Didier
Sornette
Professor &
Author
Why
Stock Markets Crash
Critical
Events in Complex Financial Systems
JIM PUPLAVA: Joining me on the program is Didier Sornette. He is a Professor of Geophysics at The University of California Los Angeles. He is also a research director at the National Center for Scientific Research (CNRS) in France. He is a specialist in the scientific prediction of catastrophes in a wide range of complex systems. He is the author of the textbook Critical Phenomena in the Natural Sciences and has authored or co-authored over 300 papers in international scientific journals. We are here today to discuss his new book, Why Stock Markets Crash: Critical Events in Complex Financial Systems.
Dr. Sornette, in your book, Why Stock Markets Crash, you take a different view from many experts in that the underlying cause of a stock market crash can often be found in preceding events, months or years before the markets crash itself, through what you call, cooperative activity. I wonder if you would explain that please.
DR. DIDIER SORNETTE: Yes, the concept of cooperative behavior is the crux of the matter. I like to take the following example. Consider a ruler or pen in your office or on your desk. Put it vertical and then release it. Most of the time it will fall on one side or the other, except if you are very talented and are able to keep it in the vertical position. The point is to ask why is the pen falling to one side or the other? You can have two explanations. One refers to its initial position. You had an imperfect position or a burst of air passed over it and pushed it onto one side. The more profound explanation is that the pen was prepared into an unstable position.
Similarly for the stock market, you have two classes or two levels of explanations. When you witness a crash, most of the time people invoke some news--some recent news--like a new tax law or interest rate increase. Something like that that just occurred a day before or a week before. What we have found in our work is that such local explanation does not actually describe the origin of the destabilization of the market. The fundamental explanation is probably going to be found similarly to the pen example, in that the market is building up into an instable position over months to years before the occurrence of the crash. We see this in the build up of specific patterns in the trajectories of such variables as the price, volume and volatility as a function of time. Typically we find evidence of an incoming instability in the precursory patterns of time trajectories of the price, volume and volatility variables. The time evolution of these variables tell us that these patterns are not sustainable and that an instability is ripening. This defines a bubble. For instance, the higher a bubble builds up, the more unstable becomes the market, until the point when any news or event could topple it.
Endogenous or Exogenous Events
JIM: You also talk about how a collapse is due to this unstable condition. In other words the collapse is an endogenous origin and it is usually some sort of exogenous shock that merely acts as a trigger factor. The true cause is systemic instability. In the opening preface to your book, you talk about how, for example in the US in the 1990s, the US stock market capitalization increased and trading volume tripled in the 1990s and the volume of securities issued was a multiple of almost six.
DR. SORNETTE: Yes, this is correct. From a scientific point of view, it is very interesting to witness this development of systemic instabilities. The systemic instabilities, of course, are scary from a personal point of view. What we have found in our studies [Johansen and Sornette, 2002] is that basically 2/3 of all shocks—of all dramatic crashes--can be attributed to an endogenous origin as opposed to exogenous origin. As example of exogenous shocks, let me cite of course, World War I and World War II. These very big events significantly perturbed the market and are exogenous in the sense that they are not part of the intrinsic dynamics of investors on the stock market. You can sometimes find significant causes that had a tremendous effect on the market and were very big surprises. These correspond to what I denote “exogenous” events. Other examples are September 11th, the Coup against Gorbachev in August 1991. Both significantly affected the western stock markets. These are example of exogenous events. However, most of the time, markets can move and even crash due to endogenous events. Quantitatively, we have found that 2/3 of the significant market drops can be qualified as endogenous by the specific patterns that preceded them and that we have been able to identify.
JIM: And when you looked at some of the various manias or bubbles in the past, the tulip mania, Florida real estate or the tech stock in the 90's, another common characteristic leading up to these bubbles or crashes was the so-called “sure thing" where someone comes up or a particular fad takes hold where it is an easy way to make money.
DR. SORNETTE: That is correct. Two examples come to mind. For example the day before the October 19, 1987 crash, another big crash of the 1980s, the Wall Street Journal wrote on that very day that "It is a sure thing that the stock market will continue to go up." Another example is the real estate bubble in Japan that terminated on the end of December 1989. At that time the Japanese banks were providing loans to private companies and individuals to buy real estate without any collateral. Actually borrowers were getting more than 100% of the total value of the property based on the idea that the market was going up so well and so fast, that it was a sure thing. The usual down payment, typically 20% that was required, was not necessary in that case because of the expected upward move of the market. That is quite incredible to learn but it seems to be a robust psychological trait of human beings, namely over-confidence.
JIM: A very common characteristic in the financial field, as much as it may appear to those who are informed, is that we are in a bubble--whether it is a real estate bubble or a stock market bubble. You make reference that in many times these crashes that occur are unforeseen, especially by economists because economists have a hard time looking at trend reversals. It is difficult to forecast even within the economic framework.
DR. SORNETTE: That is correct. There are two points here. One is that most economic approaches are really good at predicting changes in the slope of growth. Suppose that last year, the growth was 1%, well it should be 1.5% or 2% this year. But what is very difficult to do is to predict a change of trend, that is, to tell that last year the economy was growing and that next year it will do down. Predicting recessions is usually very difficult for economists, who know only after the fact. It is very difficult because the dynamics of the economy or of the stock market obey what is called, in the scientific jargon, a non-linear process. The dynamics is not linear. This means that consequences are not proportional to causes and different causes intermix in a complicated way, sometimes enhancing each other dramatically. Nonlinear dynamics is usually very difficult to predict.
Another point is that a crash often occurs after a huge appreciation of the market. Typically, a growth of the market may been going on because of positive mood and positive sentiment on the part of investors. Many news, many impacts are interpreted positively in such periods and give rise to a reinforcement of the positive mood. In such situation, most of the people including the economists and analysts justify the positive mood on the basis of apparently rational arguments. For example, in the last information bubble that terminated in March 2000 and was followed by the NASDAQ crash, analysts from famous banks were justifying the skyrocketing price of the new economy companies by invoking new theories. One such argument used the added value offered by so-called real options held by “new economy” companies. According to these analysts, the stock prices were skyrocketing because they were accruing new values from the possibility of evolving and adapting better in the near future, and this “real option” of being to be able to adapt and invade new industrial niches should be priced in the stocks and were thus justifying these skyrocketing prices. Now, of course, the theory is different and very few analysts remember these arguments. People have a tendency to find theories to justify their idiosyncratic behavior. This is actually a well-documented trait in behavioral science.
JIM: It seems so obvious when looking back that usually these crashes occur during good times because people tend to extrapolate the good times into infinity. In your book, you mention Burton Malkiel, who has authored the book “A Random Walk down Wall Street”, and who talks about these “castles in the air,” about these bubbles--whether it was 1929, 1987, 2000, or 2001--and with these castles in the air, the only difference really is in the sectors that they occurred. For example in 2000, it was the NASDAQ. In 1987, it was other sectors of the stock market. In the late 1970s it was gold and silver. And it was electronic stocks in the 1960s, Japanese stocks. It seems to me that we always have a habit of building these castles in the air.
Negative and Positive Feedback
DR. SORNETTE: That is interesting. To make the comparison more profound, people were using the same words “new economy” both recently as we all remember, but also in the 1960s. You were mentioning the electronic boom that developed in 1962. The same words were used “the new economy” during this tronic boom of 1962. The same words of a “new economy” were also used in the late 1920's. At the time, AT&T was one of the hot company. So this is really describing a commonality associated with what we call in scientific jargon, positive feedback.
The point is that in the market, most of the time, the stock market is quite efficient. It is true that it is hard to make money on it. There are not many $100 bills lying on the street. They are taken at the very time when seen. And the same thing occurs in the stock market. As soon as there is a new arbitrage opportunity as it is called, people try to devise new strategies to make profit from it and this gives rise to so-called negative feedback. Negative feedback refers to the fact that if there are too many rabbits and they are eating grass, grass will be lacking after a while and the population of rabbits will drop. That is negative feedback. The fact is that there is a loop of interaction limiting the population after a while because of the lack of food.
That is exactly what happens when you devise strategies that take opportunities and then delete these arbitrage opportunities. There are also positive feedbacks based on the fact that people believe that things are going up and it creates an optimistic mood. This gives rise to positive feedback. Positive feedback refers to the idea that "the better it is, the better it will become." Positive feedbacks mean that anticipations are self fulfilling and the realized fact further amplify the anticipations in the same direction. For example, if you invested last year or three years ago in the stock market and made 50% a year and you tell me. I am your friend and find it hard to resist being tempted to emulate such a large gain. I would be attracted to also invest in the stock market. This would pour more money into the stock market. It will raise the prices and this will feed the price up, giving rise to the positive feedback enhancing the phenomenon. This castle in the air really refers to the idea that this positive feedback leads to price appreciations that build up with no relationship with the fundamental values.
The Importance of Drawdowns
JIM: Explain the importance of draw-downs, which I found to be very significant in your research in explaining why these crashes occur. You give the example of the stock market crash of 1987, where you had four consecutive days of market losses of 2%, 3%, 6% and 22%, where you had a 30% drawdown in a period of four days. Explain how the danger of events are correlated versus independent.
DR. SORNETTE: Yes, that is a very important point. Usually, in the economic literature, in the finance literature, in risk management, people study the statistics of stock market returns to attempt to understand and predict the potential gains and risks. What they do is to use statistics of returns at fixed time intervals, such as daily or weekly returns. The problem is that fixing a time scale may bias significantly the analysis and the conclusions.
Consider that a company has lost 30% over a short period of time. Let us assume that this significant crash has occurred over three days. In this simple example, let us assume that the first day has a 10% drop, the second day has 10% drop, and the third day has also a 10% drop. That leads to a drawdown of 30%. A drawdown is a run of cumulative losses (three in my example).
When this 30% drop is analyzed statistically at the daily scale, as is usually done in standard statistical analysis, it becomes three independent events of 10% drop. A 10% drop for example on the NASDAQ occurs once every four years statistically. It is an event that has a probability of one in one thousand to occur at any given day. It is not that rare. In contrast, what is very rare is to witness three such daily losses occurring in a run of three consecutive days, cumulating into a 30% drawdown. It is such drawdown that is hurting your portfolio, not really the daily losses, but the runs of losses. This is why I stress in my research the important of looking at these drawdowns which are much better measures of the large risks and the crashes. Now the probability of three such 10% drop events to occur in a run is the probability of one event -- times the probability of the second event -- times the probability of the third event. That is one in one thousand to the power cube. That gives a probability of one in one billion. Translated into the time of recurrence of such a run or a drawdown of 30%, it corresponds to one event in four million years.
Thus, people who are not taking into account the fact that you have a run or drawdown, which gives rise to the dramatic drawdowns, are providing an incorrect description of the event because they would predict that such events would occur once in four million years. Such large recurrence time means that such an event is practically impossible and one should never observe it. In reality, drawdowns of 30% or more are seen very often these days. So, what is the solution of this puzzle? It is that the market is characterized by the existence of strong dependencies between successive days. Such strong dependencies are however rare and difficult to qualify and detect by standard statistical techniques. We have developed specific tools to quantify them. We say that these strong dependencies occur intermittently in pockets of predictability.
The mechanism underlying the appearance of these intermittent dependencies is to be found in the positive feedbacks I was referring to before. Once the market starts to drop, there may be positive feedback which may amplify this drop further down. These positive feedback mechanisms may have different origins, both technical and psychological. Technical mechanisms are for instance found in the impact of portfolio insurance, which has been invoked as the explanation for the October 1987 crash. In portfolio insurance, when the market start to drop and the portfolio starts to lose money, the automatic trading rule is to sell to cover a part of the loss. But selling may amplifying further the downward trend of the price by increasing the selling pressure. Another example of technical positive feedback is found in the hedging of options or more generally of derivatives. Psychological positive feedbacks can be found in the tendency for investors to follow trends. When you start to lose and when you lose significantly, you look around you. You speak with others. This may lead to herding and crowd effects, panics that enhance the drops. In sum, all these mechanisms may give rise at some times to transient correlations. Technically, the problem we are facing is to decipher the information of the possible unraveling of a transient correlation in an ocean of price patterns that are most of the time uncorrelated. One way is to study the drawdowns and develop methods for identifying the existence of special drawdowns that we call outliers, because they are extreme events outside the regime of normal behavior of the stock market prices.
Outliers and Extreme Cases
JIM: Let’s talk about the importance of a large body of evidence of these outliers or these extreme events. In this case we are talking about stock market crashes. Does it rest on a distribution to be smooth? You have different distributions of returns that people measure, but these outliers are extreme events, as you've just pointed out, and they are much more frequent than one would suggest by looking at financial literature.
DR. SORNETTE: Yes, this property refers to the effect of transient correlation that I was describing above and can only be detected by using flexible time scales to search for the existence of anomalous drawdowns. If you do not adopt a flexible time scale for your analysis, you would conclude incorrectly that outliers should never occur or that a 30% cumulative drop should occur only in four million years.
Mathematically, we analyze the tail of the distribution of drawdowns in a given financial time series and we find two regimes. We find typically that 99% of the distribution of drawdowns can be explained by the standard theory of more or less uncorrelated returns; this corresponds to the normal time, of business as usual. However, about 1% of all the drawdowns are found to be outliers, to be characterized by a different distribution that falls off much less rapidly that predicted by the extrapolation from the 99% bulk of the normal distribution. Typically, we find that the largest drawdown outliers occur about 100 times more often than they should, if they were predicted by the extrapolation of the body of the distribution. This is the bad news. The bad news is that these dramatic losses occur much more often then extrapolated or predicted from the distribution of the small ones. The good news, however, is the fact that they are outliers, that is, they are fundamentally different from the rest of the population. If they are different, there is the possibility to have distinguishing patterns, that may make them predictable. For instance, if they are much bigger and if they are occurring much more often than expected from the extrapolation of the 99% bulk part of the population, this is probably because amplifying mechanisms are at work to create them. If there is a mechanism amplifying the occurrence of drawdown outliers, it might suggest that we could understand this mechanism, use it maybe to predict their occurrence, at least with a partial success. This constitutes the starting point of our approach.
JIM: Let’s talk about some of the mechanisms that lead to their occurrence. I want to move on to how different financial theories look at these events and the different approach that you are taking. We have, for example, the random walk theory and we have behavioral finance theories. You are taking a look at behavioral aspects of the individual and then correlating that with how the individual acts in a group collectively that can contribute to these crashes.
DR. SORNETTE: The standard theory of finance postulates the efficiency of the market. This efficient market hypothesis holds most of the time, as I was alluding to before. How to find arbitrage opportunity? You have to really do your hard homework in order to beat the market. Malkiel in his book has documented that 80% to 90% of professional traders do less well than those who use buy and hold strategies. Most of the time, the stock market is quite efficient. However, more and more academic work and also experts find that there are some arbitrage opportunities. Quite rare they are, but apparently there are indeed pockets of predictability which indicate the existence of deviations from the perfect textbook efficient market hypothesis.
There is a strong body of research in the last decade which is blossoming more and more, putting emphasis of what is called behavioral finance. Behavioral finance the idea that people are not these rational, extraordinary super computers, able to analyze continuously all situations and anticipate perfectly. In fact, we all know that we are just limited flesh bodies and minds with limited cognitive processing abilities. We make mistakes as buyers. I want to refer for example to the last Nobel Prize in 2000 in economics, given to Professor Kahneman and Professor Smith for their work on behavioral finance and on experimental economics. They recognized that there are indeed other ways of describing what a person is, in order to understand the stock market. However, the difference between their approach and the one that we have been developing in my group is that their approaches, even if it takes into account human behavior, only focuses on the individual. These approaches focus on trying to describe as best as possible each single investor, with his or her limitations, biases, idiosyncrasies, and so on.
However, the stock market is the result of the collective behavior of all investors, who not only interact through the formation of prices, but also interact by belonging to a network of information transfer. Typically, a proficient trader calls half a dozen to a dozen or more friends or colleagues to share information, to share the sentiment of the market, to exchange their moods. They also interact through the media, as we are discussing today. This may have an influence and pertain to the information that is spread over the network of investors. So, through this network of interactions, a collective behavior develops. This transcends, in a way, the bias of the individual behavior on each other. This aspect has not been answered at all in most of the work of my colleagues.
This is the key point in order to understand the complex behavior of the stock market, which is where my work is different than that of most of my colleagues. Putting emphasis on the behavior of individual people might miss the “the forest for the trees” in the sense that collectively our group behavior has been known and is now well-documented to be different from the behavior of a single individual. That is where we are trying to put the most emphasis in our work in order to get access to the predictability of these outliers, of these extreme crashes.
JIM: Do you see, with the advent of modern communications, the inner-connectedness of markets today that move at lightening speed, and also the growing emphasis on technical analysis? For example in New York, in Bonn, in Tokyo, London, Hong Kong, traders are using the same kind of approach, technical analysis, the same computer programs and watching the same thing. Do you think this is contributing collectively to emphasizing more and more of this kind of event as being possible?
Synchronized Markets
DR. SORNETTE: Yes, this is one of the mechanism of the whole process. We have found something that astonished me very much. In one of our last technical papers, we analyzed and compared most of the largest stock markets in the world. We analyzed the 30 stock markets in the world, and especially the western stock markets, including Japan, Europe and US. And we found--remarkably woven in the technical details--we found that since 2000, especially since the summer of 2000, which is to us the start of a world-wide anti-bubble--we found that all stock markets are almost completely synchronized. You don’t need to develop highly technical mathematical analysis to see it. This, I have never seen ever in the last century, ever in any stock market in the world. Indeed we see signatures of the globalization of the single village, as we say. It is probably this sharing of this information, of computers, of algorithm, of technical skills and all this participates to the development of the global village and of this cooperativity.
JIM: Now, let’s move on to when you were examining the work of others. You took a look at why stock markets crashed and like any scientist, you formed a theory. One of them was that financial data were random. As a result of your various analyses, you rejected that the exponential model for the distribution of drawdowns (which is the prediction coming from the assumption of independent sequential returns) doesn’t apply and in particular as you looked at the crash of 1914, 1929 and 1987.
DR. SORNETTE: Yes, we find some interesting things. The problem in the academic literature has been to try and qualify speculative bubbles. The term “bubbles” refers to the “cattles in the air” that we were referring to before. Bubbles refer to the situation when prices appreciate significantly above the fundamental price, so that the price becomes completely unrelated to it. The question is: How can you say for sure that at a given time we are in a bubble and at another time the price is right or is valued correctly, based on the fundamental valuations of the cash flow of the companies, of their earnings, dividends, the interest rates and so on? It is very difficult actually, and the debate on how to detect bubbles is still going on in the academic literature. A large part of the academic community has actually lost hope of distinguishing bubbles from normal times. It is not very hard to argue against the occurrence of bubbles. You could always develop new arguments to say the price is justified. Then later, after the fact that the market has crashed, in hindsight, it becomes clear that the price was too high, and you can say that indeed there was a bubble, like the 2000 Bubble. But to decipher the existence of bubble before a crash is very difficult indeed.
Our work has focused on a new method to identify these bubbles. It is based on a combination of two patterns. One is based on the positive feedback phenomenon I was stressing before, which many times gives rise to this super exponential growth of the stock market for a while. The other pattern is based on the competition between positive feedbacks and negative feedbacks, together with inertia. Inertia is the idea that when you make a decision for investing, you make this decision based on past data and information--say from yesterday to today, last month to today--and then your investment will be realized tomorrow. Typically, there would be a delay. And this investment will impact the stock market with a delay. This delay between information processing, decision making and investment, gives rise to inertia. The same kind of inertia describes when you are driving a car and make a turn, there is a centrifugal force which is applied to your body which is just reflecting the mass inertia. In the stock market, there is also an inertia, like a mass, if you like a mass inertia in our investment decisions. The combination of positive feedback, negative feedback and inertia, gives rise to new patterns that I call log-periodicity, which together with a super exponential growth provide distinguishing patterns qualifying the existence of bubbles.
Circuit Breakers
JIM: Let’s talk about the nature of systemic risk. In reading your book, it really came through the realization of these large drawdowns that led to crashes, resulting from our run of losses over several successive days. Yet after 1987, we put in circuit breakers. But as you talk about in your book, these circuit breakers expose the illusion of market liquidity.
DR. SORNETTE: There are pros and cons for the circuit breakers. For example, speaking of October, 1987, when we make a comparative study of the largest stock markets in the world, a few stock markets had already circuit breakers and they crashed even more. The pros and cons are really complicated because nothing is clear cut. You could say circuit breakers give rise to the possibility of calming the game down. Giving the possibility of people speaking to each other, speaking to their brokers, assessing the situation, looking at the liquidity, so a better assessment of the situation should provide calm and that might stop the crash. On the other hand, you could argue and other people have argued, that this can actually enhance the frustration of not being able to make concrete the open interests or the open positions and accentuates the crash by lack of liquidity. So this is a quite complex issue and I would say that in the end, the solution would not be resolved by technical arguments, but to me would have to refer to the confidence game, to the scientific collective process. Circuit breakers interfere with it, depending on the over all mood and how the process is going on and what is the underlying instability I was referring to earlier with the example of the unstable pen. How unstable is the situation? If we understand the underlying forces, then we might be able to understand the role of circuit breakers and other regulations. The effect of circuit breakers should be studied together with the collective behavior of investors to understand the positive and negative feedbacks that may result. I am not ready with answers for that problem, more study has to be done in order to understand the interaction of circuit breakers with systemic instabilities resulting from collective phenomena.
JIM: I wonder if you might discuss the role as it relates to positive feedback, in the sense of the introduction in the 1980’s and in particular 1990’s, the role that derivatives and hedging plays in this positive feedback loop, because one thing that we did see is with the introduction of these derivative instruments, you had the introduction of low cost speculative vehicles that came into play in the market place.
DR. SORNETTE: Derivatives have often taken the position of the accused in the fact that they have been attacked as being the instrument of instability. Often big numbers are invoked to illustrate the potential power of destabilization that derivatives may exhibit. For instance, one often hears that derivative positions at any time total 100 trillion dollars. These quotes are misleading in the sense that they are not really the true amount of dollars put into the trading of derivatives but they are corresponding to the value of the underlying asset, which is an artificial statement. Having said that, it is true that anything that allows people to immediately take opportunity of feelings of arbitrage opportunities, of herding according to these collective phenomena was I referring to, might enhance the instability. On the other hand, the argument is also that the more liquidity is available, the more flexibility is present, more ability to trade will wash out the arbitrage opportunity and make the market more efficient. Most arguments are actually valid all together. The problem is that we have to see the problem at the larger scale. Most of the time, indeed the instruments work for better liquidity or better efficiency of the market, practically washing out these arbitrage opportunities allowing a lot of liquidity arbitraging risk, hedging risk and so on. However, these are the goods signs when there is chaos, if you like, when there are many people buying and selling, when the market is mostly stable. However, when we enter a phase of positive feedback, then these instruments may participate, as many others to the global instability, so I don’t see them as the devil, but as part of the global processes that may produce speculative bubbles, among many others.
You don’t need derivatives actually in order to get systemic instability; they would be just the start of the herding positive feedback phenomenon. Actually without derivatives, we have had dramatic crashes much before the 1980’s and 1990’s. I would say that it is not the derivatives that impact more and more the instability of the market. We see more and larger moves, larger volatility. To me, this points to the fact that it is more and more the signature of collective phenomena controlling the stock markets. The global village is taking over, not the derivative. The controlling factors are the herd effect, the fact that we are acting as coherent groups of investors imitating each other even not knowingly. This leads to the higher level of volatility and to the higher level of systemic risk. I was referring to the fact that, in the last two years, 13 of the most developed stock markets in the world are basically moving as one single market, in synchrony. There is such a strong correlation. This means that, if you are an investor and want to diversify internationally, it is not possible any more. You should just buy the US stocks, because buying in Europe is the same. Of course, there are still ways to diversify according to our study, you should buy Chinese and Russian stock markets, but will you do it? That is another level of risk. These are the issues that have to be kept in mind I think.
JIM: This leads me to my next question. Explain the role of the various herding plays in making markets abnormal. We have the international cascading affect, reputational herding, investigative herding, institutional herding and of course the effect of rumors.
DR. SORNETTE: Herding takes many shapes. Herding is often refered to as being a rather stupid behavior on the part of people. Lemmings herd and we are used to view with contempt herding behavior. For example, I remember the cover of the Economist on the week after the big turmoil on the US market following October 27, 1997, when the market dropped by 7% on the DOW Jones and rebounded the next day and many professionals thought that a crash was coming. The cover of the Economist was showing this very nice cartoon of people herding, based on a rumor or false information. That is not reality. People herd, not because they are stupid. If you or I want to invest, we are not going to throw our dollars or our Euros blindly. We will take some advice and think about it for a while and of course, a professional would think even harder about the job of trying not to lose. They don’t want to be losers and their income is a function of their relative successes with respect to that of their colleagues. Most people don’t herd because they are stupid like lemmings or sheep, but because it might be rational to do so. When you lack information, we remember in our school days that it was rational to look over the shoulder of our friends to find the answer when we do not have the solution. More seriously, theories in science show that when you lack information, it becomes rational to herd, to imitate your neighbor because his behavior may contain useful information you don’t have. That is one point; you can herd simply on the basis of lacking information and trying to catch up the information. Another aspect of herding or imitation is based on the recognition that imitation is actually one of the most, if not the most profound base of cognition in humans. Only the highly intelligent animals imitate, like apes or chimps, dolphins and a few birds know how to imitate. We know that culture and knowledge can be transferred by mother apes to her children by imitation. That is actually what we use most of the time, at school from infancy to adult in order to learn. This is the most important and most powerful way of learning. So, imitation is actually the highest cognitive level of learning for humans. I think that is why it is wired very strongly in our brain. Of course there are times when imitation is not that bright, but I would like only to insist on the good side of imitation and most of the time this is probably what is occurring.
JIM: In your book, you made the statement, "Crashes are made possible when order wins." Explain that concept.
DR. SORNETTE: This is very simple. Most of the time, stock markets do not crash. Why? Approximately, there is a balance between buy orders and sell orders. Price remains approximately the same, because price is based on the competition between supply and demand. This is the chaos time. This is the disorder case when people disagree. You have as many people who want to buy as there are people who want to sell, so there is chaos in the stock market. Everybody disagrees and this is the healthy state of the market. This provides liquidity. At the time of a crash, unfortunately, everyone wants to sell at the same time. It is a time of order, in the sense that everybody agrees. There are few or not dissonant opinions. Everyone goes in the same direction. There are no counterparties. No one wants to buy. Everyone agrees. It is the perfect order in the sense that everyone has the same opinion. Of course, this creates chaos at another level, but I hope you get my meaning.
Autopsy of a Crash
JIM: Sure. In your book, you did an autopsy of past market crashes, in particular, the crash of 1929, the crash of 1987, and the NASDAQ crash of 2000. What were some of the common characteristics that you found in your autopsy of these past crashes?
DR. SORNETTE: What is fascinating in all these crashes, and I could add many other crashes to this list, for example the crashes in East Asia in 1994, Hong Kong in 1997, etc. All these events are characterized by practically the same story, which can be decomposed in four or five stages.
The first stage is when the economy, the underlying economy, is doing well as a result of various factors. As a consequence, the stock market appreciates a little, as it should because the price reflects the value of the company, the earnings flow, the dividend flow, and this is a good time. The second stage is when the stock market has appreciated and is going well. This attracts new capital. Banks start to free up more availability of money or cash for investment. Money is available with smaller margins or with less collateral, (I am not referring obviously to the top of the Japanese real estate money, where in the end the banks didn’t ask for any collateral and were getting more than the value of the properties.) So, the second stage is when banks loosen the limits to the availability of money by a variety of means. Having more money available, this money has to go somewhere. With the same number of stocks, typically the price goes up. The third stage is where the price is going up more and more. This attracts less informed investors. The situation develops more and more. The bubble develops because, of course, gains are attracting more and more people. People are putting money into the stock market. International investors put more money in the stock market, with the same amount of stock and with more money. The stock price has to go up, because there are more buy orders. Until the fourth stage when the price is so large, so much higher than the fundamental price. The castle is so high in the air, to come back to that expression by Malkiel, there is no more a reasonable reference price. Now your grandmother starts to take the cash from under her mattress and put it in her stock market portfolio because it is becoming irresistible. So we come, of course, to the situation in the fifth stage where the price is so high that we have no connection anymore with the fundamental price. The consequence is that savvy investors start to take out their money before the others. This may give rise, together with any other contribution to the instability, to a depletion of the engine powering the bubble, which ultimately leads to a crash. As I said in the beginning, the fundamental reason for a crash is that the stock market has evolved into a very unstable situation and practically any news, behavior or impact may make it crash within weeks or months once the market is ripe.
JIM: In the conclusion to your book, you state proposals to mitigate these crises. In your view, they have no real chance for success. Why do you believe that is so?
DR. SORNETTE: Maybe I am not as specific as you seem to imply. The point is that these crashes are really systemic instabilities. They result fundamentally from the collective behavior of people. It is not through simple regulations that we shall stabilize stock markets, because the stability of stock markets has to be found fundamentally in the psychology of people. As long as we want to keep our freedom of investing, as long as we want to move around capital, then it is basically the effect of the positive feedback phenomena that is intrinsic to the occurrence of these systemic instabilities. In order to prevent that, we would have to take steps at the very strong macro-economic level, maybe fighting against some of the freedom or rights that we love. It is with this that I am a little pessimistic.
On the other hand, the positive view is the following. In California, where I live, there are many interconnected problems. Earthquakes are a problem. They can destroy a huge amount of property and kill people. However, if you look at the risk-return balance, I would rather live in California than in another country without earthquakes. Most of the richness of California, the geological richness, the beautiful landscape, the mountains, the oil, the agriculture, the attraction to people, comes from the fact that you have a rich geology here. I put on my geo-physicist hat; you have a rich geology coming from earthquakes. You have active tectonics that lift up mountains and create this beautiful scenery. So you have a very interesting trade-off in the fact that you have all this wealth and beauty, which comes from the fact that you have this earthquake dynamics. I see it in a similar way for the stock market. We are paying somehow by these rare extreme events, the price for our freedom for doing what we want, for having a liquid market allowing most of the time an efficient allocation of resources. Of course it is up to us to know what is going on and I hope my book will help in this direction. If we have more information, maybe we can personally take steps to protect against these instabilities and live on the good side of the complexity.
Anti-Bubble
JIM: Finally, given your knowledge and study of the markets and why they crash, what is your view of where we are in the market currently. On your website you say that basically there is widespread evidence of cooperative herding and imitation working in this bear market that began in 2000.
DR. SORNETTE: Yes, we are not in a bubble, clearly. We are in what we call in my group an “anti-bubble”. We coined this term “anti-bubble” taking as an inspiration the concept of an “anti-particle” in physics. An anti-particle is the same as a particle but it annihilates when it encounters its sister particle. A positron is the anti-particle associated with the electron for instance. There are both the same and the opposite in a certain sense. I call the present trajectory of the US stock market an anti-bubble to refer to the fact that we see the same kind of patterns reflecting the positive feedback phenomena, but this does not lead to a bullish market but rather to a bearish market. Similarly to the characteristic power law acceleration and logperiodicity patterns characterizing bubbles, we find a power law deceleration and logperiodicity patterns characterizing the present trajectory of the US stock market. The patterns again reflect the competition between inertia and positive and negative feedbacks that I was referring to before. These are the technical elements that make me think that there is indeed a herding phenomena going on. In addition, there is the fact that I was also referring to before that the 13 most developed stock markets in the world have been essentially coinciding in their structure over the last two or three years.
I have published with my collaborator W.-X. Xhou at UCLA on December 2, 2002 , a paper published in Quantitative Finance, a scholarly journal. The paper was titled, The descent of the US Stock Market, how much longer and deeper. In this paper, we have issued a formal prediction of the future trajectory of the US stock market based on the ingredients I have described above. The prediction was made in August 2002, based on the data up to the end of August 2002 and we predicted that the stock market will go up until the first to the second quarter of 2003 and will then start a long descend until around the end of the first semester of 2004. That is what I can say at the present time. Each month, we are updating this prediction which is available on my website.
I would like to stress that making such a prediction is a rare thing to do for an academic. We did it because I wanted to do a real-time experiment. It's quite hard to make an experiment in social sciences. Of course one way to show you are right is to make money with that one thing. But my passion is research and I have not much time to give to create a company for implementing this. Even if I have colleagues I am working with who do it on hedge funds, I prefer to build up more and more of the research. It is a very exciting subject as are the real-time predictions that are developed as a way to prove or disprove what we are developing. This is one of the several experiments we are running.
JIM: For our listeners of this radio broadcast, we will be putting your web address and your predictions, so our listeners can follow how you progress as you go along with this. Professor, a final question, after studying stock markets and crashes, if there was one thing in your book that you wanted to point out to your readers. What would that be?
DR. SORNETTE: One thing about my book is that, for me, writing this book was not so much about speaking of the stock market or the crashes, but to try and pass over to the readers the fascination I have for extreme events. You were mentioning that I have been dedicating my life to try to understand, to model and maybe to predict catastrophes. Here and now, we have been speaking only of stock markets, but I am working on the prediction of earthquakes, on landslides, as well as several other catastrophes. I have done a lot of work on the prediction of engineering catastrophes, like rocket crashes. I am interested in prediction of parturition, which is the act of giving birth. This might not seem like a catastrophe to you, but this is an extraordinary event in the lives of women and of all of us. It turns out that there are similar structure as we find in the other examples, even in the crashes.
To me, the subject of my book is about passing over the fascination on extreme phenomena that govern the world. To make that we live in a fundamentally dynamic world both natural and social, and extreme phenomena or crises are part of our landscape. Even if they are rare extremes, very rare, they nevertheless control most of our lives. If you think about the universe, it was created in the big bang, which is arguably the largest catastrophe we can think of. The solar system was made by energetic impacts between planetesimals. The landscape that we see around us was made by a few very rare millennium flood that move boulders the size of small mountains. The landscape that we see, mountains, was from landslide or mountain collapse that occurred once in 50,000 years. The rest of the processes, the erosion, are mostly negligible. Everywhere you look around, the country we live in was made by a few revolutionary and wars. Even our personal life is made of a few bifurcations. That is my fascination with life and this is the main message of the book which is applied to the stock market.
JIM: I tell you, you live in a beautiful part of the world, joining us from Nice, France on the French Riviera. I understand that you will be coming back to the US.
DR. SORNETTE: I would say that California is not bad either.
JIM: You have the best of both worlds, Professor. I want to thank you for joining us on Financial Sense Newshour. The name of the book is Why Stock Markets Crash: Critical Events in Complex Financial Systems -- definitely worth reading -- by Didier Sornette. Professor, once again thank you for being so generous with your time.
DR. SORNETTE: Thank you, it was a pleasure.
©
2003
Financial Sense™ is a Registered Trademark
NOTICE: This
transcription may NOT be reproduced without the expressed, written permission of
Financial Sense Online. Email
FSO Selective quotations are permissible as long as
this web site is acknowledged through hyperlink to: www.financialsense.com