The global financial crisis was not predicted by any economic model, not 10 years out, not 5 years out, not 1 year, 1 month or 1 minute out. Nearer the time some models did suggest some temporary minor blip and recovery predicted shortly thereafter but no more than this. The Head of the European Central Bank, Jean-Claude Trichlet, said of the GFC: “In the face of the crisis, we felt abandoned by conventional tools." In response, government and some researchers have begun to reexamine the foundations of traditional economic models and to ask how we can do better. For example, European nations, banks, and business are cooperating in building a major new type of complex systems simulation model of the European Economy, CRISIS.
Another example is the complex systems model of banking networks developed by Lord May and his colleagues published in Nature earlier this year. This showed how the banking network as a whole becomes increasing unstable as each individual bank diversifies its portfolio to spread its risk – a reasonable thing to do – and ends up with all being diversified in the same way. As a result, when a crisis hits they are all blown away.
Complex Systems Models (CSM) is a fast growing method revolutionising many sciences. It complements and extends existing methods. Other examples are the visionary EU funded framework project to build a complex simulation model of the world – all of it (www.futurict.eu); business systems (www.antoptima.com/site/en/index.php); epidemiology (www.medscape.com/viewarticle/709859_2). They can be vast, for example the Epicast model at Los Alamos has 300m agents, one for every person in the USA.
Economic systems are complex systems made of a multitude interacting people and organisations. Larger scale order and patterns of behaviour, such as movements in prices, incomes, trade balances and deficits, emerge in a self-organizing bottom up way from these interactions. And these large scale patterns are noticed and reported and have feedback effects on actions and interactions. Understanding and influencing the dynamics and evolution of such systems is not easy. Actually impossible. It calls for a total rethink of what we mean by control management, prediction and understanding.
We are all participants in a system we jointly co-create. Our actions and performance depend on much more than our own endeavours, they depend on what others do and how they respond. And how we respond to their responses ad infinitum. Overall behaviour and performance are not simple additive sums of the behaviour of the actors involved.
The challenge for managers and policymakers is not about the management and control of such systems but how to participate and manage in them. Government agencies and policy makers do not operate outside business networks: instead, they are part of them along with the other people, firms and organisations. No one is in overall control, though some may have more influence.
Unfortunately, past research and modeling of economic systems is dominated by non realistic models of behaviour, constructed for analytic convenience. They tend to assume highly egoistic behaviour (homo economicous), represente only average behaviour, and assume infallibility of markets (the efficient market hypothesis). Such models are of limited value in showing how real systems behave and evolve and how rare extreme events such as the GFC can emerge.
A different approach is needed, one that involves developing much more realistic and nuanced models of social and economic systems, like the ones I already referred to. They are called complex simulation models or agent-based models (ABM). We no longer have to make silly simplifying assumptions in order to make the maths and stats work out.
CSM are models formalized in computer code that could be written down as equations but that does no good as they are highly non-linear - way beyond any algebraic solution. The only way is to literally play out the rules over time and count what happens using a computer. This is why this approach is referred to as computational economics or social science.
If they are so good, why don’t we see more of them?
For one thing, mainstream economic and business researchers, the ones in charge of the top journals and grants, resist it. This is because it requires a new type of skill to build such models – programing and algorithmic thinking, which they are generally not good at. So CSM undermine the value of their established expertise – even though CSM complement and extend traditional research approaches; they do not compete with them. And the analysis of CSM models requires the same maths and stats skills they do have.
Another major factor holding back CSM is a lack of awareness and training in such methods in business schools and economics departments. Students get trained in maths and stats but not high level programming skills. However, there is good news, the programs to do it have become ever more friendly and freely available. One is NetLogo, which was started out as a way of teaching ABM to primary school kids! Surely, university researchers can learn to do it to? Aren’t they smarter than a fifth grader? Moreover, the very same methods are behind all the games kids now play on the computer, such as TheSims, SimCity, SimLife, SecondLife. Instead of developing playworlds we can be used the same methods to develop models of the real world in order to better understand it and study it, in ways that are impossible in the real world. In the real world we only have one history – our history – an effective N of 1. But in articifial worlds we can explore economic and business life as it could be.
To help develop and promote this approach to science in business, marketing and economics we held a one month intensive course on ABM and complex systems research, using NetLogo, at the University of Sydney Business School in July 2011 and we are currently developing new course options on CSM. Also, in late January 2012, we will be holding a high powered summer school and research workshop on CSM in business and biology. Where will it be? On Heron Island, in the Great Barrier Reef, in part because it is itself a wonderful example of a complex socio-economic and biological system. Check the website: http://sydney.edu.au/business/research/complexity.
Ian Wilkinson
Honorary Professor of Marketing, University of Sydney
Visiting Professor of Entrepreneurship and Relationship Management, University of Southern Denmark
August 2011(forthcoming, maybe, in TheConversation.edu.au)
No comments:
Post a Comment