Chapter 1

Chapter One

Behavioral Foundations

  1. Does the chapter present any information that would lead you to conclude that managers at Sun Microsystems were averse to a sure loss? Discuss.

ANSWER: The Behavioral Pitfalls box discussing the actions of Scott McNealy included the following passage.

On March 8, 2001, Cisco announced that because the downturn looked like it would last much longer than expected, it was going to lay off 18 percent of its workforce. Some of Sun’s executives wanted to follow suit. One stated: “When times are hard, you’ve got to shoot activities that aren’t making money.” However, McNealy refused to do so.

Refusing to shoot activities that are not making money is evidence of being averse to a sure loss.

  1. During its most profitable years in the late 1990s, Sun did not carry any debt. In 1999, Sun paid $138.6 million in taxes. Compare the situation at Sun with the situation at Merck in respect to their debt policies.

ANSWER: The situations are similar. During the late 1990s, both Sun and Merck were highly profitable firms whose market values were well in excess of their book values. The assets of both firms had large components that were intangible, residing in research and development. At the same time, both firms did have tangible assets, and could have held more long-term debt, thereby shielding some of their income from taxes. Both firms appear to have chosen to hold less debt than was optimal, thereby paying more corporate income tax than was necessary.

  1. Did Merck’s managers exhibit confirmation bias in its assessment of Vioxx? Discuss.

ANSWER: The Behavioral Pitfalls box discussing Merck includes the following passage:

Surprisingly, Merck’s post-approval study appeared to show that Vioxx actually caused heart attacks and strokes. However, the firm’s executives resisted that interpretation and invested heavily in promoting the drug.

The chapter text also contains the following passage:

Researchers at StanfordUniversity, HarvardUniversity, and the Cleveland Clinic wrote scientific articles that raised concerns about Vioxx’ safety. Merck challenged these concerns, and continued to promote Vioxx as safe. In February 2001, the FDA issued a letter to Merck’s CEO Ray Gilmartin, chastising the firm for deceptive promotional practices. In August 2004, a researcher from the FDA’s drug-safety office presented data that showed that higher doses of Vioxx correlated with a tripled risk of a heart attack or sudden cardiac death. Merck responded by issuing a press release reiterating its confidence in the safety and efficacy of its drug.

Both of the above passages suggest that executives at Merck underweighted information that disconfirmed their views about Vioxx being both safe and efficacious. That information came from Merck’s post-approval study, and the interpretation of the data from that study by prominent researchers.

  1. At the beginning of 2001, Merck’s CFO Judy Lewent predicted that Vioxx sales for the year would be between $3 and $3.5 billion. In June she qualified her prediction to say that although Vioxx sales would be closer to the lower end of the prediction range, four of their top five drugs would achieve the upper end of their prediction ranges. One month later, she stated that Merck’s research pipeline was as productive as any other time in the firm’s history. However, in August the Journal of the American Medical Association published a study linking Vioxx to an increased risk of heart attack and stroke. Immediately thereafter, sales of Vioxx began to slow. In 2001 Vioxx sales were $2.3 billion. In 2002 Vioxx sales were $2.5 billion. In light of the graph of return on equity displayed in exhibit 1-4, discuss whether the events just described reflect any behavioral biases.

ANSWER: If Judy Lewent exhibited behavioral bias in her sales predictions for Vioxx, that bias would be excessive optimism. The first indication of excessive optimism is her June 2001 qualifying statement that placed Vioxx sales at the lower end of her prediction range. Given Merck’s past glory, her statement that Merck’s pipeline was as strong as any other time is the firm’s history seems very optimistic. Exhibit 1-4 displays strong growth in ROE between 1993 and 2000, but a decline in mid-2001. There is no reason to predict that the reduction in Vioxx sales that took place after the August 2001 publication of the JAMA article was predictable before the fact. At the same time, given that Merck executives appeared to exhibit confirmation bias, it would have been reasonable to predict that at some point such an event would occur.

  1. Merck’s VIGOR study used 8000 subjects. Notably, Merck chose to include only subjects whose risk of experiencing a heart attack was low. Half the subjects in the study received Vioxx and the other half received naproxen. Of those receiving Vioxx, 20 had heart attacks. Of those receiving naproxen, 4 had heart attacks. Many medical researchers believe that naproxen does not reduce the incidence of heart attacks. Suppose that this is the case. From the vantage point of 1999, discuss whether the study outcome was just a fluke, with Vioxx actually being no riskier than naproxen when it comes to heart attacks.

ANSWER: If naproxen does not reduce the incidence of heart attacks, then in 1999 it might have been reasonable to expect that in the VIGOR study about 4 patients would have experienced a heart attack. (In fact, experts in the field suggest that on average about 3 out of 4000 patients in the VIGOR study would have experienced heart attacks, given that Merck had only included patients that were at low risk of a heart attack or stroke.)

Consider the use of a binomial probability model. Suppose that the probability that a patient in the VIGOR study would normally experience a heart attack is 4 out of 4000, or 0.001. If Vioxx did not increase the probability of a heart attack, with what probability would we expect to observe 20 patients taking Vioxx to experience heart attacks? The answer is

1 - the cumulative binomial probability associated with 19 heart attacks

given 4000 trials and a binomial probability of 0.001. The Excel function binomdist(19, 4000, 0.001, 1) gives the cumulative probability associated with 19 heart attacks. That number turns out to be 99.999999 percent. Therefore, the probability of observing that 20 patients who took Vioxx experienced heart attacks is 0.000001 percent.

In other words, the probability that the VIGOR result was a fluke is 0.000001 percent, one in 100 million.

  1. On November 2, 2004The Wall Street Journal published an article describing the manner in which Merck’s executives made decisions about Vioxx. The Wall Street Journal article pointed out that the November 2000 issue of The New England Journal of Medicine included an article describing the results of the VIGOR study. The New England Journal of Medicine article stated that Vioxx did not significantly increase the incidence of heart attacks among patients who did not appear to be at high risk of having a heart attack. Notably, Merck had excluded subjects whose risk of experiencing a heart attack was anything but low. In retrospect, the executive editor of TheNew England Journal, Gregory Curfman, told The Wall Street Journal that his journal did not have all the details that were available to the FDA. He stated that his journal concentrated its efforts on ensuring that the text of the article accurately represented the data presented in the article. The authors of the article were academics who received consulting contracts or research grants from Merck and employees of Merck. Can you detect any behavioral issues and agency conflict issues in the above discussion?

ANSWER: If the editors’ judgment reflected a behavioral bias, it would be bias stemming from availability. In this respect, the editors did not raise questions about the incidence of heart attacks in patients who took Vioxx but were not at low risk of a heart attack.

As for agency conflicts, the authors of the article had either received consulting contracts from Merck or were employees of Merck. That does not imply that they must have had a conflict of interest. However, the potential for such a conflict was definitely present. Remember that the VIGOR study took place well before November 2000, and featured 20 patients who were at low risk of having a heart attack taking Vioxx and subsequently experiencing a heart attack. Yet the authors of the article concluded that Vioxx did not increase the incidence of heart attacks among patients who did not appear to be at high risk of having a heart attack.

Case AnalysisQuestions

  1. Identify which psychological phenomena described in the chapter played a role in the Galveston disaster of 1900.

ANSWER: The key psychological phenomena involved in the minicase about Galveston are availability, excessive optimism, and overconfidence.

Availability: Prior to September 8, 1900 Isaac Cline had had no personal experience with hurricanes. Therefore his judgments prior to that date were heavily influenced by impressions formed by information that had been available to him. Cline himself says as much in his autobiography, when he states: “This being my first experience in a tropical cyclone I did not foresee the magnitude of the damage which it would do.”

Excessive optimism and overconfidence. These traits were evident in Isaac Cline’s 1891 opinion about there being no need for Galveston to erect a sea wall. The language is unequivocal: “It would be impossible for any cyclone to create a storm wave which could materially injure the city.” Notice that Cline had been successful in forecasting freezes and floods, events that would reinforce the tendency to be overconfident. Also excessively optimistic and overconfident were the crowds who flocked to Galveston on September 8, 1900 in order to watch the high waves, and ignored Cline’s personal warnings that day. In ignoring Cline’s warning, they succumbed to confirmation bias.

  1. Compare the psychological traits and experiences of Isaac Cline with those of Scott McNealy, chief executive officer of Sun Microsystems.

ANSWER: Both Isaac Cline and Scott McNealy achieved early success in the face of opposition. Cline had achieved success in accurately forecasting freezing weather warnings to local farmers twenty-four to thirty-six hours in advance. This occurred despite the chief of the Weather bureau having judged that such a feat was impossible. McNealy had achieved success in the 1980s by deciding to substitute Sun’s own microprocessors for those manufactured by Motorola. The decision turned out well for Sun, despite his executives having advised against it.

Those early successes might have encouraged overconfidence in both Cline and McNealy, not to mention excessive optimism. Cline was certain that a cyclone could not cause severe damage to Galveston and recommended against building a sea wall. McNealy was confident that the recession of 2001 would be short-lived, and resisted cost cutting at Sun.

  1. Use the concepts and examples developed in the chapter text to analyze the behavioral issues that arose in respect to hurricane Katrina that struck New Orleans in 2005. Do you see any common behavioral patterns in the situations that prevailed in Galveston and New Orleans before they were devastated?

ANSWER: Like Isaac Cline, Homeland Security Secretary Michael Chertoff was surprised by the strength of the hurricane in question. Surprise is the telltale sign suggesting overconfidence.

Availability appears to have played a significant role in both situations. Cline had never seen a tropical storm as strong as the one that struck Galveston, and so relied on events that were mentally available. As to federal government officials, Walter Maestri’s comments about psychology in 2002 proved to be prescient. Just as Scott McNealy is described as being preoccupied with the Microsoft suit in the chapter text, federal government officials appear to have been preoccupied with terrorist threats, despite the ample evidence about the threat of a category 4 or 5 hurricane striking New Orleans.

As to debiasing, proponents of behavioral finance do not contend that people fail to learn, only that they learn slowly and sometimes painfully. Over the course of a century, people did learn about how to prepare for hurricanes. In the case of Galveston, most residents did not evacuate, although to be fair, communication was slower and transportation more difficult in 1900 than in 2005. Nevertheless, keep in mind that people rushed to Galveston to view the hurricane. In contrast, one of the great successes of New Orleans’ experience with Katrina is the low death toll relative to projections, mostly because large scale evacuation of the city was part of a plan.

At the same time, availability was a root cause of that preparation being suboptimal. In the late 1800s, availability bias was a major factor in Galveston’s failure to invest in a sea wall. In the late 20th and early 21st centuries, availability bias was a major factor in government officials’ failure to invest in the fortification of New Orleans’ levee system, the breaching of which caused more damage than the hurricane itself.

McGraw-Hill/Irwin ©20071-1