Antifragile Banking and Monetary Systems
White, Lawrence H., The Cato Journal
"Fragility" is the well-known property of being easily breakable, of failing under moderate stress. The opposite property is "antifragility," a term coined by Nassim Nicholas Taleb (2012a) and the title of his recent book. Taleb (2012b) defines antifragility as the property exhibited by "things that gain strength from stressors and get stronger from failure, like evolution." An antifragile thing or system is stress-loving. What doesn't kill it makes it stronger. We exercise, for example, because our muscles grow stronger from moderate stress. Robustness, an intermediate concept, is the property of being unaffected either way by moderate stresses. Taleb illustrates the threefold distinction this way: We stamp "handle with care" on a package containing something fragile; we needn't stamp any instructions on a package containing something robust, because it won't be affected by handling; but we would stamp "please handle roughly" on a package containing something antifragile, because such handling would make it emerge stronger.
Here I consider how we might achieve ant/fragile banking and monetary systems. There are reforms that can marginally reduce fragility, but I will argue that to achieve ant/fragility will require a serious turn away from "one-practice-fits-all" centralized regulation and toward a free market's mixture of innovation and strict discipline. In banking it will require an end not only to "too big to fail" bailouts of uninsured creditors and counterparties, but also to other forms of taxpayer-backed depositor and creditor guarantees. Deposit guarantees, contrary to intention and despite their immediate run-suppressing effects, in the long run have fostered moral hazard and thereby contributed to banking system fragility. In monetary policy, it will require an end to centralized monetary policy. The centralization of money issue has eliminated the market-based disciplinary and error-correction mechanisms that once governed money creation, thereby putting all our monetary eggs in one basket and creating monetary system fragility.
The Banking System Is Not Naturally Fragile
Many economists--and certainly regulators--will object that to make the banking system antifragile is a futile undertaking because banking is naturally fragile, meaning, inherently prone to collapse in the absence of government guarantees to depositors (and by extension guarantees to all short-term creditors). (1) For example, Nobel laureate economist Robert Lucas (2011: 20), in the slides accompanying a recent lecture that is otherwise favorable to free markets, states flatly that "a fractional reserve banking system will always be fragile, a house of cards." In such a view the best that can be done is to institute government guarantees to mitigate fragility, or perhaps to outlaw fractional reserve banking, as recommended by Kotlikoff (2010) among others, by banning the modern practice of providing payments services through checking accounts whose balances are demandable debts.
The most widely cited model of banking in the economics literature today is the Diamond-Dybvig (1983) model, which depicts a very fragile bank. A depositor run will easily break it, and a run can easily occur, triggered merely by self-justifying worries that others will run. A form of deposit insurance is needed to fix the problem. Many have taken from the model (and from the large theoretical literature built on it) the lesson that any modern banking system is naturally fragile. Models depicting banks as naturally fragile seem descriptively plausible to those whose familiarity with banking history is limited to the United States. The United States did have a series of banking panics between 1873 and 1933, and the introduction of federal deposit insurance in 1933 did finally stop that decade's panics.
A more thorough look at theory and empirical evidence indicates clearly that banking is not naturally fragile. Theoretically, the fragility result of the Diamond-Dybvig model is itself fragile: it does not survive small modifications that make the model's assumptions more realistic. …