Sub-prime quantification

I have spent most of my career trying to understand what makes experts in their field so good. Looking at fireground commanders making life-and-death decisions in seconds, nurses noticing that babies are developing infection – even before the blood work comes back positive, pilots inventing ways to control “uncontrollable” airplanes, military commanders immediately spotting the focal point for an upcoming battle, the experts see things that are invisible to the rest of us.

But there are some areas where people don’t develop expertise. Danny Kahneman and I have speculated that under truly chaotic situations people don’t have a chance to identify patterns because the patterns aren’t there. In addition, people need to have some sort of feedback in order to develop accurate as opposed to superstitious beliefs. In the absence of either one of these conditions – under random conditions or zero feedback – expertise won’t be possible.

The stock market provides a good example of a domain where people don’t develop expertise. No one is particularly gifted in predicting market cycles or selecting good stocks. The problem isn’t lack of feedback. Market analysts and brokers get lots and lots of rapid feedback. Rather, the stock market acts as a random walk (see B.G. Malkiel, “A random walk down Wall Street,” Norton, 2003). Stockbrokers may act very confident. They may use jargon that sounds impressive. They may boast about the size and experience of the research staff in their home office. They may know how to execute trades very quickly. But the evidence shows that they aren’t able to accurately select stocks.

The one counter-example that people throw back at me is the new generation of computer-based quantitative methods that are able to take advantage of temporary discontinuities. It takes computers to detect them and respond quickly enough to exploit them. It’s not the same as having expertise about market cycles or promising stocks to buy. It’s just a way to pounce on temporary anomalies. So I am OK with that. Sort of.

Nevertheless, I’ve been interested in the recent perturbations stemming from the U.S. sub-prime mortgage market. There perturbations have rippled throughout the financial community. Lending institutions understood that the sub-prime mortgages were risky. No doubt about that. But by combining a number of them and by using analytical methods to calculate the degree of riskiness, lending institutions sought to trade in baskets of these sub-prime mortgages and to use them as collateral for various loans. When the sub-prime market started to collapse, this created a ripple effect that severely reduced liquidity. Banks were reluctant to grant loans until the situation sorted out. The liquidity crunch forced hedge funds to dump good investments into a falling market in order to raise cash, etc., etc. The unforeseen interactions and dependencies created enough instability that the Federal Reserve had to take action and may get even more involved.

And which hedge funds were the hardest hit? The quants. The funds that relied on quantitative analyses. The Economist (August 18, 2007) described one quantitative model that evaluated the current market as 25 standard deviations away from normal. (A likelihood of 0.000 …0006 where there are 138 zeros before the six.)

In other words, the quants were insensitive to the hidden dependencies with a result that has threatened to destabilize the U.S. economy. And the part of me rejoiced that is irritated by this loophole in the notion of a random stock market.