Wednesday, February 10, 2010

Shred the Depositor Safety Net?

Mike over at Rortybomb addresses the thankfully-never-going-to-happen proposal by some calling for the elimation of bank deposit insurance. If only depositors had to worry more about the viability of the banks holding their money, the thinking goes, banks would be more severely disciplined and less inclined to gamble on riskier investments. Besides ignoring the fact that deposit insurance is an essential protection against bank runs in a system of fractional reserve banking, the proposal also stinks of the typical policy solution for libertarians and market fundamentalists: shift the risk on to us, we can handle it.

Problem is, that's not practical for 99% of individual depositors. Being the generous debater and financial engineer that he is, Mike imagines what kind of mathematical gymnastics we ordinary depositors would have to start employing to keep our money safe:
I know the simple way you do it, some techniques that I’ve had some training in: You place out the payment structures using monte-carlo simulations with lognormal random walks; you take a metric of correlation in the market, perhaps in a gaussian copula structure and use that to run correlations at each step between the instruments; you take the distribution you generate and apply a “value-at-risk” logic to it, looking at some piece of the tail distribution.
I love it when earnestness and sarcasm intertwine. Seriously, this is how a high-powered quant analysis would work, by making some basic assumptions about bank cash flows using historical data, possibly applying data on default correlations between asset classes, and then looking for maximum losses in the tails of a simulated distribution. Thank god you got that Applied Mathematics degree; your life savings may yet survive intact.

This intensively computational analysis, though, is subject to the same problem that bedevils all financiers/economists/forecasters/astrologers. You can't know the goddamn future, no matter how nifty your models are or how rich your data is! (And Mike Konczal certainly knows this, but he's not trying to divine, just to show how such a problem might, with trepidation and book-learnin', be approached.) Felix Salmon has already deconstructed the validity of the Gaussian copula in finance, and Joe Nocera took time a year ago to investigate the dubious risk management value of VaR models. So to those financial sophisticates who would tear up the deposit insurance contract: you don't even have the firepower to foresee the black swans out there, so why should everyone else be expected to?

The notion that we should exorcise deposit insurance from our banking system is, I think, symptomatic of the tired policy solutions emanating from a bloc of thinkers itching to shift more risk onto consumers in the name of "incentives." Yes, economic incentives are powerful, but what about those faced by financial institutions themselves? Might those be a better target for reform? We could stand to bear some amount of market inefficiency in the form of a depositor guaranty if we lessened the chances of it being used. Say, with regulation of banks' leverage, size, and scale of interconnectedness. These are the things that make the financial system inherently unstable and susceptible to magnificent collapse.

Conservatives who justify limited government intervention, in all matters, by pointing to uncertainty (like David Brooks) are on to something. We should be humble in the face of uncertainty and acknowledge that this uncertainty is universal. Quants aren't immune to it (though they mistakenly confuse it with "risk" and feel they've pinned it down), and neither are investors, depositors, or regulators. We know that banks will fail in the future, and that at times the system itself will buckle, but we don't know much more than that. The best we can do is limit the damage when it happens. Let's do that by pushing banks to raise new equity, not by transforming depositors into equity investors.