Even the experts seem bewildered by the current economic crisis. Quantitative analysts (quants)--the whiz-kid financial engineers whose algorithms have dominated Wall Street trading in recent years--have watched those algorithms fail. Former Federal Reserve Chair Alan Greenspan acknowledged in October that there was "a flaw in the model that I perceived … defines how the world works."...the classical theory of finance simply does not address human psychology. It looks more like a physical science than a social science--relying on the premises that markets are "efficient,"Andrew Lo, a financial economist at MIT, is developing alternative models:
Blame has fallen on quants for various aspects of the crisis. First, mathematical models were increasingly used to determine whether someone deserved a loan, bypassing individual judgments. "In the end, there was very little sound credit judgment going into making these credit calls," says Bjorn Flesaker, a senior quant at Bloomberg in New York. Then, quant models were used to rate the riskiness of financial instruments, including the CDOs. "We never necessarily viewed the rating agencies as having the greatest rocket scientists around," says Flesaker, yet investors accepted those ratings, taking on more risk than even they realized...like many of the elements that economists and the media have focused on, the quant models are simply "proximate causes." Ultimately, experts must examine human behavior to find out why the crisis happened. Why did so many people take on mortgages that they would not be able to pay? Why did the best minds of Wall Street ignore warnings about a housing bubble? "The bottom-line question that economists, I think, still are struggling with is: 'Did anybody know that the risks were so great and, if so, why did they continue investing?'
The madness of crowds
Classical finance theory's model of speculative bubbles, such as the dot-com bubble of the late '90s and the recent housing bubble, does not match real-life observations. Classical finance contends that rational investors will always have the best possible portfolio, so they will not buy or sell unless they have extra money to invest or need to cash in their investments. However, researchers have observed that people buy and sell much more often than that during a bubble--with the rate of transactions becoming increasingly manic the bigger the bubble gets.
Lacking a good classical model for stock-market bubbles, Scheinkman, whose work is primarily classical, turned to a concept in behavioral finance. Psychologists have found that people often overestimate the precision of their knowledge. Scheinkman and his Princeton colleague Wei Xiong guessed that overconfident investors would trust their own opinions about the price of an asset, so they would consider others' opinions, if different, a little "crazy," says Scheinkman. Looking to make money off others' crazy opinions, investors would be willing to pay more than they think an asset is actually worth because they believe that they will be able to sell it in the future to an overeager buyer. This process would inflate prices and cause a trading frenzy. Incorporating investor overconfidence into a theoretical model published in 2003 in the Journal of Political Economy, Scheinkman and Xiong were able to recreate more accurately the hyperactive trading in bubbles.
Lo's species behave differently based on what part of their brains they are using. When things go well and people make money, as they did for the past decade, the experience stimulates investors' reward circuitry. This causes them to seek more profits and ignore possible risk, leading, for example, to a bubble. When things take a turn for the worse, panic overrides rational decision-making, leading to a crash. Only when the market is steady does the rational brain take over. Lo is starting to use functional magnetic resonance imaging and other tools of neuroscience to quantify these behaviors and incorporate them into his models. He also needs more real-world data on the way different funds invest money--data that are now secret or that no one bothers to collect.
Although Lo's idiosyncratic approach lies outside of the behavioral and classical theories, he says it reconciles them. "If you were an efficient-markets type, I think you'd be hard-pressed to explain what happened over the last few weeks. And if you were an irrational finance person, you'd be hard-pressed to explain what happened over the previous 10 years. So I think that the only way to reconcile the two is to acknowledge that both are different aspects of the exact same truth."
Behavioral researchers are eager to prove that their ideas mirror nature by using quantitative methods to link them directly to real-life data... Stock pricing lends itself to such studies, because valuing a stock involves conjecture--which is subject to psychological factors--and a lot of stock-market data have recently become available to academic researchers.
In a 2007 paper in the Journal of Economic Perspectives, Wurgler and co-author Malcolm Baker, a financial economist at Harvard Business School, looked for signatures of investor sentiment--irrational optimism or pessimism--in stock-market data since the 1960s. They hypothesized that certain stocks would be more subject to sentiment than others: broadly speaking, stocks for which the true value is difficult to determine. For example, a young, promising company would fit the bill. "The combination of no earnings history and a highly uncertain future allows investors to defend valuations ranging from much too low to much too high," they write.
Comparing the stock-market data with their measure of investor sentiment, they found what they had expected. In optimistic times, difficult-to-value stocks were wildly popular and therefore made much more money than average. In pessimistic times, they were wildly unpopular and therefore made much less money than average. On the other hand, easy-to-value stocks, which are considered safer, were more popular in pessimistic times than optimistic ones, but their prices stayed much closer to average. This helps explain past bubbles in certain types of stocks--say, dot-com stocks in the 1990s--and is also useful for making predictions for the future