At the Santa Fe Institute, Yale economist John Geanakoplos has teamed up with two physicists to look at the natural competition that emerged among hedge funds as they competed to attract investors. The group is examining how hedge funds took on additional leverage during that process—and how the state of the market changed fundamentally as a result of the added debt.

This way of thinking is foreign to mainstream economic theory, which assumes that people, firms, and other economic agents act rationally. Markets are presumed to exist in economic equilibrium, a more or less stable balance achieved through the players’ varying aims. “It’s completely unrealistic,” says finance professor Andrew Lo of the Massachusetts Institute of Technology. “Economists like the model because it can be solved, but people would have to be superhuman in their intelligence to fit the assumptions.”

The evident failure of these assumptions put former Federal Reserve chairman Alan Greenspan into an intellectual tailspin, as his recent testimony before Congress shows. “Those of us,” Greenspan admitted, “who have looked to the self-interest of lending institutions to protect shareholders’ equity (myself especially) are in a state of shocked disbelief.”

Other economists are finally beginning to acknowledge that basic notions such as equilibrium—an idea originally imported into economics from 19th-century physics—aren’t adequate to understanding complex markets.

Consider financial derivatives, for example. It’s taken for granted by economists that derivatives make markets more stable. They are designed to give market participants more flexibility by allowing them to take highly specific market positions. But some economists—notably William Brock of the University of ­Wisconsin and colleagues—have suggested that this view may be backward. They are exploring the consequences of adding one rather obvious fact to standard economic models: that people learn as they participate in markets and may quickly copy other investment strategies if they seem to be working. The result is a pile-on that makes the initial strategy ineffective. Brock’s results show that such adaptive learning leads to derivatives actually destabilizing markets.

A similar conclusion emerges from mathematical physics too. Over the past 30 years, physicists have developed methods for calculating the properties of what they call disordered systems, which have a range of linked components. Adapting these techniques to markets, statistical physicist Matteo Marsili of the Abdus Salam International Centre for Theoretical Physics in Trieste, Italy, showed recently that the proliferation of derivatives inevitably produces an unstable market—a finding that traditional economists have acknowledged only in hindsight.

Marsili emphasizes that we still know next to nothing about the overall consequences for markets from the use of derivatives. Both Brock and Marsili point out that derivatives can be used beneficially to hedge risk, even if they’re often used in practice to leverage positions and thereby increase risk. To know when and how derivatives can be used safely, investors must, at the very least, take the issue of systemic risk seriously.

Mauro Gallegati, of the University of Ancona, in Italy, says that even without the subprime problem, instability in the global credit markets would have eventually prompted a meltdown of some sort. The credit network had become so highly interlinked that all participants were, as he put it, “very ­fragile with respect to the possible collapse of their partners.” Traditional central-bank controls and banking regulations just weren’t up to controlling such instability.

Gallegati believes that governments will need to take a much more ambitious approach in the future. Rather than just looking at individual banks’ lending practices to see if their risks are at acceptable levels, regulators will need to take a more holistic view, monitoring the nature of the links between institutions and the overall stability of the credit network.

“These networks have to be checked,” Gallegati says.