Bookstaber, The End of Theory

“As we embrace complexity we come to the end of theory.” So writes Richard Bookstaber, author of A Demon of Our Own Design, in his new book, subtitled Financial Crises, the Failure of Economics, and the Sweep of Human Interaction (Princeton University Press, 2017). Although he casts his analysis in the context of financial crises, it works perfectly well as an account of financial markets behaving “normally.”

Four phenomena are endemic to financial crises, Bookstaber believes: emergence, non-ergodicity, radical uncertainty, and computational irreducibility. Emergence occurs “when systemwide dynamics arise unexpectedly out of the activities of individuals in a way that is not simply an aggregation of that behavior.” Non-ergodicity is a feature of financial markets throughout. That is, markets vary over time; they do not follow the same probabilities today as they did in the past and will in the future. Uncertainty is radical when it cannot be expressed or anticipated, when we’re dealing with unknown unknowns. Finally, our economic behavior is so complex, our interactions so profound that “there is no mathematical shortcut for determining how they will evolve.”

How are we to survive in a complex, ever changing environment, where the future is not like the past, where projected probabilities are fictions? One short answer is: act like a cockroach. Use coarse, simple rules that ignore most information. “The coarse response, although suboptimal for any one environment, is more than satisfactory for a wide range of unforeseeable ones. … [P]recision and focus in addressing the known comes at the cost of reduced ability to address the unknown.” Alternatively put, don’t rely on optimization based on past data. Instead, use heuristics.

As Bookstaber boldly states, “if you can model it, you’re wrong.” It’s not just that all models are inherently wrong, it’s that models as normally conceived are useless under these circumstances. “If we want to understand a crisis, we have to construct a story, and we must be willing to do so in the ‘road in the headlights’ fashion: ready to change the narrative as the story line develops. A change in narrative means a change in model, and the model changes are not simply a matter of revising the values of various parameters, be it by the statistical tool of Bayesian updating or whatever. It might be a change in heuristics, in the types of agents in the system. … Models need to be like novels, molding to twists and turns and unexpected shifts.”

Bookstaber’s analysis is rooted in the work of the Santa Fe Institute, with a smattering of George Soros’s reflexivity theory added for good measure. It is pragmatic rather than axiomatic, inductive rather than deductive. It’s definitely a worthwhile read.

By the way, the Sante Fe Institute is re-offering its popular (and, I can attest, excellent) online course "Introduction to Complexity." The course started a couple of weeks ago.