Inflation
In chapter 3 I discussed money illusion, the difficulties presented by inflation. It turns out that inflation also impacts the prediction errors of Wall Street strategists. For Goldman Sachs’s Abby Joseph Cohen, the biggest surprise for 1997 was the change in expectations about inflation and the subsequent impact on equity prices. In June 1997 she stated: “If inflation is low, there is a willingness to pay a higher price/earnings multiple for whatever earnings are being generated. More importantly, there is great confidence in the durability of the economic cycle and the profit expansion.”30
Inflation exerts an important effect on equity valuation. Why? There are two reasons. The first is money illusion; the second is that due to anchoring-and-adjustment, people underreact to changes in inflation.
Abby Cohen is correct that price/earnings ratios are higher in periods of low inflation. But this may be the result of a behavioral bias. During the 1970s, inflation was high, nominal interest rates were high, and stock prices were low. At the time, Franco Modigliani and Richard Cohn (1979) argued that financial analysts were undervaluing equities by not taking proper account of inflation in their valuation formulas. Modigliani and Cohn suggested that analysts were mismatching real and nominal variables, discounting real earnings at nominal rates. Of course, that leaves us with the question of whether the reverse phenomenon, overvaluation, occurs in periods of falling inflation. In analyzing forecasts that extend back to 1952, De Bondt (1991) finds that it does.
Robert Shiller (1995) argues that inflation is very salient for most people, but that they significantly overweight the importance of inflation. Why? Because inflation is much more salient when it comes to the prices people pay than the incomes they receive.
Summary
Wall Street strategists are prone to committing a variety of behavioral errors and biases: gambler’s fallacy, overconfidence, and anchoring.
Gambler’s fallacy stems from two sorts of confusion. First, people have very poor intuition about the behavior of random events. With gambler’s fallacy, they expect reversals to occur more frequently than actually happens. The second source of confusion stems from the reliance on representativeness. People tend to base their predictions and probability judgments on how representative an event is. Alas, predictions based on representativeness exhibit too much volatility.
Sentiment reflects the aggregate errors and biases in the market. As such, the presence of sentiment adds to the difficulty of accurately predicting the market. But most people are overconfident about their ability to complete difficult tasks successfully. Therefore, they are surprised more frequently than they anticipate. Moreover, overconfidence seems to increase with the difficulty of the task. In fact, people who regard themselves as experts tend to be the worst offenders.
Some experts use fundamental analysis and others use technical analysis. One reason why fundamental analysts tend to be surprised is that they underweight the impact of sentiment. On the other hand, technical analysts are often surprised because their reliance on representativeness leads them to make excessively volatile predictions. But fundamental analysts and technical analysts do have one major characteristic in common: They both are slow to learn.
In predicting the future, people tend to get anchored by salient past events. Consequently, they underreact. Finally, most people—including Wall Street strategists—have difficulty taking inflation into proper account.
Chapter 6 Sentimental Journey: The Illusion of Validity
Many investors believe that there is a negative relationship between the predictions made by newsletter writers and the subsequent moves in the market. Is the relationship they perceive truly valid, or is it nothing more than an illusion? Are these investors prone to the “illusion of validity?”
This chapter discusses the following:
• why people mistakenly think that it’s possible to make money by betting against the market predictions contained in advisory newsletters
• confirmation bias leading to overconfidence: why investors are prone to the illusion of validity
• how the predictions of newsletter writers have historically responded to changes in market conditions
The Logic of Going Against Sentiment
What is the supposed logic that leads some investors to bet against sentiment, or at least against the prevailing views of advisory newsletter writers?1 To gain some insight into this question, consider the behavior of the market during the first quarter of 1998.
In those three months the Dow Jones Industrial Average rose over 1000 points to surpass 9000 for the first time. This remarkable increase, of almost 14 percent, was even more astonishing because it followed three consecutive years of extraordinary increases in the market. So in April 1998, the Wall Street Journal announced the retreat of bearish sentiment. The headline read: “The Final Bears May Be Giving Up: Bulls’
Victory Prompts Fears.” Fears of what? Greg Ip, the author of the article writes:
Wall Street strategists, individual investors and investment advisers are all registering their greatest optimism on stocks in years.
And that could spell trouble.
Traditionally, market sentiment is seen as a contrarian indicator. Markets rise, the theory goes, as bears become bulls and put money into the market. The market peaks when there are no bears left and everyone is invested.2
So, there we have it, in that last paragraph: The logic behind the view. If there are no bears left, the market must top out. And how do we know how many bears there are relative to bulls? Because Chartcraft, Inc., tracks them. Chartcraft compiles the forecasts of stock market newsletters, and reports the summary data in its publication Investors Intelligence. Investors Intelligence has been publishing this data since 1963. Chartcraft classifies newsletter writers into three distinct camps: bullish, bearish, and those who are optimistic for the long term but expect a near-term correction. As we shall see below, some refer to the correction camp as “chickens.”
The relative size of the three groups provides a sense of the sentiment of newsletter writers. To measure the degree of optimistic sentiment in the market, Investors Intelligence computes an index called the Bullish Sentiment Index. To define the index, let the number of Bulls denote the number of bullish newsletter writers and the number of Bears denote the number of bearish newsletter writers. Now define the Bullish Sentiment Index as
As Ip mentioned, practitioners of technical analysis treat the Bullish Sentiment Index as a contrarian indicator. Investors Intelligence explains the rationale underlying this view as follows: “Since most advisory services are trend followers, they are most bearish at market bottoms, and least bearish at market tops.”3
Ip writes: “the bearish portion of advisory newsletters has fallen below 25% in the past two weeks, hitting a six-year low.”4 That’s why technicians who believe that the Bullish Sentiment Index is a contrarian indicator, sense there may be trouble ahead.
The Evidence Concerning the Bullish Sentiment Index
Greg Ip is no stranger to the sentiment index. A year earlier, on May 12, 1997, he described the sentiment of newsletter writers and portfolio managers when the market quickly dropped in March and April and then recovered just as rapidly.5 Ip noted that at the time of the market’s March peak, bulls outnumbered bears 51 percent to 30 percent. A month later, when the market bottomed out, bears outnumbered bulls 41 percent to 35 percent. As the saying goes, the market climbed a “wall of worry.” These events provide confirming evidence for betting against the Bullish Sentiment Index. But was this case typical?
To answer this question, consider figure 6-1. I have adapted this figure from two papers, one by Michael Solt and Meir Statman (1988), and the other by Roger Clarke and Statman (1998). The figure plots subsequent percentage annual changes in the Dow against the current value of the sentiment index. If the case described in Ip’s article is typical, then we ought to find that the data in the graph are clustered as follows: Low values of the Bulli
sh Sentiment Index, at the left-hand side of the graph, give rise mostly to subsequent increases in the Dow; and high values of sentiment, at the right, give rise mostly to decreases.
But what does figure 6-1 show? It shows that low bullish sentiment is followed about as often by increases as by decreases. The same is true for high bullish sentiment. Bullish sentiment does not serve to signal
Figure 6-1 Percentage Change in Dow Jones Industrial Average versus Percentage of Bulls
Annual percentage change in Dow Jones Industrial Average subsequent to reading of Bullish Sentiment Index. Statistical analysis indicates the absence of a significant relationship. anything at all about future market performance. The Bullish Sentiment Index is about as likely to predict how the market will move in the future as a coin toss!6
Denial
The evidence presented by Solt and Statman (1988) is now over ten years old. When this evidence first appeared, John Dorfman surveyed the reaction for the Wall Street Journal. And what was that reaction? In a word, denial.
Michael Burke, editor of Investors Intelligence, says the indicator caused the newsletter to be “100% bullish” in July 1982, a month before the beginning of the five-year bull market that saw stock prices triple. In 1987, he says, it advised subscribers to get out of stocks in August—near the market’s peak and two months before the crash. “We were up 40% for the year,” Mr. Burke says.
Profs. Solt and Statman argue, however, that such successes don’t prove the rule. “The index is useless,” they write, “not because it does not provide some good forecasts, but because it also provides so many bad forecasts.” Mr. Burke of Investors Intelligence says that while he has no specific criticisms of the Solt-Statman methodology, he feels the professors “arbitrarily used” the data. A different way of looking at the data could lead to a different conclusion, he contends.7
Louis Rukeyser’s Sentimental Journey
Why do investors continue to believe that the sentiment index is useful despite strong evidence to the contrary? They do so because they focus on the evidence that confirms their views but overlook evidence that is contradictory. Investors search only for confirming evidence; and they ignore disconfirming evidence.
Here is an entertaining example that illustrates the bias, taken from the popular television program Wall $treet Week with Louis Rukeyser. On February 6, 1998, as the Dow was rising rapidly, Louis Rukeyser took a few moments to trace the relationship between the sentiment of newsletter writers and the subsequent performance of the market. He referred to this retrospective as a “sentimental journey.” Here is the beginning of Rukeyser’s portrayal of that journey:
Let’s take one of our periodic looks at why every self-respecting market technician treats the sentiments of his colleagues with contempt, as we track the embarrassing records of market advisers. So come along as we are “Gonna Take a Sentimental Journey.”
In 1963, the very first year Investors Intelligence conducted its poll of market newsletter writers, bulls were at an all-time low of less than 9 percent; with nearly 78 percent chickens looking for a correction; and almost 14 percent outright bears.
You betcha. The Dow proceeded to rise 250 points in twenty-one months. Ten years later, nearly 62 percent of those polled thought the market would head even higher. And what came to pass? You guessed it. Down 470 points in 23 months.
Figure 6-2 provides a graphical representation of Rukeyser’s entire retrospective. One line represents the percentage of bulls, and the other line represents the point rise in the Dow Jones Industrial Average. The data is arranged so that the percentage of bulls and the subsequent point gain are vertically aligned.
Notice that figure 6-2 consists only of evidence that confirms the usefulness of the Bullish Sentiment Index. Graphically, the events traced out by Rukeyser show that whenever the curve representing the percentage of bulls declines, the curve representing the point move in the Dow goes up.
Louis Rukeyser ends his journey with a very important question: “[Y]ou can’t win them all, but can’t these guys win any of them?”
Figure 6-2Louis Rukeyser’s Sentimental Journey
Along Louis Rukeyser’s sentimental journey, the Dow Jones Industrial Average increases when newsletter writers are bearish and decreases when they are bullish.
Rukeyser doesn’t tell us. But he could have selected different events and told a very different story, such as the one illustrated in figure 6-3. This figure, which depicts percentage changes rather than points, shows that newsletter writers can get it right at least some of the time.
So What’s the Big Deal?
So what if newsletter writers are unable to predict where the market is headed? Is that a problem? Well, it means that investors who rely on the advice in these newsletters to time the market will be trading too frequently, generating trading commissions but not helping themselves. A study by Brad Barber and Terrance Odean (1998a) finds that individual investors who trade the most frequently have the poorest-performing portfolios.
Analogous remarks apply to the stock selection advice in newsletters. Economist Andrew Metrick (1999) has analyzed stock recommendations of newsletter writers. The May 12, 1998, issue of the Wall Street Journal quotes Metrick as saying: “After trading costs, on average they would have lagged behind the market. There doesn’t seem to be a lot of evidence that the stock selection of these newsletters is any good.”8
Confirmation Bias: The Illusion of Validity
Psychologists Hillel Einhorn and Robin Hogarth (1978) have studied the general issue of why people persist in beliefs that are invalid, that is, why they succumb to the illusion of validity. Einhorn and Hogarth suggest that people do so because they are prone to search for confirming evidence, not disconfirming evidence. Consequently they not only may come to hold views that are fallacious, but they may be overconfident as well.
In his sentimental journey, Louis Rukeyser effectively makes the statement, If advisors are bearish, the market is more likely to go up than go down. Notice that this statement has the form “If P, then Q.” Most people experience great difficulty assessing the validity of statements that have this form, which is why they accept false statements as true.9 The root cause is that people use a heuristic10 that leads them to search for confirming evidence (where P and Q hold) instead of the logically correct search for disconfirming evidence (where P and not-Q hold).11
Einhorn and Hogarth provide a general framework for their discussion of the illusion of validity. Here is a description of that framework, applied to the contrarian hypothesis for the Bullish Sentiment Index. Assume four types of situations. (1) periods when the Bullish Sentiment Index was high, (2) periods when the Bullish Sentiment Index was low, (3) periods when the market rose, and (4) periods when the market fell. Match these situations according to the arrangement in table 6-1.
Table 6-1 depicts four types of event combinations associated with the claim that the Bullish Sentiment Index is a contrarian indicator. Notice that the two diagonal cells are labeled as “hits.” A hit occurs when an event takes place that validates the contrarian theory. Either sentiment is bearish and the market subsequently rises (a positive hit), or sentiment is bullish and the market subsequently falls (a negative hit). Notice that all of the event combinations depicted in Louis Rukeyser’s sentimental journey are hits.
The off-diagonal entries relate to events that go against the Bullish Sentiment Index’s being a contrarian indicator. All of the events described in figure 6-3, the journey Louis Rukeyser did not take, fall into the off-diagonal cells.
Suppose that we focus our attention on the full history of the Bullish Sentiment Index and allocate events into the four cells in table 6-1. If most of the event combinations fell into the diagonal cells, we would conclude that the contrarian view of the Bullish Sentiment Index is valid. If most of the events fell into the off-diagonal cells, we would conclude that investment advisers appear to have impressive forecasting ability. If the events fall evenly across diagona
l and off-diagonal cells, then we would conclude that the Bullish Sentiment Index is useless as a tool to forecast the market.
So where do most of the events fall? As Solt and Statman (1988) showed, event combinations are evenly distributed across the diagonal and off-diagonal cells in table 6-1. This is why the sentiment index is useless as a forecasting tool. Recall that Michael Burke, editor of Investors Intelligence, is alleged to have said that Solt and Statman used the data arbitrarily. But to avoid succumbing to the illusion of validity, they followed the general procedure prescribed by Einhorn and Hogarth.
Table 6-1
Figure 6-3 Sentiment Index versus Percentage Change in Dow Jones Industrial Average
Along the alternative journey, the Dow Jones Industrial Average increases after newsletter writers have become more bullish and decreases after they have become more bearish. For each observation, the interval associated with the change in the Dow is 12 months.
Naive Extrapolation and Nervous Bullishness
The sentimental journey suggests that advisers are most bullish at market tops and most bearish at market bottoms. However, if you take a careful look at figure 6-1, you will see that the index is no more effective at the extreme right (bullishness) or extreme left (bearishness) than it is in the middle of the range.12
By now you may be wondering about the existence of a valid connection between sentiment and the market. Well, a connection does exist. Bullishness increases after the market has gone up; bearishness increases after the market has gone down. For the period 1972 through 1997, a 10 percent increase in the Dow has on average been followed by an 8.3 percent increase in the proportion of bulls. Even Michael Burke recognizes this. He is quoted in the April 13, 1998, issue of the Wall Street Journal as having said, “It’s hard to stay bearish when the market goes up all the time.”13
Beyond Greed and Fear Page 11