Jun 25, 2009 Economy
Can we pack an entire economy, with all its complex human and political interactions, into a computer? Physicist Dirk Helbing of ETH thinks so – as long as we’re bold enough in going about it.
He points out that financial systems aren’t the only monsters we’ve let out of the box. How traffic flows in and around huge cities simply cannot be grasped by mathematical analysis, but computer models let millions of virtual vehicles interact on realistic road patterns – and often discover potential problems before they occur in reality.
The complexity of today’s economy, Helbing suggests, demands a similar approach. “We’re not currently using the best capabilities of science,” he says. “We need to bring together scientists from different fields and put together tools that can be used as a kind of wind tunnel for testing out social and economic policies.”
The approach has already proved its worth on a small scale. Several years ago, when the state of Illinois decided to deregulate its electricity market, it wanted to avoid California’s disastrous experience, when the disgraced firm Enron manipulated prices, created shortages and caused rolling blackouts. The state authorities hired Charles Macal and colleagues at Argonne National Laboratory to build a sophisticated model of Illinois’ power market, incorporating suppliers, consumers and regulators, as well as the full network along which power flows.
The team first interviewed the various market participants to learn their behaviour. Each was represented in the model by virtual agents who could act on those strategies, but also learn on their own. Running the model, the team found that the initial legal framework for deregulation had flaws which made it vulnerable to Enron-like manipulation. The state successfully altered its plans, and the approach has since been used in similar cases in Croatia, Portugal and South Korea. Some stock markets, such as NASDAQ, have used it in planning changes to their trading rules.
Groups in the US and Europe are already building similar models of whole economies, including millions of individuals, families, firms and banks, as well as government regulators. Helbing and colleagues have lodged a proposal with a European Union funding body to start building the extensive communities of researchers and collaborations – between economists and other social scientists, physicists, lawyers and computer scientists – that would be needed to simulate whole economies effectively.
With growing computing capacity and greater understanding of complex systems, they suggest it should be possible within a decade to have functioning models of the global economy to which policy-makers could look for sound insights with which to penetrate the overwhelming complexity of today’s markets.