Can complex systems thinking provide useful tools for aid workers? Draft paper on some DFID pilot experiments, for your comments
Ben Ramalingam, who wrote last year’s big book on complexity and aid (Aid on the Edge of Chaos) has been doing some interesting work with DFID and wants comment on his draft paper (with Miguel Laric and John Primrose) summarizing the project. The draft is here BestPracticetoBestFitWorkingPaper_DraftforComments_May2014 (just comment on this post, and the authors will read and reply where necessary, and make sure any non-bonkers comments are reflected in the final version).
The project tries to answer a thorny question for complexity wallahs. Can the standard research tools for studying complex systems provide a useful toolkit for aid agencies, or is that an oxymoron? i.e. is the whole point of complex systems that you can’t have standard approaches, only connected, agile people able to respond and improvise?
If the answer is ‘oxymoron’, then we may have a problem in terms of thinking about and navigating complexity. But there is an issue if the answer is ‘yes’, as the ‘toolkit temptation’ is deeply rooted in the aid and development sector often unhelpfully, and could lead to unhelpful and damaging ‘complexity silver bullets’ and tick boxes.
Ben and colleagues reckon that this circle can be squared, and that the right complexity toolkit:
‘can help navigate a middle ground in the face of complex problems: to ensure development professionals neither have to surrender to uncertainty on the one hand nor construct convenient but false and potentially unhelpful log-frame ‘fictions’ on the other.’
To explore this, they have been trying out some complexity tools, using various DFID wealth creation programmes as guinea pigs (pilots). The pilots were trade facilitation and girls’ empowerment in Nigeria, private sector development in DRC, and a cross-DFID programme management review.
The Nigerian trade programme gets the most coverage in the paper, and exemplifies some of the issues. The aim of the programme was to improve the volume and value of trade. Trade is a classic complex system, full of feedback loops – e.g. if an activity becomes more profitable, and there are no significant barriers to entry, more entrepreneurs jump in. That means that linear analyses (reduce tariffs by X and trade will rise by Y) are often proved wrong in practice, because they fail to include the effects of feedback. So the researchers built a simulation of the trading system, and used it to run a simulation game for DFID staff to illustrate how tweaking different activities would ripple through the trade system.
‘The best way forward, short of trying to analyse and predict the system in advance – which is likely to be impossible, is to employ a portfolio approach: identifying possible entry points for interventions, launching multiple parallel interventions and learn in ‘real time’ to ensure the appropriate sequence and mix of activities. Indeed, the method is designed to support such an evolutionary approach to programming.’
The programme on girls’ empowerment used network analysis tools to explore the range of actors influencing girl power, and the interaction between those actors, producing an improved understanding of the ‘power ecosystem’ at play (the difference with traditional stakeholder mapping is an explicit analysis of the interactions between different stakeholders, rather than mapping them in isolation). The DRC private sector work did something similar. And I must admit I didn’t read the case study on internal DFID process reviews (can anyone blame me for that?!)
Across the four pilots, the research arrived at some findings that will ring true with many aid workers:
Lots of DFID staff are already working in ways that fit with a complexity approach, but ‘Examples of flexible and adaptable approaches in DFID were seen to happen despite corporate processes, rather than because of them.’
DFID staff want tools, mentors and permission to experiment.
They also want some way, at corporate level, for senior management to classify the nature of the problem – simple systems where traditional linear approaches are probably OK, or complex systems where they are not. ‘At present, [complex] problems are officially recognised as such only after several failed attempts to tame them.’
The researchers concluded that the benefits of a complexity toolkit include:
‘• ‘Getting inside the black box’ of the problems covered;
• Developing a sharper understanding how wider contexts shape and influence a given problem;
• Providing more sophisticated analyses of the potential causal pathways through which a change process might unfold;
• Bringing multiple perspectives together to broker a common understanding;
• Supporting the development of strategies to cope with inherent complexity, thereby giving a more systematic way of working towards ‘best fit’;
• Providing an analytical platform for experimentation and learning and supporting a more adaptive management approach with appropriate evidence-based tools.’
While I thought the paper was excellent, overall, I had to read it a couple of times to write this post – the language is often quite abstract, and I hankered for much more specific ‘so whats’, for example, what you would do differently compared to a traditional trade project.
My general take away is that the project focused at the analysis stage, helping us understand the nature and dynamics of complex systems. Even this is a big step forward on generating a log frame and blindly following it regardless of reality. But whether there are specific toolkits for what you then do, beyond ‘try lots of stuff, learn, adapt and iterate’ was less clear to me. Perhaps this should be the focus of follow-up work?
One other issue, this kind of approach seems to work best for the big guys (and gals) – organizations like DFID that are trying to influence an entire system, like a large chunk of Nigeria’s trade. For small players like Oxfam, there would have to be more focus on how other organizations are intervening (absent from this paper), and which small, simple interventions are compatible with complex systems (see the ‘responding’ bit of my recent post on complexity and small island states).
Anyway, over to you. ODI, which is publishing the final version as a Working Paper, and DFID, which sponsored the work, are interested to see if posting the draft on this blog produces useful comments and feedback – please don’t shame me up/let me down!