by David Dreman
Forecasting Follies 9: Psychological Influences on Decisions
In the preceding chapter, we saw that expert forecasts were sharply off the mark in many fields besides the stock market. Within the market, the analyst traveling with a laptop can run spreadsheets, check stock quotes, receive faxes, even tap into voluminous databases. At home base, his data input capabilities increase enormously. As one Morgan analyst stated, extracting useful information from the forty-nine databases the bank subscribes to is like finding a needle in a haystack. “The more data you get, the less information you have,” he groaned.29 His intuition coincides with the psychological findings. Increased information, as was demonstrated, does not lead to increased accuracy. A large number of studies in cognitive psychology indicate that human judgment is often predictably incorrect. Nor is overconfidence unique to analysts. People in situations of uncertainty are generally overconfident on the basis of the information available to them; they usually believe they are right much more often than they are.
These findings apply to many other fields. A classic analysis of cognitive psychologists found that it was impossible to predict which psychologists would be good diagnosticians. Further, there were no mechanical forecasting models that could be continuously used to improve judgment. The study concluded that the only way to resolve the problem was to look at the record of the diagnostician over a substantial period of time.
Researchers have also shown that people can maintain a high degree of confidence in their answers, even when they know the “hit rate” is not very high. The phenomenon has been called the “illusion of validity,” as noted before briefly.30 This also helps explain the belief that analysts can pinpoint their estimates despite the strong evidence to the contrary. People make confident predictions from incomplete and fallible data. There are excellent lessons here for the stock forecaster.
Forecasting Follies 10: Mr. Inside and Mr. Outside
Daniel Kahneman, who for several decades was a coauthor of many important scholarly pieces with Amos Tversky, wrote on this subject in collaboration with Dan Lovallo.31
Forecasters are “excessively prone” to treating each problem as unique, paying no attention to history. Cognitive psychologists note that there are two distinct methods of forecasting. The first is called the “inside view.” This method is the one overwhelmingly used to forecast earnings estimates and stock prices. The analyst or stock forecaster focuses entirely on the stock and related aspects such as growth rates, market share, product development, the general market, the economic outlook, and a host of other variables.
The “outside view,” on the other hand, ignores the multitude of factors that go into making the individual forecast and focuses instead on the group of cases believed to be most similar. In the case of earnings estimates, for example, it would zero in on how accurate earnings forecasts have been overall or how accurate they have been for a specific industry or for the company itself in deciding how precisely the analyst can estimate and the reliance that can be placed on the forecast.
If stock market forecasters are to succeed using the inside view, they must capture the critical elements of the future. The outside view, in contrast, is essentially statistical and comparative and does not attempt to read the future in any detail.
Kahneman relates a story to demonstrate the difference. In the mid-1970s, he was involved with a group of experts in developing a curriculum on judgment and decision making under uncertainty for high schools in Israel. When the team had been in operation for a year and had made some significant progress, discussion turned to how long it would take to complete the project. Everyone in the group, including Kahneman, gave an estimate. The forecasts ranged from eighteen to thirty months. Kahneman then asked one of his colleagues, an expert in curriculum development, to think of similar projects he was familiar with at a parallel stage in time and development. “How long did it take them from that point to complete their projects?” he asked.
After a long pause the expert replied with obvious discomfort that, first of all, about 40 percent of the projects had never been completed. Of the balance, he said, “I cannot think of any that was completed in less than seven years, nor any that took more than ten.” Kahneman then asked if there were any factors that made this team superior in attempting the task. None, said the expert. “Indeed we are slightly below average in terms of our resources and our potential.” As experienced as he was with the outside view, the curriculum development expert was just as susceptible to the inside view.*51
As is now apparent, the inside and outside views draw on dramatically different sources of information, and the processes are poles apart. The outside view ignores the innumerable details of the project on hand (the cornerstone of analysis using the inside view) and makes no attempt to forecast the outcome of the project into the future. Instead, it focuses on the statistics of projects similar to the one being undertaken to determine the odds of success or failure. The basic difference is that with the outside view, the problem is treated not as unique but as an instance of a number of similar problems. The outside view could be applied to a large number of the problems we’ve seen, including curriculum building, medical and psychiatric or legal diagnosis, and forecasting earnings or future stock prices.
According to Kahneman, “It should be obvious that when both methods are applied with intelligence and skill the outside view is much more likely to yield a realistic estimate. In general, the future of long and complex undertakings is simply not foreseeable in detail.” The number of possible outcomes when dozens or hundreds of factors interact in the marketplace is, for all practical purposes, infinite. Even if one could foresee each of the possibilities, the probability of any particular scenario is negligible. Yet this is precisely what analysts are trying to accomplish with a single, precise prediction.
Forecasting Follies 11: The Forecasters’ Curse
Let’s return to our analytical friends and look at their chances of success in terms of the inside view. As Table 8-3 makes clear, the probability of their being correct on their forecasts over any but the shortest periods of time is extremely low, and this means that the chances of making money consistently using precise forecasts are almost negligible.
As we saw earlier, the Street demands forecasts normally within a range of plus or minus 3 percent. Table 8-3, taken from our previous analysts’ forecasting study, shows how slim the probability of getting estimates within even the wider 5 percent range actually is. Remember, only 30 percent of forecasts made this target in any one quarter.
The table shows the chance of an analyst’s hitting the target for one quarter, four quarters, ten quarters, and twenty quarters; for all earnings surprises in column 1, negative surprises in column 2, and positive surprises in column 3. It is not reassuring. The odds against the investor who relies on fine-tuned earnings estimates are staggering. There is only a 1-in-132 chance that the analysts’ consensus forecast will be within 5 percent for any four consecutive quarters. Going longer makes the odds dramatically worse. For any ten consecutive quarters, the odds of fine-tuning the estimates for a company within this range fall to 1 in 199,000, and for twenty consecutive quarters, they fall to 1 in 40 billion. We stopped our calculation of estimates there, as the odds would have been likely to go up to the trillions twenty years out. Yet those are exactly the forecasting techniques most good investors on the Street use—and, oh yes, what the EMH theorists say keeps the market efficient. Both are guilty of significant error: For the practitioners it is not wanting to believe the odds are so high against forecasting. For the theorists it is not having an inkling of how badly one of the chief tools of security analysis performs. The bell tolls for both groups.
To put this all into perspective, your probability of being the big winner of the New York State Lottery is more than 777 times as great as your probability of pinpointing earnings every quarter for the next five years. Few people would put a couple of bucks into a lottery against odds like those, but mi
llions of investors will play them in the marketplace for big stakes anytime.
Some folks will say, “Who cares if earnings come in above estimates? In fact, I’ll applaud.” Fair enough. So we asked: what are the chances that you will avoid a 5 percent negative surprise for ten to twenty consecutive quarters? The answer is “Very poor.” The investor has only a 1-in-5 chance of not getting a negative earnings surprise 5 percent below the consensus forecast after only four quarters. After ten quarters, the chance of not receiving at least one crippling earnings surprise goes down to 1 in 62, and after twenty quarters, it is 1 in 3,800.
Yet, as we have also seen, forecasts often have to go out a decade or more to justify the high prices at which many growth companies are trading. If the odds are staggeringly high against being precise for five years, what are they for ten or fifteen? Many extremely pricey growth stocks must make their forecasts for ten- or fifteen-year periods in order to justify their current prices.
Think about it: should anyone not on something want to play against these odds? Yet, as we know, relying on accurate estimates is the way most people play the investment game. Investors who understand these prohibitive odds will obviously want to go with them if there is a way to do it, which is exactly what we’ll look at in the next section.
What we see here is a classic case of using the inside rather than the outside view. Evidence such as the above strongly supports Kahneman’s statement that the outside view is much more likely to yield realistic results. Yet, as Kahneman states, “The inside view is overwhelmingly preferred in forecasting.”
In the marketplace, the outside view does not give the stock investor the same confidence that he is in control and can use his expertise to power through to above-average returns. Nor does it provide much excitement or good “war stories,” which most clients like. It is used far less frequently than the inside view, although we have seen in the previous chapter that index funds, which are structured entirely on the outside view, easily beat most mutual funds over time. The superior returns from contrarian strategies also come from the outside view, as we’ll see in the next section.
Another Psychological Guideline is appropriate here:
PSYCHOLOGICAL GUIDELINE 16: The outside view normally provides superior returns over time. To maximize your returns, purchase investments that provide you with this approach.
Looking at the figures above, someone might ask, “Why?” The answer is again psychological. The natural way a decision maker approaches a problem is to focus all of his or her knowledge on the task, concentrating particularly on its unique features. Kahneman noted that a general observation of overconfidence is that even when forecasters are aware of findings such as the foregoing, they will still use the inside approach, disregarding the outside view, regardless of how strong its statistical documentation is.
Often, the relevance of the outside view is explicitly denied. Analysts and money managers I have talked to about high error rates repeatedly shrug them off. In sum, they ignore the record of forecasting because they have been taught and believe that investment theory, when executed properly, will yield the precise results that they require. Analysts and money managers seem unable to recognize the problems inherent in forecasting.
This situation is not unique to Wall Street. Indeed, the relevance of the statistical calculations inherent in the outside view is usually explicitly denied. Doctors and lawyers often argue against allying statistical odds to particular cases. Sometimes their preference for the inside view is couched in almost moral terms. Thus, the professional will say, “My client [or patient] is not a statistic; his case is unique.” Many disciplines implicitly teach their practitioners that the inside view is the only professional way to come to grips with the unique problems they will meet. The outside view is rejected as a crude analogy from instances that are only superficially similar.
Not to pay enormous prices for the skyrocketing earnings estimates of a company such as Google, on the cutting edge of Internet technology, many analysts would argue, is vastly unfair to current shareholders and potential buyers. Ironically, rapid technology change, accompanied by rapierlike earnings growth, makes the forecasting process even more difficult than it is for more mundane companies.
Forecasting Follies 12: Analysts’ Overconfidence
The forecasting pond is getting crowded. Let’s conclude with some recent examples in case you think that in the 2000s we turned a corner to a better understanding of the problems.
Large optimistic errors appear to be a way of life with corporate capital spending, particularly when new technologies or other projects where the firm is in an unfamiliar situation are involved. A Rand Corporation study some years back examined the cost of new types of plants in the energy field.32 The norm was that actual construction costs were double the initial estimates, and 80 percent of the projects failed to gain their projected market share.
A psychological study examining the cause of this type of failure concluded that most companies demanded a worst-case scenario for a capital spending project. “But the worst case forecasts are almost always too optimistic. When managers look at the downside, they generally describe a mildly pessimistic future rather than the worst possible future.”33
Overoptimism often results from differences in estimates made from the inside rather than the outside view. A clear-cut example is demonstrated by the behavior of large banks, investment bankers, the Federal Reserve, and the Treasury throughout the financial crisis almost until the end. As the financial system began to unravel in late 2006 and early 2007, dozens of reassuring statements were made by these large institutions that they would emerge with little damage. Those utterances were supported by Fed Chairman Bernanke, Treasury Secretary Paulson, and dozens of other senior officials.
On the outside, thousands of stories on the housing bubble and many of its excesses appeared in the media, by many astute observers, including Paul Krugman, a Nobel laureate in economics, and Gretchen Morgenson, a Pulitzer Prize winner with The New York Times, up to two years before the bubble popped. They were ignored or shrugged off by both the Fed and senior government officials.
Through 2007 and 2008, as conditions worsened, the banks and investment banks continued to be inordinately optimistic about the value of their toxic mortgage portfolios. They were substantially underreserving their losses to the spring of 2008, even though the subprime mortgage origination industry and many hedge funds had already collapsed several months before, with many dozens wiped out entirely. Only when banks and investment bankers realized their survival was at stake did reality break through, and enormous write-downs of these holdings took place. By then, of course, it was far too late.
Here is a good Psychological Guideline to train yourself to follow.
PSYCHOLOGICAL GUIDELINE 17: Be realistic about the downside of an investment; expect the worst case to be much more severe than you anticipated.
In this chapter, we have looked at the striking errors in analysts’ forecasts, errors so large that they render the majority of current investment methods inoperable. We have also seen that even though high error rates have been recognized for decades, neither analysts nor the investors who religiously depend on them have altered their methods in any way.
The problem is not unique to analysts or market forecasters. We have also seen how pervasive it is in many professions where information is hard to analyze, as well as how difficult the problem is to recognize, let alone change. Finally, we have found overoptimism to be a strong component of expert forecasts, both within and outside the stock market.
Now the question is: what can we do about it? The answer is that we need to move on to a better investing paradigm that does not rely primarily or exclusively on the efficient-market hypothesis or on the forecasts and estimates of company earnings from securities analysts. On the contrary, there is a better way.
Chapter 9
Nasty Surprises and Neuroeconomics
HISTORY IS REPLETE with unhee
ded alarm bells and dreadful surprises. Almost a hundred years ago, in 1915, the German embassy took out an ad in The New York Times. It warned U.S. citizens that they risked torpedo attacks on Allied passenger ships. The United States had not yet entered World War I. (In fact, President Woodrow Wilson would successfully run for reelection on the slogan “He kept us out of war.”) Yet when the British ocean liner Lusitania left New York bound for Liverpool, many American passengers were on board, along with 173 tons of rifle ammunition and shells for the British army. A German submarine attacked the ship off the south coast of Ireland, causing the loss of 1,198 passengers and crew, including 114 Americans. Although the ship was a legitimate target under international law, many Americans were outraged at the “surprise,” and public sentiment began to tilt against Germany. Two years later, the United States was in the war on the side of the British.
Twenty-six years later, the United States was surprised again. This time it was Pearl Harbor. The Japanese sneak attack on the military naval base in Hawaii catapulted the United States into World War II. Decades later, the evidence is fairly conclusive that in 1941 there were plenty of warning signs of what the Japanese navy intended. (Interestingly, it seems that many U.S. military leaders felt that the “inferior” Japanese people could never pull off such an attack.) That surprise also galvanized and unified a nation still reeling from the Great Depression and the widespread fears of civil strive, anarchy, and communism.
Knowing history, some twenty-three years later President Lyndon Johnson informed the American public of yet another complete surprise. To whip up support for a major escalation of the Vietnam War, in 1964 he greatly exaggerated the report of an attack on a U.S. destroyer in the Gulf of Tonkin. North Vietnamese torpedo boats had supposedly attacked the USS Maddox without provocation or cause. (Few knew at the time that the previous day the South Korean navy, under U.S. direction, had carried out clandestine raids on nearby North Vietnamese islands or that the destroyer had been on an intelligence mission.) Reacting to the announcement, Congress immediately handed the president the sweeping powers he requested and a blank check for the war.