Review of Aid on the Edge of Chaos
By Ben Ramalingam
Ben Ramalingam’s book applying complexity theory to aid is wide-ranging, well-researched, lively in style, and attractively presented. It challenges prevailing ways of thinking and doing. It deserves to be read and acted upon. The puzzle, however, is that the core ideas are actually quite familiar. That raises some big questions about the stickiness of the status quo and the barriers to change.
The problem Ramalingam addresses is summarised on Pg 361: ‘conventional aid conceives of systems and problems, behaviours, relationships and organisations, and dynamics of change, in highly abstract, idealised and simplified ways. These are . . . poorly matched to the reality of the world’. In brief, conventional aid is characterised by: simple, linear models of causation; reductionist analysis of comparative statics; and command and control approaches to management. Ever based your project plan on a logical framework? Or recommended a simple technical solution to a development problem? Or carried out a randomised controlled trial? Or planned an M and E programme based on outcome indicators? If so, Ramalingam has you in his sights! And there is no escape. The cases range widely, across locations and disciplines. Agricultural development in Bali? Malaria control in India? Humanitarian relief in Mozambique? Conflict management in the DRC? Sorry. Doomed.
The initial case is laid out over 12 chapters, six which describe how aid works and six which discuss the way the world works. These are followed by a further seven chapters, the ones I read most eagerly, which offer new approaches and ways of working.
The first part of the book takes aim at the simple narrative of development, in which cash, commodities or technologies are deployed to tackle single issue problems like hunger or disease. Best practices are avidly sought and disseminated, using a standardised, assembly line approach. Development practitioners are accountable upwards in well-established hierarchies. Their performance is measured by means of standardised indicators. All this, as Ramalingam remarks, smacks of ‘physics envy’ (Pgs 123ff). Taken as a whole, the conventional approach is ‘simple, neat and wrong’ (Pg 39).
The second part of the book contextualises the argument. It reviews the literature on complexity, covering systems, behaviours, networks and dynamics. Warren Weaver, Herb Simon, Jane Jacobs, Per Bak and many others act variously as the lodestars guiding the discussion – leading from the closed, static and linear to the open, dynamic and non-linear (Pg 142) – not just in aid, but in the murmuration of starlings, the life of a large city, the change in the climate, the Cuban missile crisis, and the global financial crash. Systems are complex and adaptive, connected in networks, frequently surviving on the edge of chaos. When change happens in non-equilibrium systems, it may be unexpected, sudden and drastic. Contradicting Darwin, evolution happens by jerks not creeps (Pg 219). In non-linear systems, like financial markets, minor disturbances are magnified in unpredictable ways with unintended consequences. Phase change can take everyone by surprise.
Is this critique of aid well-founded? Well, of course – at least as a critique of one strand of aid. I haven’t read a tenth of Ramalingam’s extensive bibliography on the science of complexity, but I do have to say that the core ideas were central to our research and teaching at IDSInstitute for Development Studies, Sussex from the mid-1980s onwards. Our principal debt, of course, was to Robert Chambers, whose book on rural development, published in 1983, first introduced many of these strands of thinking. It became commonplace to talk of ‘complex, diverse and risk-prone livelihoods’, of ‘process planning not blueprint planning’, of ‘baskets not packages’ in agricultural extension. Some of Robert’s phrases entered our domestic vocabulary: ‘we’ve made a mistake’, we would say, ‘ – good!’. We would ‘start small and grow’. Or think in a new way about the command to the firing squad: ‘Ready? Fire! Aim!’.
For me, this was a personal journey. I arrived at IDSInstitute for Development Studies, Sussex in 1981, having most recently been a farming systems economist in Bolivia. At the time, the default approach was based on uniform ‘recommendation domains’. It took some time to deconstruct and overcome such old concepts. By the mid-1980s, however, I was running the first extended rural development courses at IDS, pitching Robert Chambers’ approach against the logical framework, teaching mid-career professionals about process planning, and overseeing debates about the limitations of the World Bank’s top down and routinized Teaching and Visit approach to agricultural extension. A key reading on our courses was Tom Peters’ management handbook, ‘Thriving on Chaos’, published in 1987.
By the end of that decade and into the 1990s, I was involved in debates about the need to broaden definitions of poverty beyond income into multi-dimensionality. I remember chairing at least one debate between Robert Chambers and Michael Lipton on the epistemology of poverty: Ben Ramalingam would have enjoyed the cut and thrust. And by the mid-1990s, I was writing in a new way (for me) about food security, reading Gleick on chaos theory, deciding it relied too much on the order imposed by strange attractors, and finding solace instead in post-modernism. The art form of the movement, I remember reading, in Harvey’s book on the subject, was collage. This emphasis on mix and match led me, for example, to focus on self-targeting approaches to food-based safety nets, allowing people themselves to select from a menu of options. Eventually, I moved to ODIOverseas Development Institute (London) – but before I did had a management role at IDS. I well remember Robert Chambers stalking the corridors talking about ‘self-organised groups on the edge of chaos’. That, at the time, seemed rather challenging!
You didn’t need to sit a few doors down from Robert in the IDSInstitute for Development Studies, Sussex Octagon to imbibe these ideas. They did not all originate with Robert in the first place, as he would be the first to argue, and they spread widely through the development community. Furthermore, they continued to develop. Ben Ramalingam cites several relevant texts, including Bill Easterly on ‘searchers’ versus ‘planners’ (see my review here), Tim Harford on adaptation (see here) and Nassim Nicholas Taleb on Black Swans. There is other new literature, of course. I learned particularly from Ch 11 of Ramalingam’s book, which deals with phase transitions and other aspects of change in unstable, non-equilibrium systems.
New ideas notwithstanding, it is possible to discern a thread of continuity in development thinking which runs from the early 1980s to the present day, a period of thirty years. And yet, here we are in 2014, and the dominant development dialogue appears not be about understanding complexity, so much as about results and accountability: a narrative that seems to correspond only too closely to the model Ramalingam sets out to criticise. I’ve just read Michael Barber’s book, ‘Instruction to Deliver’: it is a fascinating account of implanting a target-driven approach at the heart of Government. In the development community, donors talk incessantly about national results frameworks, often derived from the global-scale Millennium Development Goals. Applicants for funding are required to submit logical frameworks and ‘theories of change’ in their ‘business case’. And evaluation privileges impact assessment.
Perhaps this is ungenerous. I wrote a piece in 2011 on competing approaches to results measurement, which I dubbed the Fordist or blueprint approach and the post-Fordist or process approach: Results 1.0 and Results 2.0. Ben Ramaligam is not alone in taking potshots at Results 1.0. My piece reported on a meeting in which Andrew Mitchell, then UK Secretary of State, took on critics who made similar points. It is necessary to report that he was not entirely unsympathetic, dismissive of ‘bean-counting’ in evaluation, and strongly supportive of an exploratory, risk-taking ‘venture capital’ approach.
Still, the debate rumbles on, and it’s not hard to judge who is on top. As I write this, Esther Duflo, randomista par excellence, has been nominated by Prospect Magazine as one of the world’s leading fifty thinkers. You can read my review of her book with Abhijit Banerjee here.
Why is this area difficult? Culture plays a part, and power: the status quo is hard to shift. I can see two other reasons. First, politicians and aid administrations have a legitimate interest in requiring both a strong business case before they fund projects and strong evidence of concrete results at the project end. They are, after all, accountable for public money, and have constantly to make the case for aid to a sceptical public. Philanthropic organisations in the world of impact investment are perhaps less accountable, but equally concerned with value-for-money. There is a nice cartoon in Ramalingam’s book (Pg 271), one of many specially drawn by Fran Orford, which shows the reaction when an agency leader dares to admit he ‘isn’t sure’: ‘is he mad?’, observers ask, ‘will you be resigning?’, ‘is this the end of your career?’.
The second reason is that managing risk, uncertainty and complexity are just plain difficult. What does the newly appointed middle-manager say to Robert Chambers when he calls for the team to be run as a self-organised group on the edge of chaos? I struggled with this at IDS, but also in a more senior position, as Director of ODI. In principle, fluid task cultures with flat hierarchies are just the ticket, and I was a strong advocate – but who manages recruitment, appraisal, team formation and personal development? There is an interesting discussion in Ramalingam’s book about networked organisations which do not respect hierarchy, with an illustration of an organisation in which the key player, ‘the unknown actor who played a central role in the network’ is a person called Cole, buried deep in the organisation chart, reporting to Williams, who reports to Jones (Pg 321). ‘Hm, I thought. I wonder what Williams and Jones thought about that. I hope they were confident enough to be positive. I fear they may not have been.
Examples like that help to open a discussion about how to manage organisations and programmes – and are why, as I said, I turned with special interest to the final seven chapters of the book. These are grouped together in a section entitled ‘the way aid could work’. The key themes - with specific techniques and research approaches to back them up - are systemic learning, adaptive strategies, networked organisations, and managing dynamic change.
Systemic learning is really about thinking systemically, taking a holistic and multi-disciplinary view, and adjusting the ‘level of zoom’ to combine analysis of the big picture as well as the fine detail. It is also about working with the grain rather than against it. ‘Listen to the system’, we are told, ‘to identify the spaces where change is already happening and try to encourage and nurture them’ (Pg 257).
Adaptive strategies is about learning-by-doing, holding on to ‘strategic agnosticism’ (Pg 295), and pursuing a strategy of ‘iterative adaptation’ (Pg 278). Games can be useful in this venture, and some modelling. There is a strong element of double-loop and even triple-loop learning in the analysis.
Networked organisations are where we find our old friend Cole, but also analysis (courtesy of Ricardo Hausmann and his colleagues in their work on the importance of networks in ‘product space’) of what makes industrial development happen. In networked organisations, Ramalingam concludes ‘the role of managers and leaders is not to come up with quick fixes for thorny issues such as structure or culture, but rather to accept the inevitable lack of control inherent in . . . organisational life’ (Pgs 325-6). ‘Agencies’, he says, ‘should aim to be a poised network, with some clear areas of focus and stability and some scope for flexibility and adaptation’ (Pg 325). I wonder whether a good place to look for success stories is in the large consultancy company, like PwC or McKinsey. I must ask them: are you poised?
Finally, beware the unexpected at times of dynamic change and focus on adaptation and resilience – the ability to maintain system functions in the face of stress and shocks. Diversity is a key feature, not imposed, but built from the bottom up.
Did I find all this empowering? Intriguing is probably a better word. I am certainly comfortable with the philosophy, as you might expect. I am also conscious that complexity thinking is a valuable antidote to some current tendencies in development. It is also a potential solution to some common problems. For example, I have been writing about the challenge of managing cross-Government approaches to solving global public policy problems like climate change, and have suggested that network approaches may offer a way forward. I still worry that the specific techniques may not be powerful enough to overcome organisational inertia and effect a cultural and political revolution. The good news is that Ben Ramalingam, and people like him, have put concepts like systems thinking, innovation and resilience on the agenda of development agencies. We will see how they get on.
Thanks for your review of the above. Thoughful and useful words.
You ask “Ever based your project plan on a logical framework?..... .Or planned an M and E programme based on outcome indicators? If so, Ramalingam has you in his sights! And there is no escape.”
(As an aside, ironic that ODI, through its advocacy work on outcome mapping, also re-invents wheels and oversells itself and in doing so potentially falls prey to being shot down through developing outcome indicators by “establishing a vision of the human, social and environmental betterment to which the programme hopes to contribute and then focuses monitoring and evaluation on factors and actors within that programme's direct sphere of influence.”)
Damning logframes is all very well. Ramalingam’s words mimic well-rehearsed critiques made by Chambers et al. They are also too sweeping in nature and smack of intellectual indulgence that lacks practical experience. I maintain that many of the logframe’s (and results chains and theories of change if the logframe’s simplicity offends) limits are due to mis-use. Yes they are linear and reductionist, but the method in developing and then revising them need not make them altogether redundant in yielding some intrinsic value. And what is the alternative anyway given their hegemony among those who finance the aid industry?
This is not the place to raise my own queries about Robert Chambers’ own treatment and understanding of M&E while being a beneficiary of his amazing work in the early 1980’s. Having said this, however, the bottom line is that writing books that peddles the ‘pop’ view is one thing, changing the bureaucracy of donors quite another. Do any of us have choices in a practical situation? Is not the challenge, therefore, to put to right the mis-use and limited understanding of the true intrinsic flaws to logframes? One of these, no doubt, is how an excessive focus on ex ante indicators (at any level, including outcomes) is the kiss of death to meaningful monitoring that enables us to be accountable to those we support and claim we can impact on?
In my view, good practice in log-frame formulation and M&E based on the log-frame and/or a theory of change gives the assumptions an importance barely less than the programme’s outputs and outcomes themselves. These assumptions, formulated on the back of discussions with those whose behaviours and/or decisions the project/program me seeks to trigger, begin to connect with the complexity to which Mr Ramalingam writes about. Monitoring processes can and should involve listening to and observing how different people respond to and interact with aid interventions. We make significant assumptions about these and it is these not the indicators we need to focus on. In my experience, doing this provides an antidote, albeit partial in nature and scope, to the complexities that surround the environments into which aid is put to work.
Mr. Matt Andrews from Harvard advocates a problem-driven iterative approach (dubbed PDIA) informed by the thinking and work of Rondinelli et al. This is not wholly incompatible with the logframe approach, indeed helps resolve ways it is mis-used. I am trying to use this in my work supporting the M&E of a health system reform programme in Timor Leste and a business support programme in Tanzania.
In sum, it would be useful for writers like Mr Ramalingam and Chambers to illustrate ways to overcome such limits; that is as opposed to just describing the water we all often drown in. In his sights and no escape maybe, but what is the consequence of this – simply disaffection?