A shortened version of this article was published on the Guardian Development website on 28 January 2011.
Innovation, complexity and developmen
We are surrounded by complex adaptive systems: animal biology, the internet, complex ecosystems, competitive markets, and many others. These systems have many common characteristics. They are open and adaptive. They include their own sources of innovation—for example, through sexual reproduction—by which they improve and respond to a changing environment. The whole is greater than the sum of the parts, usually because of a rich combination of specialization and exchange (of genes, ideas, or services). Success is easier to explain in retrospect than it is to predict in advance. Successful systems adapt through feedback, which amplifies successful innovation and kills off ineffective change: for example, survival of the fittest allows good genetic variations and combinations to reproduce; in markets, bankruptcy eliminates firms that fail.
Systems like this are difficult to design or manage centrally. But we can nurture institutions which make them likely to thrive, while remaining agnostic about their destination. We don’t try to pick winners among firms, but we do set rules to make the market work better: preventing monopoly, promoting transparency, encouraging innovation and trade. We don’t try to design the results of human evolution, but most human societies have a taboo against incest because it undermines the genetic benefits of sexual reproduction.
Development requires the emergence of successful economic and social systems, so we might do well to look at what we have learned about what makes adaptive systems work. Yet the conventional wisdom about aid effectiveness is not easy to reconcile with the lessons from successful complex systems: it frowns upon the proliferation of donors; it promotes harmonization and activities aligned to a blueprint for development; and it implies a predictable connection between individual activities and outcomes. The development business lacks mechanisms which are common in successful complex systems, such as specialization and exchange, and feedback which drives out poor performance and scales up success.
Donors accept that they do not know all the answers. But this self-awareness is often not reflected in their day-to-day behaviour. They recognize too rarely how often their approaches do not succeed, and there is little reliable pressure which forces the development business to learn and adapt. For example, donors continue to spend about a quarter of all aid on capacity building, despite a mountain of evidence gathered over forty years, not least from their own evaluation departments, that their usual approach to capacity building does not work.
There are signs that this is beginning to change. As Madeleine Bunting reported on 17 January, the Canadian NGO, Engineers Without Borders, has published its first failure report. By setting out openly and honestly what has not worked, and drawing lessons from those experiences, the NGO has helped to strengthening the feedback loop. The UK Government’s promotion of aid transparency and the US Government’s recently-announced new approach to evaluation are also important and welcome steps towards nurturing a better aid system, not by yet another lurching change of direction but by investing in institutions which will enable the aid system to evolve.
One of the most striking—and to my mind unexpected—innovations of the past decade in development has been the growth of social protection, especially cash transfers, as a way to increase the resilience of the poor. The idea came from Latin American countries themselves. The idea has been reproduced across the developing world mainly because Progresa, a system of cash transfers in Mexico, was subjected to rigorous evaluation which demonstrated its effectiveness. Cash transfers are good example of how change can come about from innovation within developing countries, strengthened by rigorous evidence which enables the intervention to be scaled up and adapted elsewhere.
A quite different approach to encouraging innovation is the Advance Market Commitment. So far this has been used to promote the development and production of a vaccine. Previously, donors would select and subsidise companies to develop modified vaccines for use in developing countries and produce them at the scale required. Under that approach, vaccines such as Hib (Haemophilus influenza type B vaccine) were available in developing countries about 15 years after they were widely available in rich countries. Under the Advance Market Commitment, a group of donors instead guaranteed to underpin a more lucrative market for vaccines, on the basis of which companies could make their own investment in the necessary research and production facilities. The companies competed to bring their product to market quickly to reap the promised market reward. The result: in December 2010, the rollout of the pneumococcal conjugate vaccine began in developing countries within a year of its introduction in rich countries, and at a fraction of the price. Instead of trying to prescribe the solution, donors created an institutional environment in which companies had an incentive to find their own answer to the problem.
The Advance Market Commitment is an approach which has a wide variety of possible applications, from agricultural technologies to clean energy and climate change adaptation. But it also illustrates a broader principle: that we can do better by creating the conditions in which decentralised innovations produce socially-valuable outcomes than by trying to design and manage those results centrally.
As Ben Ramalingam pointed out the other day, the results agenda can be conceived in ways which are linear and formalistic, and so stifle the innovation and flexibility needed to achieve change. But done right (for example, by trying Cash on Delivery aid) the results agenda could respect the complex, ambiguous world we live and work in.
So we are making some progress – greater transparency and better evaluation to improve feedback loops are potentially important steps forward. But the development industry has a long way to go if it is to be a successful complex system, particularly in promoting greater decentralization, specialization and exchange.
Twenty five thousand people die each day of easily preventable diseases; and about a billion people live in absolute poverty. Between us we have the resources and the knowledge to put this right: what we lack is an institutional framework which puts those assets properly to work. Our best hope for accelerating development is to build the conditions which nurture successful complex adaptive systems: decentralised decision-making; a wide array of variation and innovation; specialization and exchange; and feedback loops which promote success and dispense with failure.