Poor Economics – a rich new book from Abhijit Banerjee and Esther Duflo

May 13, 2011

Scepticism v invisibility; stop punching hippies; nutrition call; the US climate is changing; why hair shirtists are wrong; flood salvage; climate scientist rap: climate links I liked

May 13, 2011

So the world is complex – what do we do differently?

May 13, 2011
empty image
empty image

Spent yesterday discussing the implications of complexity theory for development (previous discussion on this blog here) at a seminar organized by the UKCDS, a body that promotes interdisciplinary research on development. It was totally gripping, not least complexity signbecause two of my gurus were there – Eric Beinhocker, whose brilliant book on evolution and economics, The Origin of Wealth, you absolutely must read (it took me three posts to review, here, here and here), and Robert Chambers of IDS.

The purpose of the day was not to bury old thinking (linear change, the economy as a system of static equilibrium), although Ben Ramalingam did that in passing: ‘the linear Newtonian model is staggering around the global stage like a mortally wounded Shakespearean actor’ – nice. Rather it was to delve into the ‘so what’, or as Eric Beinhocker put it, the journey of complexity theory from a ‘Sunday morning’ idea that shapes the way you see the world, to something that makes you do things differently when you get to work on Monday morning.

Here are some of the suggestions that surfaced in the discussion (other participants, please add the ones I missed):

First, the purpose of development interventions, whether by states, companies, civil society organizations or aid donors, must move from deluded attempts at ‘creation’ of development from blueprints, first principles etc to acceleration of the evolutionary process that drives development in the real world. One way to think about that shift is to look separately at the three processes that constitute evolution – variation, selection and amplification (see Beinhocker posts for more on that).

Variation: facilitate, encourage and if necessary, fund emergence of new ideas, institutions and approaches e.g. from private sector, CSOs. The legal system and other institutions can help or hinder.

Selection: Academics, media or NGOs can identify new variations, study them and spread the knowledge, acting as a lubricant in the selection process. Deliberately looking for outliers, both of success and failure – known as positive deviance – is one way to promote this within development institutions.

Amplification: Too big for most development actors, but they can do advocacy to larger bodies (states, companies) to replicate success.

Second, we need to ‘wallow in failure’ – Beinhocker’s description of the US military’s determination to learn from past defeats. Failure is the essential engine of evolution, as much as success. We need to admit it, study it, learn from it (but then stop funding it….). Lots of kudos for Engineers Without Borders for introducing a website where NGOs can discuss their failures.

Third, we really need to improve the sales pitch, starting with the word ‘complexity’. Geeks revel in using the word, but it’s toxic for politicians and normal people (not the same thing), who usually want simple messages and ‘what do I do’ checklists. Beinhocker talks to them about real world economics, evolution (at least everywhere except Kentucky), adaptiveness, resilience – anything but call it complexity.

We also need to boil down some simple rules of engagement in working in a world of complexity, equivalent to the US marines combat instructions of ‘take the high ground, stay in contact, keep moving and improvise the rest’. The Santa Fe Institute, which gave birth to a lot of this thinking, has a rather more sedentary set of rules – study whatever you want, but you must attend lunch at 12 and tea at 4, where you will interact with other members. Robert Chambers reckons that participatory, bottom up approaches are a perfect response to a complex world (he promised to write something on this for the blog – v exciting.) He also suggested the System of Rice Intensification as a model – a few basic rules, but the rest depends on responding to local conditions.

Finally, we need case studies – of success and failure, both of responses to complexity and what happens when you ignore it. Hopefully Ben Ramalingam’s forthcoming book, with the same title as his blog, Aid on the Edge of Chaos, will include a lot of these.

Back in 1997, Robert Chambers asked if the new physics provided ‘a deep paradigmatic insight, an interesting parallel, or an insignificant coincidence’ for development practitioners. He now believes the answer is that this is a new paradigm. Got a feeling I’ll be posting more on this in the coming months.

And just in case this is all too abstract, and because it’s a really cool video, here’s some bouncing jelly (jell-o if you’re the other side of the Atlantic). There is no way to describe this or predict the movements with linear, Newtonian or any other maths – we need to change our thinking.


  1. Agree very much about Beinhocker’s book, it stimulated my thinking more than any other book I’ve read recently.

    I’m working on climate change adaptation for coffee farmers and am now trying to put his ideas into practice – viz. to first generate variation; a multiplicity of ideas, tools, techniques that farmers can try out.

    Hopefully we can learn from what works and then amplify.

    But here’s the rub: this may take too much time, which we don’t have, because climate change is happening and having major effects now, e.g. Colombia.

    And doesn’t it kind of cut across what the rampant randomistas want?

    And how can you talk to a donor about wallowing in failure?

    Duncan: all good points (particularly about talking about failure!). In particular, I think it complicates the randomista approach – by emphasizing just how different and context specific evolutionary change is, it reduces the relevance of an RCT experiment in one place to what might happen at another place or time.

  2. Great, Duncan. Thanks for sharing this with those of us who weren’t there. It is all about the ‘so what’ question, absolutely. And dialogue about what is emerging is central to this in order to take a next step. Too much focus is on the data part of the equation – ‘as long as we know enough/more, we’ll be safe’. But data without working it is useless.

    Was the notion of expectations discussed at all? This is where the topic of your blog sits so very uneasily with the tightening noose of conversative interpretations of ‘value for money’, guaranteed results. Evolutionary change is anathema for the security of predictable results.

    Duncan: Good question Irene. I think evolutionary change hugely complicates attempts to do development by simply implementing linear, logframe type planning. But I’m not so clear on the implications for measuring success – after all, venture capitalists and the US marines, both of whom arguable use evolutionary models, both have clear success metrics. So it may mean measuring things differently, rather than invalidating measurement altogether.

  3. Duncan – great post. Dealing with complexity is something we need to be talking about much more in aid work including how we deal with in in our planning processes within aid organizations since we still lean heavily towards engineering approaches.

    One comment on evolution and selection. In biology this also implies de-selection or traits. If this also applies to aid then we need mechanisms by which unsuccessful approaches (and organizations?) do not survive leaving the better adapted approaches to thrive. In reality poorly adapted approaches and organizations can persist for a very long time despite evidence that they are not working.

    Part of this is a result of not having adequate feedback loops and common goals between those who pay for aid and those who benefit from it. Finding ways to close this gap could do a lot to improve the quality of aid, especially on the selection and amplification (or attenuation) part.

    Duncan: Thanks Ian, there are real difficulties with getting all Darwinian and urging development organizations to identify and drop failures. Firstly, there is a lot of disagreement about what constitutes success/failure, due to the lack of a clear equivalent of profit/loss that guides a venture capitalist (or victory/defeat in the case of the marines). But secondly, it runs very much against ideas of partnership and solidarity – INGOs are already heavily criticises for their perceived lack of loyalty and staying power in many relationships, and an evolutionary approach could massively exacerbate that.

  4. Following on from the last post – one could assert that organisations follow biological strategies already i.e. r and K strategies. The former are those that quickly exploit a new resource, e.g. an aphid that breeds very quickly (r) or the neo-con institutes that flourished under the globalisation era. The latter (K) are slow growing species like whales, three-toed sloths or venerable institutions like my own that are in it for the long-term.

    The problem is that thanks to the fads and fashions of funding, the K-strategist can go extinct before the funders regain their senses.

  5. Most academics (public health physicians, epidemiologists, sociologists, etc) consider “complexity theory” to be a pseudo-scientific construct that isn’t testable, reproducible, and lacks construct validity, particularly with regard to international development. It’s unfortunate that this nonsense is still being pushed around by the “development industry” in the UK.

  6. John – well sociologists in my experience are not the most scientifically minded folk I have met; and your list is quite short. You didn’t mention economists for instance – should they all just stick to linear models, efficient market hypothesis, general equilibrium theory and other assorted tosh that has got us into our present fix?

    If they had thought a bit more about complexity, maybe they would not have been so keen to hook together all the ‘black box’ economies of the world at millisecond response times with completely inadequate governance and expect it all to work perfectly.

    A good thing about Beinhocker’s book is that he gives them a good working over…and he wrote it before Greenspan’s astonishing admission: “I found a flaw in the model that I perceived is the critical functioning structure that defines how the world works, so to speak.”

  7. Thanks for the reply, Duncan. I find it intriguing that you assumed that I was ‘invalidating measurement’ when I was talking about expectations. Why is that assumption made – not only by you? Of course measurement is important but it is about doing it differently and having different expectations of what it tells you. I was mentioning the tension with guaranteed, predictable results, not about not measuring. I believe these automatic assumptions which occur on both sides of the measurement fence make open discussion difficult.

    Thanks again for an insightful blog!

  8. Hi duncan – really happy to see you engaging with this debate. We are trying to deal with the complexity issue in a study in Kenya by using action research methodology and continuous feedback loops to frame and reframe what we are trying to change. The idea behind this is to move the team from working at compromise to working collaboratively and take us into an iterative change model that tries to take into account the complex dynamic arena in which we are working – not sure how we are doing yet but we are trying!!

Leave a comment