How to Model the EconomyIn the August 2009 issue of Nature, J. Doyne Farmer, a professor at the Sante Fe Institute, and Duncan Foley, an economist at the New School for Social Research, published an opinion piece (pdf) in which they argue that the types of economic models most commonly used to make economic predictions predictions that businesses often use in their planning are seriously flawed.
Farmer and Foley explain the two types of macroeconomic model that are currently available:
- Econometric models that base predictions essentially on extrapolating from past economic data. If the economy experiences major changes from what has occurred in the past, predictions from these models go seriously off-track
- Idealized models that assume a well-functioning economy, i.e., one that does not experience crises.
In such a model,
... at any given time, each agent acts according to its current situation, the state of the world around it and the rules governing its behaviour. An individual consumer, for example, might decide whether to save or spend based on the rate of inflation, his or her current optimism about the future, and behavioural rules deduced from psychology experiments. The computer keeps track of the many agent interactions to see what happens over time. ... Policy makers can ... simulate an artificial economy under different policy scenarios and quantitatively explore their consequences.The article offers as an example the model Farmer and colleagues have created to explore how hedge funds' use of leverage borrowing from banks to finance their investments affects fluctuations of stock prices. The model "shows that the standard ways banks attempt to reduce their own risk can create more risk for the whole system." Admittedly, this model covers only a subportion of the larger economy, but its structure and use are nonetheless illustrative of the principles of agent-based modeling.
Framer and Foley acknowledge that there are technical issues that make creation of agent-based models a challenge, most importantly the difficulty
... in specifying how agents behave and, in particular, in choosing the rules they use to make decisions. In many cases this is still done by common sense and guesswork, which is only sometimes sufficient to mimic real behaviour. An attempt to model all the details of a realistic problem can rapidly lead to a complicated simulation where it is difficult to determine what causes what. To make agent-based modeling useful we must proceed systematically, avoiding arbitrary assumptions, carefully grounding and testing each piece of the model against reality and introducing additional complexity only when it is needed.Farmer and Foley close their article by advocating investment of public funds in creating an agent-based model of the whole economy in order to have a more reliable tool than those currently available for "quantitatively exploring how the economy is likely to react under different [policy] scenarios."
1 You can read about some of the other areas besides economics in which agent-based models are used in this earlier post about US Army counterinsurgency training, in this wiki entry on use of agent-based models in biology and medicine, and in this paper (pdf) describing agent-based modeling of the airline industry.