It’s a Gamble - MarkyBon

It’s a Gamble

Our financial elites insist that they can manage their risk and tailor their exposure to risk with a combination of their brilliant models and hedging with credit default swaps and other derivatives. In The Black Swan, Nicholas Nassim Taleb explains exactly why they are wrong. Nevertheless, the testosterone-poisoned alpha males are winning the regulatory battle. Floyd Norris, writing in the NYT, tells us that

As a new regulatory system for derivatives is shaped on Capitol Hill, the banks will try to preserve as much of both as they can. To the extent they succeed, it will be the customers, and the financial system, that are at risk.

If that happens, it will be in large part the result of the lobbying efforts of Richard Baker, president of the Managed Funds Association, and a former Louisiana congressman who left office in 2008. He was chair of the House Financial Services Subcommittee, which had supervisory authority over all the big financial players, held hearings, and did nothing. Once again, the revolving door hurts society. And the $16,791,894 in campaign contributions from hedge funds in 2008 no doubt made a lot of friends.

It seems obvious that letting the jerks who caused the problem, and those who did nothing to stop it, decide how to fix it is a stupid idea. But this isn’t just common sense. These people are, in Taleb’s view, seriously and dangerously wrong about everything important, a point he makes with excellent bombast. The basic error lies in their models.

Taleb explains that Wall Street models are based on the normal distribution, the bell curve. Wikipedia has some pretty pictures and some math for those interested. But your mental image is good enough: high in the middle, and low on both ends.

This model is great for simple stuff, like flipping fair coins, and average heights. Stephen Hsu, a physicist at the University of Oregon, provides a simple example of the use of the normal distribution in securitizations. It serves as an excellent demonstration of how our financial masters think about collateralized debt obligations. They think that the probability of loss at the far ends of the normal distribution is so low it can safely be ignored. The problem, says Taleb, is that the outliers are really dangerous, Black Swans, so you can’t ignore them. Danielle Fong provides an excellent discussion of outliers, some of which cause avalanches of snow or debt, in plain English and with cool pictures.

Taleb says that more complex models, specifically those from fractal math, can provide better modeling. The most famous fractal is the Mandelbrot Set, a beautiful mathematical object. There is a fundamental difference between the normal distribution and fractals. The normal distribution works where the variables are completely (or almost completely) independent, in the sense that the outcome of one test doesn’t affect any other tests.

For dynamic systems, those which evolve and change over time, fractal math gives better models. Dynamic systems include natural phenomena, like weather, and many social systems, like the stock market. Taleb gives us a taste of this in his notes, 326 ff. He provides a more detailed description of the failure of the normal distribution to deal with markets here, and in the technical appendix to this paper.

Take a look at this applet, which shows how fractals can be used to create a mountain. The author, Antonio Miguel de Campos, explains how it works. He sets up an algorithm, applies it to a pyramid, then repeats it over and over. This process of repeating the algorithm is called iteration, and it is the key to explaining fractals. Each time you run the applet, you get a different mountain. Averaging won’t help you at all. In fractal math, what you get is what you get, a new thing each time. However, the outcomes are a lot alike in some intuitive way. This is called self-affinity, and it is a key to understanding fractal math. It is much more difficult to use this to make predictions, if it is possible at all. I think Taleb is right. There is something pleasing about fractals: they look natural, while the bell curve looks like a math problem.

Our financial elites have decided that their line of attack on derivatives, including credit default swaps, is that they let the little guy manage risk. Taleb shows us that that is nonsense. If your models aren’t predictive, how can they be used to manage risk? And just who are these players? Five Wisconsin school districts? Google [“credit default swaps” suit] and pick your own favorite example.

Your financial masters: they can’t manage risk, but they sure can screw investors and taxpayers.