Bubbles, Human Judgment, and Expert Opinion

Financial Analysts Journal

ISSN: 0015-198X (Print) 1938-3312 (Online) Journal homepage: https://www.tandfonline.com/loi/ufaj20

Robert J. Shiller

To cite this article: Robert J. Shiller (2002) Bubbles, Human Judgment, and Expert Opinion, Financial Analysts Journal, 58:3, 18-26, DOI: 10.2469/faj.v58.n3.2535 To link to this article: https://doi.org/10.2469/faj.v58.n3.2535

Published online: 02 Jan 2019.

Submit your article to this journal

Article views: 110

View related articles

Citing articles: 13 View citing articles

Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=ufaj20 PERSPECTIVES

Bubbles, Human Judgment, and Expert Opinion Robert J. Shiller

he widespread public disagreement about endowments. Before the market peak in March whether the U.S. stock market has been 2000, those who manage university endowments undergoing a speculative bubble since had not withdrawn en masse from the stock market. T about 1995 reflects underlying disagree- According to a National Association of College and ment about how to view human judgment and University Business Officers (NACUBO) Endow- intellect. Many argue that the market, or major com- ment Study, in 1999, the median endowment had ponents of it, has been undergoing a bubble and 54.7 percent in U.S. equity and another 10.5 percent that, with the P/Es of major companies still very in non-U.S. equity. And most endowment portfo- high, the bubble has not burst completely. This lios remained in U.S. stocks, so they continued to group includes some writers for the Economist, the be involved in, not withdrawing from, the market Wall Street Journal, and Barron’s. Apparently, it is just before its peak. essential to their notion of a bubble for investors’ Those who manage university endowments actions to have been, in one way or another, foolish. have at their disposal some of the finest scholars and Others sharply disagree with the bubble stories, university trustees drawn from the highest ranks of and it is precisely this intimation of foolishness that the business world. Who would presume to call seems to bother them. Of course, some investors are these people foolish? But if one wishes to attribute foolish, they will admit, but they find the idea that the market’s behavior to human error, calling the investors by and large have been foolish to be sim- experts foolish is apparently what one must do. ply implausible. Rather, they will say, the high val- I believe that the stock market has indeed been uations the market has placed on the stock market caught in a speculative bubble in recent years recently can be attributed to the actions of rational (Shiller 2000). But the kind of less-than-perfectly- investors who are wrestling with hard-to-interpret rational behavior that underlies the bubble is not evidence about such issues as how much recent abject foolishness. It is not the error of fools. It is technological innovation will promote future earn- more the error that afflicts some of Shakespeare’s ings growth. Suggesting that investors in general tragic figures—in the sense of having subtle weak- have been irrational seems arrogant and presump- nesses or a partial blindness to reality. The kinds of tuous. errors that people have been making and that The dispute about whether we have been in a underlie the recent stock market bubble do reflect bubble is not simply about whether P/Es have human shortcomings—but exactly the kinds of become too high. One can argue that P/E multiples shortcomings that can infect the thinking of profes- have been or are too high without saying that some- sors, analysts, and trustees as much as anyone thing irrational underlies the high values. One can else’s. The current situation in financial markets is argue that the market has merely not taken certain simply a fertile ground for the amplification of the facts into account and will do so once these facts kind of errors even experts can make. have been properly disseminated. The idea that My aim is to draw attention to human foibles there has been a speculative bubble goes beyond that we all are subject to and that research in psy- high or low P/Es; the concept of “bubble” implies chology, behavioral finance, and other social sci- some less-than-rational aspect of investor behavior. ences reveals thoroughly and systematically. What Has investor behavior been irrational? One of I am doing is rather like what psychologists do the most important arguments that foolishness has when they show, using certain optical illusion not brought us the high valuations in the aggregate charts, that we all tend to make certain characteris- market comes from observing the decisions made tic visual-recognition errors and like what sociolo- by experts—for example, managers of university gists do when they point out how contagion of idea patterns underlies the spread of political ideologies. Robert J. Shiller is professor of economics at Yale Uni- In this article, I will expand on the sort of versity and principal at Case Shiller Weiss, Inc. This irrationality that seems to be at work in the mech- article derives in part from Irrational Exuberance anism by which bubbles are amplified. I will pur- (2000), winner of the 2001 Commonfund Prize. sue the nature of less-than-perfect rationality not

18 ©2002, AIMR® Bubbles, Human Judgment, and Expert Opinion only among the general class of investors but also investor feedback have been defined. For example, among experts—that is, the errors characteristic of Barberis and Shleifer (2000) developed a model of investment professionals. speculative bubbles that relates demand for a particular investing style to an average of past returns, Feedback and Bubbles but in their model, excess returns are defined as excess returns of investments in one style over The essence of a speculative bubble is the familiar another. They used their model to show that not feedback pattern—from price increases to increased only does the feedback amplify price movements, investor enthusiasm to increased demand and, it also causes assets in one style to move together hence, to further price increases. The high demand too much. They also showed that assets in different for the asset is generated by the public’s memory of styles may tend to move together too little. Their high past returns and the optimism the high returns model can explain such events as the Nasdaq bub- generate for the future. The feedback can amplify ble and burst of 1995–2001 at a time when the positive forces affecting the market, sending the nontechnology stocks showed relatively mild market to higher levels than it would go if it were bubblelike behavior (see also De Long, Shleifer, responding only directly to these positive forces. Summers, and Waldman 1990 and Hong and Stein Note that a bubble is not indefinitely sustainable: 1999). Prices cannot go up forever. And when price In this article, I want to step back from the increases end, the increased demand that the price formal models of speculative bubbles and ask (1) increases generated also ends. Then, a downward whether this kind of feedback of price change to feedback pattern can replace the upward pattern. further price change is plausible at all and (2) In one model of a speculative bubble (Shiller whether it is consistent with what we know about 1990), the demand for shares depends on a “distrib- investment experts who, apparently, commonly go uted lag” of past returns—that is, on a weighted along with bubbles or, at least, do not act to offset moving average of past returns, with weights that bubbles. decline through time into the past—and depends It is important to realize that the bubble theory on some other precipitating factors, such as earn- requires only that past price changes produce an ings growth, that affect perceptions of stocks inde- inconstancy in our judgments, not that we foolishly pendently of returns. In this model, demand is most believe that past increases must continue. The strongly boosted by past returns if the high returns theory does not require that we forecast future came in the past few months, less strongly boosted price changes by some mechanical extrapolation if the high returns came not in the past few months rule or that we place rulers to chart paper to fore- but within the past few years, and only a little cast. It requires only that our observations of the boosted if the high returns came not in the past few past price changes alter the way we resolve the years but in the past few decades. The reasons that confusing array of conflicting information that we the weights in the weighted average of past returns must all sift through in judging the market. decline into the past are that we remember the more Ultimately, people who make asset allocations distant past less clearly, we hear stories about the must use their subjective judgment about the prob- more distant past less often, and we tend to be ability that stock trends will continue. This neces- aware of institutional or cultural changes that make sity is true for experts as well as the general public. the more distant past seem less relevant than the Although experts have the help of formal statistical near past. models, we all understand that these models are If supply is relatively fixed, this demand model only as good as their specific assumptions. Assess- implies that price is itself a distributed lag on the ments of trends and probabilities that underlie other factors that affect demand. The impact of asset decisions are inherently subjective.1 these other factors is amplified by the feedback. The Intuitive probability has many aspects. One is model explains excess volatility in financial mar- the “representativeness heuristic,” identified orig- kets, explains overreaction to such factors as earn- inally by psychologists Tversky and Kahneman ings growth, and at the same time (although the (1974). They showed that in forming subjective model does not imply that stock prices are exactly judgments, people have a tendency to disregard a random walk) is consistent with the kind of base-rate probabilities and to make judgments approximately random walk of speculative prices solely in terms of observed similarities to familiar that we actually observe. patterns. For example, Tversky and Kahneman per- Now that the field of behavioral finance is formed an experiment in which subjects were becoming an active area in academic finance, a asked to guess the occupations of individuals number of explicit models of bubbles deriving from whose personalities were briefly described to them.

May/June 2002 19 Financial Analysts Journal

They tended to guess rare occupations that seemed A lot of evidence indicates that such over- to match perfectly the personality descriptions confidence in intuition is a powerful force in the without regard for the rareness of these occupa- markets. For example, a questionnaire I used to tions—that is, their low base-rate probabilities. The survey individual and institutional investors right subjects would have been closer to the mark if they after the October 19, 1987, stock market crash pro- had guessed the humdrum and common occupa- vided such evidence (Shiller 1989). I sent out 3,000 tions more often than the exotic. questionnaires within a week of the crash to a We can expect the same representativeness random sample of both wealthy individual inves- heuristic to encourage people to see patterns in tors and institutional investors asking them what stock market price changes (simple patterns such they were thinking on the day of the crash, and I as bull or bear markets), even though such received nearly 1,000 responses. Among the ques- sequences of same-sign price changes are actually tions I asked was whether they recalled that they quite rare. Thus, the representativeness heuristic’s had a “pretty good idea when a rebound was to effect on intuition encourages people to expect past occur” on that day. I found that 29.2 percent of the price changes to continue, even if they know from individual investors (47.1 percent of buyers of professional training that the trends should not. stocks) and 28.0 percent of institutional investors (47.9 percent of buyers of stocks) said yes. These This tendency to fit one’s future expectations to salient images or simple patterns is tempered percentages were remarkably high in light of the uncertainty of that day. The next question asked somewhat by a tendency toward conservatism. those who answered yes what made them think Psychologists have conducted experiments in they knew a rebound would occur. The answers which subjects’ abilities to revise probabilities in they wrote can only be described as merely intui- light of new evidence was tested (Phillips and tive. They spoke of gut feelings, of vague compar- Edwards 1966). The experiments were constructed isons with past events, or of a sense that market so that a single correct answer existed to the psychology had changed. Thus, the intuitive judg- probability-revision problem posed to the subjects; ments that the psychologists have been studying the answer could be deduced by application of are ultimately vital in determining the direction of Bayes’ Law, which is a fundamental principle from the market. probability theory.2 They found that subjects Another part of the mechanism by which past tended to revise their probabilities in the correct price increases boost demand for an asset class has direction but tended not to revise them far enough. to do with the simple attention that such price This finding that people respond insufficiently to increases attract. As long ago as the 19th century, new information has been replicated in a number the psychologist William James noted that atten- of studies and is now referred to as “conservatism tion is a fundamental aspect of human intelligence bias” (Beach and Braun 1994). and that anomalies of attention are behind many Economists have developed the twin princi- errors that people make (James 1890). People gen- ples of representativeness heuristic and conserva- erally do not know what has attracted their tism bias into a model of speculative bubble attention and cannot explain the attraction. Psy- propagation (Barberis, Shleifer, and Vishny 1998). chologists have documented a social basis for atten- In this model, the representativeness heuristic tion; that is, people tend to pay attention to what encourages people to react to price changes in an others are paying attention to. Not surprisingly, exaggerated manner, but the conservatism bias speculative assets whose prices have gone up a lot spreads this pattern out through time. Their model recently gather a great deal of attention. People are of the propagation of a speculative bubble, relying more likely to buy assets that have their attention as it does on only a couple of subjective probability simply because they are thinking about them more biases, is undoubtedly oversimplified, but it offers than assets that have not garnered the attention of useful insights. big price increases. Interpreting the model more broadly, I think The capriciousness of attention is apparently we can say that investors have overconfidence in a characteristic even of professional investors. In one complex culture of intuitive judgments about study, John Pound and I surveyed institutional expected future price changes and an excessive will- investors who had reported holdings of stocks ingness to act on these judgments. This overconfi- whose price had increased greatly in the recent dence is a powerful force in the market; the intuitive past, the experimental group, and compared their judgments are ultimately behind both the upward answers with those of institutional investors in ran- feedback that underlies the bubble and the down- dom stocks, the control group (Shiller and Pound ward feedback that signals the end of a bubble. 1989). In regard to their holdings, we asked both

20 ©2002, AIMR® Bubbles, Human Judgment, and Expert Opinion groups if they agreed with the statement: “My ini- believe that wishful thinking plays a role in the tial interest was the result of my, or someone else’s, propagation of a speculative bubble. After a bubble systematic search over a large number of stocks has continued for a while, many people have (using a computerized or similar search procedure) become committed to the investments—emotion- for a stock with certain characteristics.” In the ally and reputationally as well as financially. experimental group, only 25 percent agreed, whereas in the control group, 67 percent agreed. Judging the Judgments of Others The design of this experiment reveals that institutional investors, as well as individual investors, In making asset-allocation decisions, long-term have their attention ultimately attracted by past investors should realize the essentially judgmental price increases. nature of the task. Investing for the long term Another part of the mechanism by which past means judging the distant future, judging how his- price increases affect the judgments that are actu- tory will be made, how human society will evolve, ally made about investing for the future has to do how the world economy will change. Reaching with the feelings of confidence and self-esteem that decisions about such issues cannot proceed from past successes in investing give successful inves- analytical models alone. Major input must come tors. Success in investing usually involves some from judgment that is essentially personal and acquired skills in understanding the particular cat- intellectual in origin. Asset-allocation decisions egory of investment and in the strategy of dealing involve a bewildering complex of relevant factors, with it. When a substantial number of people some represented by quantitative data, others only acquire such skills regarding a category of invest- suggested by cases or events, still others suggested ments, the demand for those investments tends to by intangible intellectual currents in society. One increase. Psychologists Heath and Tversky (1991) must decide among a wide array of investment showed through experiments that, after probabilities strategies, each of which is described by some are held constant, people prefer to bet in situations advocates as “supported by evidence” but for in which their perceived competence is high. In which no clear proof of success exists. With such a their experiments, Heath and Tversky asked sub- confusion of factors, it is hard for anyone to make jects to answer general knowledge questions, such objective judgments without being influenced by as questions about football, or to give political pre- the recent behavior of the market and the recent dictions and then to assign probabilities that their success of one’s investments. answers were right. The subjects were next asked The complex judgments that institutional whether they would prefer to bet on their answers investors and portfolio managers must make about or on a chance lottery with the same probability of these factors is inevitably influenced by the judg- winning. They found that subjects were likely to ments of others. The reason is that in making major want to bet on their own answers when they had allocation decisions, managers almost inevitably assigned high probabilities that their answers were wind up trusting to a common view or consensus right or when they thought they knew a lot about view about the future. No one person can be at once the subject. a historian, political scientist, economist, and psy- In addition, people who recently benefited chologist rolled into one. Few of us, even the best from a price increase in an investment class tend to investment professionals, can even make much of have a personal association with the price increase, a beginning at the task. Most professionals have too and investment positions in appreciated assets many things to do in a day, too many other claims tend to lead investors to an expectation of future on their time. Institutional investors who attend price increases. Having been emotionally involved seminars and keep abreast of research make an with an investment strategy causes us to overesti- effort to synthesize all this knowledge, but unfor- mate the benefits of the strategy. Julius Caesar said, tunately, synthesizing the knowledge, deciding “Men willingly believe what they wish.”3 His who is “right,” is an even more difficult task than insight was on target. A number of psychological trying to add a piece of scholarly evidence. Profes- studies have shown that people suffer a “wishful sionals ultimately end up assuming that what their thinking” bias; that is, they overestimate the prob- colleagues believe is true. ability of success of entities with which they feel an The errors in judgment that underlie the bub- association. For example, soccer fans give exagger- ble do not stem from naiveté or credulousness. ated probabilities that “their” team will win (Babad Major speculative bubbles are always supported by 1987), and supporters of political candidates give some superficially plausible popular story, a “new- strikingly upwardly biased probabilities that their era theory,” that is widely viewed as having sanc- candidate will win (Uhlaner and Grofman 1986). I tion from some authoritative figure(s). The

May/June 2002 21 Financial Analysts Journal judgment errors arise from the difficulties of assess- egos of investment managers, who often make the ing where the public prominence of the new-era error of hubris. For example, Boston University theory came from and assessing how accurate the invested nearly 20 percent of the university’s port- theory is. Despite some concrete facts as part of its folio in one biotech company, Seragen, with disas- story, a new-era theory has, in fact, no solidity; the trous results. Yale University, in the 19th century, concrete facts cannot lead to a new-era conclusion invested virtually its entire endowment in one without the insertion of some outright guesses. The firm, Eagle Bank, and lost it. We must tell fiducia- error people make is to presume that someone else ries not to do such things. has verified the conclusion carefully when, The challenge is to know how to tell them just although some will have tried, no one has really what it is they are to do and not do. We cannot been able to do so. In other words, the error people simply tell fiduciaries to be smart or sensible. We make is to assume that the currency of the new-era can tell fiduciaries that they must do certain tasks theory is evidence that many people have com- of due diligence, but we cannot ask them to be pleted all the missing links in the argument when, invulnerable to essentially human limitations. in truth, its currency is evidence of the bubble Many years ago, the presumption was that fiducia- itself.4 ries should invest primarily in fixed-income securities, but after decades of debt underperforming Prudent Person Standard. Conforming equity, that stricture became inconsistent with con- one’s actions to conventional wisdom, the truths cepts of prudence. The result is that the prudent that one assumes on the basis of having heard them person standard is interpreted as telling fiduciaries reiterated again and again, is not only a natural not to deviate too far from what others think is thing to do, but for fiduciaries, it is required by law. right, which often means (given the feedback that Fiduciaries have been held by common law for afflicts other investors’ judgments) following the hundreds of years to the prudent person standard, investment that has outperformed in recent years. which obliges them to invest in a way that would The prudent person standard tells fiduciaries be generally regarded as prudent. The prudent per- to follow conventional wisdom. The problem with son rule was enshrined for pension funds in the the rule is, of course, that it makes fiduciaries inter- Employee Retirement Income Security Act of 1974 preters of conventional wisdom rather than invest- (ERISA), which states that investments must be ment professionals. They cannot take action made with without showing that it is conventional. the care, skill, prudence, and diligence, under the circumstances then prevailing, that a pru- The News Media. The news media play a dent man acting in a like capacity and familiar prominent role in generating conventional wis- with such matters would use in the conduct of dom. They do so more among nonprofessionals an enterprise of a like character and with like than professionals, but they do also affect invest- aims. ment professionals. After ERISA became law, pension managers And the news media help reinforce conven- were at a loss to know how to translate this stan- tional wisdom along some dimensions and help dard into concrete actions; many apparently change it along others. The media are in a fiercely thought that the standard meant that fiduciaries competitive business for survival as news media. should simply do whatever people had been doing So, they cannot be indifferent to public resonance for a long time (O’Barr and Conley 1992). One with the stories they write. Success thrives on good manager spoke of his reaction to ERISA at the time: writing. A well-written story can have a powerful “If you can find a guy who works in a building impact on public thinking; indeed, it can become a that’s got granite on the outside and it says ‘Estab- news event itself. One well-written news story that lished a long, long time ago,’ then you’re probably succeeds in grabbing public attention begets a long complying.”5 But as the prudent person standard sequence of follow-up stories in competing media evolved, it became clear that the rule does not outlets and reinforces its impact on public thinking. specifically mean that the prudent person is some- The news media are, therefore, generators of one who lived 100 years ago or someone who is attention cascades—as one focus of attention in pathologically risk averse. The prudent person public thinking leads to a related but slightly dif- standard refers to someone who does what most of ferent focus, which leads, in turn, to yet another us think is sensible. Ultimately, it must refer to focus of attention. Thus, shifts in public attention conventional wisdom. to economic issues are rather like the shifts in topics That fiduciaries have been given such a rule is of conversation at a dinner party. During a party, understandable. The need does exist to restrain the the focus of attention meanders and jumps as one

22 ©2002, AIMR® Bubbles, Human Judgment, and Expert Opinion person after another is reminded of a related should focus their energies on alternative asset interesting story, and there is no telling where the classes where the opportunity for active managers conversation will be in another 10 minutes. The is greater than in the broad stock market. succession of attention-grabbing topics in the These writers are apparently advocating, as media is similar, although spread out over days regards listed stocks, free riding on the judgment and weeks rather than minutes. of others who are supposedly looking at the long- Such cascades of public attention are an impor- run outlook. The advocates have decided that their tant part of the reason for the unpredictable nature game is so well played, there is no point in trying of speculative price movements. Stock market price to compete. This attitude involves a curious irony: increases generate news stories, which generate If such astute investors as these writers never try to further stories about new-era theories that explain judge the future course of the market, who is assess- the price increases, which in turn, generate more ing whether the market is appropriately priced and news stories about the price increases. For example, who is providing guidance to the market? stories about the “new economics” surged around The basic problem with the efficient markets 1997 after publication of an influential Business hypothesis is that it is a half-truth. Presenting mar- Week cover story (which also alleged that Alan ket efficiency as a concept to students and amateur Greenspan was a recent convert to this new-era investors is useful lest they come to believe that it faith). The story appeared after the first really enor- is easy to get rich quickly. It is not easy to get rich mous increases in the stock market had captured quickly by trading in speculative markets. The everyone’s attention. All of the follow-up stories short-run day-to-day or month-to-month profit about the new economy prominently mentioned opportunities that many people imagine they have the stock market. Thus, feedback mechanisms found are probably not there. But we should not include not only price-to-price feedback but also extrapolate from this simple notion of short-run price-to-news-to-price feedback.6 approximate market efficiency to the idea that markets are also efficient in the long run. In fact, data Efficient Markets Theory. The expert’s task for long intervals of time indicate that the stock of disentangling the source of public credence in market is anything but efficient. new-era theories is made more difficult by an intel- For example, I have shown that 10-year real lectual theory that has come to dominate much returns on the Standard & Poor’s Composite Index thinking about financial markets—the efficient have been substantially negatively correlated with markets hypothesis. This theory (in the “semi- P/Es at the beginning each 10 year period (Shiller strong form” that is most widely accepted) asserts 2000; Campbell and Shiller forthcoming 2002). that prices optimally incorporate all publicly avail- When the market has been high, it has tended to able information at all times. If one believes in come down. efficient markets, one believes that the marketplace Thus, the ratios indicate that we should not 7 of ideas somehow works out optimally. Therefore, expect stock market returns over the long run to be one who believes in efficient markets also believes unchanging. Yet, using P/Es as forecasters of that broad diversification, without any regard for returns, although enshrined in quite a number of the current market situation, is the ideal. Public papers in finance journals going back to Basu in information is already in prices; therefore, asset 1983, has never become a part of conventional wis- picking (let alone stock picking) based on individ- dom about investing. Arguments have been made ual judgment is useless. that the apparent forecastability of returns based on Certainly, many investment professionals these ratios is not necessarily unexplainable in make no effort at all to take into account informa- terms of some rational model of human behavior. tion about the long-run outlook. Many of them are Moreover, most studies of the forecasting of not in the business of asset allocation at all, their returns used very short return horizons, and ques- charge being to attend to other issues such as track- tions of statistical significance are never finally ing an index. Others may have the discretion to resolved because one might base an analysis of make allocations but do not do so as a matter of these ratios on many different statistical models. principle. Ellis put it bluntly in his popular book For all these reasons, the conventional wisdom Winning the Loser’s Game: “Market timing is a seems to be stuck on the idea that the stock market wicked idea. Don’t try it—ever” (1998, p. 10). can be expected to provide the same real returns Swensen, in his book Pioneering Portfolio Manage- regardless of the P/Es. ment, asserts that in the major securities markets, “market timers run the risk of inflicting serious Conformity Pressures. In his book about damage” ( 2000, p. 68) and that portfolio managers professionals’ herd behavior, Groupthink, Janis

May/June 2002 23 Financial Analysts Journal

(1982) discussed several reasons professionals concerns events that are essentially unprecedented operating in groups may be unwilling to deviate in nature, whose probability must be judged from the group consensus and described a number through analogous thinking and induction, by of case studies in which professional groups made thinking globally rather than specifically. He serious errors because of this “groupthink.” Janis argued that an essential reason for the success of identified one of the reasons for reluctance to offer the free enterprise system is its ability to deal with “different” opinions as people’s desire to preserve true uncertainty. Bureaucracies, he thought, are their status within the group. The result is what fundamentally ill equipped to deal creatively with Janis calls the “effectiveness trap.” uncertainty. Enterprise and entrepreneurship relish The term derives from interviews Janis later true uncertainty, and entrepreneurs pursue advan- made with people in the Lyndon Johnson adminis- tages that are seen only by human insight and can tration during the escalation of the Vietnam War never be proven objectively. (one of the case studies of serious errors committed In addition to the economy, examples of true by experts). They told him that there was a phrase uncertainty concern national issues—whether to applied to dissenting members of the administra- go to war, whether to invest more in infrastructure tion: “I’m afraid he’s losing his effectiveness.” Any- or education, and the like. But decisions about such one who dissented from the prevailing view in the national issues tend to be debated in committees, administration gradually began to be regarded as with expert testimony. So also are the decisions a “has been” and was gradually excluded from about institutional portfolio allocations. voice and power. People were allowed to express Uncertainty characterizes the present situa- some dissension without losing effectiveness, as tion in the stock market. In judging whether the long as it was presented in a suitably detached way stock market remains a good investment despite and in good humor. The dissenter would have to high P/Es, organizations must somehow judge accept mild jokes at his expense. For instance, Bill whether we are entering a new era, as some have Moyers, a close advisor to President Johnson, was claimed. Organizations and their committees are referred to as “Mr. Stop-the-Bombing” and Under- fundamentally ill equipped to make such judg- secretary of State George Ball, as “the in-house ments, however, just as organizations are ill devil’s advocate on Vietnam.” But, Janis concludes, equipped to write novels. An organization cannot their subdued and collegial criticism of the policy duplicate the creative energy of one independently served more to sustain conventional wisdom than thinking mind. to challenge it. Their weak presence gave decision makers the mistaken impression that they had con- Making a judgment about the situation in the sidered the dissenting view and rejected it. More- stock market means weighing evidence about cur- over, the dissenters were forced to remain silent rent technological advances and recent changes in publicly about their dissension, which blunted national and world institutions and comparing the their ability to pursue their arguments. current situation with past innovations and past Janis reported that President Johnson said in changes in institutions. We must make judgments his memoirs that he felt there had been substantial about the importance of computer networking rel- and open dissension within the group. The actual ative to the importance of the telephone, the impor- dissension that he remembered and reported, how- tance of recent changes in free market institutions ever, was limited to disagreements about tempo- versus New Deal institutions, and so on. The judg- rary halts in bombings. Johnson did not recall any ments we make will depend on answers to ques- fundamental disagreements about the wisdom of tions that have animated historians, philosophers, the bombing campaign itself. He emphasized, and political scientists throughout history. No com- instead, the unanimous assent that his advisors mittee is competent to answer such questions. often gave to his decisions. The recent empirical literature suggests that institutional or professional investors have been True Uncertainty and able to do little better than the market. One reason that institutional investors may not do much better Organizations than the market is that they have (or believe they Knight (1964), a University of Chicago professor in have) clients who have expectations of them that the first half of the 20th century, highlighted the make it difficult to pursue their own best judgment. distinction between risk and what he considered Clients often expect the professionals to invest in uncertainty. Risk, he explained, concerns events accordance with certain fads. Clients may expect about which the probability law is known and them to trade frequently or, at least, may not be which have quantifiable probabilities. Uncertainty willing to pay high management fees unless they

24 ©2002, AIMR® Bubbles, Human Judgment, and Expert Opinion do so. These effects dilute the advantages that pro- tional wisdom and past procedures so dear to com- fessional investors naturally have. mittees. Committees have great difficulty taking Another reason that the differences between action to alter their decisions on the basis of the professionals’ performance and the market’s per- changing weight of evidence. To have an impact in formance are small is that institutional investors challenging conventional wisdom thus requires believe they must have reasons for what they do, more than one’s intuitive assessment of probabili- reasons that can be justified to a committee. They ties. One needs a striking argument that is tren- feel they do not have the authority to make trades chant and on target—salient. If there is little in accordance with their own best judgments, prospect of having an impact, one tends to hold which are often intuitive. Their obeisance to con- one’s peace or make only perfunctory objections. ventional wisdom hampers their investment performance. Summing Up Psychologists have argued that human thinking—individual or group—that leads to action A number of factors help us understand the propa- tends to be motivated by qualitative reasons and gation of bubbles—the feedback mechanism from justifications rather than abstract weighing of prob- price change to further price change—and the inter- abilities and scenarios. Psychologists Shafir and action of this feedback with changing conventional Tversky (1993) conducted some experiments that wisdom. I have given particular attention here to showed people need some salience to a justification the factors that relate to professional investors. before they take action. In one experiment, partici- Many of the elements of the propagation mech- pants in one group were asked to decide which of anism have to do with the nature of subjective a pair of divorcing parents to award child custody probability and with intuitive and personal judg- to and participants in another group were asked to ments. Other elements have to do with the social decide which of a pair of divorcing parents to deny environment in which decisions are made, the child custody to. Subjects were given brief descrip- prominence of the news media, and the nature of tions of both parents, but one parent was described human interactions within organizations. vividly with both extremely bad and extremely Clearly, human patterns of less-than-perfectly- good characteristics and the other had a bland rational behavior are central to financial market description. Both groups tended to pick the parent behavior, even among investment professionals, with the vivid description—so, in one group the but there is little outright foolishness among inves- vividly described parent was awarded custody, tors. Writers may describe the behavior of invest- and in the other, the same parent was denied cus- ment professionals as if it were foolish, but only tody, which means the groups were contradicting because of the difficulty the writers about financial each other. These groups tended to choose the par- markets have in conveying the nature of irrational- ent for which more salient reasons were given, both ity. They cannot review all the relevant social sci- pro and con, apparently because saliency provided ence literature in their news articles. They are left a reason to offer for the decision. with punchy references to pop psychology that are The need for justifiable authority to change convincing to the undiscriminating reader but that investing behavior that has been successful in the also discredit them to the more careful reader, and past imposes on professional investors a sort of that only amplify our sense of strong public dis- conservatism—the broadly perceived conven- agreement about the nature of speculative bubbles.

Notes

1. The study of subjective probability has been a fertile field 4. The mistakes people make for this reason may be com- for psychologists for decades. See, for example, the volume pletely rational; they result simply from lack of information of survey articles edited by Wright and Ayton (1994). about others’ information (see Demarzo, Vayanos, and 2. Bayes’ Law determines a conditional probability (for exam- Zweibel 1998). ple, the probability that a person is in a certain occupation 5. Quoted by O’Barr and Conley. conditional on some information about that person’s per- 6. Even more complicated feedback mechanisms involving sonality) in terms of other probabilities, including the base- the news media are no doubt at work—for instance, price rate probabilities (for example, the unconditional probabil- to news to confidence to economy to news to price. ity that a person is in an occupation and the unconditional 7. By inference, one might suppose that the prominent theo- probability that the person has a certain personality). ries that appear to move investors’ decisions are also based 3. Julius Caesar, De Bello Gallico, iii, 18. on the best possible information.

May/June 2002 25 Financial Analysts Journal References

Babad, E. 1987. “Wishful Thinking and Objectivity among Sport Financial Markets.” Journal of Finance, vol. 54, no. 6 Fans.” Social Behavior: An International Journal of Applied Social (December):65–91. Psychology, vol. 4:231–240. James, William, Principles of Psychology, New York: Henry Holt, Barberis, Nicholas, and Andrei Shleifer. 2000. “Style Investing.” 1890. National Bureau of Economic Research Working Paper W8039. Janis, Irving L. 1982. Groupthink. 2nd ed. Boston, MA: Houghton Barberis, Nicholas, Andrei Shleifer, and Robert Vishny. 1998. “A Mifflin. Model of Investor Sentiment.” Journal of Financial Economics, vol. Knight, Frank. 1964. Risk, Uncertainty, and Profit. Chicago, IL: 49, no. 3 (September):307–343. University of Chicago Press. Basu, Sanjoy. 1983. “The Relationship between Earnings’ Yield, O’Barr, William M., and John M. Conley. 1992. Fortune & Folly: Market Value, and Return for NYSE Common Stock: Further The Wealth and Power of Institutional Investing. Homewood, IL: Evidence.” Journal of Financial Economics, vol. 12, no. 1 Business One Irwin. (March):129–156. Phillips, L., and W. Edwards. 1966. “Conservatism in a Simple Beach, Lee Roy, and Gary P. Braun. 1994. “Laboratory Studies Probability Inference Task.” Journal of Experimental Psychology, of Subjective Probability: A Status Report.” In Subjective vol. 72:346–354. Probability. Edited by George Wright and Peter Ayton. New Shafir, Eldar, and Amos Tversky. 1993. “Reason-Based Choice.” York: Wiley. Cognition, vol. 49, no. 2:11–36. Campbell, John Y., and Robert J. Shiller. 1998. “Valuation Ratios Shiller, Robert J. 1989. Market Volatility. Cambridge, MA: MIT and the Long Run Stock Market Outlook.” Journal of Portfolio Press. Management, vol. 24, no. 2 (Winter):11–26. ———. 1990. “Market Volatility and Investor Behavior.” ———. Forthcoming 2002. “Valuation Ratios and the Long Run American Economic Review, Papers and Proceedings, vol. 80, no. 2 Stock Market Outlook: An Update.” In Advances in Behavioral (May):58–62. Finance II. Edited by Richard H. Thaler. New York: Sage ———. 2000. Irrational Exuberance. Princeton, NJ: Princeton Foundation. University Press. De Long, Brad, Andrei Shleifer, Lawrence Summers, and Shiller, Robert J., and John Pound. 1989. “Survey Evidence of Michael Waldman. 1990. “Positive Feedback Investment Diffusion of Interest among Institutional Investors.” Journal of Strategies and Destabilizing Rational Speculation.” Journal of Economic Behavior and Organization, vol. 12, no. 1 (August):47–66. Finance, vol. 45, no. 2 (June):379–395. Shleifer, Andrei. 2000. Inefficient Markets. Oxford, U.K.: Oxford Demarzo, Peter, Dimitri Vayanos, and Jeffrey Zweibel. 1998. “A University Press. Near Rational Model of Persuasion with Implications for Swensen, David. 2000. Pioneering Portfolio Management: An Financial Markets.” Paper presented at the National Bureau of Unconventional Approach to Institutional Investing. New York: Economic Research Asset Pricing Program Meeting, University Free Press. of Chicago, IL (October 30). Tversky, Amos, and Daniel Kahneman. 1974. “Judgment under Ellis, Charles D. 1998. Winning the Loser’s Game: Timeless Uncertainty: Heuristics and Biases.” Science, vol. 185, no. Strategies for Successful Investing. 3rd ed. New York: McGraw- 4157:1124–31. Hill. Ulhaner, C.J., and B. Grofman. 1986. “The Race May Be Close Heath, J., and Amos Tversky. 1991. “Preference and Belief: But My Horse Is Going to Win: Wish Fulfillment in the 1980 Ambiguity and Competence in Choice under Uncertainty.” Presidential Election.” Journal of Political Behavior, vol. 8, no. Journal of Risk and Uncertainty, vol. 4, no. 1 (January):5–28. 2:101–129. Hong, Harrison, and Jeremy Stein. 1999. “A Unified Theory of Wright, George, and Peter Ayton. 1994. Subjective Probability. Underreaction, Momentum Trading and Overreaction in New York: Wiley.