Thought Contagions in the Stock Market

Aaron Lynch

Published in the Journal of Psychology and Financial Markets 1: 1, p. 10-23

Received: August 20, 1999
Copy Edited: January 24, 2000.
Publication Date: March 24, 2000.

Paper presented on December 11, 1999
At the Institute For Psychology and Financial Markets
Meeting at the Aspen Institute, Aspen, Colorado.

[Post-publication note: The term �meme,� given a technical definition herein, has been dropped forthcoming papers in favor of more self-explanatory terms such as "idea," "belief," "behavior," "artifact," "thought contagion," "doctrine," "opinion," "belief system," "rumor," etc. Credit for early work on self-spreading ideas has also been given to a 1973 article by F. T. Cloak, available at www.thoughtcontagion.com/cloak1973.htm.]

ABSTRACT

The evolutionary epidemiology of ideas, or thought contagion theory, is introduced and applied to possible examples in the stock market. It is suggested that differences in transmissivity, receptivity, and longevity of belief may contribute numerous irrational influences on the stock market, generating sources of inefficiency. These include a wide variety of mechanisms that may generate both positive and negative market overreactions. The soaring prices of Internet stocks during 1998-1999 are used as an example of how investment ideas correlating with new communication behaviors may affect share prices, and how contagion effects in general can affect the broader market. New avenues of empirical investigation are proposed to test the types of hypotheses presented.

INTRODUCTION

The theoretical paradigm of thought contagion theory examines how the recursive transmission of beliefs results in some ideas becoming widespread while others become rare, ultimately affecting whole societies. For example, thought contagion analysis might reveal that certain fad diets propagate because they fail to achieve lasting results. By leading dieters into repeat episodes of dramatic weight loss, the least effective fad diets can attract the most attention, causing adherents to be asked more frequently �how did you do it?� � and leading to more word of mouth transmission.[1] Thus, in addition to considering how cognitively or emotionally susceptible a person is to persuasion, the theoretical paradigm encompasses novel hypotheses about how ideas �program� their own retransmission or otherwise assist their own spread.

Thought contagion theory holds that ideas propagate in ways that do not depend upon truth, utility, rationality, or even emotional appeal. These include mechanisms of greater peer-to-peer transmission, abundant parent-to-child transmission, prevention of adherents dropping out, and interference with the spread of competing beliefs. Ideas harnessing these human functions most effectively win out over weaker variants. Often evolving like life forms, through evolution by natural selection, thought contagions vie for ever stronger influence in human lives (Lynch [1996]). As an area of prospective empirical research, this broad thesis of the evolutionary epidemiology of ideas is currently considered a theoretical paradigm rather than confirmed theory.[2]

Analogies between psychological crowd contagion and biological contagion have been drawn for over a century (e.g., Le Bon [1895]), as have analogies between cultural evolution and biological evolution (James [1880]). Building upon those older lines of work is one aspect of the evolutionary epidemiology of ideas paradigm. Yet the evolutionary epidemiology of ideas has an added dimension not found in nineteenth-century works: evolution of ideas happening through distinctly epidemiological mechanisms.

Similarly, Ewald describes the evolutionary epidemiology of microorganisms as �an emerging discipline� (Ewald [1994, p. 7]). Evolutionary epidemiology is not simply contagion, nor simply evolution, nor even simply evolution plus contagion. A well-known biological example is the emergence of drug-resistant bacteria resulting from the wide use of antibiotics. With ideas, a comparable example is the emergence of refutation-resistant strains of belief. One class of refutation resistance comes with grandly conspiratorial ideologies for which all objections can be dismissed as either part of the conspiracy or as harm done by the conspiracy (e.g., Lynch [1996, 1999], in press).

Another example is how movements such as religions can evolve from mainly peer-to-peer transmission to mainly parent-to-child transmission after spreading so widely that few persuadable peers remain (Lynch [1996]). In the evolutionary epidemiology of ideas, the term �epidemiology� refers to helpful, harmful, and neutral ideas.[3]

As with older works, the evolutionary thought contagion paradigm proposes numerous processes generating forms of irrationality and delusion that can only be achieved by a population, and that add to the irrational aspects of human psychology at the individual level. Political ideologies, major religions, common myths, health fads, prevailing mores, national controversies, and ideas about investing are all affected. In particular, thought contagions appear to affect the prevalence of beliefs about the stock market and specific stocks, thus playing a major and frequently irrational role in determining share prices. Bubbles, panics, market rumors, sharp rises and declines in the Internet stocks, and a wide variety of other beliefs overvaluing or undervaluing stocks can be investigated from this novel and quantifiable theoretical framework. With such important phenomena affected, we hope further research into thought contagions in the stock market will follow.

Irrational manias and panics in financial markets have been documented for over a century, as with the vivid description of the seventeenth-century Dutch �tulipmania� in the classic work of Mackay [1841]. That crowd psychology differs from individual psychology has been documented for over a century by both Mackay and Le Bon [1895]. That population intelligence differs from individual intelligence continues as a theme of the evolutionary epidemiology of ideas (Lynch, in press).

Thought contagion analysis also moves beyond describing the mass phenomena and eventually toward identifying the modes of transmission that determine which beliefs attain self-sustaining propagation and which ones decline or stagnate. We are interested in how the nature of what is believed affects propagation differentials, which are then recursively compounded to form a mass movement or delusion.

Such propagation differentials are associated with specific forms of transmission events that can be written symbolically. If we label a given idea as �idea A,� then typical transmission events would be A + ~A � 2A (host of A plus non-host of A yields two hosts of A); 2A � 2A + ~A � 3A (two hosts of A have a baby non-host of A who is subsequently inculcated with A by the parents); A � ~A (host of A drops out); and so forth. Events involving combinations of ideas can be represented in similar fashion (Lynch [1998a]).

Part of the importance of such event diagrams is that their form determines much of their mathematical implications for host population growth or decline. Once measured, the propagation differentials from each kind of transmission event can be analyzed either by using systems of non-linear differential equations modeling the changes in host populations over time, or by using computer simulations with event rate parameters (Lynch [1998a]).

Others have also used quantitative methods to model various aspects of cultural evolution and transmission (Boyd and Richerson [1995]; Cavalli-Sforza and Feldman [1981]). This article, however, discusses propagation differentials as areas of prospective investigation, and does not reintroduce the equations or use the event diagrams extensively.

Some propagation differentials arise from the effects of beliefs upon one�s own ability to communicate beliefs. The beliefs in profitable futures for Internet companies, for instance, may correlate strongly with personal abilities to send and receive ideas electronically, thus initially favoring the transmission of optimism about specific Internet companies and the Internet sector as a whole. Preliminary evidence suggests that online chat about eBay exceeded online chat about Ford in the 1998-1999 period, as discussed below. And online chat about the Internet economy as a sector, combined with company-specific online chat, may have given EToys a much higher market capitalization than Toys �R� Us in its first days of trading.[4]

Belief and investment in a company that has recently grown (which will tend to validate feelings of self-confidence) may make the shareholder more proud to discuss his or her holdings than belief and investment in a recently devalued stock. This can further explain how some companies become overvalued while others become neglected.

Belief and investment in highly visible enterprises may cause the investor to be more frequently reminded of his or her investment, leading to more belief transmission. One factor in the visibility of a company is simple proximity, which has been found to correlate positively with investment rates (Coval and Moskowitz [1997]; Huberman [1997]; Lewis [1998]; Shukla and van Inwegen [1995]), turnover rates (Coval and Moskowitz [1997]), and investment performance (Coval and Moskowitz [1997]; Shukla and van Inwegen [1995]). As Coval and Moskowitz [1997] and Shukla and van Inwegen [1995] suggest, proximity can improve information flow to the investor, resulting in more profitable decisions. Yet visibility advantages for a company from factors such as advertising need not correlate perfectly with knowledge advantages to the investor. Visibility, therefore, might affect prices in ways not necessarily commensurate with earnings prospects.

Meanwhile, those who hear rumors about a company may retransmit them simply to find out if they are true � providing attention to a company that, again, may not correlate with earnings prospects. In the market as a whole, ideas of overall boom or cataclysm may make believers feel the need to spread the word in order to help others � still another influence that may not correlate well with earnings. These and other hypotheses about the effects of thought contagions on share prices are therefore proposed to help stimulate empirical investigation and quantification.

THOUGHT CONTAGION THEORY

Although thought contagion theory has applications to beliefs about religion, politics, sex, family life, war, health, consumer fads, fashions, and many other important topics, we focus on its applications to beliefs that affect the stock market. With its emphasis on population psychology, thought contagion theory does not replace the extensive work on the contribution of individual psychology to market forces. Similarly, the study of infectious diseases and symbiotic bacteria does not preclude study of non-infectious conditions.

Thought contagion theory also does not seek to supplant existing methods of identifying companies with lucrative new patents, solid business plans, or good cash flow. Rather, it aims to provide a more inclusive picture of how mass psychology combines with individual psychology to affect market behavior. In doing so, it often draws upon knowledge of individual psychology to help understand processes happening at the population level.

In 1976, zoologist Richard Dawkins noticed that self-propagating ideas bear some similarities to evolving life forms and genes. He gave them the new name meme (pronounced �meem�) in a short chapter on cultural evolution (Dawkins [1976]). He did not claim to have discovered a new kind of entity, but merely coined a convenient shorthand to label the subclass of ideas whose occurrence results from replication.

More technically, I define the term to mean �a memory item, or portion of an organism's neurally stored information, whose occurrence depends critically on causation by prior occurrence of the same memory item in one or more other organisms' nervous systems� (Lynch [1998a]). The scientific study of how memes evolve and spread is now known as memetics. Still, the more self-explanatory terms thought contagion and thought contagion theory often suffice for the topics here.

Thought contagions exert their effects on financial markets both directly and indirectly. Those affecting financial markets indirectly may include doomsday beliefs, political ideologies, the movements leading to and from war, religions, and economic ideas. For instance, proliferation of religious and moral beliefs might affect the market by raising the fertility rate, which influences investment rates by shifting the ratio of workers to retirees, while also affecting the fundamentals of everything from real estate companies to baby food makers.

In the late 1990s, a belief that the Y2K (year 2000) computer bug would cost trillions of dollars and even billions of lives has proliferated by inducing believers to go forth and warn others, while more realistic appraisals of the problem stimulated less thought and less urge to communicate. Those beliefs may have played a temporary role in raising the share prices of companies working on Y2K solutions. This hypothesis can be investigated by systematically tabulating Y2K Internet postings and looking for correlations with share prices of the Y2K companies. Y2K catastrophe memes have also slightly affected July to October 1999 markets by causing a small fraction of investors to pull money out of their investment accounts to �rescue� the funds from supposedly doomed computers.[5]

In a rather different mechanism, the spread of socialism in a given country can lower share prices in its stock market by making investors nervous about taxes and expropriation. Large numbers of idle people with plenty of time to spread their beliefs and a large anticipated reward for spreading ideas and winning political power can help the belief system spread. Beliefs leading a nation to war and war preparations can affect stock markets in general and the shares of defense contractors in particular. For example, a belief that �there is a dangerous foreign enemy who must be stopped� may spread because those who have the idea want to warn others in order to save their own lives and their compatriots� lives.

A variety of ideasthat spread parent to child and peer to peer can cause periods of changing family sizes such as the �baby boom,� the �baby bust,� and the newer �echo baby boom� as well.[6] The demographic fluctuations in turn may exert a delayed long-term effect on stock prices as people reach life stages of house buying, retirement saving, and retirement spending at varying rates (Bakshi and Chen [1994]).

Finally, economic ideas, such as belief that the economy has entered a permanent era of prosperity, can affect markets accordingly. Those holding such beliefs may invest more heavily and less carefully. They may also feel they have a relatively new, distinct, and vivid message to convey to others, leading them to retransmit their beliefs. These are just a few examples of beliefs that affect the financial markets by indirect but often powerful means.

Other thought contagions pertain more directly to the value of securities, as evidenced by an emerging subfield of behavioral finance research using such terms as herding, contagion of opinion, informational cascades, and mimetic contagion.[7] For contagion effects, these include mathematical models (e.g., Topol [1991]; Orl�an [1992], and Lux [1995]) and computational models (e.g., Lux [1998]). Using somewhat different quantitative methods, informational cascade theory also applies to stock market phenomena (Welch [1992]; Bikhchandani, Hirshleifer, and Welch [1998]). Moreover, survey evidence indicates that the epidemic spread of interest in specific stocks does play a major role in causing rapid price increases (Shiller and Pound [1989]).

These findings apply to both individual and institutional investors. Herding behavior has been investigated in stock analyst forecasts (De Bondt and Forbes [1999]), as has herding in general (e.g., Smith and S�rensen [1999]; Shiller [1995]). Collectively, they provide a basis for developing and empirically testing further hypotheses about the effects of thought contagions on individual stocks, market sectors, and entire stock markets. Further development of a theoretical framework for such hypotheses is thus in order.

TRANSMISSIVITY, RECEPTIVITY, AND LONGEVITY OF BELIEF

With any thought contagion, net propagation depends on such things as how many offspring are inculcated per host per year, how many non-offspring are persuaded per host per year, dropout rate, mortality rate, resistance to conversion to alternative memes, interference with the spread of alternative memes, intellectual appeal, emotional appeal, and so forth. These can be distilled into three general factors for each meme: transmissivity, receptivity, and longevity (Lynch [1998a]).

Transmissivity measures how much dissemination behavior the meme elicits from its existing hosts. In other words, it is a measure of how widely and often existing hosts attempt to spread an idea to others or exhibit behaviors that may incidentally spread the idea to others. A meme that spreads this way is the belief that you need to find a romantic partner of a "compatible" astrological sign. This idea causes singles who have it to talk about astrological signs with each new potential partner in order to determine compatibility. So the idea exploits human mating drives to get itself transmitted to more listeners.

Incidental transmission can happen, for example, with cigarette smoking: A parent may not want to pass on the idea, but may nevertheless exhibit lighting and puffing behavior that incidentally conveys the idea to children. Still another example is the pre-war Nazi idea of using violence to promote the movement. Those who disagreed were often intimidated into silence, which helped the Nazi movement to become a mass phenomenon.

An example of transmissivity in financial markets is the spread of an unconfirmed and unrefuted merger rumor. Those who have heard the rumor want to know if it is true, especially if it affects their work as traders or investors. So they may ask others what they know about the rumor, potentially transmitting it to previously unexposed non-hosts in the process. The newly exposed ask still others, resulting in rampant transmission that can dramatically affect share prices.

Another transmissivity effect can be seen in the recent phenomenon of day trading. The great majority of day traders lose money (Shellenberger et al. [1999]). But even when a majority are forced to give it up due to heavy losses, those who happen to win may be more visible and more vocal. Hence, they have more opportunities (and more incentive) to retransmit their ideas about day trading to reporters and potential new customers. Moreover, those engaged in day trading tend to think about their investments very frequently, which might lead them to talk about it more often to friends and family. Those forced out by heavy losses may be too embarrassed to talk about it. Thus, day trading need not be generally lucrative to enjoy episodes of self-sustaining transmissivity.

Within a meme�s host population, transmissivity levels can vary. It may be represented as a distribution function or broken down into propagation parameters corresponding to specific types of meme transmission events such as parent to child, peer to peer, etc.

Receptivity refers to how easily the non-hosts accept the meme. All the transmission attempts in the world don�t matter if the message always falls on deaf ears. The merger rumor, for instance, must enjoy not only transmissivity, but also at least some receptivity to spread. Subfactors of receptivity may include rates of sensory perception, (e.g., hearing or seeing a message), comprehension (understanding the message), cognitive and emotional reactions, and, finally, adoption.[8]

A meme showing high receptivity is the idea that friendly angels live in our midst and watch over us. The idea�s emotional appeal leads people to accept it more easily upon exposure to it, contributing to its wide prevalence in past and present societies. In the investment arena, almost any tip that promises rich rewards can enjoy high receptivity. People are generally eager to learn new ways of making money, so they often listen intently to assertions that a stock is �hot,� for instance. Yet intelligent people also form immune reactions to investment claims that sound too good to be true. This limits receptivity and provides at least some damper to the spread of boundless hyperbole � especially in populations where most people have been exposed to investment scams.

People may also be highly receptive to loss prevention memes that warn against imprudent investment, further countering the spread of boundless hyperbole by the action of competing memes.

Receptivity need not be uniform even across populations of those previously unexposed to a meme: Some may be highly susceptible to conversion while others may be extremely resistant. The others fill the various gradations of receptivity between the extremes.

Longevity refers to how long a host of the meme remains a host before dropping out or dying. A meme complex (set of mutually supporting memes) exhibiting great longevity is Freudian psychoanalytic theory. If someone tries to challenge part or all of the theory, the Freudian psychoanalyst can analyze the doubter�s motives in rather unflattering terms that quickly dispense with the challenge and leave the Freudian belief system intact. This feature may have helped Freudian theory achieve its enormous prevalence in the twentieth century.

In the realm of investment beliefs, as with many other areas of life, memorable concepts persist longer than unmemorable ones. Thus, one of the most prevalent concepts from technical analysis is called the �head and shoulders,� a supposed pattern of stocks climbing to a main peak, called the �head,� bracketed by two lesser peaks called the �shoulders� (see, e.g., Martin [1985]). Because it conjures a simple and memorable image, the idea can persist for years, regardless of whether it is empirically supported. Moreover, the �head and shoulders� idea refers to a supposedly ever-recurring chart pattern rather than just current market conditions, giving its hosts more rationale for remembering it over the long term than the simple idea �company A is hot� (discussed later).

Indeed, the �head and shoulders� idea may persist much longer than painful memories of losses incurred by using it. The longer an idea persists, the more chances its host will have to retransmit it, creating an added transmissivity advantage that usually accompanies longevity advantages. Note, too, that the simplicity of the �head and shoulders� meme makes it easier to communicate � another transmissivity advantage that combines with the meme�s longevity advantage.

Day trading may similarly exhibit some special longevity effects even when the trader is slowly losing money. Because day trading resembles gambling (Shellenberger et al. [1999]), it may exhibit a similar addiction potential. Such an addiction could tend to preserve both the behavioral pattern of day trading and the cognitive beliefs behind it, at least until losses become too great to continue.[9] That, in turn, keeps the day trader active and propagating belief in day trading for as long as his personal savings and credit lines can support it. It also helps the day trading firm make more money to spend on misleading advertising, offsetting customer failures. That helps preserve and expand the day trading company itself as a source of meme transmission. As with receptivity, longevity of hostship varies from person to person, and may be measured as a distribution function.

Memes need each of these factors, transmissivity, receptivity, and longevity, in order to propagate. Differences in the three factors determine how widely a meme spreads in comparison to alternative memes.

Transmissivity, receptivity, and longevity can affect not just decentralized communications such as ordinary conversation, but also the ideas spreading in highly centralized communications, such as ticker messages of share valuation or journalists getting ideas from each other�s media output. Although the truth and utility of an idea can contribute to the receptivity, longevity, or even transmissivity it enjoys, truth and utility alone do not assure high levels of these propagation factors. Likewise, wide propagation alone does not indicate either the truth or utility of an idea. In financial markets, this means that the success of a good investment strategy does not assure wide adoption and the wide prevalence of an investment idea alone does not assure effectiveness at earning money.

SIMPLICITY SPREADS

Transmissivity, receptivity, and longevity all tend to benefit from the simplicity of an idea. As we have noted, a simple idea is easier to express by someone who already has it, easier to comprehend and adopt by someone newly exposed to it, and easier to remember by someone who is converted to it. In the stock market, a simple idea that can spread easily is the idea that �company A is hot.� If its price has gone up for awhile, then one who believes that �company A is hot� can easily convince someone else merely by pointing to the charts or reciting prices.

Making a more realistic case for the growth prospects of company A requires knowing and expressing a far more complicated picture. But the difficulty of knowing and expressing the complex realities gives such knowledge much lower transmissivity. On hearing such information, one is less likely to comprehend and learn it as well, giving complex but realistic ideas less receptivity.

Finally, one is less able to remember complex and elaborate information about a company than to remember a simple idea that the company �is hot.� Since one cannot act upon or retransmit what one has forgotten, this again favors oversimplifications like �company A is hot� over more complex truths.

The spread of such simple ideas can send demand up, again causing share prices to rise far above a level justified by subsequent earnings. Similarly, a belief that �company B is a loser� can spread more effectively than the more complex realities, leading to a cycle of infectious undervaluation of company B. Reality eventually catches up, however, as earnings surprises come in and send the overvalued share prices sharply lower and the undervalued share prices sharply higher (Dreman [1998]).

Since the idea that �company A is hot� enjoys the greatest receptivity when company A has recently appreciated in price, an idea like this tends to offer less potential reward than a strategy for picking stocks that are likely to start a period of market outperformance. Yet sound strategies are again more complicated than such simple ideas as �A is hot,� so we have a mechanism by which simple but less lucrative ideas can outpopulate more complex but rewarding strategies as investment memes. The �company A is hot� idea may suffer at least one longevity disadvantage, however: It refers only to supposed present conditions, and does not have any built-in mechanisms favoring long-term belief. This can allow the meme to be dropped as readily as it was accepted.

STOCK AND NON-STOCK CONTAGION DIFFERENCES

The study of how economic and memetic forces combine and interact is a broad interdisciplinary field I call econo-memetics. Some of the questions it addresses are why certain beliefs and mores correlate with socioeconomic status and why consumer preferences spread or decline.[10] Econo-memetics also includes the contagion of memes within employment and financial markets. In markets, thought contagions exert different effects in different spheres: They may lead to relative efficiencies in employment markets, while generating inefficiencies in stock markets. In employment markets, decisions that yield high rewards can attract more imitators � a matter of high, often active, receptivity.

For instance, if there is a high demand for a particular kind of employee in relationship to the supply of applicants, the pay will be high. These highly paid people can attract more imitators to their profession until the supply of available workers drives down their pay scales. At that point, there is no longer such a high rate of imitators, and the influx of new workers to the profession dwindles. This favors some degree of market efficiency in matters of employment. Yet this market is also affected by parentally spread memes, the ones whose propagation on a mass scale results from large mean family sizes. Such parentally spreading memes can generate a surplus of workers and consequent unemployment � a form of employment market inefficiency.

For example, beliefs in parts of Africa and Asia that powerful ancestral spirits want many descendants may outpropagate alternative beliefs simply by influencing adherents to have more children � a result that could be compounded over many generations. The mechanism may incidentally produce more job seekers, even in areas that already have high unemployment rates.

Another source of possible inefficiency in employment markets is that technological progress and business cycles can suddenly change the demand for people with specific skills. The demand for people with good car-painting skills may decline as a result of the introduction of car-painting robots, for instance � or from a cyclical slump in the automotive industry.

With investments such as stocks, we may again see abundant imitation for those choices that pay unusually well. Each trade on the NYSE, AMEX, Nasdaq, etc. helps publicize investment decisions and the results they achieve. For example, two traders on an exchange do not trade stocks in secret, but make a widely seen and recorded announcement about the value and value change of those stocks. This provides a basis for wide imitation. But a high imitation rate for increased valuation means high demand at the increased price, which drives the stock price up further and the yield down. Ideally, this should in turn lower the imitation rate, allowing the influx of investors to a particular company to level off or fluctuate.

Operating solely in this rather idealized manner, thought contagions can favor relatively high efficiency. Yet this idealized process is only one of many thought contagion effects going on simultaneously. Moreover, the imitation of successful stock choices generally comes with a delay. It takes time for the stock choice to prove successful, and, by that time, it may no longer have nearly as much potential to appreciate further. Yet it is precisely after the stock has already appreciated dramatically that the idea of choosing it may command the most favorable attention and perhaps the most imitation among other potential investors.

Stocks have other serious differences from occupations in their meme transmissivity. Those who hold well-paid jobs may want to transmit the relevant skills to friends and family in order to help them. But they may often wish to exclude most other people, in order to avoid competition and pay cuts. That tends to limit the transmissivity of occupational memes. Yet a shareholder may feel motivated to spread his or her investment idea to others, whether to help friends or to help drive up the share prices. Seldom do they have economic motives for excluding others. So they often give stock tips favoring whatever companies they decide to own.

AIR TIME

Additionally, people may feel more inclined to talk about their holdings in companies that have recently outperformed the market than holdings in companies that have underperformed the market. Suppose company A has risen dramatically in price over the past year, while company B has declined in price. The owner of company A may feel more inclined to talk openly and often about holding those shares: Doing so can amount to a way of revealing or implying one�s wealth and wisdom to others. The new shareholder who has not even made any money on the investment can feel the same motive: In social settings, projecting the illusion of wealth and success is often as important as attaining it in reality. Meanwhile, the owner of company B may hesitate to discuss his or her holdings. Even the new investor in company B might unconsciously fret that revealing the investment could give others the impression that he or she has lost money or could not afford more expensive stocks. This investor might thus decide that it is better to wait until B goes up in price before revealing the investment to a circle of friends.

All of this means that just as company A attains higher prices and lower yields, it receives more �air time,� or transmissivity, in informal investment discussions. Company B, on the other hand, must wait until its price rebounds before receiving similar air time. This can have the antiefficiency consequence of increasing the demand and price for a company�s shares just as its yield declines. That in turn can fuel further increases by giving people even more status motivation to discuss their ownership of recently appreciated shares. The price, of course, does not run to infinity, as people still have limited cash reserves and limited time for talking about stocks. But the price can soar far above what rational methods of valuation alone would allow.

Another possible �air time� effect is that successful companies become more visible and remain visible for much longer times than do new companies that fail. Companies that failed or were bought out at low prices are forgotten with the passage of time, while the successful ones continue to make their success known to investors and the general public. This may give individual investors a false sense of the prudence of investing in new companies.

THE GREATER FOOL THEORY

Adding to the transmissivity effect is a receptivity effect resulting from �the greater fool theory.� Investors who can clearly see the increased price and diminished yield of company A may actually become more receptive to the idea of investing in company A due to the expectation that they can sell later to a �greater fool.� They are, in effect, gambling that the thought contagion will continue to spread, and that they will be able to pull out before it spreads to a majority of potential investors. Unless the company finds exceptional ways to generate new earnings, the �greater fool� strategy becomes a Ponzi-like investment scheme.

DIFFERENTIAL COMMAND OF ATTENTION

Another contagion mechanism comes from differential command of attention among stocks. If investors average sixty minutes a week thinking about company A, but just six minutes a week thinking about company B, they may spend more time talking about company A to their friends. The difference might result from a company product being more widely used, more visible or audible when used, or more widely advertised. Or it could result from one company or product capturing the imagination better than another. Such differences in communication rates and preoccupation rates can be measured by surveys (Shiller and Pound [1989]).

CORRECTIONS, CRASHES, AND DOOMSDAYS

When contagion effects send individual stocks and entire markets soaring to levels that cannot be justified by more rational earnings expectations, the stage is set for corrections and crashes, which may themselves be fueled by thought contagions working in the opposite direction and with greater speed. When a negative surprise hits a roaring market, an initial wave of investors may start selling their holdings and sending the market sharply lower, as has been observed for individual stocks (Dreman [1998]). When the sharp decline hits the news,it may cause a race for the exit in many other investors. They in turn also act accordingly, setting up a recursive contagion of panic selling. The idea of a race for the exit sometimes becomes a self-fulfilling thought contagion.

Some of the transmissivity of panic selling thought contagions probably comes from mechanisms similar to those that drive outbreaks of doomsday beliefs in society at large. Those who believe in an imminent cataclysm, whether from divine forces or from Y2K malfunctions, feel strongly motivated to warn anyone they care about in order to save them. With religious apocalypse memes, the motive is to save people from hell by having them prepare spiritually. With Y2K apocalypse memes, the motive has been to save friends and loved ones by persuading them to prepare for the end of civilization (Lynch [1999]).

When a market crash is in progress, the idea of a race for the exit can become a kind of financial doomsday belief. It may motivate its hosts to retransmit to any other investors they care about in order to save them from financial ruin. An investor who has already pulled his or her money out of the market may, for instance, try urgently to persuade close friends, relatives, and associates to do the same.

When contagion mechanisms help drive share prices up and then down, they combine with individual psychology mechanisms to cause the kind of overreactions that make stocks overvalued and then undervalued in relation to earnings and realistic earnings prospects. Similarly, contagion mechanisms may help drive shares down and then up, contributing to overreactions in the opposite direction.

Dreman [1979, 1998] and De Bondt and Thaler [1985, 1987] have produced evidence of overreaction phenomena in real stock markets, and Smith, Suchanek, and Williams [1988] have studied overreactions in experimental laboratory markets. Other contagion mechanisms that drive share prices in one direction and then the other are examined in relation to Internet stocks.

THE INTERNET STOCKS

Along with transmissivity and receptivity effects that can arise for nearly any company whose share prices rise or fall are effects resulting from the nature of the business itself. The recent soaring prices of Internet stocks provide examples worthy of further investigation.

One such sector-specific effect is that Internet enthusiasts may be more likely to know about, use, understand, and invest in Internet companies. That means investors in these companies are more likely to be Internet users than are investors in other kinds of companies. They may spend more time online and know more about how to use online communications. They may give and get more of their stock tips in this globalized, fast medium, and have greater than usual capacity to electronically spread their beliefs worldwide. Many people may even learn more about Internet communications just by studying Internet companies as investment options, since researching these companies can make one aware of electronic forums, web promotion methods, and so on. These factors can contribute non-truth-contingent transmissivity to beliefs in the great futures of Internet firms.

Many Internet companies have devised brilliant new uses for today�s technology, and that brilliance becomes a factor in the investor frenzy by increasing receptivity to the idea of investing in Internet companies. Yet this might not be the only receptivity factor at work. Internet skills may also make it easier for believers in those companies to reach people interested in receiving their messages (e.g., in specialized forums), distinct from people chosen by factors such as mere proximity. So without needing any advantages in quality, their investment advice may enjoy greater receptivity than is enjoyed by those favoring non-Internet stocks.

Armed with an enriched concentration of good electronic communicators, the shareholders of Internet companies may routinely outdo the shareholders of other young companies in the arena of belief transmission. For example, news of a favorable earnings report or merger negotiation involving an Internet company might spread farther and faster over the Internet than would comparable news for a low-tech company.

The bull market in Internet stocks may therefore come from the faster and wider communications of its participants. This hypothesized phenomenon results from memes spreading in large host populations of individual investors, who come to outnumber the institutional investors. The rising intensity of their belief also leads them to price many institutions out of the market. This leaves the stocks of Internet companies widely held by individuals rather than large institutions (Anders [1998]). Among individuals and institutions alike, those with the most bullish memes about Internet companies can price others out of the market. This causes share prices to be selectively set by the host populations of very bullish memes, driving prices higher.

Adding to the thought contagion fueled by individual investors is the possibility that brokers favoring Internet stocks may also know more about how to use the Internet for promoting stocks electronically. For example, they would know more about how to create multiple email and web aliases for posting in online investment forums than would those brokers who favor non-Internet industries. Internet communications may have also accelerated the spread of investment ideas fueling the rest of the bull market, but the effect is logically more pronounced for the stocks of Internet companies, whose fans probably use the Internet more. This may mark a new kind of contagion advantage fueling the fast price increases in the stocks of Internet companies in 1998 and 1999. As a relatively new phenomenon, it has understandably surprised those analysts looking at industry fundamentals.[11]

Many of the messages sent about stocks are now recorded on electronic databases, providing a possible means of researching the transmissivity of interest in stocks through the Internet and other media. Word searches can systematically reveal whether Internet stocks are more often mentioned in the Internet media than in print media, for instance. Such searches can also reveal whether Internet companies get disproportionate �air time� in electronic forums as compared to non-Internet companies. Searching the forum �misc.invest.stocks� from December 1997 to December 1998 yields a striking difference between the online auction company eBay and the Ford Motor Company: About 5,900 documents mention the word �eBay,� while only about 1,300 mention the word �Ford�[12] � even though Ford is a much larger company with many more shareholders. More sophisticated semantic software may detect whether messages are running in a bullish or bearish direction in research on this subject.

Other high-technology companies besides the explicitly Internet firms may also enjoy more air time in the online forums, although probably to a lesser extent. This would again result from possible correlations between one�s general interest in technology and one�s mastery of the Internet. It could affect IC manufacturers, software companies, computer makers, drug companies, and so forth. The Internet may thus play a role in accelerating investment in technology, especially in the heavily wired United States.

MULTIMEDIA FEEDBACK EFFECTS

As online thought contagions apparently helped propel the share prices of Internet companies upward, their spectacular appreciation became a news item in not just the business media, but the mainstream media as well. That further increased the number of people learning about Internet companies and in turn investing in them. Yet even these �offline� communications became amplified by online media in a way that affected the Internet companies disproportionately.

Consider what happens if a financial story about a company is broadcast on one of the leading evening TV news programs, for instance. Most potential investors in that company will miss the broadcast either by watching a different channel or by not watching TV at that time. But investors who subscribe to an electronic forum dedicated to that company or its market sector are apt to hear about the broadcast anyway. And those with an interest in Internet companies are far more likely to be signed up to such newsgroups than are people interested in low-technology firms. So a higher fraction of potential investors in Internet companies will hear about the publicity their company has received, making them more likely to react by buying shares. Even news stories transmitted by older technologies come to disproportionately favor investment in Internet companies, because news about news has more possible routes of transmission to those who use the Internet regularly.

Conventional print and broadcast news sources are also increasingly carrying copies or transcripts of news stories on their web sites. These second copies of financial news stories again tend to disproportionately reach avid Internet users. Some of the recipients are people visiting web sites of financial journals and TV programs, while others get the web addresses through postings and emailings in forums or from friends. As before, people who are frequent Internet users are more likely to get such information.

Because of the likely correlation between Internet usage and interest in Internet companies, those interested in Internet stocks may receive an unusually high proportion of the news items about their favorite companies. Thinking about their favorite companies more often can in itself lead to more frequent investment, because having a company on one�s mind is essential to perceiving it as an investment opportunity. Yet the effect may compound itself through thought contagion as well, because having a company more frequently on one�s mind can lead to more frequent communication about that company. Having a company more frequently on one�s mind is essential to viewing it as a conversation subject, an email subject, a forum topic, etc. These added collateral routes of communication may feed on themselves to produce a more widespread interest in the companies most associated with new collateral communication channels: Internet companies.

As with other new companies, the new Internet companies that succeed and experience rising share prices become far more visible to investors than those that fail. Whether in a discussion forum or on the nightly news, one hears more often about a company that is still in business and still has a large market capitalization than about a company that went bust. This may favor the spread of unrealistically optimistic ideas about Internet investment through both conventional and online media.

In addition to broadening the number of routes by which communicable investment information can spread, the Internet also enhances the persistence of stories carried by older broadcast media and even print media. If you missed a nightly news item about Amazon.com, chances are you can pick up the transcript days, weeks, or months later at the web site of the broadcast network that carried the story. If you no longer have a Wall Street Journal article about Yahoo!, chances are you can still get it online.

Yet those interested in Internet stocks may again be more likely to know how to do this than people interested in, say, railroad stocks. News articles about Internet companies thus have more effective longevity for their segment of the investing public than news articles about older lines of business for their segments of the investing public. This can help the stories to ultimately reach a broader fraction of Internet enthusiasts than, say, railroad enthusiasts, contributing still further to the contagion effect fueling interest in Internet stocks.

AMPLIFICATION FROM ONLINE INVESTMENT

Those who embrace online technology not only have more and faster ways of catching investment thought contagions, but they also have faster and easier ways of changing their holdings accordingly. Online brokerages and day trading centers are presumably more commonly used by those most comfortable with Internet technology. This in itself may help to further increase investment in Internet companies. It may also make Internet enthusiasts much more likely to respond to the financial news about favorite companies that they already receive more quickly through Internet usage.

By investing more quickly and easily, they more quickly and easily progress to the stage of active belief transmission about the stocks they hold. For instance, when bullish news about Lycos reaches investors interested in the Internet sector, they are likely to see it as relatively easy to invest quickly through an online brokerage. The average delay from receiving the news to making the investment may thus be shorter than it would be for people interested in, say, the oil sector.

Compared to people receiving bullish news about Lycos, individual investors receiving bullish news about Exxon may react more slowly. The Exxon investor may be more likely to have to call a brokerage and talk to someone during business hours before ordering shares. If they feel that too much time has passed and the opportunity is already missed, they may become less likely to invest. So those hearing bullish news about Lycos may be more likely to invest and to do it sooner. As a result, they are more likely to enter the phase where they become financially motivated to pass their investment ideas on to others. So online brokerages and day trading operations may accelerate financial thought contagions in the Internet sector more so than in traditional sectors of the market.

The growth of online trading can be not only a cause, but also a consequence of contagious investment in the Internet sector. Thought contagions may raise the share prices for online investment companies for the same reasons they can raise the prices of other Internet companies. Yet those high share prices of online investment companies have enabled them to promote their services aggressively not just on the Internet, but through television and other mass media as well.

In addition, the selectively quoted high returns of the most bouyant Internet stocks lend apparent credibility to the private tropical islands and lottery-like winnings in even the most surreal advertising from online brokerages.High share prices of Internet companies in general and online investment firms in particular also help spur conventional brokerages to open and promote their own online trading operations. The greater the market share they can gain, the greater the anticipated �Internet price boost� on these brokerages� own stocks, creating an incentive for advertising that goes beyond the per trade commission. All of this promotion may lead to more online investment, particularly in the Internet sector. The arrival of online investment to the Internet may thus have fostered positive feedback that further raised share prices in the Internet sector and the online investment subsector.

BEARISH INTERNET THOUGHT CONTAGIONS

Not all online meme transmission is bullish, however. Investors looking for opportunities to short-sell Internet stocks would also have unusual levels of Internet mastery, and they also have motives to retransmit their opinions. So thought contagions of bearish ideas can also spread rapidly online, and drive share prices down as quickly as they rise. As another disproportionately online crowd, Internet company short-sellers should have additional ways of catching bearish news about Internet companies for the same reasons that other Internet investors have extra ways of catching bullish news. This may contribute to outbreaks of bullish and bearish memes and add to market volatility.

Certain bearish ideas may be particularly adapted to online transmission. The belief in a Y2K cataclysm, for instance, has spread particularly well on the Internet (Lynch, 1999). Its clash with bullish Internet stock ideas may have added some volatility to the Internet stocks during July to October of 1999. The Y2K apocalypse thought contagion need only have generated fear, uncertainty, or doubt in some investors to become an excuse for selling high-priced Internet shares. Fearful anticipation of selling by others may have caused more selling than did fears of potential century rollover computing problems.[13] Still, those with the strongest Y2K doomsday beliefs may have typically have held smaller portfolios than other investors. Meanwhile, a widespread reckoning that doomsayers already held few stocks may have limited the year-end effect of Y2K beliefs.[14]

Perhaps of greater importance is that people who bought Internet stocks at high price-to-earnings ratios in hopes of an eventual bonanza will generally not wait forever. Unlike certain ideological systems, belief that a certain Internet company will eventually earn vast fortunes does not have a built-in mechanism for long-term meme preservation.

In biological epidemiology, there is little evolutionary pressure for a fast-spreading microorganism such as influenza to have mechanisms that keep hosts infected over the long term. So many contacts catch the virus after several days that it can afford to lose its host to recovery in a relatively short time. For common non-lethal strains, waves of infection can thus lead to delayed waves of immune reactions. Fast-spreading thought contagions often lack long-term preservation mechanisms for similar evolutionary reasons (Lynch [1997]).

Since the idea of investing in Internet companies has spread very quickly in 1998 and 1999, we should expect that it would not come with a built-in ideology preventing dropouts. This means that hosts of an Internet stock meme would drop their belief in the company�s prospects if low returns persist, and waves of �immune reactions� against ideas that Internet companies make good stock investments can follow after a delay from the thought contagion�s upsurge. This alone can lead to waves of selling.

A related factor is that investment memes that have spread so rapidly often consist of beliefs that the stocks would yield huge short-term gains. Thus, most of those who invested accordingly may have already been planning to sell after a relatively short term. If so, this tends to produce a rather short-delayed wave of selling regardless of earnings reports, even among those who do not convert to bearish ideas about Internet stocks. Such waves of thought contagion followed by either delayed waves of deconversion or short-term investor selling may further contribute to the market overreaction phenomena observed by Dreman [1979, 1998] and De Bondt and Thaler [1985, 1987] in non-Internet stocks or whole stock markets as well.

Internet stock investors who do convert from bullish to bearish ideas about companies due to persistent low earnings, tips from others, and so on, are still as disproportionately connected to the Internet as they were before. This affords them a better ability to spread their newly bearish ideas about Internet companies for the same reasons that they could better spread their bullish ideas earlier. This can amplify a downturn in prices. Bullish Internet stock ideas may subsequently enjoy renewed propagation during such downturns, possibly sending prices back up.

People viewing themselves as bargain hunters may also help slow or reverse a downturn even if the price-to-earnings ratios are still extremely high. All of this leads to a complex quantitative problem calling for further research and measurement of propagation parameters.

CONTAGION AND INNOVATION AS ONLINE PROFIT LIMITERS

Some of the same forces propelling the sharp rising phase in share prices of Internet companies may, ironically, also limit their profitability. With heavy investments made in Internet companies, some of the bullish thought contagions may turn into self-fulfilling prophesies of success. With capitalizations in the billions overnight, online companies can afford expensive commercials to attract business away from their offline competitors.

Online retail sales, for example, have apparently benefited from heavy advertising made possible by contagiously rising share prices.[15] And the Internet allows a company to gain in a few months or years a level of market share that used to require decades or generations for conventional firms to achieve.

Yet working against these advantages is one of the very forces that would favor market efficiency under ideal circumstances: the disproportionate imitation of success. Not only do investors seek to copy investment ideas that earn money, entrepreneurs seek to imitate successful ways of raising capital. Periods of high share prices for Internet companies bring floods of �me too� startups and initial public offerings (IPOs). The successful IPO can make company executives instantly rich regardless of whether they have a solid business plan. So Internet IPO memes spread rampantly among entrepreneurs and their underwriters, causing competition for investor dollars. Much as the ability to reproduce new tulip bulbs helped drive prices down from their seventeenth-century highs, so may the relative ease of starting new Internet companies eventually drive down prices in the Internet sector.

Meanwhile, the arrival of new Internet companies leads to levels of retail price competition that may be more severe than in conventional retail. With conventional retail, the ability to saturate a local market limits the number of competitors in a given town. That in turn allows them to price merchandise more profitably. But on the Internet, the �local� market might be a country, a continent, or the world. This allows more companies to compete and operate at a loss while seeking market share. Regardless of where they are headquartered, all the �dot.com� companies are about equally easy to reach from anyone�s home computer. IPO thought contagions thus limit the ability of Internet stock investment thought contagions to become self-fulfilling prophesies of wealth for the shareholders.

Compounding this effect is that well-capitalized brick and mortar stores such as Wal-Mart are opening online operations in the quest to protect market share. They, too, add to the retail price competition that tends to limit profitability.

Finally, the march of technology itself may limit how long today�s Internet retailers can retain their online advantages. In ten years, the Internet may be replaced with something much better, and electronic shopping may be done through virtual reality schemes in which today�s domain name format is quaint and meaningless. And while many useful functions are being developed for web servers, the emergence of artificial intelligence and �expert systems� in vastly more powerful desktop or remotely shared computers may render many web companies obsolete.

Humans, for instance, currently use brand identities to avoid recalculating the value of products and services. In the future, we might use a smart �robot� program based at home or a local network to impartially surf the web and find the best deals while rating company trustworthiness. This could bypass Internet companies such as eloan.com, priceline.com, etc., that thrive on providing a multisource price shopping function. It could also bypass the brand loyalty advantage of companies like Amazon.com and MotherNature.com, which thrive on the convenience of not having to judge retailers each time we shop.

Portal functions such as those provided by Yahoo!, Lycos, etc., might also become duplicated to a large extent on desktop computers and the servers of other companies. Their advertisements would hold little or no value to software �robots� doing the online shopping.

Finally, anything approaching widely accessible and cheap artificial intelligence would end the advantage of web companies that have more memorable names, such as drugstore.com and eToys.com. Future computers may remember and surf millions of the most nondescript domain names. All of this change could force more severe price competition onto Internet companies while reducing their individual advantages.[16]

The present Internet companies may have to make handsome earnings for decades in order to justify their still high share prices. Fast-changing computer technology and online competition give further reason to suspect that present Internet company share prices are largely a contagion-fueled bubble that is likely to eventually deflate or burst.[17]

The 1998 to 1999 run-up in Internet stock prices bears comparison to phenomena ranging from historic stock market bubbles to tulipmania. But are there any similar forces at work in the Internet and the tulip market phenomena? One possible similarity is that simple ideas like �tulip bulbs are hot� may have had a transmissivity advantage over more complex but realistic ideas. And while tulip bulbs do not enhance one�s capacity for electronic communication, investment in them may have actually helped cause more communication of this investment idea. When one exchanged gold for a tulip bulb, the new investment needed to be put out in the open where it could gather sunlight. Those who held gold, on the other hand, were more inclined to keep it hidden and locked away from view. As an openly displayed investment, the tulips would have commanded more attention from potential imitators.

Moreover, the owner of a tulip bulb needed to attend to its needs by watering it. This meant thinking about it regularly. And, again, the more often one thinks about an investment, the more likely one is to talk about it often. So the tulip bulb investment idea may have been doing something for its own retransmission � a principle that warrants investigation as a possible enduring force in market manias through the centuries.

CONCLUSION

Thought contagion theory has broad potential for explaining market phenomena. But most of the arguments we give here are simply hypotheses awaiting empirical testing. Further research in this field is essential to developing a more comprehensive understanding of how the evolutionary epidemiology of beliefs affects the up and down movements of prices in the financial markets. Specific thought contagion hypotheses can also be tested.

For example, data banks of online postings may be treated as raw historical data in discerning whether the numbers and nature of postings about a company correlate with its subsequent stock price moves. (Aliases can be filtered out by examining a subset of distinct names that can be tracked to specific people in telephone databases.) One can also study the extent to which such online participation correlates with market sectors such as the Internet sector.

At a more ambitious level, one can design surveys intended to measure propagation parameters for company-specific investment ideas: numbers of non-hosts converted per host per month, distribution functions of belief duration before �spontaneous� dropout, and similar parameters for contrary ideas. These might then be used mathematically or computationally to project host populations versus time for beliefs about a company or its stock. The actual and projected host populations can be compared as a test of the model.

The host population as functions of time can also be tested for correlations to the stock�s price movements in relation to overall market movements during periods of scant news about the company. Positive correlations would be consistent with the thesis that thought contagions exert an influence on share prices. The possible effect of more general ideologies and investment ideas on the broader market can be empirically researched using similar methods. Expanded research into stock market thought contagions thus holds considerable potential as a useful subfield in the expanding domain of behavioral finance.

Notes