Economic Models Ignore Dependency and Irrationality - Is Agent Based Computing the Answer?

by Nic Plum on Tuesday 23 November, 2010 - 10:58 GMT

Posted in BusinessComplexityEconomicsModellingSystems Thinking

Tags: agenteconomicsmagazinemodellingnew scientist

Stock price ticker - what next?

There is a nice article in the New Scientist Magazine (30th October 2010) by Philip Ball that looks at economic modelling.

It starts out by looking at the popular cause of the current economic crisis, the bankers and the American sub-prime mortgage market but points out the error in simply blaming the bankers , the opaque financial mechanisms and inducements to sell:

it ignores the complexities of a system that led the initially small perturbation of the "sub-prime" mortgage crisis to morph into systemic collapse. Moreover, concentrating on the specific causes of that one event fixates on the past at the expense of the future. When the next crisis comes, the disturbance will probably ripple out from another quarter of the economy, taking us completely unawares

It then asks if there is a better way in which we could prepare ourselves

Instead of making unrealistically simplistic assumptions about human behaviour and the properties of markets, we can harness the number-crunching power of modern computing, coupled with our emerging understanding of the physics of complex systems, to rebuild economic theory from the bottom up. Extending that approach to the social sciences more generally could help us develop forecasting tools to assess a whole range of problems threatening human society

The trouble is that a lot of the economic models are based on theories that hark back to Adam Smith that are far too simplistic:

  • statistical models are fitted to past data and assume that markets always find stability left to their own accord - they never crash on their own
  • Individuals and institutions are assumed to have access to all the information that exists relating to the price of an object
  • Individuals and institutions are assumed to make rational decisions

Of course the world is a far more complex place and these assumptions ignore the effect of the choice that others make on our own choice.

The article then provides a brief history of agent-based computing starting with the 1970s and John Conway's "Game of Life" where individual cells interact with adjacent cells based on a few simple rules to produce complex behaviour for "the whole". In agent-based models individual agents represent the participants in an economy and are assigned rules representing interactions with other agents. It then goes on to ask is it feasible to build an agent-based model on a scale capable of simulating a nation's or indeed the world's economy?

Obviously this isn't just a case of vast amounts of computing power being thrown at the problem but also one of getting access to the information needed. There is also a lot of work needed to build and verify the many models needed. Clearly then a work in progress but a potentially useful and far more realistic one than has been used to date.



Be the first to comment on this article - Economic Models Ignore Dependency and Irrationality - Is Agent Based Computing the Answer?

(Logged-in site members can choose to receive email notifications of comments made on this article:  Login · Register)

Return to Previous Page

All articles/posts © of the respective authors

Site design and architecture © 2010 - 2019 Eclectica Systems Ltd.