In “All Systems Will be Gamed,” chapter six of Complexity and the Economy, Brian Arthurs argues that economics lacks a failure mode analysis of how systems work analogous to stress testing in structural engineering. The reason economics doesn’t have such a mode of analysis, Arthur argues, is that economists work with equilibrium assumptions that assume away the exploitive behavior that such stress tests are trying to predict:
Economic analysis assumes equilibrium of the system in question, and by definition equilibrium is a condition where no agent has any incentive to diverge from its prevent behavior. It follows that for any system being studies no invasive or exploitive behavior cannot happen. If such a system could be invaded, some agents would be initiating new behavior, and the system could not have been in equilibrium. Equilibrium economics then, by its base assumptions, is not primed to look for the exploitation of systems, and as a result systemic studies of how systems might fail or be exploited are not central to how the discipline thinks. (Arthur 2014: 104).
As will be clear to those who read the book, Brian Arthurs really has an issue with equilibrium that becomes a consuming obsession. There is much to be said about the economics profession’s probable abuse of the notion of equilibria. Nevertheless, in attacking the concept, Arthurs does exactly what those who abuse equilibrium do: He takes equilibrium too seriously. As a conceptual tool, equilibrium is a useful way of constructing systems that can improve our understanding of the world. It’s only when we presume that we therefore assume away the unknown unknowns of life, to borrow Donald Rumsfeld’s illuminating phrase, through our systems that we err, yet that is an error Arthur makes himself with his own complexity approach.
Such system building isn’t bad per se. Indeed, it is neither good nor bad per se. Building systems, however primitive they may be, is one of the ways that the human mind comes to understand the world surrounding it. The men of system who helped to set up the 2008 Financial Crisis, be they investors who utilized a ridiculous degree of leverage or regulators who pushed value at risk as the way of doing risk management, made the fatal mistake of thinking that their systems were near perfect approximations of the real economy. They were the ones who assumed away tail risks and presumed that they knew enough to end history. The presumption of knowing all of the facts, not the presumption that an economy tends towards equilibrium is what did those men of the system in. (After all, it is very possible for an economy to be heading towards equilibrium and the people within it being wrong about what equilibrium it is tending towards.)
Brian Arthur has nothing to say about such systems in his article. Instead, he wrongly believes that the sin was in the assumption of equilibrium, rather than in the presumption of knowledge. “Failure-mode studies are not at the center our our discipline for the simple reason that economics’ adherence to equilibrium analysis assumes that the system quickly settles to a place where no agent has an incentive to diverge from its present behavior, and so exploitive behavior cannot happen” (Ibid: 117).
Yes, the assumption of equilibrium does bring much presumption of knowledge if one takes the assumption too seriously. But as a modeling tool, it is more thematic, providing the opportunity for a system builder to provide a coherent narrative, than descriptive. It is only when we assume, as do many equilibrium theorizers do, that Even the use of engineering as the chosen parallel over the course of the article betrays Arthur’s perspective as one who believes that it is possible for economists to reliably foresee how systems are exploited: “Once we have identified where and how exploitation might take place, we can break open the overall economic model of the policy system at this location, and insert a module that ‘injects’ the behavior we have in mind” (Arthur 2014: 111).
In his very own vision of stress testing economic systems, Arthur still presumes that how the system will be exploited can become a known known, manipulable by analysts, but it’s the unknown unknowns that should be keeping policymakers awake at night. Brian Arthur’s conviction that policy can be improved by using stress-test analysis ignores that policymakers are already trying to figure out how the incentives impel people to take certain courses of action in different policy environments. Contra Arthur, the reason those policymakers make mistakes aren’t that they are working with equilibrium models, or that they aren’t sufficiently scientific. The reason is that no policymaker can’t know everything.
Bibliography
Arthur, Brian W. 2014. Complexity and The Economy. Oxford: Oxford University Press.