More than three years after the global financial crisis, economists are still wondering why some of their forecasting models failed to alert them. A growing number of them is asking whether new kinds of models are needed.

One rethinking effort led by SFI Professor Doyne Farmer is among 27 projects to receive initial funding from the Institute for New Economic Thinking, which seeks to promote changes in economic theory and practice through conferences, grants, and education initiatives. INET was founded in 2009 with a $50 million pledge by George Soros.

Doyne and External Professors Rob Axtell and John Geanakoplos want to build an agent- based model of the U.S. economy. Unlike traditional economic models, agent-based models don’t make top-down assumptions about how the whole economy behaves.

An agent-based model instead builds behaviors from the bottom up, assigning particular behavioral rules to each decision-making agent in the economy. This enables the emergence of more complex, life-like market behaviors, such as the copycat behavior that leads to “herding” among investors, or investors learning from experience or switching their strategies.

Agent-based models also can incorporate the interactions among different sectors of the economy – such as housing and finance – at different scales, something the traditional models don’t do very well.

Doyne spoke at INET’s April 2010 inaugural meeting about the need for new kinds of models that depict truly rational behavior: that of agents making decisions based on incomplete information in complex, changing environments.