Irrational Exuberance

Robert J. Shiller

Princeton University Press (2000)

The News Media

The history of speculative bubbles begins roughly with the advent of newspapers.l One can assume that, although the record of these early newspapers is mostly lost, they regularly reported on the first bubble of any consequence, the Dutch tulip mania of the 1630s.2

Although the news media-newspapers, magazines, and broadcast media, along with their new outlets on the Internet-present themselves as detached observers of market events, they are themselves an integral part of these events. Significant market events generally occur only if there is similar thinking among large groups of people, and the news media are essential vehicles for the spread of ideas.

In this chapter, I consider the complexity of the media's impact on market events. As we shall see, news stories rarely have a simple, predictable effect on the market. Indeed, in some respects, they have less impact than is commonly believed. However, a careful analysis reveals that the news media do play an important role both in setting the stage for market moves and in instigating the moves themselves.

The Role of the Media in Setting the Stage for Market Moves

The news media are in constant competition to capture the public attention they need to survive. Survival for them requires finding and defining interesting news, focusing attention on news that has word-of-mouth potential (so as to broaden their audience), and, whenever possible, defining an ongoing story that encourages their audience to remain steady customers.

The competition is by no means haphazard. Those charged with disseminating the news cultivate a creative process, learning from each others' successes and failures, that aims to provide emotional color to news, to invest news stories with human interest appeal, and to create familiar figures in the news. Years of experience in a competitive environment has made the media professions quite skillful at claiming public attention.

The news media are naturally attracted to financial markets because, at the very least, the markets provide constant news in the form of daily price changes. Certainly other markets, such as real estate, are sources of news. But real estate does not typically generate daily price movements. Nothing beats the stock market for sheer frequency of potentially interesting news items.

The stock market also has star quality. The public considers it the Big Casino, the market for major players, and believes that on any given day it serves as a barometer of the status of the nation-- impressions that the media can foster and benefit from. Financial news may have great human interest potential to the extent that it deals with the making or breaking of fortunes. And the financial media can present their perennial lead, the market's performance, as an ongoing story-one that brings in the most loyal repeat customers. The only other regular generator of news on a comparable scale is sporting events. It is no accident that financial news and sports news together account for roughly half of the editorial content of many newspapers today.

Media Cultivation of Debate

In an attempt to attract audiences, the news media try to present debate about issues on the public mind. This may mean creating a debate on topics that experts would not otherwise consider deserving of such discussion. The resulting media event may convey the impression that there are experts on all sides of the issue, thereby suggesting a lack of expert agreement on the very issues that people are most confused about.

I have over the years been called by newspeople asking me if I would be willing to make a statement in support of some extreme view. When I declined, the next request would inevitably be to recommend another expert who would go on record in support of the position.

Five days before the 1987 stock market crash, the MacNeil/ Lehrer NewsHour featured Ravi Batra, author of The Great Depression of 2990: Why It's Got to Happen, How to Protect Yourself. This book took as its basic premise a theory that history tends to repeat itself in exact detail, so that the 1929 crash and subsequent depression had to repeat themselves. Despite Batra's significant scholarly reputation, this particular book of his is not one that would be viewed with any seriousness by most reputable scholars of the market. But it had been on the New York Times best-seller list for fifteen weeks by the time of the crash. On the NewsHour, Batra confidently predicted a stock market crash in 1989 that would "spread to the whole world"; after it, he declared, "there will be a depression."3 Batra's statements, made as they were on a highly respected show, may--even though they predicted a crash two years hence-have contributed in some small measure to an atmosphere of vulnerability that brought us the crash of 1987. Although Batra's appearance on the NewsHour just before the crash might be considered a coincidence, one must keep in mind that predictions of stock market crashes are actually quite rare on national news shows. The proximity of his appearance to the actual crash is at the very least highly suggestive.

Should the media be faulted for presenting debates on topics of little merit? One can argue that they ought to focus on a variety of topics of interest to general audiences, so that the public can refine their views. Yet in doing so the media seem often to disseminate and reinforce ideas that are not supported by real evidence. If news directors followed only their highest intellectual interests in judging which views to present, the public might indeed find its consciousness constructively broadened. But that is apparently not how the media see their mission nor do competitive pressures encourage them to rethink the matter.

Reporting on the Market Outlook

There is no shortage of media accounts that try to answer our questions about the market today, but there is a shortage within these accounts of relevant facts or considered interpretations of them. Many news stories in fact seem to have been written under a deadline to produce something--anything--to go along with the numbers from the market. The typical such story, after noting the remarkable bull market, focuses on very short-run statistics. It generally states which groups of stocks have risen more than others in recent months. Although these stocks are described as leaders, there is no good reason to think that their performance has caused the bull market. The news story may talk about the "usual" factors behind economic growth, such as the Internet boom, in glowing terms and with at least a hint of patriotic congratulation to our powerful economic engine. The article then finishes with quotes from a few well-chosen "celebrity" sources, offering their outlook for the future. Sometimes the article is so completely devoid of genuine thought about the reasons for the bull market and the context for considering its outlook that it is hard to believe that the writer was other than cynical in his or her approach.

What are the celebrity sources quoted as saying in these articles? They typically give numerical forecasts for the Dow Jones Industrial Average in the near future, tell stories or jokes, and dispense their personal opinions. For example, when Abby Joseph Cohen of Goldman Sachs & Co. coins a quotable phrase-as with her warnings against "FUDD" (fear, uncertainty, doubt, and despair) or her phrase "Silly Putty Economy"-it is disseminated widely. Beyond that, the media quote her opinions but pay no critical attention to her analysis. In fact, although she no doubt has access to a formidable research department and performs extensive data analysis before forming her opinions, they are ultimately reported as just that her opinions. Of course she should not be faulted for this, for it is the nature of the sound-bite-driven media that superficial opinions are preferred to in-depth analyses.

Record Overload

The media often seem to thrive on superlatives, and we, their audience, are confused as to whether the price increases we have recently seen in the stock market are all that unusual. Data that suggest that we are setting some new record (or are at least close to doing so) are regularly stressed in the media, and if reporters look at the data in enough different ways, they will often find something that is close to setting a record on any given day. In covering the stock market, many writers mention "record one-day price changes" measured in points on the Dow rather than percentage terms, so that records are much more likely. Although the media have become increasingly enlightened about reporting in terms of points on the Dow in recent years, the practice still persists among some writers.

This record overload--the impression that new and significant records are constantly being set--only adds to the confusion people have about the economy. It makes it hard for people to recognize when something truly and importantly new really is happening. It also, with its deluge of different indicators, encourages an avoidance of individual assessment of quantitative data-a preference for seeing the data interpreted for us by celebrity sources.

Do Big Stock Price Changes Really Follow Big News Days?

Many people seem to think that it is the reporting of specific news events, the serious content of news, that affects financial markets. But research offers far less support for this view than one would image.

Victor Niederhoffer, while he was still an assistant professor at Berkeley in 1971 (before he became a legendary hedge fund manager), published an article that sought to establish whether days with news of significant world events corresponded to days that saw big stock price movements. He tabulated all very large headlines in the New York Times (large type size being taken as a crude indicator of relative importance) from 1950 to 1966; there were 432 such headlines. Did these significant-world-event days correspond to big movements in stock prices? As the standard of comparison, Niederhoffer noted that the S&P Composite Index over this period showed substantial one-day increases (of more than 0.78%) on only 10% of the trading days, and substantial one-day decreases (of more than 0.71%) on only another 10% of the trading days. Of the 432 significant-world-event days, 78 (or 18%) showed big price increases, and 56 (or 13%) showed big decreases. Thus such days were only slightly more likely to show large price movements than other days .4

Niederhoffer claimed that, on reading the stories under these headlines, many of the world events reported did not seem likely to have much impact on the fundamental value represented by the stock market. Perhaps what the media thought was big national news was not what was really important to the stock market. He speculated that news events that represented crises were more likely to influence the stock market.

Defining a crisis as a time when five or more large headlines occurred within a seven-day period, Niederhoffer found eleven crises in the sample interval. These were the beginning of the Korean war in 1950, the capture of Seoul by the Communists in 1951, the Democratic National Convention of 1952, Russian troops' threatening Hungary and Poland in 1956, the Suez crisis of 1956, Charles de Gaulle's taking office as French premier in 1958, the entry of U.S. marines into Lebanon in 1958, Russian premier Nikita Khrushchev's appearance at the United Nations in 1959, Cuban tensions in 1960, the Cuban arms blockade in 1962, and President John Kennedy's assassination in 1963. During these crises, so defined, 42% of the daily price changes were "big" changes, as compared with 20% for other, "normal" time periods. Thus the crisis periods were somewhat, but not dramatically, more likely to be accompanied by big stock price changes.

Note that there were only eleven such weeks of "crisis" in the whole sixteen years of Niederhoffer's sample. Very few of the aggregate price movements in the stock market show any meaningful association with headlines.

Tag-Along News

News stories occurring on days of big price swings that are cited as the causes of the changes often cannot, one suspects, plausibly account for the changes-or at least not for their full magnitude. On Friday, October 13,1989, there was a stock market crash that was clearly identified by the media as a reaction to a news story. A leveraged buyout deal for UAL Corporation, the parent company of United Airlines, had fallen through. The crash, which resulted in a 6.91% drop in the Dow for the day, had begun just minutes after this announcement, and so it at first seemed highly likely that it was the cause of the crash.

The first problem with this interpretation is that UAL is just one firm, accounting for but a fraction of 1% of the stock market's total value. Why should the collapse of the UAL buyout have such an impact on the entire market? One interpretation at the time was that the deal's failure was viewed by the market as a watershed event, portending that many other similar pending buyouts would also fail. But no concrete arguments were given in support of this view; rather, dubbing it a watershed seemed to have been nothing more than an effort to make sense after the fact of the market's move in response to the news.

To try to discover the reasons for the October 13,1989, crash, survey researcher William Feltus and I carried out a telephone survey of 101 market professionals on the Monday and Tuesday following the crash. We asked: "Did you hear about the UAL news before you heard about the market drop on Friday afternoon, or did you hear about the UAL news later as an explanation for the drop in the stock market?" Only 36% said they had heard about the news before the crash; 53% said they had heard about it afterward as an explanation for the drop; the rest were unsure when they had heard about it. Thus it appears that the news story may have tagged along after the crash, rather than directly caused it, and therefore that it was not as prominent as the media accounts suggested.

We also asked the market professionals to interpret the news story. We queried:

Which of the following two statements better represents the view you held last Friday:

1. The UAL news of Friday afternoon will reduce future takeovers, and so the UAL news is a sensible reason for the sudden drop in stock prices.

2. The UAL news of Friday afternoon should be viewed as a focal point or attention grabber, which prompted investors to express their doubts about the market.

Of the respondents, 30% chose 1 and 50% chose 2; the rest were unsure. Thus they were mostly reacting to the news as an interpretation of the behavior of investors.5 It may be correct to say that the news event was fundamental to this stock market crash, in that it represented a "story" that enhanced the feedback from stock price drops to further stock price drops, thereby preserving the feedback effect for a longer period than would otherwise have been the case. Yet it was unlikely to have been its cause.

The Absence of News on Days of Big Price Changes

We can also look at days of unusually large price movements and ask if there were exceptionally important items of news on those days. Following up on Niederhoffer's work, in 1989 David Cutler, James Poterba, and Lawrence Summers compiled a list of the fifty largest U.S. stock market movements, as measured by the S&P Index, since World War II, and for each tabulated the explanations offered in the news media. Most of the so-called explanations do not correspond to any unusual news, and some of them could not possibly be considered serious news. For example, the reasons given for large price movements included such relatively innocuous statements as "Eisenhower urges confidence in the economy," "further reaction to Truman victory over Dewey," and "replacement buying after earlier fall."6

Some would argue that perhaps we should not expect to see prominent news on days of big price changes, even if markets are working perfectly. Price changes in a so-called efficient market occur, so the argument goes, as soon as the information becomes public; they do not wait until the information is reported in the media. (This is a topic to which I return in Chapter 9.) Thus it is not surprising, according to this line of reasoning, that we often do not find new information in the newspaper on the day of a price change: earlier information, appearing to the casual observer as tangential or irrelevant, has already been interpreted by perceptive investors as significant to the fundamentals that should determine share prices.

Another argument advanced to explain why days of unusually large stock price movements have often not been found to coincide with important news is that a confluence of factors may cause a significant market change, even if the individual factors themselves are not particularly newsworthy. For example, suppose certain investors are informally using a particular statistical model that forecasts fundamental value using a number of economic indicators. If all or most of these particular indicators point the same way on a given day, even if no single one of them is of any substantive importance by itself, their combined effect will be noteworthy.

Both of these interpretations of the tenuous relationship between news and market movements assume that the, public is paying continuous attention to the news-reacting sensitively to the slightest clues about market fundamentals, constantly and carefully adding up all the disparate pieces of evidence. But that is just not the way public attention works. Our attention is much more quixotic and capricious. Instead, news functions more often as an initiator of a chain of events that fundamentally change the public's thinking about the market.

News as the Precipitator of Attention Cascades

The role of news events in affecting the market seems often to be delayed, and to have the effect of setting in motion a sequence of public attentions. These attentions may be to images or stories, or to facts that may already have been well known. The facts may previously have been ignored or judged inconsequential, but they can attain newfound prominence in the wake of breaking news. These sequences of attention may be called cascades, as one focus of attention leads to attention to another, and then another.

At 5:46 A.M. on Tuesday, January 17,1995, an earthquake measuring 7.2 on the Richter scale struck Kobe, Japan; it was the worst earthquake to hit urban Japan since 1923. The reaction of the stock markets of the world to this event provides an interesting case study since in this case we know without doubt that the precipitating event, the earthquake, was truly exogenous and not itself generated by human activity or business conditions-not a response to a subtle hint of economic change nor the result of a confluence of unusual values of conventional economic indicators. In the Cutler-PoterbaSumrners list of media explanations for the fifty largest postwar movements in the S&P Index in the United States, discussed earlier, not a single one of the explanations referred to any substantial cause that was definitely exogenous to the economy.?

The earthquake took 6,425 lives. According to estimates by the Center for Industrial Renovation of Kansai, the total damage caused by the earthquake was about $100 billion. The reaction in financial markets was strong, but delayed. The Tokyo stock market fell only slightly that day, and prices of construction-related companies generally rose, reflecting the expected increased demand for their products and services. Analysts reported at that time that the probable effects of the earthquake on corporate value were as yet ambiguous, since the wave of rebuilding after the quake might stimulate the Japanese economy.

The biggest reaction to the earthquake did not come until a week later. On January 23, the Japanese Nikkei index fell 5.6% on no apparent news except the gradual unfolding of numerous news accounts of earthquake damage. Over the ten days following the earthquake, the Nikkei lost over 8% of its value. If viewed as the direct result of the earthquake damage alone, the loss of value would be an overreaction.

What was going on in investors' minds over the ten days following the earthquake? Of course, there is no rigorous way to find out. We know only that over this period the Kobe earthquake dominated the news, created new and different images of Japan, and may have led to very different impressions about the Japanese economy. Moreover, the quake sparked discussions about the risk of an earthquake centered in Tokyo. Despite the fact that geological evidence suggesting that Tokyo is at risk for a major earthquake was already known, greater attention was now focused on this potential problem. The damage that an earthquake of the severity of the 1923 quake could cause to modern-day Tokyo was put at $1.25 trillion by Tokai Research and Consulting Inc. 8

Even more puzzling than the direct effect of the Kobe earthquake on the domestic Japanese markets was its effect on foreign stock markets. On the day that the Nikkei fell 5.6%, the FT SE 100 index in London fell 1.4%, the CAC-40 in Paris fell 2.2%, and the DAX in Germany fell 1.4%. The Brazilian and Argentine stock markets both fell about 3%. These diverse countries around the world suffered no earthquake damage on this occasion.

The best interpretation of the effects of the Kobe earthquake on the stock markets of the world is that news coverage of the earthquake, and of the accompanying stock market declines, engaged tile attention of investors, prompting a cascade of attentions that brought to the fore some more pessimistic factors. Another market reaction to news illustrates how media attention may, through a cascade of attentions, lead many investors to eventually take seriously news that would normally be considered nonsense and irrelevant. A sequence of news stories about Joseph Granville, a flamboyant market forecaster, appear to have caused a couple of major market moves. The only substantive content of these media stories was that Granville was telling his clients to buy or sell, and that Granville himself was influential.

Granville's behavior easily attracted public attention. His investment seminars were bizarre extravaganzas, sometimes featuring a trained chimpanzee who could play Granville's theme song, "The Bagholder's Blues," on a piano. He once showed up at an investment seminar dressed as Moses, wearing a crown and carrying tablets. Granville made extravagant claims for his forecasting ability. He said he could forecast earthquakes and once claimed to have predicted six of the past seven major world quakes. He was quoted by Time magazine as saying, "I don't think that I will ever make a serious mistake in the stock market for the rest of my life," and he predicted that he would win the Nobel Prize in economics .9

The first Granville episode took place on Tuesday, April 22, 1980. With the news that he had changed his recommendation from short to long, the Dow rose 30.72 points, or 4.05%. This was the biggest increase in the Dow since November 1,1978, a year and a half earlier. The second episode occurred on January 6,1981, after Granville's investor service changed from a long recommendation to a short recommendation. The Dow took its biggest dive since October 9,1979, over a year earlier. There was no other news on either of these occasions that might appear responsible for the market change, and on the second occasion both the Wall Street Journal and Barrons squarely attributed the drop to Granville's recommendation.

Can we be sure that media reporting of Granville and his supposed powers of prognostication caused these changes? Many people wondered if the Granville effect was not just a coincidence that the news media exaggerated. We can be sure that a sequence of news stories about Granville's pronouncements, with their substantial word-of-mouth potential, had a cumulative effect on national attention, and that public reactions to his pronouncements and to market declines at the time of his announcements were fundamentally altered by this cascade.10

News during the Crash of 1929

The role of the news media in causing the stock market crash of 1929 has been debated almost since the crash itself. In fact the puzzle facing historians and economists has been, by some interpretations, that just before the crash there was no significant news at all. But, people have wondered ever since, how could this record stock market crash get under way with no news? What common concerns were on the minds of sellers that caused so many of them to try to sell at the same time?

The Monday, October 28,1929, stock market crash was the biggest single-day drop (measured between the closing price the previous trading day and the closing price on the day) in the Dow until the October 19,1987, crash. On October 28,1929, the Dow fell 12.8% in one day (13.01% measured from the high to the low on that day). The second-biggest drop in history (until 1987) occurred the following day, when the Dow dropped 11.7% (15.9% measured from the high to the low on that day). The combined close-to-close drop in those two days in 1929 was 23.1%. What news had arisen that might rationally account for such a sizable stock market decline?

On reading the major newspapers over that weekend and on into the morning of Tuesday, October 29, one is easily led to conclude that nothing of any consequence for the fundamentals of the market was happening. Indeed that was the conclusion reported in the newspapers themselves. On the morning of October 29, newspapers around the country carried an Associated Press story that said in part, "In the absence of any adverse news developments over the week end, and in the face of the optimistic comments on business forthcoming from President Hoover and leading industrial and banking executives, Wall Street's only explanation of today's decline was that a careful checking up of accounts over the week-end disclosed numerous weak spots, which had been overlooked in the hectic sessions of last week." The New York Times attributed the drop only to a "general loss of confidence." The Wall Street Journal reported that "business in general shows no signs of disintegration" and that the decline was due to "necessitous liquidation of impaired accounts."11

What else was in the news on those days? As of Monday morning there was news that the Interstate Commerce Commission would proceed with its plan to recapture some excess railroad income. There was a favorable report on the earnings of U.S. Steel. New information was reported on charges that the Connecticut Manufacturer's Association had succeeded in introducing into a tariff bill provisions favoring Connecticut. Mussolini had made a speech saying that the "men and institutions of fascism can face any crisis, even if it is sudden." A new aspirant to the French premiership, Edouard Daladier, announced the foreign minister of his prospective cabinet. A British airliner was lost at sea with seven aboard. The Graf Zeppelin planned a trip to explore the Arctic. Richard Byrd's party was making progress toward the South Pole.

After Black Monday, early on Tuesday morning, the second day of the crash, it was reported that prominent financiers had asserted that heavy banking support would come into the market that day, in search of bargains. If this was significant news at all, one would think it was good news. Other news on Tuesday morning was that two senators had called on President Hoover to declare his position on duties on agricultural and industrial products, that Senator Hiram Bingham had complained that the Lobby Inquiry had treated him unfairly, a Hungarian count and countess had been given the right to enter the country, and another airliner had been lost with five aboard.

All of these stories sound very typical. If there really was a good reason for the drop in the market, then certainly there must have been something happening at the time that people knew about. And one would think that such concerns would have made it into the news in some form. Perhaps one must read the papers more carefully. One author, Jude Wanniski, indeed claimed that there was a story in the New York Times on the morning of Monday, October 28, 1929, that might conceivably account for such a decline. This front-page story was an optimistic report on the likelihood of passage of the Smoot-Hawley tariff, then still in committee. The story was picked up by the Associated Press and United News the following day and given front-page treatment around the country on Tuesday, October 29.12

It is conceivable that the Smoot-Hawley tariff might have been expected to hurt the outlook for U.S. corporate profits. One could have thought that it would generally benefit corporations, many of whom actively sought the tariff. But it has been argued by historians of the 1929 crash that the tariff might have been expected to have the opposite effect, given the retaliation from other countries that it would engender. Allan Meltzer in fact argued that the tariff could be the reason "why the 1929 recession did not follow the path of previous monetary contractions but became the Great Depression."13 However, other economists, including Rudiger Dornbusch and Stanley Fischer, pointed out that exports were only 7% of the gross national product (GNP) in 1929 and that between and 1931 they fell by only 1.5% of 1929 GNP This hardly seems cause of the Great Depression. Moreover, they pointed out t it is not clear that the Smoot-Hawley tariff was responsible for the decline in exports. The depression itself might be held responsible for part of the decline. Dornbusch and Fischer showed that the 1922 Fordney-McCumber tariff increased tariff rates as much as the Smoot-Hawley tariff, and the Fordney-McCumber tariff was of course followed by no such recession..14

Even if we were to allow that the possibility of passage of the Smoot-Hawley tariff was important enough to account for a decline in share values of this magnitude, one must still ask ,whether there was any news over the weekend that would substantially alter one's estimation of the likelihood that the tariff would be passed. Just what was the content of the story in the New York :Times? On Saturday, October 26, Senator David Reed declared that the Smoot-Hawley tariff bill was "dead" in committee. This provoked denials by Senators Reed Smoot and William Borah. The Times quoted Senator Smoot as saying, "If that is Senator Reed's opinion, I suppose he has a right to express it. But it isn't the view of the Finance Committee." Senator Borah said, "My opinion is that the tariff bill is not going to die." The next morning, October 29, ,the Times reported that Senator Reed had reiterated his conviction that the bill was dead and went on to cite other opinions on both sides of the issue. Although the original Times story had sounded optimistic for the bill, the United News version of the story published on October 29 was pessimistic. The Atlanta Constitution, when it ran the story on October 29, carried the headline, "Senate Gives Up Hope of Enacting New Tariff Bill."

Nonetheless, it is hard to see that this interchange among senators, so typical of political wrangling, amounts to important news. The same sort of news accounts had been coming out all along with regard to the tariff bill. A week earlier, on October 21, the Times had quoted Senator James Watson, Republican leader of the Senate, offering his view that the Senate would pass the bill within another month. On October 13, Senator Smoot was reported as telling President Hoover that there was a chance the bill would pass by November 20. Alternately optimistic and pessimistic news on the tariff bill had been coming in since Hoover's election.

Far more significant than news about fundamentals among the newspaper stories on Monday, October 28,1929, are clues to the importance attached in people's minds to the events of just a few days earlier, when the stock exchange had seen a record decline in share prices. That was the so-called Black Thursday, October 24, 1929, when the Dow had fallen 12.9% within the day but recovered substantially before the end of trading, so that the closing average was down only 2.1% from the preceding close. This event was no longer news, but the memory of the emotions it had generated was very much part of the ambience on Monday. The New York Times noted in its Monday morning edition that Wall Street, "normally deserted and quiet on Sunday as a country graveyard, hummed with activity as bankers and brokers strove to put their houses in order after the most strenuous week in history . . . . When the bell clangs at 10 o'clock this morning for the resumption of trading, most houses will be abreast of their work and ready for what may come." The atmosphere of that Sunday on Wall Street was described: "Sightseers strolled from street to street, gazing curiously at the Stock Exchange Building and the Morgan banking offices across the way, centers of last week's dramatic financial happenings. Here and there a sightseer picked up from the street a vagrant slip of ticker tape, as visitors seize upon spent bullets on a battlefield as souvenirs. Sightseeing buses made special trips through the district. "15

Indeed, on that same Monday morning of the crash the Wall Street journal saw fit to run a front-page editorial stating that "everybody in responsible positions says that business conditions are sound."16 The editorial staff of the Journal must have had reasons to suspect that reassurance was needed if the market was to remain stable. Presumably they had heard snippets of popular conversation, or could at least guess how people might react following the weekend, given the huge debacle on Thursday.

So perhaps what happened on Monday, October 28,1929, was just an echo, albeit a very exaggerated one, of what had happened the previous week. What had the media said about this? Again, the newspapers seemed to think that there was no important news. The sago Tribune wrote, on Sunday, October 27,1929, "It has been collapse of a vastly inflated bubble of speculation, with little or %' cause in the country's general situation. A top-heavy structure has collapsed of its own weight-there has been no earthquake." The New York Times said, "The market smash has been caused by technical rather than fundamental considerations." The Guaranty Survey, published by the Guaranty Trust Company, remarked that "to suppose that the selling wave of the last few weeks was due to adverse developments of corresponding importance in the gen.~ :al business situation would be a fundamental error."17

Let us go back in time and look at the news on the morning of Black Thursday, October 24,1929. Once again, the news does not ,. teen to be very significant. President Hoover had announced a plan to develop inland waterways. Atlantic Refinings' earnings for the dear were reported to be its highest ever. The president of a sugar

company had told a Senate committee investigating lobbying that $75,000 had been spent by the sugar lobby since December in a campaign to reduce duties on sugar. Negotiators had reported a setck in efforts to establish the Bank for International Settlements. A Carnegie Fund report decried the subsidization of college ath.tes. The America's Cup committee had announced the rules for the next running of the yacht race. An amateur pilot attempting solo flight across the Atlantic was reported lost. President Hoover had taken a trip on a picturesque river boat down the Ohio River.

Nothing here seems remotely to suggest anything fundamental about the outlook for the stock market. But let us look back yet another day. There was news on the Wednesday before Black Thursday that there had been a major drop in the market (the Dow closed on Wednesday down 6.3% from Tuesday's close) and that total transactions had had their second highest day in history. Should we then look for the cause in the news of October 23,1929? Again there was no national news of any apparent significance, but again there were references to past market moves. The most significant concrete news stories in the newspapers seem consistently to have been about previous moves of the market itself. The most prominent content in the news appears to have been interpretations of the reasons for these previous moves, often in terms of investor psychology.

There is no way that the events of the stock market crash of 1929 can be considered a response to any real news stories. We see instead a negative bubble, operating through feedback effects of price changes, and an attention cascade, with a series of heightened public fixations on the market. This sequence of events appears to be fundamentally no different from those of other market debaclesincluding the notorious crash of 1987, to which we now turn.

News during the Crash of 1987

When the stock market crashed on October 19,1987--setting a new record one-day decline that nearly doubled that of either October 28 or October 29,1929 (to this day it is the all-time record one-day price drop)-- I considered it a unique opportunity to inquire directly of investors what they considered to be the significant news on that day. It was no longer necessary, as it had been for those who studied the 1929 crash, to rely on media interpretations suggesting what was the important news on investors' minds. As far as I have been able to determine, no one else took advantage of this opportunity. The results of my questionnaire survey, sent out to a sample of institutional investors and a sample of individual investors the week of the crash, were the only published findings of a survey asking investors what they were thinking on the day of the crash.18

In my 1987 survey, I listed all the news stories published in the few days preceding the crash that seemed at all relevant to the Changing opinions of the market, ending with news that had appeared in the papers on the morning of the crash. I asked the investors:

Please tell us how important each of the following news items was to you personally on October 19,1987, in your evaluation of stock market prospects. Please rate them on a one-to-seven scale, 1 indicating that the term was completely unimportant, 4 indicating that it was of moderate importance, 7 indicating that it was very important. Please tell how important you then felt these were, and not how others thought about them.
I included ten news stories, and in the eleventh positions space marked "Other" where respondents could write in their own choices.

The results were broadly similar between institutional and individual investors, and between those who had actually bought or sold on October 19. Respondents rated everything as relevant. They thought that most of the news stories rated at least a 4, that is, they were of moderate importance. The only news story that merited an average score less than 3 was the sell signal that investment guru Robert Prechter was reported to have given on October 14, and even that received a score around 2. Even the news that the United States attacked an Iranian oil station, a minor skirmish reported on October 19, received a rating over 3. Respondents were not very forthcoming with other news stories in the "Other" category. They tended to mention concerns, rather than news stories that broke at the time of the crash. The most common write-in answer was a concern about too much indebtedness, referring variously to the federal deficit, the national debt, or taxes. Such a response was offered by a third of the individual investors who wrote in answers and a fifth of the institutional investors.

But the striking result was that the most highly rated news stories among those I listed were those about past price declines themselves. The most important news story, according to the respondents, was the 200-point drop in the Dow on the morning of October 19, a news story that yielded an average score of 6.54 among individual sellers on October 19 and 6.05 among institutional sellers on October 1.9. The preceding week's news of the record (in terms of points lost) stock market declines was considered the second most important story.

One of the questions asked respondents to give their recollecions of the interpretations they had attached to the price declines on the day of the crash: "Can you remember any specific theory you had about the causes for the price declines October 14-19, 1987?" Respondents were given space to write answers in their own words, which I read and categorized. Odd as it may seem from the perspective of today's much higher market, the most common theme in the answers to this open-ended question was that the market had been overpriced before the crash. Overpricing was mentioned by 33.9% of the individual investors and 32.6% of the institutional investors. Although this response accounts for fewer than half the answers, it is noteworthy that so many thought to mention this in answer to an open-ended question. (I also asked them directly elsewhere on the questionnaire whether they thought, just before the crash, that the market was overpriced, and 71.7% of the individual investors [91.0% of those who had sold on October 19] and 84.3% of the institutional investors [88.5% of those who had sold on October 19] said yes.)19 Another important theme in answer to the open-ended question was one of institutional stop-loss, identified by the presence of the words institutional selling, program trading, stop-loss, or computer trading; 22.8% of the individuals and 33.1% of the institutional investors mentioned such a theme. There was also an investor irrationality theme, identified by statements to the effect that investors were crazy or that the fall was due to investor panic or capricious changes in opinion; 25.4% of the individuals and 24.4% of the institutional investors touched on this theme. None of these major themes had anything to do with breaking news events other than the crash itself.

Immediately after this question, I asked on the questionnaire, "Which of the following better describes your theory about the declines: a theory about investor psychology [or] a theory about fundamentals such as profits or interest rates?" Most-67.5% of the institutional investors and 64.0% of the individual investorspicked a theory about investor psychology.

Thus it appears that the stock market crash had substantially to do with a psychological feedback loop among the general investing public from price declines to selling and thus further price declines, along the lines of a negative bubble, as discussed in Chapter 3. The crash apparently had nothing particularly to do with any news story other than that of the crash itself, but rather with theories about other investors' reasons for selling and about their psychology.

President Ronald Reagan, reacting to the crash, set up a study commission headed by former Treasury Secretary Nicholas Brady. He asked the Brady Commission to tell him what had caused the crash and what should be done about it. Investment professionals are generally uncomfortable going on record to explain the causes of such events, and many reports about the crash tended to focus inquiry away from its ultimate causes. But the members of the Brady Commission were under orders from the president of the United States to face the matter head on. As a result, we have in their report the only major effort to collect all the relevant facts and explain the crash of 1987. They wrote in their summary the following explanation for the crash:

The precipitous market decline of mid-October was "triggered" by specific events: an unexpectedly high merchandise trade deficit which pushed interest rates to new high levels, and proposed tax legislation which led to the collapse of the stocks of a number of takeover candidates. This initial decline ignited mechanical, price-insensitive selling by a number of institutions employing portfolio 3~, insurance strategies and a small number of mutual fund groups reacting to redemptions. The selling by these investors, and the prospect of further selling by them, encouraged a number of aggresvsive trading-oriented institutions to sell in anticipation of further market declines. These institutions included, in addition to hedge funds, a small number of pension and endowment funds, money management firms and investment banking houses. This selling, in turn, stimulated further reactive selling by portfolio insurers and mutual funds.20

This conclusion by the Brady Commission sounds in some ways very much like the one I drew from my own survey-based study of the crash. By "price-insensitive selling" they mean selling that comes in response to a price drop but is insensitive to how low the price goes before the sale is concluded-selling at any price. The commission was saying here, most prominently, that the crash was caused by what I have called a feedback loop, with initial price declines influencing more investors to exit the market, thereby eregating further price declines. The Brady Commission was saying, in effect, that the crash of 1987 was a negative bubble.

A strength of the Brady Commission's study of the crash relative to my own was their unparalleled access to major investing institutions. Their study complements my own in reaching the conclusion that a feedback loop was at work in the crash. However, their conclusion sounds a bit different from mine in that it gives prominence to the substantive content of news stories. In addition, theirs suggests that much of the selling was "mechanical" or "reactive," rather than psychological or herdlike.

Based on the results of my study, the news stories that the Brady Commission mentions about the merchandise trade deficit and about new highs in interest rates cannot be considered central to investors' thinking. In my survey, I included these in my list of news stories and got a lukewarm response from respondents (mostly 4s). Moreover, if one looks at long-term plots of both the trade deficit and interest rates, it is very clear that there was no sudden break in either of these series that could possibly be seen as standing out in a historical perspective. Virtually nothing happened to either the trade deficit or interest rates.

The proposed tax legislation that the Brady Commission mentions had completely escaped my notice as an important news story to include on my list. The news had broken on October 14, five days before the crash, and it had not seemed to me to be the subject of significant public comment in the days leading up to the crash. Representative Dan Rostenkowski's House Ways and Means Committee was considering tax changes that would have had the effect of discouraging corporate takeovers. Changing capital gains tax provisions struck many would-be interpreters of the crash after the fact as having fundamental importance for stock prices in an efficient market.

When I learned of the potential importance of this news story, I went back over the questionnaires I had received to see how many respondents had mentioned it in their answers under "Other." I found no mention at all among the 605 individual responses, and only three mentions among the 284 institutional responses. Clearly, this news story does not deserve to be singled out as a major cause of the crash.21

The Brady Commission puts quite a bit of stress on a tool of institutional investors called "portfolio insurance." Portfolio insurance is a strategy for limiting losses that was invented by Professors Hayne Leland and Mark Rubinstein at the University of California at Berkeley and successfully marketed by them to many institutional investors in the 1980s. Portfolio insurance is really a misnomer; the strategy is merely a plan for selling stocks. It involves impressive mathematical models, but in fact it is nothing more than a formalized procedure for getting out of the market by selling stocks when they start to go down. Leland himself, in his classic 1980 article on portfolio insurance, admits as much: "Some 'rules of thumb'such as 'run with your winners, cut your losses' and 'sell a new high, buy at a new low,' will be shown to approximate optimal dynamic trading strategies for certain types of estors."22 So, by using portfolio insurance, investors are merely doing what has always come naturally, only with a little more athematical precision and careful planning. But with the fancy dew name portfolio insurance, which suggests that the strategy is prudent and sensible, and with its high-tech image, the advent of this strategy quite likely made many investors more reactive to past price changes.

The adoption of portfolio insurance by many institutional testors was a sort of fad-a sophisticated fad, but a fad none~less. Since it has a distinctive name (the term portfolio insurance had essentially not been used before 1980), it is possible to trace the course of this investor fad by means of word counts in the press. I performed such a count on ABI/INFORM, a database of business s periodicals, and found no more than 1 reference to portfolio insury :: ice in each of the years 1980-83, 4 in 1984, 6 in 1985, 41 in 1986, and 75 in 1987. References to portfolio insurance were growing along the type of steady growth path that characterizes simple word-of-mouth epidemic models, which will be discussed in Chapter 8.23

So the development of portfolio insurance changed the way some investors reacted to past price changes just before the crash of 1987. There were probably other changes in the nature of the feedback loop that, because they were not so concretely programmed as portfolio insurance, we could not observe directly. But the important point is that it was the changed nature of the feedback loop, not the news stories that broke around the time of the crash, that was the essential cause of the crash.

Feedback can be modified by many factors, and the news media themselves can certainly have an impact on it. The Wall Street journal, on the morning of the 1987 crash, ran a plot showing the Dow in the 1980s and, just below it, a plot showing the Dow in the 1920s up to and for a month after the crash of 1929?4 The two plots were aligned so that the current date lined up with the date of the 1929 crash, and so the plot suggested that the crash of 1929 might be about to repeat itself. Investors had the opportunity to see this plot at breakfast a matter of minutes before the crash of 1987 actually started. The Journal was openly suggesting the possibility of a crash starting that day. True, this was not a front-page story, and no one story by itself is decisive in causing a crash. But this little story and the accompanying plot, appearing as they did on the morning of the crash, probably did help prime investors to be more alert to suggestions of a crash.

When the big price declines on the morning of October 19,1987, began, the archetype that was the 1929 crash encouraged many people to question whether "it" was happening again-the "it" being the Great Crash as illustrated in the journal, not the crash of 1907, nor the upcrash of 1932, nor any of the numerous other historical stock market events that by then had been almost completely forgotten. The mental image of the biggest crash in history possibly happening on that very day had the potential to enhance the feedback from initial price declines to later price declines. The image also provided a suggestion of how far the market would decline before it rebounded, a crucial factor in determining how far the market actually did fall. In fact, in the crash of October 19,1987, the Dow actually fell in one day almost the same amount as it did on October 28-29,1929-22.6% in 1987 versus 23.1 % in 1929. That it fell roughly the same amount on both occasions might be regarded as just a coincidence, especially since the 1987 crash took two days rather than one, and few investors in 1987 even knew exactly how far the market fell in 1929. On the other hand, many did have a rough impression of the extent of the 1929 plunge, and there was little other concrete information available to investors on October 19,1987, to suggest when the market should stop falling.

The changed feedback that occurred at the time of the 1987 crash should be thought of as just one example of continually changing price-to-price feedback, as investors' theories and methods change over time. It would be a mistake to describe the changed feedback as the result only of the technological innovation represented by portfolio insurance. Despite the use of computers in executing port-folio insurance strategies, it is still people who decide to deploy tool and who decide how quickly it will take effect in a declining market. And there are of course many other people who, aware that portfolio insurance is being used, adjust their own informal responses to past price changes depending on their perceptions of other investors' use of the strategy. Portfolio insurance is of interest to us in this context only because it shows us concretely how people's thinking can change in ways that alter the manner in which feedback from stock price changes affects further stock price changes, thereby creating possible price instabilities.

The Role of News Media in Propagating Speculative Bubbles

The role of the news media in the stock market is not, as commonly believed, simply as a convenient tool for investors who are reacting directly to the economically significant news itself. The media actively shape public attention and categories of thought, and they create the environment within which the stock market events e see are played out.

The examples given in this chapter illustrate that the news media are fundamental propagators of speculative price movements through their efforts to make news interesting to their audience. They sometimes strive to enhance such interest by attaching news stories to stock price movements that the public has already observed, thereby enhancing the salience of these movements and focusing greater attention on them. Or they may remind the public of past market episodes, or of the likely trading strategies of others. Thus the media can sometimes foster stronger feedback from past price changes to further price changes, and they can also foster another sequence of events, referred to here as an attention cascade.

This is not to say that the news media are a monolithic force pushing ideas onto a purely passive audience. The media represent a channel for mass communication and the interpretation of popular culture, but popular culture has an inherent logic and process of its own. We turn next to a study of some of the basic ideas in our culture, whose transformation over time bears a relation to the changing speculative situation in stock markets.