Slight, soft-spoken Didier Sornette is a master of disaster, a scientist who has made the forecasting of catastrophes in space, on land, in cities and in the market his life's work. He started two decades ago in France with seminal work on the predictability of rocket-engine ruptures and failures, then earned a professorship in geophysics at UCLA by applying his complex-systems research to the more grounded task of predicting waves of tectonic ruptures that lead to earthquakes. Next week, he meets with the Dutch government to explain how his models predict the sort of societal ruptures that result in waves of urban criminality. But all of those efforts are child's play next to the work he has undertaken on behalf of investors. In a new book,
Why Stock Markets Crash , the polymathic physicist proposes that complex-system theory can explain and predict both the emergence of financial bubbles as well as their rupture and path of recovery. So far, so good. The book was just published this month, but Sornette has been issuing papers in academic journals that have quietly made accurate financial predictions, ranging from the 1999 resurgence of the Japanese stock market and its follow-on decline, as well as much of the U.S. market action last year. His latest predictions, made without reliance on any of the usual technical, seasonal or economic indicators in vogue on Wall Street, suggested in August that U.S. stocks would sink into October, but then rally sharply into the start of 2003. Sornette's fractal model suggests that the current rally will stall shortly and that prices will subsequently retreat much lower over the next 12 months to 18 months, punctuated from time to time by strong countertrend rallies. After the S&P 500 index reaches the low- to mid-600s from its current perch at around 925, his model predicts a multiyear period of convalescence for U.S. stocks before a new bull phase pushes prices back toward and beyond their 2000 highs. He promises to provide a monthly update on his prediction at his UCLA faculty Web site .
Follow the LeaderWhat is risky and novel about his work is that it suggests that investors conform to "herding behavior," much like the particles in a magnet, and that long periods of "imitative cooperation" ultimately lead first to bubbles, then to short periods of extreme instability, rupture and catastrophic failure. Even though individuals appear to act of their free will when viewed under a microscope, when viewed macroscopically as a system he believes they act in non-random patterns that -- much like molecules in a pressurized container of gas -- can be captured with the precision of a mathematical formula. A less brave scientist would be content to let the ideas provide grist for financial practitioners, but Sornette has taken the unusual step of issuing forecasts, because it is the only way to prove to skeptics that his work has the necessary scientific rigor to be reproducible. "I am risking my reputation because I am convinced that there is an important truth in the theory that I have described, and the best way to convince my colleagues is to issue forward predictions," Sornette said. There's a paradox here, of course. He points out that if his theory merely does a good job of describing the past, it will be considered to be just a figment of pattern recognition -- interesting, maybe, but not revolutionary. And yet, if he does such a great job that investors change their behavior and avoid a forecast result, then his prediction will be considered incorrect. He notes in the book that he faces three unattractive scenarios: Nobody believes the prediction for a coming decline. It is correct, and the market crashes. This would be a victory for the predictor, but skeptics would probably put it down as a lucky shot without statistical significance.
Everyone believes the warning. It causes panic, and the market crashes as a consequence. The prediction thus seems self-fulfilling, and success is attributed more to the panic-effect than to real predictive power. So many investors believe that the predictions might be correct that they make reasonable adjustments. Then, the steam goes out of the bubble. The prediction thus disproves itself. Sornette says that this Catch-22 is the inescapable lot of scientific investigations into systems that are made up of humans with learning and reflective abilities -- and stands in contrast to the usual inanimate and unchanging physical laws of nature. If he just published his findings in obscure physics journals, later on he could emerge with a press release and say he told us so. But as part of a new breed of scientists who are engaged with the public, he believes that an increased awareness of the potential for market instabilities will help construct a more stable, efficient stock market.
Every Bubble's StorySornette studied and characterized 50 bubbles in various financial markets around the world -- gold, currencies, stocks and more -- en route to developing his theory with a team of researchers. The story of the stock bubbles, he said, is always the same: In the first stage, the economy is good, corporate earnings are peaking and analysts are bullish that good times will continue upward in a straight line. In a second stage, companies managed by people who are interacting with a positive media and their fellow leaders feel good about borrowing money from banks -- and their expansion attracts more investment. This additional influx of funds fuels the engine and throttles the collective market up more, creating a positive feedback loop that advances until "even your grandmother" cannot resist the attraction and will take money from under the mattress to put into stocks.
As prices begin to reach a parabolic stage of accelerated increase, which can be characterized in every case with a signature slope, the market reaches an unstable stage at which any perturbation -- an interest-rate increase, geopolitical change or something relatively prosaic -- can cause it to go "critical," or crash. Sornette's work is aimed at understanding the time frame for that window of instability. "Think of a ruler standing on your finger," he explains. "It's very unstable. You need to keep moving the finger to stabilize it. What will cause it to fall? It could be the slightest movement of your finger, or the wind. Those things would just be the proximate cause, however. It actually fell because it was in an unstable situation." In his book, which is readable despite being replete with mathematical equations, Sornette shows how the interactions between market players -- investors, bankers, analysts, corporate managers, government officials -- are like the interactions between molecules in a rocket engine. None are truly isolated from the others, as they exchange information and emerge as a global macroeconomic entity. After a crash, Sornette shows mathematically that the same herding effects that blow up a bubble then lead to a quantifiable "network of remorse" that creates a signature "anti-bubble" of negative feedback and declining prices. The archetype for an anti-bubble is the Japanese stock market since 1989. And Sornette says that it is virtually inescapable that the U.S. market will follow the same path of long-term decline.
The Herd Mentality Sets Sornette's Work ApartIt is this holistic view of a world full of investors acting together that distinguishes Sornette's work from both traditional economic theory as well as the new wave of behavioralism. The old school approach, called "utility theory," holds that each investor is able to optimize his attempt to accumulate wealth by making rational, accurate judgments about the future. That view was essentially replaced by "prospect theory," which holds that people are more sensitive to changes in their level of wealth than they are to their absolute level of wealth. The latest view, crowned with the 2002 Nobel Prize for Economics to Princeton's Daniel Kahneman and Vernon Smith of George Mason University, holds that people are not terribly rational about their relationship with the future. Instead, they make mistakes by overestimating their ability to beat the market by taking risks.
Sornette's work suggests that all of these approaches are similar to "one body" theories of physics in that they deal only with the preferences and choices of a single person. It's as if a physicist tried to understand a gas by scrutinizing the actions of a single molecule. In contrast, he thinks it is more valuable to understand how individuals' misapprehensions about the future aggregate into the emergence of global behavior that ultimately can be categorized and analyzed mathematically. The bottom line is that Sornette's work does not anticipate another crash in the U.S. stock market; that has already occurred. Instead he simply anticipates the continuation of a bearish phase of cooperative herding as individuals collectively continue to wish unproductively for an end to the decline. It will come, he says, but just not until the middle of 2004. I'll stay in touch with Sornette over the next 18 months and keep you updated on his predictions. Let me know what you think of his approach by mailing me at
firstname.lastname@example.org ; if you've got a scientific bent, let me urge you to read his book before asking me any hard questions.