thinking into a particular – Newtonesque – way. That said, I still think that if the overarching political (or private suburban donor) incentives were different this would be much more easily remedied, and that good old fashioned social science would be sufficient for *much* of the subsequent intellectual task of change making.

      I completely agree with you that we have far too little analysis of how and why aid agencies work the way they do. I have a slightly biased view on this matter as my wife (who is a fan of Ramalingham’s work) is currently doing her PhD on the forces that have changed NZ ODA in recent years and as she can tell you, the literature is scarce (there are more books than The Samaritan’s Dilemma though). What I am not so sure of is why in the first instance this means we need a new type of social science to study how aid is given. Surely in the first instance we might just be able to try using the existing tools? Something we’ve done too little of to date?

      Thanks again for the comment


    • I really like April’s summary. Complex adaptive systems approaches (which were one of the frameworks that informed the IAD methodology used in the Samaritan’s Dilemma) are one way of improving our frameworks and theories. The problem with many social science methodologies, especially those prevalent in aid, is that they are not designed to understand and navigate complexity, but rather to reduce it.

      As Prof Ostrom put it in her Nobel speech: “To explain the world of interactions and outcomes occurring at multiple levels, we also have to be willing to deal with complexity instead of rejecting it. Some mathematical models are very useful for explaining outcomes in particular settings. We should continue to use simple models where they capture enough of the core underlying structure and incentives that they usefully predict outcomes. When the world we are trying to explain and improve, however, is not well described by a simple model, we must continue to improve our frameworks and theories so as to be able to understand complexity and not simply reject it.”

    • While a simpler analytical framework to Ostrom’s Institutional Analysis and Development (which Gibson, Andersson, Ostrom and Shivakumar used in the Samaritan’s Dilemma), I enjoyed ‘The Institutional Economics of Foreign Aid’ by Martens, Mummert, Murrell and Seabright. They apply the principal-agent model to the question of why aid agencies rarely achieve their goals, examining incentives from the donor agency, to consultants, to recipients.

Leave a Comment