A crash is a precipitous collapse of listed assets. It must be violent and spectacular, and must cause serious collateral damage.
Its suddenness arises from a recurrent process: an upturn in the economy attracts investors of every sort, and this inrush of capital feeds the further expansion of the market. The process then becomes uncontrollable if investors borrow money to invest in the market. The value of the securities thus becomes detached from the value of the listed companies, as determined according to the basic criteria employed by market analysts.
This discrepancy in valuation feeds an over-valuation of the market, called a speculative bubble. In this case investors decide both to quit the market and abandon any possible remaining speculative profits, or to stay in while nevertheless awaiting any data, news, information, or even rumors that might deliver "the signal" that will burst the bubble and bring down the markets.
A number of phenomena are responsible for crashes, but the mimetic aspects of investor behavior and standards outweigh the valuation of securities on their fundamentals. The dynamics of crashes thus appear to be behavioral in nature. (3) Daniel Khaneman received the 2002 Nobel Prize in Economics for his work on decision-making under uncertainty. His work gave birth to behavioral finance (4), one of the two themes of this article, which seeks to explain the current financial crisis (5). Traditional finance has evidently become incapable of explaining our successive crashes. On the one hand, the various explanations put forward for these phenomena are inadequate (for example, explanations for the 1987 crash are unsatisfactory in that they ignore the interpretational aspect of data, news, information, and rumors, which is a central feature of this type of event). On the other hand, the assumption that they are caused by rational actions is very much open to question, being unable to explain why bubbles appear, and still less what causes them to burst.
The first part of the article proposes a model that incorporates behaviors that are more "realistic" and familiar to market actors, so as to provide answers concerning the mechanisms that bring about crashes. This model is derived from the theory of informational cascades (Bikhchandani, Hirshleifer, and Welch, 1992), which we have adapted for application to financial markets. The constraint of rationality will thus be removed and behaviors involving over- and under-confidence will be introduced. Overconfidence is one of the behavioral biases most frequently discussed in the academic literature, and for some authors (De Bondt and Thaler, 1995) the fact that individuals may be overconfident is perhaps the most robust finding in the field of the psychology of judgment. Overconfidence is defined in its classic form (6) as the over-estimation by individuals of the relevance of their private information. Underconfident behavior, often observed simultaneously (Kirchler and Maciejowsky 2002) will in contrast be defined as the over-estimation by individuals of any public information. Modeling of these behavioral theories sheds light on the origin of crashes, since underconfident behavior does indeed lead to mimicry and to speculative bubbles.
The second theme of this article relates to the development of new, worldwide accounting standards whose use and interpretation have accompanied-and perhaps contributed to-this same crash.
II. PROPOSAL FOR A MIMETIC MODEL WITH BEHAVIORAL ASSUMPTIONS
Since the end of the 1980s and up to the present day, a stock-market crash, and especially the one in 1987 or the internet crash in 2000, has been seen as the bursting of a speculative bubble (Miller, 1991). We therefore say (Garber, 1990) that a speculative bubble is simply a rapid rise in "basic value".
But this goes against what most researchers recognize, namely that bubbles are an essential feature of markets simply because technology evolves, and retires sections of economic activity that were formerly dominant. …