The great Larry Lessig emails me to point to a Newsweek article he wrote back in October, which I missed at the time. It’s entitled “Why The Banks All Fell Down”, and Lessig asks whether there’s a connection between his theory there, on the one hand (properly Lessig ascribes it to Avinash Persaud), and my story about the Gaussian copula function, on the other.
The models governing our financial system were developed in the 1950s. They were built, as Persaud explains, upon the assumption that “each user is the only person using them.” That assumption may have made sense when only the Rand Corporation (which developed the models) had the data to act upon them. But as Persaud wrote in the Financial Times in March, in “today’s flat world, [where] market participants from Argentina to New Zealand have the same data,” the assumption is nothing short of nuts. As the market shifts, the models used by all the major investment banks “throw up the same portfolios to be favored and those not to be.” This is the Heisenberg uncertainty principle for financial markets: by being observed, the observed gets changed. The global, simultaneous and practically instantaneous “adjustments” lead “the herd” (a.k.a. our financial system) right off the cliff.
The financial system is indeed full of models which make sense if only one person uses them, but which become extremely dangerous if everybody starts to use them. Here’s the relevant bit of my Wired story:
The real danger was created not because any given trader adopted it but because every trader did. In financial markets, everybody doing the same thing is the classic recipe for a bubble and inevitable bust.
The financial markets were reminded of this in the summer of 2007, when simultaneously all the formerly-very-successful quant and stat-arb hedge funds all blew up at the same time. Bryant Urstadt wrote the definitive article on the blow-up for the MIT Technology Review, which is hidden behind a registration firewall here; the gist is that all the funds were following much the same strategy (something called “pairs trading” is mentioned), and that as a result asset classes become highly correlated which were never correlated in the past.
In 2007, the damage was limited: there was a lot of fallout in the world of hedge funds, but the stock market continued to rise, and the systemic implications seemed to be contained. But the adoption of the Gaussian copula function was much more widespread than a bunch of hedge funds: it was embraced at pretty much every CDO origination and trading desk on Wall Street. And when the entire financial system starts using essentially the same model, the systemic devastation which can result is enormous.
So yes, when Persaud talks about “the market-sensitive risk models used by thousands of market participants”, he’s very much referring to things like the Gaussian copula function, which proved to be one of the best ways ever invented of shunting risk off into the tails. He’s probably also talking about the near-universal adoption of Value-at-Risk, which as Joe Nocera explains does a great job of allowing managers to ignore the tails, no matter how much risk is in them.
Persaud says that there are important implications for regulatory policy:
This brings us to the philosophical problem of the reliance of supervisors on bank risk models. The reason we regulate markets over and above normal corporate law is that from time to time markets fail and these failings have devastating consequences. If the purpose of regulation is to avoid market failures, we cannot use, as the instruments of financial regulation, risk-models that rely on market prices, or any other instrument derived from market prices such as mark-to-market accounting. Market prices cannot save us from market failures. Yet, this is the thrust of modern financial regulation, which calls for more transparency on prices, more price-sensitive risk models and more price-sensitive prudential controls. These tools are like seat belts that stop working whenever you press hard on the accelerator.
The problem is that it’s very unclear what the alternative is. The lesson of the failure of Basel II is, indeed, that you can’t trust banks to judge their own internal risks very well. But the whole reason why Basel II was created in the first place is that financial instruments are so complicated these days that it’s very easy to do an end-run around simpler regulations.
My feeling is that the best way to go is to set some very clear and simple rules, much as the Spanish central bank did, and refuse to allow banks to build enormous businesses doing things that the regulators don’t understand. And secondly to place a cap on banks’ balance sheets — I think something around $300 billion is reasonable, and that there’s no reason why any bank should be bigger than that. Alternatively, if you are bigger than that, then you have to become much more constrained in what kind of activities you’re involved in: you should basically just be doing plain-vanilla deposit-taking, borrowing, and lending.
That wouldn’t prevent future financial crises outright. But it would probably stop them from causing a major global recession.