SENSATION SEEKING AND FIRM RISK Evidence from CEO Pilots *

MATTHEW D. CAIN Mendoza College of Business University of Notre Dame Notre Dame, IN 46556 [email protected]

STEPHEN B. MCKEON Krannert School of Management Purdue University West Lafayette, IN 47907 [email protected]

This version: July 30, 2010

                                                             *

Preliminary draft; comments welcome.

Sensation Seeking and Firm Risk: Evidence from CEO Pilots

Abstract This study analyzes the relation between the sensation seeking psychological attribute of CEOs and firm risk. We proxy for sensation seeking behavior using one of the predominant survey components in Zuckerman’s Sensation Seeking Scale: the desire to fly an airplane. Using data from the FAA registry of certified pilots, we find that CEO pilots are associated with greater stock return volatility at their firms and are more likely to engage in merger and acquisition (M&A) activity. Age has been shown to be inversely correlated with sensation seeking behavior, and we find that CEOs are less likely to pursue M&A transactions as they age, producing a lower level of return volatility. The sensation seeking behavior of pilot CEOs results in higher quality acquisitions at firms with few measurable growth opportunities, implying that sensation seekers may be desirable candidates among certain firms.

1   

1.

Introduction

“Sensation seeking is related to risk taking in all kinds of risk areas. In fact, the sensation seeking trait may be the common factor that accounts for the relationships among different kinds of risk taking.” – Psychologist Marvin Zuckerman (2007, p.65) Over twenty five years ago, Hambrick and Mason (1984) concluded that managerial characteristics have an impact on organizational outcomes. Pinning down exactly which characteristics lead to certain outcomes is the subject of considerable research effort across several business disciplines. Over the past decade the finance literature has taken an interest in this line of inquiry as it relates to economic outcomes. Our study contributes to this line of literature. Specifically, our study extends the behavioral corporate finance literature, in which firm policies are influenced by managers possessing behavioral biases, but who operate within efficient markets. We take the further step of moving beyond policies alone by examining the effect of these managers on overall firm risk. In this study, we examine a particular characteristic known as “sensation seeking” in the psychology literature, and its effect on firm financial policies and outcomes. Sensation seeking is a decidedly different behavioral bias than overconfidence, a subject that is discussed in detail in section 2.3. We proxy for sensation seeking using a novel dataset of CEOs who hold small aircraft pilot’s licenses. Prior psychology studies have documented that the desire to fly an airplane represents one of the highest predictors of sensation seeking personalities.

We demonstrate that operating small aircraft is a

particularly risky activity, especially the type of flying undertaken by CEOs who hold airman certificates. Apparently, even flight instructors perceive an increased risk of flying with business managers. One flight instructor stated that he “did not like instructing business executives because they are too busy to concentrate on flying; they are always thinking about business” (McFall, 1992). Zuckerman’s (1971) Sensation Seeking Scale is correlated with a “need for change and variety”, “change seeking” behavior, and “a high aversion to conditions of ‘sameness’” (Pearson, 1970; Zuckerman and Link, 1968). Psychology studies have also documented a negative relation between sensation seeking 2   

and age (Blackburn, 1969; Brownfield, 1966; Kish and Busse, 1968). Thus, older CEOs may be less inclined to pursue risky strategies as they migrate towards a quiet life in later years of leadership. The sensation seeking trait has also been linked to behaviors that may be less than desirable for corporate leaders: psychopathy, habitual drug use, violent crimes, and antisocial behaviors. 1 Fortunately, though, psychologists feel that these problems tend to manifest in sensation seekers who lack adequate outlets for creativity and stimulus (Farley, 1981). To the extent that managing a large corporation represents a suitable outlet for sensation seekers, these issues should be of secondary interest. We document that sensation seeking CEOs are associated with higher levels of firm risk; specifically, firms with CEO-pilots have an increased level of stock return volatility. Older CEOs are associated with lower firm-level volatility. Further, we trace the genesis of the differential levels of firm risk to one activity: the propensity of these CEOs to engage in frequent acquisition activity. CEO-pilots are significantly more likely to complete mergers and acquisitions of other firms. In some models, we find that CEOs are less likely to complete M&A transactions as they age. These results are consistent with the relation between personal characteristics and firm risk. In fact, when we purge successful bidders from the sample, the relation between CEO characteristics and firm risk disappears, implying that M&A activity may be the primary channel through which sensation seeking CEOs alter the riskiness of their firms. While high-flying CEOs tend to increase the riskiness of their firms, the implication for the value effects of their acquisitions remains vague. In the full sample, we do not find a significant relation between CEO pilots and M&A announcement returns. However, in low market-to-book (M/B) firms, commonly known as “value” firms, CEO pilots are associated with higher announcement returns. One possible explanation for this pattern is that among value firms with relatively few recognizable growth opportunities, the sensation seeking behavior by CEO pilots actually creates new valuable growth

                                                             1

Zuckerman (2007) provides a thorough summary of the psychology research on risky behavior by sensation seeking individuals.

3   

prospects for their merging firms. This is consistent with prior psychology studies which document that sensation seekers exhibit higher levels of creativity and cognitive innovation. In addition to the age variable, we also proxy for the degree of sensation seeking based on prior life experiences by flagging CEOs that spent part of their childhood during the Great Depression years. A number of studies use this variable to proxy for managerial beliefs and faith in external capital markets (e.g., Malmendier, Tate, and Yan, 2010; Malmendier and Nagel, 2009; Schoar, 2007; Graham and Narasimhan, 2004). We find that CEOs growing up during the Great Depression experience lower bidding returns among low M/B firms, but higher returns among high M/B firms. These results are opposite in sign from those for the sensation seekers, providing some support for both proxies. They also imply that CEOs who rate lower on the sensation seeking scale may create value through conservative acquisition decisions at glamour firms while reducing value through overly conventional transactions at value firms. We also explore military experience as a proxy for risk-taking but find no significant relation between this measure and key variables of interest, though it is possible that our sensation seeking proxies absorb the explanatory power of military experience. One possible concern with our study is that firms may intentionally hire CEOs of a particular personality trait in order to initiate changes within the firm. If this is the case, then our proxy for sensation seeking may be picking up unobservable, time-varying firm characteristics. To address this concern, we examine the compensation structure for CEOs in our sample. If firms hire CEOs in order to alter their status quo, then we would expect boards to explicitly incentivize these CEOs through their compensation contracts. Thus, we would expect sensation seekers to have higher exposure to changes in the firm’s risk. We measure CEO incentive exposure (Tumarkin, 2010), but find no evidence that these behavioral characteristics are rewarded in the compensation structure. Thus, we conclude that the effects on firm risk, M&A activity, and value-creation do not appear to be driven by this type of CEO-firm matching.

4   

Another concern with the data is that CEOs of rural firms may obtain piloting licenses in order to facilitate business travel.

This is unlikely because CEOs at rural firms can easily charter

business/executive flights with professionally-trained crews, and we report that this form of travel is much safer than travel by noncommercial piloting. Moreover, we note that firms with CEO pilots are predominantly located in large metropolitan statistical areas (MSAs). These firms also engage in a wide range of industries. Thus, our proxy does not appear to be driven by a firm location or industry effect. This study expands upon a growing field of work that explores the link between CEO personal characteristics and corporate decisions. Bertrand and Schoar (2003) find that older CEOs are associated with lower levels of firm leverage while CEOs with MBA degrees pursue more aggressive corporate strategies, including higher leverage, lower dividends, and more capital expenditures. Malmendier and Tate (2005) show that overconfident CEOs, or those who hold options longer than theoretically optimal, invest more heavily when internal resources are abundant. Malmendier and Tate (2008) further show that overconfident CEOs are more likely to engage in M&A activity, and when they do, these transactions tend to be value-destroying for bidders. Malmendier, Tate, and Yan (2010) find that: overconfident CEOs use less external financing; CEOs who grew up during the Great Depression view capital markets with skepticism and rely more heavily on internal funds; and CEOs with military experience pursue more aggressive corporate strategies. Kaplan, Klebanov, and Sorensen (2010) show that LBO performance is positively related to individual CEO characteristics including general ability and execution skills. In a different context, Grinblatt and Keloharju (2009) find that investors who exhibit characteristics of overconfidence and sensation seeking tend to trade more frequently. TD Ameritrade became the largest online discount brokerage in the world by catering specifically to frequent traders.

Perhaps not

coincidentally, J. Joseph Ricketts, the founder of Ameritrade, is a pilot. Collectively, the results indicate that prior life experiences, qualifications, personalities, and psychological attributes have an important and measurable effect on both investment and corporate decisions.

5   

Our work has policy implications to the extent that we provide evidence on a particular bias that can lead to specific outcomes. A key element of sensation seeking that allows for applicability in practice is that it can often be observed ex ante by firms during the hiring process through driving records. Actuarial research has documented an increased rate of driving infractions such as speeding tickets among risky pilots (McFall, 1992). According to Zuckerman (1994) and Jonah (1997), driving behavior is one of the best indicators of the sensation seeking trait and is the main proxy for sensation seeking used by Grinblatt and Ketoharju (2009).

Directors may obtain prior driving records during the candidate

screening process; thus, our evidence can be utilized by practitioners in the boardroom to better understand the behavioral tendencies of alternative chief executive candidates prior to selection. In their study of CEO characteristics, Kaplan et al. (2010) use data gathered from assessment interviews administered to CEO candidates for private equity firms. Zuckerman’s Sensation Seeking Scale V (SSSV) remains a popular behavioral assessment tool among psychologists and has been used in hundreds of studies to date. Conceivably, even in the absence of driving records, hiring firms could incorporate SSSV into the types of assessments detailed by Kaplan et al. (2010) during the candidate screening process. While we would enjoy access to wide-scale reporting of CEO personal hobbies and activities such as skydiving, surfing, and other sensation seeking behaviors, we are unaware of such systematic data. Thus, we focus our attention on the subset of data which we can systematically observe, but note that directors and recruiters may wish to explore other behaviors of interest that fall under the broader sensation seeking category. The next section summarizes related literature on the psychology of individual risk-taking (2.1) and CEO behavioral proxies (2.2). Section three explains the data collection process and section four presents our empirical results. Section five discusses the results and proposes several policy implications, and section six concludes the paper.

6   

2.

Related Literature

2.1

The Psychology of Individual Risk-Taking Although there is no universal profile of individuals with high risk tolerance, the psychology

literature suggests that one personality characteristic often present in such individuals is a need for stimulation and novelty, often referred to as ‘sensation seeking.’ The desire to pilot an aircraft is well accepted in the psychology literature as evidence of thrill seeking behavior or low risk aversion. Sensation seeking is “a trait defined by the seeking of varied, novel, complex, and intense sensations and experiences, and the willingness to take physical, social, legal and financial risks for the sake of such experiences” (Zuckerman, 1994).

Sensation seeking tendencies also decline as individuals age

(Blackburn, 1969; Brownfield, 1966; Kish and Busse, 1968). Zuckerman (1971), who is credited with developing the ‘Sensation Seeking Scale’ widely used in the psychology literature, observes that the phrase “I would like to learn to fly an airplane” reports the fourth highest loading on the thrill seeking factor in his study of 113 such phrases. Appendix A, republished from his 1971 study, reports the loadings of various activities. The loading for the desire to pilot an aircraft is just slightly behind skydiving and well ahead of activities such as motorcycle riding or downhill skiing. Further, while most pilots likely have increased risk tolerance, this condition may be exaggerated in CEOs that are pilots. Anderson and Galinsky (2006) find that that the acquisition of power leads to a greater tolerance for and preference for risk. The life insurance industry also recognizes piloting small aircraft as risky behavior. A Society of Actuaries study (McCall, 1992) examines the effects on mortality from piloting small aircraft, focusing narrowly on civilian aviators not flying for pay, the precise category into which most CEO pilots fall. He reports that the mortality rate of this group is increased over 100% for a 40-year old male that qualified for standard (average life expectancy) policies. Mortality rates increased over 200% for individuals that were otherwise qualified for substandard (high risk) policies. The report specifically singled out business

7   

executives as pilots that were especially high risk. Sadly, the popular press contains many examples of CEOs that have lost their lives piloting small aircraft. 2 To further highlight the riskiness of flying small aircraft, we compare fatality rates for general aviation pilots to fatality rates associated with a variety of other activities. Many readers are likely familiar with the old adage “driving is riskier than flying,” so it is important to differentiate alternative forms of flying. We obtain fatality data primarily from the National Transportation Safety Board for alternative forms of flying and the National Highway Traffic Safety Administration for operating motorcycles and passenger cars; a detailed list of sources is listed below Figure 1. One challenge in comparing the data is that flying data is reported on a per hour basis while driving data is reported per mile; thus, an assumption about the average speed of automobile traffic is required to convert the data to fatalities per hour. We assume an average speed of 55 miles per hour which we consider to be a conservatively high figure. Lower speed assumptions would result in driving fatality rates that are lower than those reported. Figure 1 plots fatality rates over the past decade for (i) personal/business flying, (ii) motorcycles, (iii) hot air balloons, (iv) personal helicopters, (v) ‘crop dusters,’ (vi) commercial helicopters, (vii) corporate/executive flying, (viii) passenger cars, and (ix) commercial airlines. Most CEO pilots fall into category (i). Note the term ‘business flying’ in category (i) is defined as “flights made in furtherance of the pilot’s own livelihood or in support of business endeavors”, 3 whereas executives riding on a plane with a professional crew fall into category (vii). The data indicates that the latter form of travel is not much different than driving passenger cars in terms of the risk of mortality, and flying commercial airlines is even safer. Operating small aircraft is substantially more dangerous than driving passenger cars. At 21.5 fatalities per million hours, personal/business flying is over 30 times more dangerous than                                                              2

3

Some examples include: Daniel Dorsch, former CEO of Checkers Drive-In Restaurants, Inc., 2009; Douglas J. Sharratt, CEO of ProSoft Technologies, 2008; Jeanette Symons, CEO of Industrious Kid, Inc., 2008; Bruce R. Kennedy, former CEO of Alaska Airlines, 2007; Michael F. Wille, CEO of The Title Company, Inc., 2006; David R. Burke, Sr., CEO of CeleXx Corp., 2001; Michael A. Chowdry, CEO of Atlas Air, Inc., 2001; J. Wesley Rogers, CEO of Oceaneering International, Inc., 1986. 2006 Nall Report: Accident Trends and Factors for 2005, published by the AOPA Air Safety Foundation.

8   

driving and ranks as the most dangerous activity among the nine forms analyzed, well ahead of even cropdusting. Taken together, the life insurance and fatality analyses suggest that operating small aircraft is indeed a very risky activity and a suitable proxy for high levels of risk tolerance, induced by the sensation seeking personality trait. While necessarily noisy, our proxy for risk tolerance is advantageous in the sense that it is not derived from compensation or firm-specific variables. To our knowledge, this is the first study to measure CEO risk tolerance on an individual level and relate this tolerance to overall firm risk. 2.2

CEO Characteristics In a seminal paper, Bertrand and Schoar (2003) document systematic behavioral differences in

corporate decision-making across managers and find evidence that these differences impact financing and investment policies in firms that are otherwise operating in similar environments. They obtain data on two specific characteristics, birth year and whether the CEO completed an MBA degree, and find significant relations between each characteristic and particular firm policies. Specifically, they find that older managers are associated with more conservative levels of firm leverage while CEOs with MBAs pursue more aggressive strategies that include higher leverage, lower dividends and more capital expenditures. They do not find any significant relation between these characteristics and M&A activity, but note that the characteristics in their study are just two of a myriad set of possible characteristics that may influence managerial decision-making. They lament the fact that due to data constraints they are unable to observe differences in characteristics such as personal psychology, among others. Malmedier and Tate (2005) begin to bridge this gap by constructing variables that proxy for differences in personal psychology related to overconfidence. CEOs that persistently expose themselves to excessive levels of idiosyncratic firm risk, rather than exercising vested options and diversifying their personal wealth, are labeled as overconfident. They document that overconfident CEOs’ investment decisions are more sensitive to cash flow realizations. In other words, there is no significant relationship 9   

between investment levels and overconfidence, but the interaction between investment and the overconfidence dummy is significantly positive.

The result is driven predominantly by the most

financially constrained firms. The inclusion of other observable characteristics, similar to the ones utilized by Bertrand and Schoar (2003) reduces the level of significance on the interaction variable, but it remains significant at the 10% level. There are at least two empirical issues with the overconfidence proxy for personal psychology, leaving it exposed to alternative explanations. The first concern is the endogenous relationship between the proxy and the firm, whose policies it seeks to explain. Specifically, the overconfidence measure is constructed from CEO option portfolio holding patterns, but the value of the portfolios can be altered by corporate policies and at least some CEOs who possess private information may have good reason for their portfolio choices. Malmendier and Tate acknowledge this alternative explanation and run several robustness tests suggesting that the overconfidence measure is not related to inside information, because the CEOs often take losses on their personal portfolios by not exercising.

These robustness measures mitigate this concern although we posit that a personal

psychology proxy that is exogenous to firm characteristics remains first best. The second issue with the overconfidence measure is that it is not observable ex ante. Rather, the authors state that their “measures require a long tenure within the firm in order to identify a CEO as overconfident,” suggesting that directors would be unable to identify this trait prior to (or even shortly after) the hiring decision. 4 Several studies have linked overconfidence to merger and acquisition activity, beginning with Roll’s (1986) hubris hypothesis of corporate takeovers, which rests on the bidder’s presumption that its valuation is correct, even in situations in which it is not. Malmendier and Tate (2008) link a subgroup of CEOs classified as overconfident to increased M&A activity. Risk seeking is ruled out as an alternative explanation because they observe a preference for cash financing and diversifying mergers. In this study, we extend their results by examining this alternative in greater detail. Rather than ‘risk seeking,’ we use the terms ‘risk tolerant’ and ‘sensation seeking’ which we feel are more accurate descriptions of this                                                              4

p. 2680.

10   

motivation and are well documented in the psychology literature. While there are examples in the psychology literature for which authors use the terms ‘risk seeking’ and ‘sensation seeking’ interchangeably, Zuckerman (2007, p.58) states “although sensation seeking is defined by a tolerance for risk, risk is not the point of it.” Using data from Finland, Grinblatt and Keloharju (2009) construct separate overconfidence and sensation seeking measures for individual investors and study the effects on trading activity. The authors lack data for Zuckerman’s sensation seeking measure, but are able to estimate this behavioral attribute using investors’ driving records in Finland. They observe (p.533): “The correlation between our sensation seeking and overconfidence measures is very low, so both behavioral attributes have relatively independent influence on trading. Sensation seeking is less related to the decision of whether to trade at all and more related to the decision of how much to trade.” We hypothesize that if sensation seeking tendencies are present in CEOs, a logical corollary to the increased trading frequency observed in individual investors will manifest in higher acquisition frequency. Our hypothesis is also motivated by Graham, Harvey & Puri’s (2009a) evidence from survey data. They state (p. 18): “The effect of the CEO’s personal risk aversion is very significant. More risk tolerant CEOs are more likely to make acquisitions in general. This is an interesting result consistent with the idea that CEO characteristics matter in acquisition activity.” As noted by Grinblatt and Keloharju (2009), sensation seeking is distinct from overconfidence and thus represents a departure from the CEO characteristics and biases previously studied in the behavioral corporate finance literature. 5

                                                             5

Other studies and models of CEO characteristics include Kaplan, Klebanov and Sorenson (2010), Hackbarth (2008), Adams et al. (2005), and Bennedsen et al. (2008).

11   

3.

Data We draw the initial sample of CEOs from the Compustat Executive Compensation (ExecuComp)

database, which primarily covers firms in the S&P 1500 index. Because ExecuComp coverage begins around 1992, we include CEOs in the sample only if the ExecuComp “Became CEO” date is on or after January 1, 1991. This mitigates any survival bias that could affect CEOs included in the dataset. This first-pass produces 4,012 CEO-firm combinations. We then search for these CEO names on the Federal Aviation Administration’s (FAA) Airmen Certification database. 6 If a given name does not produce a match in the FAA’s database, then this observation is coded as a non-pilot and no further validation is necessary. If a given name produces at least one name-match in the FAA’s database, we take further steps to confirm whether the pilot certificate belongs to the sample CEO. We use LexisNexis, Bloomberg, and public records searches to obtain birth dates, home addresses, and other personal information on the CEOs which can be used to validate the FAA certificate information. Doing so eliminates false-positive name matching and re-assigns those observations into the non-pilot group.

We are able to locate

sufficient personal information to confidently accept or reject the FAA name matches on over 77% of the initial sample.

The manual data-checking results in a final sample of 173 CEO pilots and 2,951 CEO

non-pilots. Six of the pilot CEOs worked for more than one company and 12 firms employ more than one pilot as CEO during the sample period. In total, the full panel covers 1,003 CEO pilot-firm years and 14,407 CEO non-pilot-firm years. Table 1 reports descriptive statistics on the airmen certificates linked to the sample CEOs. The vast majority, 68%, fall under the Private Pilot designation which is the level of certification required to solo pilot an aircraft under 12,500 lbs. Another 16% further reached the Commercial Pilot certification, which requires more rigorous examination and training and allows the holder to accept compensation                                                              6

Available at https://amsrvs.registry.faa.gov/airmeninquiry/. The FAA website also provides a downloadable database of active airmen certificate information. We search the online registry in order to locate both current and expired certificates.

12   

from passengers. It is important to note that this level of certification does not imply that the holder flies as a profession; a private pilot may seek a commercial license after having accumulated enough flight hours simply because a commercial license can reduce insurance costs, as it is evidence of experience. The highest level of certification, Airline Transport Pilot (ATP), is held by 13 CEOs, comprising 7.5% of the pilot sample. A pilot with a commercial license is qualified for an ATP license after attaining 1,500 flight hours and multi-engine ratings. Fourteen CEO pilots are classified as student pilots. A student license is required to exercise solo flights during the training portion prior to earning a private license. The CEO pilots hold a variety of ratings which provide additional flight privileges. For example, about half the pilots hold an instrument rating, allowing them to fly under conditions in which view is obstructed. Further, the pilots in our sample hold a diverse range of class ratings which allow them to operate multi-engine airplanes (55 CEOs), helicopters (10), gliders (4), experimental aircraft (3), hot air balloons (2) and planes that land on water (10). The most distinguished pilot in the sample is John W. Wood, Jr., CEO of Analogic Corp., who holds every single rating listed above except experimental aircraft, and is also a certified flight instructor. Given the substantial amount of time that must be dedicated to achieving these ratings and the demands of a typical S&P1500 CEO, we were concerned whether the pilot was in fact the same person as the corporate executive. Our selection methodology was confirmed when we found an interview with Mr. Wood published recently in a corporate newsletter. It runs alongside a picture of him flying an open cockpit WWII replica plane. Table 2 provides descriptive statistics on the initial CEO sample. One concern in using pilot license data as a proxy for risk tolerance is that if sample firms are disproportionally located in rural locales, the choice to obtain a pilot’s license may be driven by a lack of alternative options for air travel. To address this possibility, we collect geographical data on firm headquarters and the corresponding Metropolitan Statistical Area (MSA) as defined by the U.S. Census Bureau. 7 Panel A reports the most common MSAs for pilot CEO firms. The 20 most common MSAs, all of which have international                                                              7

MSA data is available at http://www.census.gov/population/www/estimates/metrodef.html.

13   

airports, account for nearly 70% of all observations. Further, 150 of the 170 firms fall within an MSA ranked in the top 100 by population, which corresponds to at least half a million people. Of the 20 that do not, 10 are ranked between 100 and 150 and two are foreign, leaving only eight firms that could be considered rural. Panel B reports descriptive statistics on CEO pilot firm industries by two-digit SIC code. The CEO pilots lead firms in a diverse range of industries. The two most prevalent SIC codes are Business services (10.6%), and Electronic equipment (7.6%) followed by industries including machinery, utilities and food products. It does not appear that CEO pilot firms are disproportionately operating in aviationrelated industries or that any single industry dominates the sample. Nonetheless, to control for the possibility that industry characteristics affect the results in later sections, we employ industry fixed effects in some models. 4.

Empirical Results

4.1

CEO pilots and firm risk Corporate risk taking can take a variety of forms such as investment in high risk projects,

engaging in leverage-increasing transactions, or pursuing aggressive tax strategies. Financial economists have documented that firm risk-taking is related to various aspects of the firm’s operating environment. For example, risk-taking increases in the presence of better investor protection (Acharya, Amihud and Litov, 2009), diversified blockholders (Faccio et al., 2010), and decreases following legislation which increases takeover protection (Low, 2009) and the Sarbanes-Oxley legislation (Bargeron, Lehn and Zutter, 2010). We note that investor protection is of importance primarily for non-U.S firms, and we control for the impact of legislation by including year fixed effects in some models. A separate line of literature evaluates the relation between individuals and firm risk.

For

example, Anderson and Galinsky (2006) find that increased power leads to increased risk tolerance. Since tenure can be a source of CEO power (Shivdasani and Yermack, 1999), we control for tenure in the

14   

regressions. The bulk of this literature focuses on the effect of compensation mechanisms, specifically employee stock options (ESO). Early literature on ESOs suggested that this form of compensation reduces the principal-agent problem by aligning incentives between managers and owners. The notion is that if risk-averse managers hold options that only become valuable if equity value increases, then they are more willing to take on risky, value-enhancing projects. Rajopal and Shevlin (2002) report evidence consistent with ESOs increasing the willingness of managers to engage in risky projects. Coles, Daniel and Naveen (2006) find a significantly positive relation between the exposure of a CEO’s wealth to stock volatility and the pursuit of risky corporate policies. However, recent literature suggests that the effect of ESOs on risk-taking is ambiguous. ESOs expose CEOs to firm-specific risk and when the options are inthe-money, a significant portion of the CEO’s wealth can be tied to the firm. Lewellen (2006) reports evidence that CEOs choose debt levels more conservatively when their personal portfolios are highly exposed to stock price volatility. Because our proxy for risk tolerance is not derived from compensation measures, we are able to cleanly measure the impact of this CEO characteristic on firm risk. We measure firm risk following Becker and Strömberg (2010), and several papers cited above, by using the annualized standard deviation of monthly equity returns. 8 In Table 3 we explore the link between CEO characteristics and firm risk. The table reports results from OLS regressions in which the dependent variable equals the annualized standard deviation of firm-level monthly stock returns. Firms with high return volatility are “riskier” than firms with low return volatility, although these models do not allow us to identify the corporate policies that drive this risk. The models include several firm-level control variables which have been shown to affect return volatility (e.g., Faccio et al., 2010): leverage, sales growth, return on equity (ROE), firm size measured as the natural log of total assets, and the natural log of firm age. The construction of these variables is detailed in Appendix B. The independent variables also include CEO characteristics of interest: an indicator variable that equals one if the CEO has a pilot’s license in a given firm-year (Pilot), an indicator variable that equals                                                              8

In unreported results, we also repeat the tests with daily returns, which produce qualitatively similar results.

15   

one of the CEO grew up during the Great Depression (Depression), an indicator variable that equals one if the CEO has prior military experience (Military), the age of the CEO, and the tenure of the CEO within a given firm. We find CEOs’ prior military experience (or lack thereof) from biographical information listed in BoardEx. Firm-level data is obtained from Compustat. Our main hypothesis concerns the impact of sensation seekers on firm risk. In the prior sections, we motivate the use of the Pilot variable as a proxy for sensation seekers, and Age and Depression as proxies individuals that would measure lower on the sensation seeking scale. The Military variable does not necessarily proxy for the sensation seeking attribute, but we include it to control for other aspects of risk tolerance that are not picked up by our primary variables. In Column (1), the coefficient on the Pilot variable is positive and significant, indicating that CEO pilots are associated with a greater degree of stock return volatility within their firms. The Age coefficient is negative and significant, indicating that firm risk tends to decrease as CEOs grow older. An age differential of 15 years would produce an effect that is similar in magnitude to but offsetting the effect of the Pilot variable (-0.002 * 15yrs vs. 0.033). Both of these results are consistent with theoretical predictions from the sensation seeking psychology literature. The Depression variable is not significantly related to firm volatility, and the coefficient on the Military variable is significant but negative. In Columns (2) through (4), we further control for time effects and industry effects by including year fixed effects in Column (2), industry fixed effects in Column (3), and both year and industry fixed effects in Column (4). One might be concerned that Pilot CEOs tend to self-select into more volatile industries. However, the coefficients on both the Pilot and Age variables remain significant and similar in magnitude across all models. Thus, the results do not appear to be driven by certain time periods or differences across industries. Although we cannot conclusively rule out other explanations, we interpret the results to indicate that sensation seeking pilot CEOs pursue riskier corporate policies, while aging CEOs become less inclined towards sensation seeking and pursue a “quiet life” that reduces firm risk.

16   

4.2

CEO pilots and acquisition activity As Graham, Harvey, and Puri (2009b) document, CEOs are more likely to retain the decision-

making authority for mergers and acquisitions relative to other corporate policies such as internal investment, payout, and capital structure. Thus, we focus on M&A as a reasonable source of corporate volatility. Table 4 reports logit models in which the dependent variable equals one if a firm announces a successful bid in a given year and zero otherwise. M&A transaction data comes from the Thomson Reuters SDC Platinum database. We include a number of control variables which have been shown to impact the likelihood of engagement in M&A activity (Bauguess and Stegemoller, 2008; Harford, Humphèry, and Powell, 2010; Malmendier and Tate, 2008): leverage, dividend yield, an indicator for net loss firm years (Loss Dummy), the natural log of total assets, free cash flow, Tobin’s Q, and capital expenditures (CapEx). The construction of these variables is explained in Appendix B. Coefficients are reported as odds ratios; thus, a coefficient greater than one is positive and less than one is negative. In Column (1) of Table 4, the coefficient on the Pilot variable is greater than one and statistically significant at the 1% level, indicating that CEO pilots are more likely to complete acquisitions. Older CEOs appear to engage in fewer acquisitions as well, as the coefficient on the Age variable is less than one. CEOs are more likely to make their acquisitions later during their tenure with the firm, with the coefficient on Tenure greater than one and significant at the 1% level. 9 The coefficients on the other CEO characteristics are not statistically different from zero. In Columns (2) through (4), we control for industry fixed effects and year fixed effects, with Column (4) reporting results from a model that includes both controls. The coefficient on the Pilot variable remains positive and significant at the 5% level, but Age is not significant in Columns (3) and (4). It is possible that the inclusion of both year and industry fixed effects reduces the power of the test on the Age variable in Column (4). In Column (5) we include these controls but drop the Pilot, Depression, and Military variables from the model, which increases the sample size. In the larger sample results reported in Column (5), the Age variable remains negative but is                                                              9

Identification for both the age and tenure variables requires that some CEOs switch firms during the sample period, which resets Tenure but not Age.

17   

now statistically significant at the 10% (two-sided) level. Thus age, which has been shown to attenuate sensation seeking tendencies, leads older CEOs to pursue fewer acquisitions, while sensation seekers, as measured by the Pilot variable, are more likely to exhibit their “change-seeking” attribute at a corporate level through an increased tendency to engage in M&A activity. 4.3

The link between acquisition activity and firm risk The effect of M&A transactions on firm risk is ambiguous. If acquisitions are diversifying, firm

risk could decrease (Amihud and Lev, 1981). However, a low beta firm acquiring a high beta firm could increase the volatility of returns to equity. We next turn to exploring the extent to which M&A activity drives changes in corporate risk as measured by return volatility. To do so, we calculate models that are similar to those produced in Table 3, except that we purge firms from the regressions if the firms engaged in M&A activity during the sample period. This leaves only firms that pursue organic growth through internal investment. Table 5 reports results from OLS models in which the dependent variable is the same as that in Table 3: stock return volatility. This significantly reduces the sample size to 478 firms over 2,451 firm-years. 10 In these models, the Pilot and Age variables are now insignificant. One interpretation is that sensation seekers influence the riskiness of their firms primarily through M&A activity. When we purge M&A activity from the sample, the relation between sensation seeking and firm risk disappears. We note, however, that an alternative explanation is that the testing power is reduced in these models due to the reduced sample size. Thus, our results must be interpreted with caution. Also, the Military and Tenure variables remain negative and statistically significant in several of the models, raising the possibility that these CEO characteristics influence firm risk through channels other than M&A, such as leverage choices.

                                                             10

One concern is that the restricted sample will contain a number of CEO pilots that is insufficient to adequately test the hypothesis on sensation seeking. We note that in the full sample, about 6.5% of firm-years are associated with pilot CEOs and in the restricted sample for Table 5, about 5.4% of the firm-years are associated with pilot CEOs. Thus, the restricted sample appears to have sufficient power to evaluate the coefficient on the Pilot variable.

18   

4.4

Acquisition announcement returns The predicted effect of the sensation seeking trait on overall firm value is not obvious. On the

one hand, if sensation seeking behavior induces CEOs to make overly aggressive decisions and pursue targets that are not within the firms’ core competencies, then this characteristic could reduce the value of the firm. This would be related to the value-reducing acquisitions made by overconfident managers as documented by Malmendier and Tate (2008). On the other hand, the finance literature is frequently concerned with the task of adequately incentivizing risk-averse agents to undertake value-increasing, but risky, projects for the firm. Equity-based compensation is one such incentive mechanism. Thus, to the extent that sensation seekers are more willing take on risky projects and acquisitions, then the M&A decisions made by these CEOs may be of greater benefit to certain firms, particularly those firms that are in greater need of risky growth opportunities. In Table 6, we evaluate the announcement period returns of sample bidders relative to CEO characteristics and other controls. Table 6 reports OLS regressions using bidder announcement returns as the dependent variable. As in Malmendier and Tate (2008), cumulative abnormal returns are formed over the (-1, +1) window around merger announcements, using the S&P 500 Index as the benchmark expected return. 11 If a firm announces more than one acquisition during the window, only the larger acquisition is retained in this sample. We include transaction characteristics as control variables: an indicator for cash payment, the natural log of transaction value, an indicator for private targets, and an indicator for transactions with different bidder and target three-digit SIC codes (Diversifying). We also include bidder characteristics from prior tables, but lagged by one year since announcement returns occur prior to year-end. All independent variables are defined in greater detail in Appendix B. In Column (1) of Table 6, bidder returns are unrelated to all of the CEO characteristics. Thus, on average, CEO characteristics do not appear to significantly impact the quality of their acquisitions in our                                                              11

Our results are not sensitive to this choice of benchmark; qualitatively similar results are obtained using the CRSP equal-weighted and value-weighted indices as the benchmark for expected returns.

19   

sample. Column (2) adds an indicator variable for bidders with an above-median market-to-book (M/B) ratio, and an interaction term of this variable with the CEO Pilot variable. We speculate that sensation seekers may produce greater benefit from M&A activity within firms that have poor growth prospects. We proxy for this using M/B, with below median M/B firms being “value” firms and above median firms being “glamour” firms. The Pilot variable is positive and significant in Column (2) while the interaction of Pilot with above median M/B is negative and significant, partially offsetting the beneficial impact of sensation seeking acquisitions. To further explore this effect, we split the sample into low M/B bidders (below median) and high M/B bidders (above median). Column (3) reports results from the bidder return regression in the low M/B subsample, and shows a positive and significant effect of Pilot CEOs. Additionally, the coefficient on the Depression indicator is negative and significant. The results are consistent with the hypothesis that sensation seekers pursue value-increasing acquisitions at bidders with few recognizable organic growth prospects. It is possible that CEOs who are less inclined to engage in outside stimuli, as measured by Depression, engage in lower quality acquisitions at value firms, which could potentially benefit from a more aggressive pursuit of growth options. Column (4) includes year fixed effects, and the coefficient on Pilot remains positive and significant, though the Depression variable is no longer statistically significant. In Column (5), we compute a similar model of bidder returns in the high M/B “glamour” subsample of bidders. The results for Pilot and Depression are opposite in sign to those in the “value” subsample. While the coefficient on the Pilot variable is not statistically significant, the Depression indicator is positive and significant. Results are similar in Column (6), which includes year fixed effects. One interpretation of these results is that CEOs who grew up during the Great Depression are more conservative in their acquisition decisions, and this conservatism is beneficial in high market-to-book “glamour” firms, but detrimental in low market-to-book “value” firms. We conclude that in at least certain firm types, sensation seekers can create value by pursuing a policy of heightened merger and acquisition activity.

20   

4.5

Robustness We interpret our primary findings to indicate that sensation seeking CEOs are associated with

greater firm risk, and that these CEOs tend to increase the volatility of their firm’s stock by pursuing corporate growth through increased M&A activity. The results do not appear to be driven by a rural vs. urban firm location effect, or by an industry effect. One might be concerned that the results are influenced by some industries more than others. The most plausible industry of concern would be that containing airline-related firms, since these firms may tend to attract CEOs with flying experience. While it is unlikely that this effect would produce a positive relation between our Pilot variable and firm risk or M&A activity, it is possible. To address this concern, we purge all airline-related firms from the sample (SIC codes 4512 and 3721) and re-calculate all models. The results remain qualitatively similar; thus, our main findings do not appear to be driven by an airline industry effect. General industry dynamics should be adequately controlled for through the inclusion of industry fixed effects in the models. Another possibility is that sensation seeking CEOs are hired by boards who wish to institute dramatic shifts in corporate development. To the extent this occurs, it is plausible that the boards of these firms are advocating for increases in firm risk and/or increases in M&A activity. Thus, the CEOs who institute these changes are doing so not out of their own behavioral tendencies, but rather in response to board directives. This interpretation does not fully explain the fact that these boards tend to hire CEOs with piloting licenses. To put it another way, why else would directors seek out CEO pilots if not for their beneficial sensation seeking skills? Thus, this alternative CEO-firm matching explanation does not discount the documented relation between CEO characteristics and corporate policy. Nonetheless, we conduct a robustness check to examine the compensation structure for CEOs of varying characteristics.

If boards hire CEOs with certain characteristics to carry out risky, new M&A

programs, then we would expect the compensation structure for these CEOs to reflect this incentive. One way to incentivize agents to take risk-increasing projects is through an increase in equity-based compensation. We measure the CEO’s incentive exposure by calculating the fraction: Black-Scholes

21   

Delta from option and stock exposure times stock price, divided by total compensation (Tumarkin, 2010). We predict that if directors wish to incentivize CEOs to pursue riskier strategies, they will structure the CEOs’ compensation package with a greater incentive exposure. Table 7 reports results from OLS regressions in which the dependent variable is CEO incentive exposure*1,000. Various firm-level controls and CEO characteristics are included as independent variables, as well as year and industry fixed effects across the different models. None of the CEO characteristics are significant in any of the models. In fact, the coefficient on the Pilot variable is negative, indicating that if anything, CEO pilots receive less explicit incentive compensation than nonpilot CEOs. Thus, we do not find any evidence to indicate that CEO-firm matching drives our main results. 5.

Discussion and policy implications Malmendier and Tate (2008) rule out a risk seeking explanation for the increased M&A activity

by overconfident managers because they observe a preference for cash financing and diversifying mergers.

However, diversifying mergers would not represent evidence inconsistent with sensation

seeking. Rather, mergers outside of the firm’s core competency are exactly the type of transactions one would expect from sensation seekers looking to satiate a desire for change. Nor is cash financing inconsistent with sensation seeking acquisitions. As stated earlier, sensation seekers are not seeking risk in and of itself; rather, they are seeking diverse experiences and are tolerant of increased risk in the pursuit of these experiences. Ample cash reserves would allow the sensation seeking tendencies to manifest in higher acquisition frequency without attracting the scrutiny involved with raising external capital. If free cash flow allows a CEO to pursue value-destroying acquisitions, then Jensen’s (1986) free cash flow problem is at hand. However, in our analysis of announcement returns, we find no evidence that value is being destroyed by sensation seekers. Our study is related to contemporaneous work by Malmendier, Tate and Yan (2010), who use military experience as a proxy for risk tolerance. It is important to note that the CEO Pilot proxy is 22   

fundamentally different from one based on life experiences. Sensation seeking is a stable personality trait that individuals are born with, which differentiates it from military experience. 12 Thus, our findings complement their work by highlighting the fact that risk tolerance is a heterogeneous condition. While both combat veterans and sensation seekers can be categorized as risk tolerant, the manifestation of this tolerance on corporate policies can be quite different. For example, Malmendier, Tate, and Yan (2010) report evidence that CEOs with past military experience choose higher levels of leverage than their predecessors or successors. In unreported results, we test leverage choices for sensation seeking CEOs and find no significant evidence of higher or lower debt utilization. Our findings are consistent with Graham, Harvey & Puri’s (2009b) evidence that CEOs engaging in high frequency M&A decisions tend to delegate capital structure decisions; thus, it is not clear that we would expect sensation seeking behavior by CEOs to have any effect on leverage. Although flying airplanes may be relatively rare, nearly all CEOs have a driving record. Driving records are not observable to the econometrician studying US CEOs, but are readily available to firms during the CEO hiring process. Zuckerman (1994) notes that driving violations are an excellent way to observe sensation seeking, which manifests primarily through speeding violations; this is the motivation behind Grinblatt and Keloharju’s (2009) data.

The important policy implication is that unlike an

overconfidence measure based on option holding behavior, which reveals itself only many years after a chief executive is hired, sensation seeking is often observable ex ante to the CEO’s tenure. A board of directors can observe this driving record and develop a sense of whether or not a candidate may possess a high degree of sensation seeking. We refrain from painting this trait as universally positive or negative, as some firms may or may not desire and benefit from a CEO with a high risk tolerance. However, recognizing that a candidate has a propensity for acquisitions, before they occur, is unquestionably valuable information from the standpoint of corporate governance.

                                                             12

Sensation seeking is highly correlated with the ‘novelty seeking’ dimension in Cloninger's Tridimensional Personality Questionnaire (McCourt et al. 1993) which has been conclusively linked to specific genes.

23   

As mentioned in the introduction, sensation seeking is associated with many different behavioral outcomes, some of which are detrimental. On the other hand, sensation seeking exhibits a significant positive relationship with general intelligence, higher levels of creativity, and originality (Zuckerman, 1994).

Farley (1981) finds that although sensation seeking correlates with both delinquency and

creativity, it does not necessarily do so in the same individual. The environment one is exposed to during the formative years can dictate whether the sensation seeking will be expressed in criminality or creativity, and we hypothesize that our sample of S&P 1500 CEOs are comprised largely, if not entirely, of the latter group. The psychology literature suggests that sensation seekers think non-linearly and can see past distractions while focusing on a task; their performance is not negatively impacted by competing stimuli. In other words, the sensation seeking trait is associated with several desirable traits in CEOs, even outside of the higher degree of risk tolerance. From a principal-agent perspective, hiring a sensation seeker as CEO may be a form of efficient contracting. Much of the literature on employee stock options revolves around risk-neutral principals attempting to induce risk-averse agents to pursue positive NPV, but risky, projects. To the extent that some candidates come hard-wired with risk tolerance, high levels of equity based compensation would not be required to encourage the desired behavior. Our evidence is consistent with this interpretation. 6.

Conclusion This study analyzes the relation between the sensation seeking psychological attribute of CEOs

and firm risk. We proxy for sensation seeking behavior using one of the predominant survey components in Zuckerman’s Sensation Seeking Scale: the desire to fly an airplane. Using data from the FAA registry of certified pilots, we find that CEO pilots are associated with greater stock return volatility at their firms and are more likely to engage in merger and acquisition activity. Age has been shown to be inversely correlated with sensation seeking behavior, and we find that CEOs are less likely to pursue M&A transactions as they age, producing a lower level of return volatility. The sensation seeking behavior of pilot CEOs results in higher quality acquisitions at firms with few measurable growth opportunities, 24   

implying that sensation seekers may be desirable candidates among certain firms. While relatively few CEOs hold piloting licenses, board members can obtain other personal information that is of use during the candidate screening process. Driving records, current age, and psychological profiles can similarly reveal sensation seeking tendencies and inform boards about appropriately qualified candidates based on firm type and desires for future growth. A promising area for future research will be to better understand the myriad behavioral biases that lie behind differential preferences for risk. Shedding light on these biases and the corporate policies associated with them will ultimately lead to better corporate decisionmaking.

25   

References Acharya, Viral V., Yakov Amihud, and Lubomir Litov, 2009, Creditor rights and corporate risk-taking, Working Paper, London Business School, New York University, and Washington University. Adams, Renee, Heitor Almeida, and Daniel Ferreira, 2005, Powerful CEOs and their impact on corporate performance, Review of Financial Studies 18, 1403-1432. Amihud, Yakov and Baruch Lev, 1981, Risk reduction as a managerial motive for conglomerate mergers, Bell Journal of Economics 12, 605-617. Anderson, Cameron and Adam D. Galinsky, 2006, Power, optimism, and risk taking. European Journal of Social Psychology 36, 511-536. Bargeron, Leonce L., Kenneth M. Lehn, and Chad J. Zutter, 2010, Sarbanes-Oxley and corporate risktaking, Journal of Accounting and Economics 49, 34-52. Bauguess, Scott and Mike Stegemoller, 2008, Protective governance choices and the value of acquisition activity, Journal of Corporate Finance 14, 550-566. Becker, Bo and Per Strömberg, 2010, Equity-debtholder conflicts and capital structure, Working Paper, Harvard Business School and Stockholm School of Economics. Bennedsen, Morten, Francisco Pérez-González, and Daniel Wolfenzon, 2008, Do CEOs matter? Working paper, Columbia University. Bertrand, Marianne and Antoinette Schoar, 2003, Managing with style: The effect of managers on firm policies, Quarterly Journal of Economics 118, 301–330. Blackburn, R., 1969, Sensation seeking, impulsivity, and psychopathic personality, Journal of Consulting and Clinical Psychology 33, 571-574. Brownfield, Charles A., 1966, Optimal stimulation levels of normal and disturbed subjects in sensory deprivation, Psychologia 9, 27-38. Coles, Jeffrey L., Naveen D. Daniel, and Lalitha Naveen, 2006, Managerial incentives and risk-taking, Journal of Financial Economics 79, 431-468. Faccio, Mara, Maria-Teresa Marchica, and Roberto Mura, 2010, Large shareholder diversification and corporate risk-taking, Working Paper, Purdue University and University of Manchester. Farley, F. H., 1981, Basic process individual differences: A biologically based theory of individualization for cognitive, affective, and creative outcomes. In F.H. Farley & N.J. Gordon (eds.), Psychology and Education: The State of the Union, McCutchon Publishing: Berkeley, CA. Graham, John R., Campbell R. Harvey, and Manju Puri, 2009a, Managerial attitudes and corporate actions, Working Paper, Duke University.

26   

Graham, John R., Campbell R. Harvey, and Manju Puri, 2009b, Capital allocation and delegation of decision-making authority within firms, Working Paper, Duke University. Graham, John R., and Krishna Narasimhan, 2004, Corporate survival and managerial experiences during the Great Depression, Working Paper, Duke University. Grinblatt, Mark and Matti Keloharju, 2009, Sensation seeking, overconfidence, and trading activity, Journal of Finance 64, 549-578. Hackbarth, Dirk, 2008, Managerial traits and capital structure decisions, Journal of Financial and Quantitative Analysis 43, 843-881. Hambrick, Donald C. and Phyllis A. Mason, 1984, Upper Echelons: The Organization as a Reflection of Its Top Managers, Academy of Management Review, 9, 2, 193-206. Harford, Jarrad, Mark L. Humphèry, and Ronan Powell, 2010, The sources of value destruction in acquisitions by entrenched managers, Working Paper, University of Washington. Jensen, Michael C., 1986, Agency costs of free cash flow, corporate finance and takeovers, American Economic Review 76, 323–329. Jonah, Brian, 1997, Sensation seeking and risky driving: A review and synthesis of the literature, Accident Analysis and Prevention 29, 651–665. Kaplan, Steven N., Mark M. Klebanov, and Morten Sorensen, 2010, Which CEO characteristics and abilities matter?, Journal of Finance, forthcoming. Kish, George B. and William Busse, 1968, Correlates of stimulus seeking: Age, education, intelligence, and aptitudes, Journal of Consulting and Clinical Psychology 32, 633-637. Lewellen, Katharina, 2006, Financing decisions when managers are risk averse, Journal of Financial Economics 82, 551-589. Low, Angie, 2009, Managerial risk-taking behavior and equity-based compensation, Journal of Financial Economics 92, 470-490. Malmendier, Ulrike and Geoffrey A. Tate, 2005, CEO overconfidence and corporate investment, Journal of Finance 60, 2660-2700. Malmendier, Ulrike and Geoffrey A. Tate, 2008, Who makes acquisitions? CEO overconfidence and the market’s reaction, Journal of Financial Economics 89, 20-43. Malmendier, Ulrike, Geoffrey A. Tate, and Jon Yan, 2010, Managerial beliefs and corporate policies, Working Paper, UC Berkeley, UCLA, and Stanford. Malmendier, Ulrike, and Stefan Nagel, 2009, Depression babies: Do macroeconomic experiences affect risk-taking? Working Paper, UC Berkeley and Stanford. 27   

McCourt, William F., Ronald J. Gurrera, and Henry S. G. Cutter, 1993. Sensation seeking and novelty seeking: Are they the same?, Journal of Nervous & Mental Disease 181, 309–312. McFall, Melvin C., 1992, Lincoln National aviation mortality and claim analysis, North American Actuarial Journal 33, 6-8. Pearson, Pamela H., 1970, Relationships between global and specified measures of novelty seeking, Journal of Consulting and Clinical Psychology 34, 199-204. Rajopal, Shivaram and Terry Shevlin, 2002, Empirical evidence on the relation between stock option compensation and risk taking, Journal of Accounting and Economics 33, 145-171. Roll, Richard, 1986, The hubris hypothesis of corporate takeovers, Journal of Business 59, 197–216. Schoar, Antoinette, 2007, CEO careers and style, Working Paper, MIT. Shivdasani, Anil and David Yermack, 1999, CEO involvement in the selection of new board members: An empirical analysis, Journal of Finance 54, 1829-1853. Tumarkin, Robert, 2010, How much do CEO incentives matter?, Working Paper, New York University. Zuckerman, Marvin, 1971, Dimensions of sensation seeking, Journal of Consulting and Clinical Psychology 36, 45-52. Zuckerman, Marvin, 1994, Behavioral Expression and Biosocial Bases of Sensation Seeking, Cambridge University Press: New York, NY. Zuckerman, Marvin, 2007, Sensation Seeking and Risky Behavior, American Psychological Association: Washington, D.C. Zuckerman, Marvin and Kathryn Link, 1968, Construct validity for the Sensation-Seeking Scale, Journal of Consulting and Clinical Psychology 32, 420-426.

28   

Appendix A. Survey Evidence on Sensation Seeking This table reproduces factor analysis results from Zuckerman (1971, Table 1). Survey responses to 113 items are from 160 male and 172 female undergraduate students at Temple University. The subset of significant items for the “Thrill and Adventure Seeking” factor are reported below in a descending sort on the item loadings from male respondents. Emphasis on fourth item added by authors. Item Sensation Seeking Item Choice, Thrill and Adventure Seeking Factor Loading 72

I would like to try surfboard riding.

67

I would like to take up the sport of water-skiing.

65

I would like to try parachute jumping.

63

I would like to learn to fly an airplane.

63

I would like to go scuba diving.

57

I think I would enjoy the sensations of skiing very fast down a high mountain slope. 

55

I would like to sail a long distance in a small but seaworthy sailing craft.

54

I would like to drive or ride on a motorcycle.

52

I sometimes like to do things that are a little frightening.

52

Sometimes I like to swim far out from the shore.

51

I like to dive off the high board.

46

I often wish I could be a mountain climber.

44

I like to drive in open convertibles.

35

I enjoy many of the rides in amusement parks.

29   

Appendix B. Variable Definitions CEO Characteristics Age CEO’s age, updated annually; Source: Compustat Executive Compensation Depression =1 if CEO born between 1920 and 1929, 0 otherwise Military =1 if CEO has military experience, 0 otherwise; coded using BoardEx biographical information Pilot =1 if CEO has had at least one certificate in FAA records, 0 otherwise; Source: Federal Aviation Administration Tenure Years of service as CEO at given firm; Source: Compustat Executive Compensation Firm Characteristics (Source: Compustat) Assets (AT) CapEx (CAPX) Dividend Yield (DVPSP_F / PRCC_F) Firm Age Cumulative number of firm years listed in Compustat Free Cash Flow ([OIBDP – XINT – TXT – CAPX] / ATt-1) Leverage ([DLC + DLTT] / AT) Loss Dummy =1 if negative net income (NI) in given year, 0 otherwise M/B ([PRCC_F * CSHO] / SEQ) Q ([AT – SEQ + [PRCC_F * CSHO]] / AT) ROE (EBITDA / ATt-1) Sales Growth (REVT / REVTt-1) Merger Characteristics (Source: Thomson Reuters SDC Platinum) Cash Payment =1 if transaction consideration is cash, 0 otherwise Diversifying =1 if bidder and target in different 3-digit SIC code, 0 otherwise Private Target =1 if target private, 0 otherwise Trans. Value Transaction value in $mms

30   

25.00 21.50

15.50 15.00

13.70 11.50

10.00

0.98

0.67 passenger cars

4.70

5.00

corporate / executive aircraft

Fatal crashes per million hours

20.09 20.00

0.06 commercial airlines

helicopter (commercial)

aerial application aircraft

helicopter (personal)

hot-air balloon

motorcycle

personal / business flying

-

Figure 1. Fatal crash rates per million hours of various forms of transportation. Data sources are as follows: •

Personal/business flying, corporate/executive aircraft, aerial application aircraft: National Transportation and Safety Board Annual Review of Aircraft Accident Data, U.S. General Aviation, Calendar Year 2005; General Aviation and Air Taxi Activity and Avionics Surveys CY2005, Table 1.6. Corporate/executive aircraft category includes a paid, professional flight crew, whereas personal/business flying does not.

•

Motorcycle, passenger cars: National Highway Traffic Safety Administration (NHTSA), Fatality Analysis Reporting System Encyclopedia. To convert miles driven to hours driven, we assume an average speed of 55 miles per hour.

•

Hot-air balloons: National Transportation and Safety Board Annual Review of Aircraft Accident Data, U.S. General Aviation, “Lighter-than-air” craft category.

•

Helicopter (personal, business): 2009 Nall Report, An AOPA Air Safety Foundation Publication.

•

Commercial airlines: National Transportation and Safety Board, Aviation Accident Statistics, Table 6.

31   

Table 1. CEO Airman Certificate Descriptive Statistics Panel A reports the highest proficiency level attained by CEO pilots in the sample. Panel B reports the various pilot certificates held by sample CEOs. CEOs may hold multiple certificate ratings. Airman certificate information is obtained from the Federal Aviation Administration’s (FAA) Airmen Certification database (https://amsrvs.registry.faa.gov/airmeninquiry/), which includes both active and inactive certificates. Panel A: Pilot Certificates N 118 28 14 13 173

Private Pilot Commercial Pilot Student Pilot Air Transport Pilot Total Panel B: Certificate Ratings Single Engine Airplane Instrument Multiengine Airplane Helicopter Water Landing Flight Instructor Glider Experimental Aircraft Hot Air Balloon

157 87 55 10 10 9 4 3 2

32   

% 68.2% 16.2% 8.1% 7.5% 100.0%

Table 2. CEO Pilot Sample Location and Industry Distribution Panel A reports the location distribution of firm headquarters for the CEO pilot sample, by Metropolitan Statistical Area (MSA). Panel B reports the industry distribution of the sample, by two-digit SIC code. Panel A: Pilot Firm Locations MSA# 41940 35620 26420 33460 16980 19100 31100 14460 41860 41180 47900 12060 17460 17140 19740 29820 37980 38900 41700 41740

N 17 16 10 10 9 8 7 6 5 4 4 3 3 2 2 2 2 2 2 2

% 10.0% 9.4% 5.9% 5.9% 5.3% 4.7% 4.1% 3.5% 2.9% 2.4% 2.4% 1.8% 1.8% 1.2% 1.2% 1.2% 1.2% 1.2% 1.2% 1.2%

MSA Description San Jose-Sunnyvale-Santa Clara, CA New York-Northern New Jersey-Long Island, NY-NJ-PA Houston-Sugar Land-Baytown, TX Minneapolis-St. Paul-Bloomington, MN-WI Chicago-Naperville-Joliet, IL-IN-WI Dallas-Fort Worth-Arlington, TX Los Angeles-Long Beach-Santa Ana, CA Boston-Cambridge-Quincy, MA-NH San Francisco-Oakland-Fremont, CA St. Louis, MO-IL Washington-Arlington-Alexandria, DC-VA-MD-WV Atlanta-Sandy Springs-Marietta, GA Cleveland-Elyria-Mentor, OH Cincinnati-Middletown, OH-KY-IN Denver-Aurora, CO Las Vegas-Paradise, NV Nashville-Davidson--Murfreesboro--Franklin, TN Portland-Vancouver-Beaverton, OR-WA San Antonio, TX San Diego-Carlsbad-San Marcos, CA

Panel B: Pilot Firm Industries SIC2 73 36 35 49 20 37 38 33 28 60 80 29 48 13 39 87

N 18 13 11 11 10 10 8 7 6 6 6 5 5 4 4 4

% 10.6% 7.6% 6.5% 6.5% 5.9% 5.9% 4.7% 4.1% 3.5% 3.5% 3.5% 2.9% 2.9% 2.4% 2.4% 2.4%

SIC Description Business services Electronic equipment and components, except computer equipment Industrial and commercial machinery and computer equipment Electric, gas, and sanitary services Food and kindred products Transportation equipment Instruments and related products Primary metal industries Chemicals and allied products Depository Institutions Health services Petroleum and coal products Communications Oil and gas extraction Miscellaneous manufacturing industries Engineering and management services

33   

Table 3. CEO Pilots and Firm Risk OLS regressions with annualized standard deviation of firm-level monthly stock returns as the dependent variable. A constant is included in all models. Independent variables are defined in Appendix B. Standard errors are clustered by firm and year, and p-values are in parentheses with ***, **, and * representing significance at the 1%, 5%, and 10% levels, respectively.

(1)

(2)

(3)

(4)

CEO Characteristics Pilot

0.033 ** (0.010)

0.027 ** (0.031)

0.037 *** (0.001)

0.032 *** (0.001)

Depression

0.015 (0.726)

0.039 (0.252)

0.013 (0.769)

0.026 (0.450)

Military

-0.020 * (0.096)

-0.022 ** (0.042)

-0.005 (0.608)

-0.011 (0.197)

Age

-0.002 *** (0.001)

-0.002 *** (0.001)

-0.001 *** (0.004)

-0.001 ** (0.020)

Tenure

-0.001 (0.635)

-0.001 (0.167)

-0.002 (0.531)

-0.001 (0.156)

Leverage

0.006 (0.852)

-0.026 (0.476)

0.127 *** (0.000)

0.078 *** (0.009)

Sales Growth

0.049 (0.352)

0.061 * (0.050)

0.043 (0.410)

0.055 * (0.059)

ROE

-0.213 *** (0.000)

-0.193 *** (0.000)

-0.187 *** (0.000)

-0.167 *** (0.000)

Ln(Assets)

-0.032 *** (0.000)

-0.031 *** (0.000)

-0.032 *** (0.000)

-0.032 *** (0.000)

Ln(Firm Age)

-0.065 *** (0.000)

-0.057 *** (0.000)

-0.048 *** (0.000)

-0.036 *** (0.000)

Industry Fixed Effects

No

No

Yes

Yes

Year Fixed Effects

No

Yes

No

Yes

10,692

10,692

10,692

10,692

1,556

1,556

1,556

1,556

28.53%

43.17%

38.27%

52.66%

Firm Characteristics

Observations Firms 2

R

34   

Table 4. Acquisitiveness of Pilot-CEOs Logit models in which the dependent variable equals one if the firm announces a successful merger bid in a given year and zero otherwise. A constant is included in all models. Independent variables are defined in Appendix B. Coefficients are reported as odds ratios. P-values are in parentheses with ***, **, and * representing significance at the 1%, 5%, and 10% levels, respectively. (1) CEO Characteristics Pilot

(2)

(3)

(5)

1.434 ** (0.011)

1.403 ** (0.017)

1.369 ** (0.018)

1.344 ** (0.027)

Depression

1.142 (0.767)

1.290 (0.577)

0.888 (0.785)

0.960 (0.926)

Military

0.897 (0.425)

0.893 (0.413)

0.956 (0.730)

0.946 (0.669)

Age

0.990 * (0.094)

0.990 * (0.099)

0.997 (0.640)

0.998 (0.692)

0.992 * (0.079)

Tenure

1.032 *** (0.002)

1.038 *** (0.001)

1.023 ** (0.021)

1.031 *** (0.005)

1.036 *** (0.000)

0.760 (0.153)

0.700 * (0.068)

1.179 (0.396)

1.109 (0.603)

1.310 * (0.090)

Dividend Yield

0.000 *** (0.000)

0.000 *** (0.000)

0.001 *** (0.002)

0.004 ** (0.011)

0.001 *** (0.000)

Loss Dummy

0.850 * (0.060)

0.850 * (0.065)

0.756 *** (0.001)

0.757 *** (0.001)

0.725 *** (0.000)

Ln(Assets)

1.257 *** (0.000)

1.264 *** (0.000)

1.360 *** (0.000)

1.374 *** (0.000)

1.359 *** (0.000)

Free Cash Flow

6.236 *** (0.000)

2.584 *** (0.000)

5.850 *** (0.000)

6.405 *** (0.000)

3.278 *** (0.000)

Q

1.061 ** (0.031)

0.029 (0.144)

0.995 (0.858)

0.975 (0.357)

1.004 (0.846)

CapEx

8.018 *** (0.000)

4.845 *** (0.000)

20.595 *** (0.000)

23.808 *** (0.000)

8.641 *** (0.000)

No No

No Yes

Yes No

Yes Yes

Yes Yes

9,912 1,463

9,912 1,463

9,912 1,463

9,912 1,463

15,152 2,174

Firm Characteristics Leverage

Industry Fixed Effects Year Fixed Effects Observations Firms

35   

(4)

Table 5. CEO Pilots and Firm Risk, Excluding M&A Activity OLS regressions with annualized standard deviation of firm-level monthly stock returns as the dependent variable. Observations are excluded from the sample if the firm announced at least one successful acquisition during the sample period. A constant is included in all models. Independent variables are defined in Appendix B. Standard errors are clustered by firm and year, and p-values are in parentheses with ***, **, and * representing significance at the 1%, 5%, and 10% levels, respectively.

(1)

(2)

(3)

(4)

Pilot

0.015 (0.542)

0.006 (0.799)

-0.001 (0.960)

-0.012 (0.568)

Depression

0.227 (0.169)

0.168 (0.133)

0.189 (0.363)

0.127 (0.432)

Military

-0.014 (0.530)

-0.023 (0.326)

-0.027 (0.100)

-0.034 ** (0.016)

Age

-0.001 (0.443)

-0.000 (0.656)

-0.000 (0.659)

-0.000 (0.848)

Tenure

-0.004 (0.173)

-0.004 *** (0.008)

-0.003 (0.274)

-0.003 * (0.052)

Leverage

0.071 * (0.050)

0.044 (0.275)

0.145 *** (0.000)

0.105 *** (0.009)

Sales Growth

0.011 (0.775)

0.021 (0.445)

0.001 (0.968)

0.012 (0.600)

ROE

-0.178 *** (0.000)

-0.166 *** (0.000)

-0.151 *** (0.000)

-0.137 *** (0.000)

Ln(Assets)

-0.040 *** (0.000)

-0.040 *** (0.000)

-0.039 *** (0.000)

-0.037 *** (0.000)

Ln(Firm Age)

-0.055 *** (0.000)

-0.051 *** (0.000)

-0.029 ** (0.014)

-0.021 ** (0.033)

Industry Fixed Effects

No

No

Yes

Yes

Year Fixed Effects

No

Yes

No

Yes

2,451

2,451

2,451

2,451

478

478

478

478

27.67%

40.59%

42.68%

54.94%

CEO Characteristics

Firm Characteristics

Observations Firms 2

R

36   

Table 6. CEO Pilots and M&A Announcement Returns OLS regressions with bidder announcement returns as the dependent variable. Abnormal returns are calculated over the window from one day prior to one day following merger announcements (-1, +1), using the S&P 500 Index as the expected return. A constant is included in all models. Independent variables are defined in Appendix B; all bidder characteristics are lagged by one year. Standard errors are clustered by firm and year, and p-values are in parentheses with ***, **, and * representing significance at the 1%, 5%, and 10% levels, respectively. Full Sample (1) CEO Characteristics Pilot

(2)

Low M/B Bidders (3) (4)

0.003 (0.413)

0.011 ** (0.048)

0.012 ** (0.023)

0.010 ** (0.047)

-0.004 (0.568)

-0.006 (0.366)

Depression

-0.003 (0.852)

-0.003 (0.862)

-0.017 ** (0.026)

-0.008 (0.247)

0.024 * (0.066)

0.034 ** (0.012)

Military

-0.001 (0.745)

-0.001 (0.749)

0.001 (0.816)

-0.002 (0.636)

-0.003 (0.576)

-0.003 (0.580)

Age

-0.000 (0.838)

-0.000 (0.830)

-0.000 (0.705)

-0.000 (0.582)

0.000 (0.933)

0.000 (0.827)

Tenure

-0.000 (0.102)

-0.000 (0.100)

-0.000 (0.366)

-0.000 (0.989)

-0.001 (0.203)

-0.000 (0.761)

Bidder Characteristics M/B > Median

0.002 (0.555)

Pilot * M/B > Median

-0.016 * (0.061)

Ln(Firm Age)

-0.000 (0.973)

0.000 (0.938)

0.004 (0.106)

0.005 ** (0.045)

-0.004 *** (0.000)

-0.003 ** (0.021)

Free Cash Flow

-0.006 (0.334)

-0.006 (0.337)

-0.015 (0.599)

-0.008 (0.749)

-0.005 (0.370)

-0.005 (0.398)

CapEx

0.023 ** (0.047)

0.022 * (0.060)

-0.040 (0.193)

-0.047 (0.175)

0.028 ** (0.027)

0.022 (0.104)

Ln(Assets)

-0.003 *** (0.000)

-0.003 *** (0.000)

-0.003 ** (0.017)

-0.002 ** (0.025)

-0.003 ** (0.018)

-0.003 ** (0.023)

Loss Dummy

0.004 (0.651)

0.004 (0.628)

-0.002 (0.735)

-0.002 (0.672)

0.009 (0.642)

0.013 (0.521)

Leverage

0.011 (0.266)

0.010 (0.277)

0.007 (0.554)

0.003 (0.793)

0.011 (0.490)

0.011 (0.504)

Dividend Yield

-0.008 (0.956)

-0.012 (0.941)

-0.078 (0.680)

-0.117 (0.551)

0.158 (0.488)

0.098 (0.673)

Transaction Characteristics Cash Payment 0.003 (0.281)

0.003 (0.299)

-0.002 (0.696)

-0.000 (0.933)

0.006 * (0.074)

0.011 *** (0.001)

Ln(Trans.Value)

-0.000 (0.764)

-0.000 (0.725)

-0.002 ** (0.049)

-0.002 * (0.053)

0.001 (0.636)

0.001 (0.753)

Private Target

0.020 *** (0.000)

0.020 *** (0.000)

0.021 *** (0.000)

0.021 *** (0.000)

0.019 *** (0.000)

0.020 *** (0.000)

Diversifying

0.000 (0.875)

0.000 (0.875)

-0.001 (0.791)

-0.000 (0.908)

0.000 (0.986)

-0.001 (0.726)

No 2,426 742 3.97%

No 2,426 742 4.08%

No 1,140 483 6.27%

Yes 1,140 483 8.36%

No 1,286 407 3.89%

Yes 1,286 407 4.94%

Year Fixed Effects Observations Firms R2

37   

High M/B Bidders (5) (6)

Table 7. Robustness: Incentive Exposure and CEO Characteristics OLS regressions with CEO incentive exposure*1,000 (Tumarkin, 2010) as the dependent variable. Independent variables are defined in Appendix B. Standard errors are clustered by firm and year, and p-values are in parentheses with ***, **, and * representing significance at the 1%, 5%, and 10% levels, respectively.

(1)

(2)

(3)

(4)

Pilot

-2,733.8 (0.286)

-2,702.7 (0.288)

-2,247.4 (0.302)

-2,230.4 (0.304)

Depression

-3,278.1 (0.438)

-1,633.2 (0.625)

-3,057.5 (0.462)

-1,250.4 (0.692)

Military

15,179.3 (0.301)

15,790.7 (0.299)

20,081.9 (0.277)

20,658.1 (0.276)

Age

-0.6 (0.983)

-1.2 (0.964)

73.8 (0.449)

68.9 (0.463)

Tenure

95.4 *** (0.009)

-48.5 (0.718)

135.7 ** (0.018)

-21.3 (0.871)

Leverage

3,859.1 (0.406)

4,300.8 (0.388)

128.5 (0.944)

565.8 (0.778)

Sales Growth

6,561.8 (0.322)

7,157.6 (0.315)

5,108.6 (0.355)

5,687.3 (0.344)

ROE

-503.1 (0.338)

-105.1 (0.702)

-682.0 (0.381)

-291.8 (0.576)

Ln(Assets)

300.0 (0.132)

231.7 (0.141)

962.9 (0.178)

836.2 (0.177)

Ln(Firm Age)

1,440.9 (0.324)

1,389.5 (0.329)

195.8 (0.691)

-7.1 (0.988)

Industry Fixed Effects

No

No

Yes

Yes

Year Fixed Effects

No

Yes

No

Yes

Observations

8,927

8,927

8,927

8,927

Firms

1,528

1,528

1,528

1,528

0.98%

1.08%

5.10%

5.18%

CEO Characteristics

Firm Characteristics

R

2

38