This is how security analysts react to earnings announcements: They do not revise their earnings estimates enough to reflect the new information. Consequently, positive earnings surprises tend to be followed by more positive earnings surprises, and negative surprises by more negative surprises. Of Course, the unexpected surprises in store for analysts are also a manifestation of overconfidence because overly narrow confidence bands mean people get surprised more frequently than they anticipate.
Aversion to Ambiguity
Imagine that I offered you the choice between accepting a sure $1,000 or an even gamble in which you either win $0 or $2000. When I pose this question in MBA classes, about 40 percent of the students say they would take the gamble.
I describe this choice to students by telling them that there is a bag containing 100 poker chips, 50 black chips and 50 red chips; they can choose a sure $1,000, or a lottery ticket that pays $2,000 if a black chip is drawn at random from the bag but $0 if a red chip is drawn.
Now consider this variation. Imagine the bag contains 100 colored chips that are either red or black, but the proportions are unknown. Many people who are willing to gamble when the odds are even prefer to play it safe and take the sure $1,000 when the odds are unknown. This phenomenon is known as the aversion to ambiguity. People prefer the familiar to the unfamiliar.
Remember the Wall Street proverb about greed and fear? I note that the emotional aspect of aversion to ambiguity is fear of the unknown. The case of Long-Term Capital Management, discussed in chapter 1, provides an apt example of this phenomenon. Recall that on September 23, 1998, a $3.6 billion private rescue of LTCM was arranged. The Federal Reserve Bank of New York orchestrated this plan because of a concern that the failure of LTCM might cause a collapse in the global financial system. The November 16, 1998, issue of the Wall Street Journal describes the scene as the participants departed the meeting at which the deal was struck. The article attributes an interesting remark to Herbert Allison, then president of Merrill Lynch, a remark that typifies aversion to ambiguity as fear of the unknown. “As they filed out, they were left to ponder whether all this was necessary, and whether a collapse would really have jolted the global financial system. ‘It was a very large unknown,’ Merrill’s Mr. Allison says. ‘It wasn’t worth a jump into the abyss to find out how deep it was.’”6
Emotion and Cognition
The issues discussed in this chapter involve cognitive errors, that is, errors that stem from the way that people think. But in describing ambiguity to aversion in terms of fear of the unknown, I suggest that some phenomena involve a combination of cognition and emotion. Of course, both involve mental processes, and may be physiologically linked, as opposed to being separate from each other. Scholars have produced ample evidence that emotion plays an important role in the way people remember events. So, phenomena involving the availability heuristic may reflect both cognitive and emotional elements. Here is an example.
In 1972, the Dow closed at 1020. In 1982 it closed at 1047, just 27 points higher than the value achieved a decade earlier. In between, it gyrated wildly, recording four years of negative growth. During this period, inflation reduced the purchasing power of a dollar by over 66 percent. A 1995 article in the Wall Street Journal quotes Russell Fuller, president of RJF Asset Management (now Fuller & Thaler Asset Management) in San Mateo, California, as follows: “‘People like myself, who have been in the business since before the 1973–74 crash, we were terrified by that crash,’ says Mr. Fuller, the money manager. ‘That’s a very low probability event. But many of the people in this business have spent the last 20 years worrying about that happening again.’”7
Summary
This chapter described the first theme of behavioral finance, heuristic-driven bias, and introduced some of the main heuristics upon which financial practitioners rely. Throughout the book, readers will encounter many instances of representativeness, anchoring-and-adjustment, overconfidence, availability bias, and aversion to ambiguity. These heuristics surface in many different contexts, such as analysts’ earnings forecasts, investors’ evaluation of mutual fund performance, corporate takeover decisions, and the types of portfolios selected by both individual and institutional investors. Because of their reliance on heuristics, practitioners hold biased beliefs that render them vulnerable to committing errors. In addition to the heuristics described in this chapter, readers will come across a host of others, such as excessive optimism, the illusion of validity, hindsight bias, the illusion of control, and self-attribution error. There are many examples of such errors in this book.
Chapter 3 Frame Dependence: The Second Theme
This chapter discusses the following:
• loss aversion, loss realization, and losing projects
• mental accounting, frame dependence, and facing risk
• hedonic editing and tolerance for risk
• self-control and dividends
• regret and pension fund allocation
• money illusion and inflation
Frame independence lies at the heart of the Modigliani-Miller approach to corporate finance. Merton Miller has a succinct description of frame independence. When asked to explain, in twenty-five words or less, the essence of his contributions with Franco Modigliani, he said: “If you transfer a dollar from your right pocket to your left pocket, you are no wealthier. Franco and I proved that rigorously.”1
It is a matter of form whether a person keeps a dollar of wealth in the right pocket or in the left pocket. The form used to describe a decision problem is called its frame. When I speak of frame independence, I mean that form is irrelevant to behavior. Proponents of traditional finance assume that framing is transparent. This means that practitioners can see through all the different ways cash flows might be described. Yet many frames are not transparent but rather are opaque. When a person has difficulty seeing through an opaque frame, his decisions typically depend on the particular frame he uses. Consequently, a difference in form is also a difference in substance. Behavior reflects frame dependence.
Loss Aversion
In their landmark work on prospect theory, a descriptive framework for the way people make choices in the face of risk and uncertainty, Daniel Kahneman and Amos Tversky (1979) provide evidence of frame dependence. The starting point in their work is the role of “loss,” an issue explored by Harry Markowitz (1952b). Kahneman and Tversky studied how people respond to the prospect of a loss. Here is one of their examples. Suppose you face a choice between (1) accepting a sure loss of $7,500, or (2) taking a chance where there is a 75 percent chance you will lose $10,000 and a 25 percent chance you will lose nothing. The expected loss in both choices is $7,500. Would you choose to take the guaranteed loss or take a chance? Most people opt for the latter. Why? Because they hate to lose! And the uncertain choice holds out the hope they won’t have to lose. Kahneman and Tversky call this phenomenon loss aversion. They find that a loss has about two and a half times the impact of a gain of the same magnitude.2
It is not difficult to find real-world illustrations of loss aversion. In a manual for stockbrokers, Leroy Gross (1982) describes the difficulties investors face in coming to terms with losses.
Many clients, however, will not sell anything at a loss. They don’t want to give up the hope of making money on a particular investment, or perhaps they want to get even before they get out. The “get-evenitis” disease has probably wrought more destruction on investment portfolios than anything else. … Investors who accept losses can no longer prattle to their loved ones, “Honey, it’s only a paper loss. Just wait. It will come back.” (p. 150)
Some people learn about “get-evenitis” the hard way. Take the case of Nicholas Leeson. In 1995, Leeson became famous for having caused the collapse of his employer, 232-year-old Barings PLC. How? He lost over $1.4 billion through trading. In 1992, Leeson began to engage in rogue trading in order to hide errors made by his subordinates. Eventually, he incurred losses of his own, and “get-
evenitis” set in. He asserts that he “gambled on the stock market to reverse his mistakes and save the bank.”3
“Get-evenitis” also afflicts corporate executives’ ability to terminate losing projects. For example, 3Com’s popular Palm Computing products, the handheld organizers that access data with a stylus, had a predecessor—Apple Computer’s more sophisticated Newton.4 Apple CEO John Sculley was thoroughly committed to the Newton, and made it the center of his personal vision for the computer industry. He coined the term “personal digital assistant” to describe the concept and argued that it would be a pivotal step in the convergence of three industries: computing, communications, and entertainment.5
Development of the Newton began in 1987, and the product was launched in 1993. But at $1000, it was much too expensive for the mass market. Moreover, because of initial failures in its handwriting recognition capability, cartoonist Gary Trudeau lampooned the Newton in his comic strip Doonesbury. Given the size and demographics of Gary Trudeau’s readership, think about the impact the availability heuristic had on Newton’s potential market.
By January 1994, it was apparent that sales were disappointing and the Newton was a losing project. But Apple did not terminate it. The company was committed to personal digital assistants. A year later, in January 1995, Apple had added enhanced features, and the year after that it came out with a backlit screen, but to no avail. In March 1997 Apple spun the Newton off into its own division, but this did little good, and six months later Apple folded the division back into its own organization. Through all this, the Newton remained a loser.
CEOs may come and go, but losing projects stay on. John Sculley “went”; he was replaced by Gil Amelio, who also came and went. In a dramatic comeback, Steve Jobs, Apple’s cofounder and first CEO, replaced Amelio. Years before, Sculley had ousted Jobs. In January 1998, about ten years after its inception, CEO Jobs announced his decision to terminate the Newton project.
Concurrent Decisions
Here is another Kahneman-Tversky decision problem:
Imagine that you face the following pair of concurrent decisions. First examine both sets of choices, then indicate the option you prefer for each.
First decision: Choose
A. a sure gain of $2,400, or
B. a 25 percent chance to gain $10,000 and a 75 percent chance to gain nothing.
Second decision: Choose
C. a sure loss of $7,500, or
D. a 75 percent chance to lose $10,000 and a 25 percent chance to lose nothing.
The way that people respond to this problem tells us a lot about their approach to making decisions. Consider your own responses. Choosing A in the first decision would be the risk-averse choice. Most people find a sure $2,400 difficult to pass up. Although $10,000 is a lot more than $2,400, the odds of collecting it are only one in four. Hence, the expected value of B is $2,500, considerably less than $10,000. In fact, $2,500 is just a tad more than the guaranteed $2,400 offered in A.
Did you recognize the second decision? We encountered it before, in the previous section. Did you respond the same way as before? In my own experience, about 90 percent choose D in the second decision problem. They want the chance to get even.
The two decision problems together constitute a concurrent “package.” But most people do not to see the package. They separate the choices into mental accounts. And that brings us to frame dependence.
Suppose you face a choice. You can take a 75 percent chance you will lose $7,600 and a 25 percent chance you will win $2,400. Or you can take that same chance and accept an additional $100. Which choice would you make? A no-brainer, right? It should be: This decision frame is transparent.
But sometimes the frame is opaque. Consider the decision problem at the beginning of this section. When I administer this problem to my MBA students, about half choose A &D: A in the first decision problem and D in the second. People who choose A &D end up facing a 25 percent chance of winning $2,400 and a 75 percent chance of losing $7,600. However, they could do better: They could choose the B &C combination, which would offer them a 25 percent chance of winning $2,500 and a 75 percent chance of losing $7,500. But most people don’t see through the opaque frame. Therefore, they act as if they don’t value $100. The opaque frame makes for a “brainer” instead of a no-brainer.
Hedonic Editing
In his stockbroker manual, Gross (1982) implicitly raises the issue of frame dependence within the context of realizing a loss. His essential point is that investors prefer some frames to others, a principle known as hedonic editing. Consider Gross’s advice to stockbrokers:
When you suggest that the client close at a loss a transaction that you originally recommended and invest the proceeds in another position you are currently recommending, a real act of faith has to take place. That act of faith can more easily be effected if you make use of some transitional words that I call “magic selling words.” The words that I consider to have magical power in the sense that they make for a more easy acceptance of the loss are these: “Transfer your assets.” (p. 150)
Why are “transfer your assets” magic selling words? Because they induce the client to use a frame in which he or she reallocates assets from one mental account to another, rather than closing a mental account at a loss.
Thaler and Eric Johnson (1991) propose a theory of hedonic editing for mental accounts. As part of a study, they administered a series of choice problems to subjects. You will find two of these problems below. Read the first problem, record your answer, and then move on to the next problem.
1. Imagine that you face the following choice. You can accept a guaranteed $1,500 or play a stylized lottery. The outcome of the stylized lottery is determined by the toss of a fair coin. If heads comes up, you win $1,950. If tails comes up, you win $1,050. Would you choose to participate in the lottery? Yes or no? Yes means you take your chances with the coin toss. No means you accept the guaranteed $1,500.6
2. Imagine that you face the following choice. You can accept a guaranteed loss of $750 or play a stylized lottery. The outcome of the stylized lottery is determined by the toss of a fair coin. If heads comes up, you lose $525. If tails comes up, you lose $975. Would you accept the guaranteed loss? Yes or no? Yes means you accept a $750 loss. No means you take your chance with the coin toss.
Let’s consider how people usually respond to these questions. In the first choice problem, the majority prefer to take the guaranteed $1,500 over the gamble where they might get less. This could be viewed as a typical risk-averse response, because the average payoff to the lottery ticket is $1,500, the same amount involved in the riskless option. However in the second choice problem, many people choose the lottery over the guaranteed loss. This is decidedly a risk-seeking response, in that the expected payoff to the coin toss is a $750 loss, the same amount involved in the riskless option.
There is a lesson here: People are not uniform in their tolerance for risk. It depends on the situation. Some appear to tolerate risk more readily when they face the prospect of a loss than when they do not.
It is common for financial planners and investment advisers to administer risk tolerance quizzes in order to determine a degree of risk that is suitable for their clients. However behavioral finance stresses that tolerance for risk is not unidimensional. Rather it depends on several factors, one being recent experience facing risk. Here are two more examples developed by Thaler and Johnson that bring out the complexity of these issues.
3. Imagine that you have just won $1,500 in one stylized lottery, and have the opportunity to participate in a second stylized lottery. The outcome of the second lottery is determined by the toss of a fair coin. If heads comes up, you win $450 in the second lottery. If tails comes up, you lose $450. Would you choose to participate in the second lottery after having won the first? Yes or no?
4. Imagine that you have just lost $750 in one stylized lottery, but have the opportunity to participate in a second stylized lottery. The outcome of the second l
ottery is determined by the toss of a fair coin. If heads comes up, you win $225 in the second lottery. If tails comes up, you lose $225. Would you choose to participate in the second lottery after having lost in the first? Yes or no?
Now that you have recorded your yes or no answers, compare your response to choice 3 with your response to choice 1. From a dollar perspective, choices 1 and 3 are equivalent. In the framework of traditional finance, people should respond the same to both. Yet in practice, many “switch” their choices. When replicating the Thaler-Johnson study I have found that about 25 percent of the respondents are more willing to take the gamble in choice problem 3 than they are in the dollar-equivalent choice problem 1. Why?
Thaler and Johnson suggest that the answer involves hedonic editing, the way people organize their mental accounts. In choice problem 3, if people lose $450 they combine it with the $1,500 gain and experience the net position of $1,050—exactly the situation they are presented with in choice problem 1. But if they win, they do not net their two gains; instead, they savor them separately. According to Thaler and Johnson, the added attraction of experiencing gains separately inclines people to be more willing to gamble.
Thaler and Johnson found that in choice problem 2, over 75 percent chose to gamble rather than accept a sure $750 loss.7 However, although example 4 is dollar-equivalent to choice problem 2, almost 50 percent switch their choice from taking a gamble in example 2 to playing it safe in choice problem 4. Thaler and Johnson suggest an explanation based on the way people experience losses. They note that people seem incapable of netting out moderately sized losses of similar magnitudes. So, a loss of $225 coming on top of a prior loss of $750 is especially painful. The added pain leads people to shy away from taking the gamble as framed in choice problem 4, relative to the frame in choice problem 2.
Beyond Greed and Fear Page 7