293 | Doyne Farmer on Chaos, Crashes, and Economic Complexity
What this covers
Doyne Farmer sits down with Sean Carroll to argue that standard economic models fail to capture how real economies work because they rest on assumptions—rational expectations, equilibrium, perfect information—that prevent them from generating the endogenous crises and irregular cycles we observe. Farmer, director of the Complexity Economics program at Oxford and a physicist by training, contends that economies are better understood as out-of-equilibrium complex systems where millions of boundedly rational agents interact through nonlinear feedback, much as species do in an ecology. The conversation traces how agent-based models—which drop the rationality assumption and let agents follow simple rules, learning algorithms, or trial-and-error heuristics—can reproduce phenomena that equilibrium models cannot: endogenous business cycles, housing-market imbalances an order of magnitude out of line with supply and demand, and the 2008 financial crisis as an inevitable consequence of poorly understood mortgage-backed securities spreading risk through the system.
Farmer moves through specific predictions and applications: his group used trophic levels from ecology to forecast which products would get cheaper over fourteen years, and showed that technology improves slower in complex, less modular systems by extending Wright's law with a design structure matrix. He argues that markets function as ecologies of specialized strategies producing swings around efficiency rather than perfect allocations, and that the economy behaves like the metabolism of civilization—processing resources into goods through a web of interconnected specialists. The discussion turns critical of the discipline itself: complexity economics remains marginalized in academic economics despite predictive successes at central banks and in government (his team advised the British government on lockdown economics using a disequilibrium model), partly because mainstream journals will not publish it and students are warned away. Farmer sketches a path forward: complexity economics will gain mainstream acceptance only when its models consistently beat conventional ones in hard empirical terms—adoption already underway at the Bank of Canada, Bank of Italy, and through his company Macrocosm.
Farmer argues that economies are out-of-equilibrium complex systems whose endogenous, often chaotic dynamics are better captured by agent-based complexity models than by mainstream equilibrium economics built on rational expectations and supply-equals-demand assumptions.
- Mainstream models assume rational agents and equilibrium, which forces them to treat crises as exogenous and cannot generate endogenous business cycles
- Agent-based models with millions of boundedly rational agents can capture disequilibrium dynamics and have made successful empirical predictions (COVID lockdowns, trophic-level price improvement)
- Markets behave like ecologies of specialized strategies that produce swings around efficiency, especially when novel instruments are introduced
Complexity economics represents an academic market inefficiency departments can exploit for competitive advantage.
- An economics department could leap upward in standing by taking complexity economics more seriously—framed as an academic market inefficiency that an open-minded institution could exploit.
“It's a market inefficiency. A department can leap upward by taking this stuff more seriously.”