CEO Overconfidence and the Effects of Equity-Based Compensation on Strategic Risk-Taking in the U.S. Restaurant Industry

Abstract

The purpose of this study was to investigate (a) the moderating effect of CEO overconfidence on the relationship between equity-based compensation and strategic risk-taking and (b) the relationship between franchising and strategic risk-taking in the U.S. restaurant industry. Given wide use of a franchise system among U.S. restaurant firms, an understanding of the association between equity-based compensation and strategic risk-taking relative to CEOs’ risk behaviors seems particularly important. We conducted our empirical analysis in the U.S. restaurant industry using a sample of 659 firm-year observations from 1992 to 2013. Our findings showed that (a) overconfident CEOs, while holding equity-based compensation, tended to take on more strategically risky investments, and (b) there was a positive relation between franchising and risk-taking. Considering the behavioral and industry-specific characteristics, study findings could provide a more comprehensive understanding of how equity-based compensation influences strategic risk-taking in the U.S. restaurant industry.

Keywords

overconfidence equity-based compensation strategic risk-taking prospect theory loss aversion franchising

Introduction

Over the past few decades, equity-based compensation (EBC) has become a major component in CEO compensation. In general, EBC allows executives to benefit more based on their performance. According to a recent study, almost 50% of compensation for executives in S&P 500 companies were paid in the form of EBC (Hall & Murphy, 2002). For instance, in 2013, a CEO of Starbucks received about $163 million in compensation, including more than $128 million from stock options. Hence, significant academic and practical research attention has been directed toward investigating the implications of executive compensation plans: does EBC really work as designed? Given today’s dynamic business environments, however, it is unreasonable to assume that there is a single best practice that can be universally applied to all CEOs and all industries. Literature further suggests external factors need to be considered to better understand the influence of EBC on CEOs (Devers, Cannella, Reilly, & Yoder, 2007). Our primary object was, therefore, to provide a model that can accommodate the behavioral and industry-specific characteristics and address the challenging questions concerning the effect of EBC. In particular, in this article, we analyzed whether managerial overconfidence and franchising can help explain the relationship between EBC and risk-taking in the U.S. restaurant industry.

Without any real downside, EBC has been considered an efficient engine of risk-taking. Agency theory argues that CEOs tend to be more risk-averse because they are relatively underdiversified than shareholders (Eisenhardt, 1989). Shareholders, however, favor more risk-taking based on the risk-return tradeoff that greater risk must be taken to pursue greater returns (Fama, 1976; Sharpe, 1970). As an attempt to maximize their returns, hence, they generally prefer executives to take bigger risks. Agency theory predicts that EBC could align this divergence in risk preferences by encouraging more risk-taking by CEOs (Jensen & Meckling, 1976). In a managerial decision-making context, Sanders and Hambrick (2007) defined risk as “the degree to which potential outcomes associated with a decision are deemed consequential to the extent they have the potential to alter positively or negatively the health and vitality of the firm” (p. 7). One caveat, then, is risk-taking induced by EBC is not necessarily constructive. Nevertheless, agency theorists emphasized the importance of making larger strategic risk-taking (SRT) such as R&D investment, capital expenditures, and acquisitions to bring about extreme corporate performance (Core, Guay, & Larcker, 2003). Devers, McNamara, Wiseman, and Arrfelt (2008) argued that SRT reflects the degree to which CEOs engage in uncertain but potentially value-enhancing strategic actions.

The effect of EBC on SRT has not been investigated in the hospitality literature. There are unique business environments that separate the hospitality industry from other business sectors. These unique business environments provide industry-specific research opportunities. In particular, franchising has been widely used to support growth and expansion in the U.S. restaurant industry. A recent report indicated that the restaurant industry is the largest franchise business line with the largest output of $266.43 billion in the franchise sector in 2012 (IHS Global Insight, 2012). Given this wide use of a franchise system by U.S. restaurant firms, understanding the association between EBC and SRT relative to CEOs’ risk behavior seems particularly important. In this study, therefore, we focus on the restaurant industry to investigate whether the phenomenon could apply in this industry context, especially given its recent growth through franchising, an industry specific characteristic.

The restaurant industry has been identified as a high-risk industry (De Noble & Olsen, 1986; Huo & Kwansa, 1994; Parsa, Self, Njite, & King, 2005) and research shows that one of the main reasons for firms to adopt franchising is to share risk (Lafontaine & Bhattacharyya, 1995). Also documented in the literature is evidence that franchisors force relatively large operational and financial risk to franchisees in an attempt to avoid risk (Alon, Drtina, & Gilbert, 2004; Combs & Ketchen, 1999; Oxenfeldt & Kelly, 1968). For example, franchising allows the firm to franchise less profitable and riskier locations as a result of simple risk-averse behavior (Martin, 1988). Increased risk may lead to more franchising as risk-averse CEOs are more likely to shed risk onto franchisees (Lafontaine & Bhattacharyya, 1995; Martin, Wiseman, & Gomez-Mejia, 2013). This implies that franchising may be closely associated with executives’ risk-taking behavior. Thus, under the assumption that the franchisors are risk-averse (Lafontaine & Bhattacharyya, 1995), the tests of the empirical predictions on the relationship between EBC and SRT may provide important implications for the restaurant industry.

According to agency theory, executives can be compensated for their performance as a consequence of the strategic actions by linking executive pay directly to the company’s stock price (Sanders & Hambrick, 2007). Wiseman and Gomez-Mejia (1998), on the other hand, argued that the accumulated value of EBC will negatively influence SRT because CEOs perceive their wealth to be at risk of loss. That is, EBC can create more risk for CEOs as they have more to lose. As a result, the positive impact of EBC on SRT may gradually decrease as CEOs perceive high probable gains in the accumulated value of their EBC holdings. Empirical evidence on the effect EBC on SRT has also been somewhat inconsistent (Devers et al., 2008; Devers, Wiseman, & Holmes, 2007; Larraza-Kintana, Wiseman, Gomez-Mejia, & Welbourne, 2007; Lim, 2011; Sanders & Hambrick, 2007; Seo & Sharma, 2013; Wiseman & Gomez-Mejia, 1998). This study attempted to clarify these inconsistencies in the literature regarding the effect of EBC on SRT by incorporating the behavioral aspect of CEOs—overconfidence. As important corporate decisions are generally determined by CEOs, it is critical to consider the behavioral aspects of those who make a final call (Roll, 1986). There is strong support for the hypothesis that overconfidence may induce more risky investment decisions (Ben-David, Graham, & Harvey, 2007; Camerer & Lovallo, 1999; Heaton, 2002; Gervais, 2010; Gervais, Heaton, & Odean, 2011; Gervais & Odean, 2001; Goel & Thakor, 2008; Malmendier & Tate, 2005a, 2005b, 2008; March & Shapira, 1987; Roll, 1986).

We argue that overconfidence while interacting with EBC could encourage CEOs to make more strategically risky investments. We demonstrate this by examining the moderating effect of CEO overconfidence on the relationship between EBC and SRT: (a) when EBC negatively affects risk-taking, this negative effect of EBC is expected to be mitigated by overconfidence through further risk-taking, and (b) when EBC positively influences risk-taking, this positive effect of EBC is expected to be intensified by overconfidence. In this study, the former is framed as the gain domain and the latter is framed as the loss domain. Devers et al. (2008) argued that rising stock prices can create gain frames while falling stock prices can create loss frames. In the context of EBC, for example, gain domains refer to a situation where CEOs frame increase in the value of their EBC as a gain while loss domains refer to a situation where they frame decrease in the value of their EBC as a loss (Devers et al., 2008; Holmes, Bromiley, Devers, Holcomb, & McGuire, 2011; Lim, 2011).

The purpose of this study was to (a) investigate the moderating effect of CEO overconfidence on the relationship between EBC and SRT and (b) examine the relationship between franchising and SRT in the U.S. restaurant industry. We examined these relationships using a sample of 659 firm-year observations from 1992 to 2013. Contributions of this study are twofold. First, we advanced previous compensation studies by incorporating a behavioral aspect of CEOs’ overconfidence, into analysis of the relationship between EBC and SRT. This study provided a more comprehensive understanding of the effect of EBC on SRT, assuming CEOs can confront two different decision situations: gain and loss domains. The majority of the existing studies so far have explored the relationship between EBC and SRT in the domain of gain (Devers et al., 2008; Devers, Wiseman, et al., 2007; Larraza-Kintana et al., 2007; Lim, 2011; Sanders & Hambrick, 2007; Wiseman & Gomez-Mejia, 1998). We further extended agency theory predictions by recognizing the influence of CEO overconfidence on their risk behaviors associated with EBC. Our findings showed that restaurant firms can create more efficient incentive mechanisms by accounting for executives’ behavioral characteristics.

Second, focusing on the U.S. restaurant industry, our study explored the role of industry specific characteristic, franchising, in examining managerial risk behavior. Unlike our predictions, we found that franchising was positively related to risk-taking. This suggests that franchising might be considered an alternative device to motivate risk-taking. Our findings provided a more thorough understanding of how industry specific elements can influence managerial risk behavior.

The rest of the article is organized as follows. We review the evidence of the behavioral and industry-specific influences on risk-taking behavior. Next, the data and research design are introduced. Last, study findings are interpreted and discussed in detail.

Literature Review

Compensation Studies In Hospitality Literature

As agency theory proposed, this risk-aversion tendency can be alleviated using EBC. However, there has been little attempt to investigate the association between managerial risk behaviors, compensation, and strategic investments in the hospitality literature. Several studies have investigated the determinants of CEO cash compensation in the restaurant and casino firms (Dalbor, Oak, & Rowe, 2010; Gu & Choi, 2004; Guillet, Kucukusta, & Xiao, 2011; Kim & Gu, 2005). Other studies have focused on assessing the relationship between CEO compensation and firm performance (Barber, Ghiselli, & Deale, 2006; Dalbor et al., 2010; Madanoglu & Karadag, 2006; Madanoglu, Lee, & Castrogiovanni, 2011; Ozdemir, Kizildag, & Upneja, 2013). However, many scholars criticized that this pay-to-performance relationship is complex and ambiguous because performance is affected by various other factors (Tosi, Werner, Katz, & Gomez-Mejia, 2000; Yermack, 1997). Instead, recent studies adopted SRT as a more direct outcome of CEO compensation reflecting risk preferences of CEOs (Aggrawal & Samwick, 2003; Coles, Daniel, & Naveen, 2006; Devers et al., 2008; Lim, 2011; Sanders & Hambrick, 2007). They maintained that examining risk preferences of CEOs via SRT could provide a clearer understanding of how CEO compensation influences corporate policies. Therefore, this study proposes to examine the relationship between EBC and SRT in the context of prospect theory. Next, a broad overview of decision making under risk in traditional and prospect theory is discussed.

Conventional Approach to Agency Problems

Agency theory explains agency problems by offering the opportunity to understand the risk preferences of managers and shareholders (Berle & Means, 1932; Jensen & Meckling, 1976; Jensen & Murphy, 1991; Tosi, Katz, & Gomez-Mejia, 1997). Agency theory assumes the divergence of the risk preferences between shareholders and managers with respect to investment decisions (Devers, Cannella, et al. 2007). In particular, shareholders are assumed to be risk-neutral as they are widely diversified while executives are assumed to be risk-averse as their personal wealth is mainly tied to their respective firms (Jensen & Meckling, 1976; Milgrom & Roberts, 1992). This implies that shareholders are more likely to take greater risks when making investment decisions based on the positive risk-return correlation that large risks must be taken to obtain large gains (Sanders & Hambrick, 2007). Executives, on the other hand, might not act in the best interests of shareholders and attempt to avoid risk at the expense of returns (Eisenhardt, 1989; Wiseman & Gomez-Mejia, 1998). Prior studies have discussed ways to align the risk preferences relative to specific investment choices such as mergers (Bliss & Rosen, 2001), acquisitions (Datta, Iskandar-Datta, & Raman, 2001; Sanders, 2001), and the riskiness of policy choices (Coles et al., 2006; Makri, Lane, & Gomez-Mejia, 2006; Wright, Kroll, Krugg, & Pettus, 2007).

EBC such as stock options has been argued to be an efficient means to align these risk preferences between managers and shareholders (Devers, Wiseman, et al., 2007; Jensen & Meckling, 1976; Wiseman & Gomez-Mejia, 1998). While agency theorists contend that EBC could effectively reduce agency problems, empirical evidence still remains ambiguous (Devers et al., 2008; Kim & Gu, 2005; Lim, 2011; Rajgopal & Shevlin, 2002; Sanders, 2001; Sanders & Hambrick, 2007; Seo & Sharma, 2013). Given that managers with high levels of EBC holdings benefit from increased firm value, they are expected to make large investments on uncertain projects in anticipation of dramatically boosting firm value (Sanders & Hambrick, 2007). However, studies showed that some EBC failed to align risk preferences by exacerbating risk aversion among CEOs (Devers et al., 2008; Devers, Wiseman, et al., 2007; Rajgopal & Shevlin, 2002; Sanders & Hambrick, 2007; Seo & Sharma, 2013; Wiseman & Gomez-Mejia, 1998).

For example, Devers et al. (2008) investigated the effects of different forms of EBC granted to CEOs on SRT. They argued that CEOs are more likely to make suboptimal investment decisions by avoiding excessive risk-taking that might pose a threat to their personal wealth. The mixed findings of the extant research suggest that an alternative approach may be required to better explain the relationship between EBC and SRT. In particular, Roll (1986) emphasized the importance of CEOs’ behavioral styles and characteristics in corporate behaviors. Some scholars maintain that the agency theory approach fails to take such behavioral aspects into consideration in explaining corporate decision-making processes due to its restrictive and simplistic assumptions (Bertrand & Schoar, 2003; Coffee, 1988; Wiseman & Bromiley, 1996).

Behavioral Approach to Aligning Risk Preferences

Drawing on the prospect theory developed by Kahneman and Tversky (1979), the behavioral agency model (BAM) explains how CEOs’ risk-taking decisions depart from predictions of normative models such as agency theory. There are several fundamental differences between prospect theory and agency theory.

First, prospect theory posits that individuals are loss-averse while agency theory assumes consistent risk aversion (Kahneman & Tversky, 1979). Loss aversion is defined as the tendency for individuals to prefer avoiding losses to maximizing gains (Kahneman & Tversky, 1979; Tversky & Kahneman, 1992). This is because the magnitude of displeasure from losses is argued to be greater than the same magnitude of pleasure from gains (Hastie & Dawes, 2001). BAM also allows changes in risk preferences relative to some neutral reference point (Wiseman & Gomez-Mejia, 1998). Individuals may change their risk-taking behaviors across different decision situations. BAM argues that individuals are risk-averse in the gain domain but risk-seeking in the loss domain (Hastie & Dawes, 2001; Kahneman & Tversky, 1979). The gain domain refers to situations in which an individual anticipates an outcome to exceed one’s reference point while the loss domain refers to situations in which an individual anticipates an outcome to be below one’s reference point. The shift in risk preferences between gain and loss domains occurs because individuals are more sensitive to losses than gains (Tversky & Kahneman, 1986). For example, loss aversion leads CEOs to become less risk-seeking to avoid large wealth losses in the gain context. However, they may behave in a risk-seeking fashion in the loss context in an attempt to avoid an immediate loss (Tversky & Kahneman, 1986).

Second, value is defined as deviations from a reference point in prospect theory whereas utility is only contingent on final wealth in agency theory (Kahneman & Tversky, 1979). Thus, individuals perceive outcomes as gains and losses relative to some neutral reference point rather than final states of wealth (Thaler, 1980; Tvede, 1999; Tversky & Kahneman, 1974, 1992). Wiseman and Gomez-Mejia (1998) argued that executives’ compensation packages can create reference points for them. For example, CEOs can use the exercise price of their stock options as a reference point. This implies that CEOs will adopt gain domains when the price of the stock underlying their stock options moves beyond the exercise price resulting in increase in the value of their stock options (Devers, Wiseman, et al., 2007; Holmes et al., 2011; Larraza-Kintana et al., 2007; Lim, 2011; Zhang, Bartol, Smith, Pfarrer, & Khanin, 2008). On the other hand, they will adopt loss frames when the price of the stock underlying options moves closer to the exercise price causing decrease in the value of their stock options. Holmes et al. (2011) argued that declining stock prices can elicit loss frames for CEOs even if it only reduces gains in their stock options. Unlike stock options, however, restricted stock does not have an exercise price. CEOs holding restricted stocks may use the “status quo (current wealth)” as the reference point (Kahneman, 2003). That is, when a firm performs above (below) its past performance CEOs will frame the resulting accumulated value of their restricted stock as a gain (loss). Holmes et al. (2011) maintained that CEOs can use multiple reference points which can change over time.

Prior compensation studies have mainly investigated the effect of EBC on SRT in the gain domain. Their findings partially supported BAM arguments that EBC may not promote excessive risk-taking in the gain domain where CEOs perceive significant increases in the accumulated value of their EBC holdings (Bryan, Hwang, & Lilen, 2000; Devers et al., 2008; Devers, Wiseman, et al., 2007; Larraza-Kintana et al., 2007; Lim, 2011; Parrino, Poteshman, & Weinbach, 2005; Sanders, 2001; Sanders & Hambrick, 2007; Seo & Sharma, 2013; Wiseman & Gomez-Mejia, 1998; Wright et al., 2007). In particular, Lim (2011) found an inverted U-shaped relationship between in-the-money unexercisable stock options¹ and SRT. She argued that in-the-money unexercisable stock options will provide incentives for CEOs to engage in risky investments only up to the point where doing so creates more concerns about protecting their wealth in their option holding (Devers et el., 2008; Sanders, 2001; Sanders & Hambrick, 2007; Wright et al., 2007). After this point, CEOs will become heavily endowed with the value of these stock options and prefer less risky strategies to prevent losses. However, other studies found a positive linear relationship (Devers et al., 2008) and a negative linear relationship (Larraza-Kintana et al., 2007) between the accumulated value of in-the-money unexercisable stock options and SRT.

Similarly, Devers et al. (2008) showed a curvilinear relationship between the accumulated value of in-the-money exercisable stock options and SRT. Assuming CEOs are loss averse, they argued that low to moderate levels of accumulated value of exercisable stock options will lead to greater risk-seeking because they have less to lose. However, significantly highly valued exercisable options will promote risk-aversion as they attempt to avoid placing their personal wealth at risk of loss. In other words, CEOs are more likely to pursue low-risk strategies to reduce the chances of losing their personal wealth (Devers et al., 2008; Wright et al., 2007). Lim (2011), however, found that the accumulated value of in-the-money exercisable stock options is positively associated with SRT. She argued that accumulated gains of exercisable stock options reflect strong firm performance over time and successful CEOs tend to be less-risk averse.

In a recent investigation of CEOs in U.S. restaurant firms, Seo and Sharma (2013) found an inverse U-shaped relationship between the accumulated value of in-the-money exercisable stock options and SRT. This suggests that the positive effect of exercisable stock options on SRT increases only at a decreasing rate. As the accumulated value of these options increases CEOs become more sensitive to losses and start to diminish SRT in an attempt to protect their wealth.

Other researchers further examined the effect of EBC using restricted stock (Devers et al., 2008; Lim, 2011). BAM argues that the values of different types of EBC may not be equally present in calculations of prospective wealth (Devers et al., 2008; Larraza-Kintana et al., 2007; Lim, 2011; Martin et al., 2013). For example, unlike stock options, the value of restricted stock could be endowed into perceived current wealth instantly on granting, allowing CEOs to be exposed to downside risk.² Ofek and Yermack (2000) posited that the risk properties of restricted stocks resemble those of equity ownership. Equity ownership generally promotes more careful risk-taking by CEOs because it involves both upside and downside risk.

In a similar vein, CEOs holding restricted stocks might be more cautious about risk-taking as they need to bear downside risk (Parrino et al., 2005). Therefore, the negative impact of positively valued restricted stock on SRT is expected to be much stronger than that of stock options in the gain domain. This implies that restricted stock will exacerbate risk-aversion rather than providing risk-taking incentives (Bryan et al., 2000). Devers et al. (2008) found a negative linear relationship between restricted stock and SRT, suggesting the level of risk aversion aggravated as the accumulated value of restricted stock increased.

Although previous research explored the links between different forms of EBC and SRT, their associated relationships are still unclear (Devers et al., 2008; Lim, 2011). This study, hence, examined these relationships while taking into consideration CEO overconfidence to develop a better understanding of the effect of EBC on SRT.

CEO Overconfidence and Risk Behaviors

BAM has also drawn attention to overconfidence as an important behavioral factor that might influence the risk behaviors of CEOs. Overconfidence refers to individuals’ tendency to believe their knowledge is more accurate than it actually is (Langer, 1975; March & Shapira, 1987). It is argued that highly confident CEOs tend to be more aggressive in investments because they overestimate returns on risky investments and/or underestimate the probability of failure (Baker, Ruback, & Wurgler, 2007; Heaton, 2002; March & Shapira, 1987; Malmendier & Tate, 2005a, 2005b, 2008; Roll, 1986). Consistent with this view, recent empirical studies showed that overconfidence is positively associated with increases in risk-taking with respect to corporate financing and investment policies such as capital structure, R&D expenses, acquisition, leverage, and growth opportunities (Ben-David et al., 2007; Gervais, 2010; Gervais & Odean, 2001; Goel & Thakor, 2008; Hackbarth, 2008; Heaton, 2002; Malmendier & Tate, 2005a, 2008; Roll, 1986).

This provides important implications for the extant compensation research in that the effects of EBC on SRT may vary depending on a CEO’s level of confidence. In theoretical studies, Gervais et al. (2011) showed that firm investment decisions are ultimately driven by the interaction of EBC and overconfidence. They argued that greater risk-taking tendency induced by overconfidence may lead CEOs to choose risky investment projects even when EBC fails to promote their appetite for risk. Goel and Thakor (2008) also predicted that overconfidence may help mitigate the negative effects of the executives’ risk aversion on investments. This suggests the effect of performance-based compensation on SRT needs to be understood with discussion of CEO overconfidence (Gervais et al., 2011).

Despite aforementioned theoretical predictions, however, empirical evidence on the effect of overconfidence in the agency relationship context is scarce. Several empirical studies documented that CEO overconfidence and their compensation may jointly influence strategic investment decisions (Brown & Sarma, 2007; Devers et al., 2008; Lim, 2011). These studies suggested that high levels of the accumulated value of EBC may drive CEO confidence, in turn promoting risk-seeking behaviors (Devers et al., 2008; Lim, 2011). For example, stock option gains accumulated from their success can enhance CEO confidence (Hayward & Hambrick, 1997). High levels of confidence may lead them to believe that they can increase the firm’s value by engaging in risky activities such as SRT (Sanders & Hambrick, 2007). Therefore, it was argued that increased confidence resulting from strong firm performance was partially responsible for SRT (Devers et al., 2008; Lim, 2011). However, the impact of CEO overconfidence on SRT was not directly tested in these studies. It still remains unclear whether overconfidence will induce CEOs to become greater risk seekers.

This dynamic interaction between overconfidence and EBC provides a great opportunity to contribute to the existing literature. Prior compensation studies have mainly looked at the relationship between EBC and SRT (Devers et al., 2008; Devers, Wiseman, et al., 2007; Lim, 2011; Sanders & Hambrick, 2007; Seo & Sharma, 2013; Wiseman & Gomez-Mejia, 1998). This study advanced the extant literature in examining the moderating effects of overconfidence on the relationship between EBC and SRT. Although a theoretical model by Gervais et al. (2011) discussed the influence of the interaction between CEO compensation and overconfidence and corporate policies, they analyzed this in a setting in which managers were constantly risk-averse. We extended their discussion by allowing for changes in managers’ risk preferences between the domain of gains and losses.

The Moderating Effect of Overconfidence in the Gain Domain

Consistent with previous BAM studies, different types of EBC were analyzed in this study: stock option and restricted stock. Full details about EBC are provided in Appendix A. Despite inconclusive evidence, most extant BAM studies argue that CEOs will pursue lower-risk strategies in the face of significant increases in the accumulated value of their EBC holdings (Devers et al., 2008; Devers, Wiseman, et al., 2007; Lim, 2011; Sanders & Hambrick, 2007; Seo & Sharma, 2013; Wiseman & Gomez-Mejia, 1998).

In particular, we expected the relationships between the accumulated value of stock options and SRT to be inverted curvilinear. The positive effect of the accumulated value of stock option on SRT will increase only up to a point after which CEOs become more concerned about losing their wealth (Martin et al., 2013). We further predicted that overconfidence encourages CEOs to take greater risks after this point where highly valued stock options start to exacerbate risk aversion bias (Devers, Wiseman, et al., 2007; Lim, 2011). Specifically, the negative effect of stock options on SRT on the right hand side of the inverted U-shaped curve will be reduced (positively moderated) by CEO overconfidence. In addition, we expected to find a negative linear relationship between the accumulated value of restricted stock and SRT. BAM suggests that CEOs will become risk-averse because they face downside risk. We predicted that this negative relationship will be mitigated (positively moderated) by CEO overconfidence. The following hypotheses were proposed:

Hypothesis 1a: In the gain domain, overconfidence will positively moderate the inverted U-shaped relationship between the accumulated value of in-the-money unexercisable stock options and SRT, such that overconfident CEOs will become less risk-averse while holding highly valued in-the-money unexercisable stock options.

Hypothesis 1b: In the gain domain, overconfidence will positively moderate the inverted U-shaped relationship between the accumulated value of in-the-money exercisable stock options and SRT, such that overconfident CEOs will become less risk-averse while holding highly valued in-the-money exercisable stock options.

Hypothesis 1c: In the gain domain, overconfidence will mitigate the negative effect of the accumulated value of restricted stock on SRT.

The Moderating Effect of Overconfidence in the Loss Domain

BAM argues that CEOs who have accumulated their EBC over long periods of time may incorporate the accumulated value of these stock options into their personal wealth (Wiseman & Gomez-Mejia, 1998). This suggests that they may perceive losses in their personal wealth when the accumulated value of their EBC decreases (Holmes et al., 2011). For instance, decreased stock prices will effectively create loss frames for CEOs who are holding positively valued stock options. Although BAM predicts a positive association between EBC and SRT in the loss domain, empirical evidence has been very limited.

Several experimental studies have shown evidence for loss-averse behaviors in the loss context. In an experiment with 292 MBA students, Sawers Wright, and Zamora (2006) showed that more participants selected risky projects over riskless projects in the loss domain. Devers, Wiseman, et al. (2007) studied how managers’ risk preferences change in response to their subjective option valuations. The findings of their experiments revealed that managers exhibit risk-seeking behavior in the domain of losses in attempts to reduce the prospect of loss. In the context of CEO compensation, CEOs will engage in risky actions under the loss condition, such as earnings manipulation (Zhang et al., 2008) and SRT (Seo & Sharma, 2013). We extended these findings and predicted that the positive impact of EBC on SRT in the loss domain would be augmented by overconfidence. As a result, overconfident CEOs will continue to increase SRT even when they perceive high levels of loss in the accumulated value of EBC. We proposed the following hypotheses:

Hypothesis 2a: In the loss domain, overconfidence will intensify the positive effect of the perceived loss in the accumulated value of in-the-money unexercisable stock options on SRT.

Hypothesis 2b: In the loss domain, overconfidence will intensify the positive effect of the perceived loss in the accumulated value of in-the-money exercisable stock options on SRT.

Hypothesis 2c: In the loss domain, overconfidence will intensify the positive effect of the perceived loss in the accumulated value of restricted stock on SRT.

Franchising and Risk-Behavior

Restaurant companies are well-known for using a franchise system to support growth. Franchise systems provide firms with an efficient mechanism for expansion at a relatively low risk (Combs & Ketchen, 1997, 1999; Hsu, Jang, & Canter, 2010; Roh, 1998; Sul & Khan, 2006). According to a recent report by PriceWaterhouseCoopers (2011), restaurants have the largest number of franchise establishments than any other single business line. Several scholars noted the importance of a franchisor’s entrepreneurial strategy and that the entire franchise organization is dependent on the managerial decisions of the franchisor (Harrington, 2005; Khan, 1999; Morrison & Macmillan, 2000; Sul & Khan, 2006). This suggests that the behaviors of the franchisors may play a critical role in strategic activities of the restaurant firms (Morrison, 2000; Sul & Khan, 2006). Some of the major franchise strategies include investments in aggressive marketing and expansion of units (Roh & Kwag, 1997; Sul & Khan, 2006). Hence, understanding managerial behaviors in franchising restaurants can provide a basis for the analysis of strategic investments.

In particular, Rubin (1978) argued that franchising may induce a higher tendency for risk aversion among franchisors. Risk-sharing theory supports this view, positing that franchising contracts arise mainly because franchisors seek to share their investment and capital risks with franchisees (Combs & Castrogiovanni, 1994; Koh, Lee, & Boo, 2009; Martin, 1988; Roh, 2002; Rubin, 1978). Franchising allows franchisors to reduce uncertainties by spreading risks across multiple units. Lafontaine and Bhattacharyya (1995) also postulate that risk sharing is the primary drive for franchising. For instance, franchisors can significantly reduce risks by having franchisees operate in relatively risky locations with uncertain revenues (Combs & Castrogiovanni, 1994). Hence, we predict that the level of risk-taking will decrease as the level of franchising grows. To test this relationship between franchising and SRT in our behavioral model, the following hypotheses were proposed:

Hypothesis 3a: In the gain domain, there is a negative relationship between franchising and SRT.

Hypothesis 3b: In the loss domain, there is a negative relationship between franchising and SRT.

Method

We specifically focused on the restaurant industry because it is one of the few franchising industries that can provide a meaningful sample for empirical analysis. Researchers also suggested that focusing on a single sector could enhance the internal validity because different franchising industries have different characteristics (Alon, 1999; Alon, Drtina, & Gilbert, 2004; Dant, Paswan, & Stanworth, 1996). Data for publicly traded U.S. restaurant firms were collected from Compustat, CRSP, and Execucomp between 1992 and 2013. In addition, proxy statements that offered information about CEO stock option packages were collected. The final sample for this study included 659 firm year observations of 45 U.S. restaurant firms.

To control for heteroskedasticity and autocorrelation in the panel data models, we used fixed-effects regression models with standard errors clustered by firm and year (Gow, Ormazabal, & Taylor, 2010; Thompson, 2011). The fixed-effects model uses variation between overconfident and non-overconfident CEOs within firms to estimate the moderating effect of overconfidence on the relationship between EBC and SRT.³ Including the fixed-effects parameters could effectively partial out unobserved fixed-effects while clustering could correct for any remaining within group correlation. For each regression model, the Hausman specification test (1978) was conducted to confirm whether the use of the fixed-effects model was appropriate.

To test for a relationship between EBC and SRT, moderated by CEO overconfidence, this study used multiple interaction and quadratic terms. Aiken and West (1991) argued that a regression equation with interaction and/or quadratic terms suffer from potential multicollinearity problems. Multicollinearity generally inflates variance among variables while actually adding very little independent information to the model (Belsley, Kuh, & Welsch, 1980). Therefore, we mean-centered relative variables before creating interaction and quadratic terms. Many studies have demonstrated that centered variables produce low intercorrelations (Aiken & West, 1991; Dawson & Richter, 2006; Tabachnick & Fidell, 1996). Finally, outliers were removed using the three standard deviation cutoff for standard errors.

The following regression model was used in the domain of gain to test Hypotheses 1a to 1c. This lagged data design allows independent variables temporal precedence over dependent variable.

\begin{array}{l} S R T_{i, t} = β_{0} + β_{1} U N E X_{i, t - 1} + β_{2} E X_{i, t - 1} + β_{3} {(U N E X_{i, t - 1})}^{2} + β_{4} {(E X_{i, t - 1})}^{2} + β_{5} R S_{i, t - 1} + β_{6} O C_{i, t - 1} + \\ β_{7} (U N E X_{i, t - 1} O C_{i, t - 1}) + β_{8} (E X_{i, t - 1} O C_{i, t - 1}) + β_{9} (R S_{i, t - 1} O C_{i, t - 1}) + β_{10} [{(U N E X_{i, t - 1})}^{2} O C_{i, t - 1}] + \\ β_{11} [{(E X_{i, t - 1})}^{2} O C_{i, t - 1}] + β_{12} F R A N_{i, t - 1} + β_{13} S I Z E_{i, t - 1} + β_{14} Q_{i, t - 1} + β_{15} T E N_{i, t - 1} + β_{16} A G E_{i} + β_{17} G E N_{i} + τ_{i} + ε_{i, t}, \end{array}

where SRT is strategic risk-taking, UNEX is the accumulated value of unexercisable stock options, EX is the accumulated value of exercisable stock options, (UNEX)² is the accumulated value of unexercisable stock options squared, (EX)² is the accumulated value of exercisable stock options squared, RS is the accumulated value of restricted stock, FRAN is the degree of franchising, OC is CEO overconfidence (=1 if overconfident, =0 otherwise), SIZE is firm size, Q is Tobin’s q, TEN is the number of years employed as CEO, AGE is CEO’s age, GEN is gender of a CEO, τ_i is an unobserved, time-invariant component which contains firm-specific factors, and ϵ_i,t is an idiosyncratic error term that is uncorrelated with all explanatory variables in the model. The quadratic terms were included in the model to test for a curvilinear relationship between SRT and the independent variables of interest.

In the domain of loss, the following regression model was analyzed to test Hypotheses 2a to 2c.

\begin{array}{l} S R T_{i, t} = λ_{0} + λ_{1} P U N E X_{i, t - 1} + λ_{2} P E X_{i, t - 1} + λ_{3} P R S_{i, t - 1} + λ_{4} O C_{i, t - 1} + λ_{5} (P U N E X_{i, t - 1} O C_{i, t - 1}) \\ + λ_{6} (P E X_{i, t - 1} O C_{i, t - 1}) + λ_{7} (P R S_{i, t - 1} O C_{i, t - 1}) + λ_{8} F R A N_{i, t - 1} + λ_{9} S I Z E_{i, t - 1} + λ_{10} Q_{i, t - 1} \\ + λ_{11} T E N_{i, t - 1} + λ_{12} A G E_{i} + λ_{13} G E N_{i} + η_{i} + ω_{i, t}, \end{array}

where SRT is strategic risk-taking, PUNEX is the perceived loss in the accumulated value of unexercisable stock options, PEX is the perceived loss in the accumulated value of exercisable stock options, PRS is the perceived loss in the accumulated value of restricted stock, OC is CEO overconfidence (=1 if overconfident, =0 otherwise), FRAN is the degree of franchising, SIZE is firm size, Q is Tobin’s q, TEN is the number of years employed as CEO, AGE is CEO’s age, GEN is gender of a CEO, η_i is an unobserved, time-invariant component which contains firm-specific factors, and ω_i,t is an idiosyncratic error term that is uncorrelated with all explanatory variables in the model.

Dependent Variable

Consistent with the previous studies, we measured SRT as the sum of R&D investment, capital investment, acquisition investment, and long-term debt (Beckman & Haunschild, 2002; Devers et al., 2008; Hoskisson, Hitt, & Hill, 1993; Larcker, 1983; Sanders & Hambrick, 2007). R&D investment is the total annual expense spent on research and development⁴; capital investment is the total capital expenditures of the company; acquisition investment is the total transaction values for all acquisitions during the year; long-term debt is a long-term debt divided by the total assets. Following prior studies (Lim, 2011; Sanders & Hambrick, 2007), each variable was log-transformed before they are summed. This helps equal weighing with respect to importance among different constructs in the dependent variable (Welbourne & Andrews, 1996).

Independent Variables

Our main independent variable was CEO overconfidence. Following the study by Malmendier and Tate (2005a, 2005b, 2008), we developed an overconfidence measure based on the personal portfolio decisions of CEOs using their stock option exercise information. The payoff of the stock option is determined as the difference between the exercise price and the current market price of underlying stock. CEOs, therefore, should exercise their stock options only when this difference is positive. In general, they instantly resell the shares acquired by stock option exercise (Malmendier & Tate, 2008; Ofek & Yermack, 2000). This is because unlike outside investors, executives cannot hedge the risk of their stock option holdings (Hall & Murphy, 2002). For example, CEOs are not allowed to (a) freely trade, sell, or short-sell company stock and (b) diversify their investment portfolios by purchasing shares of stock in a large number of companies. In addition, they may perceive higher firm risk as their personal wealth and human capital are mostly invested in their firms (Malmendier & Tate, 2005a, 2005b). Many scholars maintain that risk-averse CEOs exercise their stock options early to hedge against the risk of holding company stock (Gervais, 2010; Hall & Murphy, 2002; Heaton, 2002; Malmendier & Tate, 2005a, 2005b, 2008).

However, this logic does not explain why some executives persistently fail to exercise their options when the payoff is significantly high. Malmendier and Tate (2005a, 2005b, 2008) argued that overconfidence can provide alternative explanations for the failure to exercise positively valued stock options. For instance, overly positive prospects about future returns lead CEOs to believe that the payoff of their stock options will grow as the prices of underlying stocks will continue to rise. Hence, they may choose to hold onto their stock options even when their stock options are positively valued. That is, overconfident CEOs will continue to hold their stock options beyond rational thresholds for exercise (Malmendier & Tate, 2005a, 2005b, 2008). Malmendier and Tate (2005a) developed the rational threshold for exercise based on the Hall and Murphy model (2002). Using this threshold, this study constructed a measure for overconfidence: a dummy variable that equals one if the CEO fails to exercise the 67% in-the-money stock option with 5 years remaining before expiration.

BAM argues that the values of different types of EBC may not be equally perceived by CEOs (Devers et al., 2008; Lim, 2011). Hence, this study included three types of EBC: unexercisable stock option, exercisable stock option, and restricted stock. To measure the accumulated value of stock options, we used data for the aggregate value of in-the-money stock options that are vested (exercisable) and not vested (unexercisable) at the fiscal year end. The aggregate value is the difference between the product of option exercise price and the number of option shares and the market value of option shares on the last day of the fiscal year. The accumulated value of restricted stock was measured as the aggregate value of restricted stock at the fiscal year end. This value represents the market value based on the closing market price of the company’s common stock on the last day of the fiscal year. All EBC variables were scaled by total compensation. Total compensation is the sum of the annual salary, cash bonus, other annual compensation, restricted stock, stock options, long-term incentive payouts, and all other compensation.

Under the decreasing stock price trends, we argued that CEOs will frame a reduction in gains in the accumulated value of their EBC holdings as a loss (Holmes et al., 2011). This study developed a proxy for CEO’s perceived loss by comparing the accumulated value of EBC in the current year with the accumulated value in the previous year. The perceived loss is a difference between the value of the current year and the value of the previous year divided by the previous year’s value. We confirmed that the perceived loss was resulted from decreasing stock price trends by examining the quarterly stock prices, high and low prices during the period that the accumulated values of EBC decreased.

As noted above, CEOs in the restaurant industry may use franchising as a means to minimize the variance of investment risks (Lafontaine & Bhattacharyya, 1995). In this case, we predict that CEOs’ risk-taking behavior may be closely related to the level of franchising employed. In particular, we expect a higher tendency of risk aversion among franchisors and such tendency should be stronger in firms with high levels of franchising. The ratio of the number of franchise units to the total number of units was used as a proxy for a restaurant’s degree of franchising in a given year.

Control Variables

We included several variables to control for firm and CEO characteristics. To control for firm characteristics, this study used firm size and firm performance. Previous studies argued that CEOs in larger and profitable firms are more likely to take greater risk (Bolton, Chen, & Wang, 2011; Sanders, 2001; Sanders & Hambrick, 2007; Wiseman & Gomez-Mejia, 1997; Wright et al., 2007). Firm size was measured as the natural log of assets while Tobin’s q was used as a proxy for firm performance. In addition, CEO characteristics such as age, gender, and tenure were controlled. CEO tenure is referred to as the total number of years he/she has been employed as a CEO.

Results

The findings from analyses are presented in this section. First, the descriptive statistics are discussed. Next, the results of the multiple regressions to test hypotheses are provided.

Descriptive Statistics

The descriptive statistics on variables in our models are shown in Table 1. The average tenure of our sampled CEOs was 13.94 years. This relatively long tenure indicates that they might be holding a large amount of positively valued exercisable stock options. Indeed, on average, the accumulated value of exercisable stock options was $5.26 million whereas the accumulated value of unexercisable stock options was $2.08 million. We predicted that, if not overconfident, these CEOs with positively valued stock options would be more risk-averse to protect personal wealth attached to stock options.

Table 1

Summary of Descriptive Statistics

	M	Mdn	SD	Min	Max
SRT_t	3.77	3.65	2.26	0	12.77
UNEX_t-1	0.46	0.56	1.43	0	28.99
EX_t-1	1.59	0.87	3.25	0	33.19
RS_t-1	0.17	0	0.56	0	5.58
PUNEX_t-1	0.28	0	0.42	0	2.86
PEX_t-1	0.24	0	0.41	0	5.29
PRS_t-1	0.07	0	0.22	0	1
FRAN_t-1	0.35	0.21	0.33	0	1
Ln(ASSET_t-1)	5.94	6.08	1.62	1.88	10.47
Q_t-1	2.02	1.70	1.15	0.40	8.72
TEN_t-1	13.94	11	7.25	1	32
AGE_t	54.10	54	8.45	31	78
GEN	0.07	0	0.25	0	1
OC	0.50	1	0.50	0	1

Note: SRT_t-1 is strategic risk taking, constructed by the sum of the log of R&D investment, capital investment, acquisition investment, and long-term debt. UNEX_t-1 is the accumulated value of unexercisable stock options. EX_t-1 is the accumulated value of exercisable stock options. RS_t-1 is the accumulated value of restricted stock. PUNEX_t-1 is the perceived loss in the accumulated value of unexercisable stock options. PEX_t-1 is the perceived loss in the accumulated value of exercisable stock options. PRS_t-1 is the perceived loss in the accumulated value of restricted stock. The perceived loss variables are presented in percentage as a difference between the value in the current year t and the value in the previous year t − 1 divided by the value in the previous year t − 1. FRAN_t-1 is the degree of franchising, measured by the ratio of the number of franchise units to total units. ASSET_t-1 is firm assets. ROA_t-1 is defined as net income divided by total assets measuring firm performance. Q_t-1 is the market value of assets over the book value of assets. TEN_t-1 is the number of years employed as CEO. AGE_t is CEO’s age. GEN is the gender of a CEO (=0 if male, =1 female). OC is CEO overconfidence (=1 if overconfident, =0 otherwise).

Table 2 shows the results of the Pearson’s correlation analysis. As agency theory argues, the accumulated values of three types of EBC, unexercisable and exercisable stock options, and restricted stock, were positively correlated with SRT in the gain domain. The perceived loss in the accumulated values of two types of EBC, unexercisable stock option, and restricted stock, were also positively associated with SRT in the loss domain. Furthermore, the correlation between overconfidence and SRT was significant and positive. This is consistent with our arguments that overconfident CEOs will be more likely to pursue risky investments.

Table 2

Summary of Pearson Correlations

Variable	N	SRT_t	UNEX_t-1	EX_t-1	RS_t-1	PUNEX_t-1	PEX_t-1	PRS_t-1	FRAN_t-1	SIZE_t-1	Q_t-1	TEN_t-1	AGE_t	GEN	OC
SRT_t	653	1.00
UNEX_t-1	658	0.05	1.00
EX_t-1	658	0.08*	0.16**	1.00
RS_t-1	658	0.25**	0.05	−0.03	1.00
PUNEX_t-1	612	0.05	−0.14**	−0.01	−0.00	1.00
PEX_t-1	612	−0.02	−0.02	−0.13**	−0.00	0.40**	1.00
PRS_t-1	612	0.14**	−0.02	−0.05	−0.02	0.10*	0.12**	1.00
FRAN_t-1	655	0.08*	0.05	0.10*	−0.13**	0.01	0.01	−0.03	1.00
SIZE_t-1	640	0.82**	0.08*	0.02	0.20**	0.04	−0.03	0.14*	−0.08*	1.00
Q_t-1	612	0.09*	0.32**	0.18**	0.19**	−0.10*	−0.08*	−0.06	0.11*	0.15**	1.00
TEN_t-1	658	−0.20**	−0.11**	−0.02	−0.09*	−0.05	−0.04	−0.10*	−0.27**	−0.19*	−0.11**	1.00
AGE_t	658	0.19**	−0.04	0.00	−0.05	0.03	0.02	0.08*	−0.13*	0.27**	−0.07*	0.16**	1.00
GEN	658	−0.05	−0.01	−0.05	0.03	0.02	0.02	0.02	0.08*	0.25**	0.04	−0.04**	−0.17**	1.00
OC	658	0.28**	−0.10*	−0.11	−0.07	−0.02	−0.02	−0.02	0.10*	−0.26**	−0.27*	0.12*	−0.08*	0.07	1.00

Note: SRT_t-1 is strategic risk taking, constructed by the sum of the log of R&D investment, capital investment, acquisition investment, and long-term debt. UNEX_t-1 is the accumulated value of unexercisable stock options. EX_t-1 is the accumulated value of exercisable stock options. RS_t-1 is the accumulated value of restricted stock. PUNEX_t-1 is the perceived loss in the accumulated value of unexercisable stock options. PEX_t-1 is the perceived loss in the accumulated value of exercisable stock options. PRS_t-1 is the perceived loss in the accumulated value of restricted stock. The perceived loss variables are presented in percentage as a difference between the value in the current year t and the value in the previous year t − 1 divided by the value in the previous year t − 1. FRAN_t-1 is the degree of franchising, measured by the ratio of the number of franchise units to total units. SIZE_t-1 is the natural log of assets at the beginning of the year. Q_t-1 is the market value of assets over the book value of assets. TEN_t-1 is the number of years employed as CEO. AGE_t is CEO’s age. GEN is the gender of a CEO (=0 if male, =1 female). OC is CEO overconfidence (=1 if overconfident, =0 otherwise).

Significant at .05. **Significant at .01.

Empirical Findings

Overconfident CEOs in the Gain Domain

Regression models to test the moderating effect of OC on the relationship between EBC and SRT in the gain domain are displayed in Table 3. We first performed an OLS regression with the full model (Column 1). Analyzing the variation inflation factors from this regression confirmed the lack of significant multicollinearity issues. Next, we ran two fixed-effects regressions with standard errors clustered by firm and year. The regression model in Column 2 excludes the interaction terms and shows the main effects of EBC on SRT in the gain domain. Consistent with BAM arguments, we found an inverse U-shaped relationship between unexercisable stock options and SRT. The unexercisable stock options variable had a significantly positive coefficient (t = 2.51; p < .05) and its squared term was significantly negative (t = −2.55; p < .05). Although not statistically significant, the coefficient of the exercisable stock options variable was positive while the coefficient of its quadratic term was negative. These findings indicate that the positive effect of stock options may be limited in that SRT diminishes as the accumulated values of both unexercisable and exercisable stock options increase. However, we did not find a negative relationship between restricted stock and SRT.

Table 3

Results of Pooled OLS and of Fixed-Effects Analyses: The Moderating Effect of Overconfidence on the Relationship Between EBC and SRT in the Gain Domain

	(1)	(2)	(3)	Variation Inflation Factors
	OLS	FE	FE	Variation Inflation Factors
OC	0.282** (2.50)	0.279** (2.70)	0.253* (2.24)	1.49
UNEX_t-1	0.124 (1.45)	0.248* (2.51)	0.260* (2.07)	4.67
EX_t-1	0.013 (1.91)	0.002 (0.54)	−0.011 (−0.33)	3.89
RS_t-1	0.150 (0.97)	0.157 (1.06)	0.102 (0.74)	4.24
UNEX2_t-1	−0.042 (−1.09)	−0.401* (−2.55)	−0.408* (−2.19)	5.85
EX2_t-1	0.002 (1.03)	0.001 (−1.39)	−0.021 (−1.00)	5.73
UNEX_t-1 * OC	0.066 (0.96)		0.241 (1.91)	4.87
EX_t-1 * OC	0.009* (2.14)		0.058 (1.07)	3.21
RS_t-1 * OC	−0.083 (−1.45)		0.064 (0.39)	2.27
UNEX2_t-1 * OC	0.005 (0.13)		0.129* (2.23)	5.59
EX2_t-1 * OC	−0.006 (−1.58)		−0.006 (−1.79)	5.99
FRAN_t-1	0.761*** (5.05)	0.789* (1.96)	0.867 (1.71)	1.22
SIZE_t-1	1.166*** (35.18)	1.030*** (13.05)	1.007*** (12.58)	1.31
Q_t-1	−0.077 (−1.65)	0.035 (0.63)	0.037 (0.66)	1.38
TEN_t-1	−0.002 (−0.32)	−0.029* (−2.45)	−0.030* (−2.53)	1.28
AGE_t	−0.017** (−2.78)	−0.019* (−2.19)	−0.019* (−2.16)	1.16
GEN	−0.959** (−3.87)	−1.079* (−2.38)	−0.961 (−1.96)	1.14
N	607	607	607
R ²	0.72	0.67	0.68

Note: The dependent variable in the regressions is strategic risk taking, SRT_t-1, constructed by the sum of the log of R&D investment, capital investment, acquisition investment, and long-term debt. UNEX_t-1 is the accumulated value of unexercisable stock options. OC is CEO overconfidence (=1 if overconfident, =0 otherwise). EX_t-1 is the accumulated value of exercisable stock options. RS_t-1 is the accumulated value of restricted stock. UNEX2 is the accumulated value of unexercisable stock options squared. EX2 is the accumulated value of exercisable stock options squared. UNEX * OC is the product of the accumulated value of unexercisable stock options and overconfidence. EX * OC is the product of the accumulated value of exercisable stock options and overconfidence. RS * OC is the product of the accumulated value of restricted stock and overconfidence. UNEX2 * OC is the product of the accumulated value of unexercisable stock options squared and overconfidence. EX2 * OC is the product of the accumulated value of exercisable stock options squared and overconfidence. FRAN_t-1 is the degree of franchising, measured by the ratio of the number of franchise units to total units. SIZE_t-1 is the natural log of assets at the beginning of the year. Q_t-1 is the market value of assets over the book value of assets. TEN_t-1 is the number of years employed as CEO. AGE_t is CEO’s age. GEN is the gender of a CEO (=0 if male, =1 female).

Significant at .05. **Significant at .01. ***Significant at .001.

In Column 3, the findings from the regression examining interactions are presented. The results of the Hausman specification test confirmed that the fixed-effects model was more appropriate than the random-effects model (χ² = 241.66; p < .001). Hypothesis 1a predicted that CEO overconfidence would reduce the negative influence of highly valued unexercisable stock options on SRT. The coefficient of the interaction of overconfidence and unexercisable stock options squared was positive and significant (t = 2.23; p < .05). Thus, Hypothesis 1a was supported, suggesting that high levels of overconfidence will encourage CEOs to take greater risks while holding highly valued unexercisable stock options. Hypothesis 1b argued that overconfidence will positively moderate the relationship between the high (accumulated) value of exercisable stock options and SRT. This moderating effect was not supported; there was no significance to the coefficient in the interaction term of the quadratic term for exercisable stock options and overconfidence. Finally, in Hypothesis 1c, we predicted that CEO overconfidence would ameliorate the negative influence of positively valued restricted stock on SRT. However, the interaction of overconfidence and restricted stock was not significant, so no support was found for Hypothesis 1c.

Overconfident CEOs in the Loss Domain

Table 4 presents the results of the regression model that examined the positive moderating effect of overconfidence on the relationship between EBC and SRT in the loss domain. In Column 1, an OLS regression with the full-model was conducted. Variance inflation factors were all below acceptable levels, showing no sign of multicollinearity. The results of the main effects model in Column 2 indicated a positive linear relationship between the perceived loss in the accumulated value of unexercisable stock options and SRT (t = 2.73; p < .01). This implies that CEOs may increase SRT even when they perceive increased losses in the value of unexercisable stock options. However, the coefficients for the exercisable stock option variable and restricted stock variable were negative and insignificant.

Table 4

Results of Pooled OLS and of Fixed-Effects Analyses: The Moderating Effect of Overconfidence on the Relationship Between EBC and SRT in the Loss Domain

	(1)	(2)	(3)	Variation Inflation Factors
	OLS	FE	FE	Variation Inflation Factors
OC	0.328** (3.01)	0.362** (3.32)	0.362** (3.28)	1.21
PUNEX_t-1	0.058 (1.28)	0.303** (2.73)	0.269* (2.53)	3.04
PEX_t-1	−0.399 (−1.68)	−0.260 (−1.28)	−0.221 (−1.12)	2.77
PRS_t-1	−0.522 (−1.44)	−0.409 (−1.69)	−0.449 (−1.41)	1.75
PUNEX_t-1 * OC	0.297 (1.10)		0.052* (2.63)	2.91
PEX_t-1 * OC	0.059 (0.18)		−0.071 (−0.27)	2.65
PRS_t-1 * OC	0.103 (0.24)		−0.099 (−1.20)	1.82
FRAN_t-1	0.848*** (5.39)	0.134 (0.28)	0.138 (0.26)	1.22
SIZE_t-1	1.178*** (34.00)	1.150*** (13.15)	1.151*** (13.09)	1.29
Q_t-1	−0.096* (−2.02)	0.065 (1.09)	0.063 (1.05)	1.17
TEN_t-1	−0.006 (−0.84)	−0.038** (−3.13)	−0.038** (−3.12)	1.24
AGE_t	−0.016* (−2.50)	−0.024** (−2.68)	−0.024** (−2.65)	1.16
GEN	−0.959*** (−4.70)	−1.546*** (−4.94)	−1.547*** (−4.92)	1.13
N	563	563	563
R ²	0.73	0.70	0.70

Note: The dependent variable in the regressions is strategic risk taking, SRT_t-1, constructed by the sum of the log of R&D investment, capital investment, acquisition investment, and long-term debt. OC is CEO overconfidence (=1 if overconfident, =0 otherwise). PUNEX_t-1 is the perceived loss in the accumulated value of unexercisable stock options. PEX_t-1 is the perceived loss in the accumulated value of exercisable stock options. PRS_t-1 is the perceived loss in the accumulated value of restricted stock. The perceived loss variables are presented in percent as a difference between the value in the current year (t) and the value in the previous year (t − 1) divided by the value in the previous year (t − 1). PUNEX_t-1 * OC is the product of the accumulated value of unexercisable stock options and overconfidence. PEX_t-1 * OC is the product of the accumulated value of exercisable stock options and overconfidence. PRS_t-1 * OC is the product of the accumulated value of restricted stock and overconfidence. FRAN_t-1 is the degree of franchising, measured by the ratio of the number of franchise units to total units. SIZE_t-1 is the natural log of assets at the beginning of the year. Q_t-1 is the market value of assets over the book value of assets. TEN_t-1 is the number of years employed as CEO. AGE_t is CEO’s age. GEN is the gender of a CEO (=0 if male, =1 female).

Significant at .05. **Significant at .01. ***Significant at .001.

Based on the results of the Hausman specification test (χ² = =229.76; p < .001), a fixed-effects regression model was conducted (see Column 3). Hypothesis 2a suggested that overconfidence would positively moderate the relationship between perceived loss in unexercisable stock options and SRT. This interaction between overconfidence and perceived loss in unexercisable stock options was found to be positive and significant, providing support for Hypothesis 2a (t = 2.63; p < .05). This finding indicates that CEO overconfidence will strengthen the positive influence of the perceived loss in unexercisable options on SRT when CEOs perceive themselves to be in the loss domain. For Hypotheses 2b and 2c, we predicted that overconfidence would accentuate the positive influence of perceived loss in the values of exercisable stock options and restricted stock on SRT. In contrast, although not statistically significant, our results showed that overconfidence appeared to negatively moderate the relationship between perceived loss in exercisable stock options and SRT. Hypotheses 2b and 2c, therefore, did not receive support.

In summary, we found support for Hypothesis 1a that CEO overconfidence ameliorates the negative effect of the accumulated value of unexersiable stock options on SRT in the gain domain. That is, overconfident CEOs are less likely to become risk-averse even when they perceive significantly increased value in their unexercisable stock options. Hypothesis 2a received support, suggesting that overconfident CEOs are more likely to become risk-seeking when they perceive significantly decreased value in their unexercisable stock options.

Franchising and CEO Risk Behavior

Although not statistically significant, we consistently found that CEOs in firms with high levels of franchising were more risk-seeking in Table 3 and 4. These results were inconsistent with our predictions that CEOs in highly franchised firms are more likely to be risk-averse. To further test the intuition that franchising is associated with CEOs’ risk behavior, we separated firms according to the level of franchising they employ and ran regression on each group of firms. In particular, we ranked values of franchising variables and divided them into 5 equal percentage groups, quintiles (Almeida, Campello, & Weisbach, 2004; Malmendier & Tate, 2005b). As a result, the sample firms were split into 5 quintiles from least-franchised to most-franchised: firms in Quintile 1 (Quintile 5) were considered the least-franchised (most-franchised). The level of franchising for the firms in the bottom two quintiles ranged from 0 to 0.14, while that of the top two quintile firms ranged from 0.57 to 1.

Table 5 shows the results of the quintile regression in the gain domain. To save space, the results from the full model (Equation 1) were only reported. Unlike we expected, it was found that franchising was positively related to risk-taking. Thus, Hypothesis 3a was not supported. The coefficients of franchising for firms in the top two quintiles varied between 1.9 and 3.4 and were statistically significant at the 5% and 0.1% level, respectively. This suggested that an additional increase in franchising would result in around 0.019% to 0.034% change in SRT. Furthermore, CEOs in the most-franchised firms (Quintile 5) displayed the strongest risk-taking behavior (t = 3.29; p < .00).

Table 5

Results of Quintile Regression: Franchising and SRT in the Gain Domain

	Least Franchised				Most Franchised
	Quintile 1	Quintile 2	Quintile 3	Quintile 4	Quintile 5
OC	1.110 (1.50)	1.133 (1.13)	0.642 (1.27)	1.079 (1.61)	1.234** (2.82)
UNEX_t-1	0.836** (3.36)	0.017 (0.04)	0.625 (1.38)	0.154 (1.01)	0.886 (1.66)
EX_t-1	0.019 (1.18)	0.367 (1.49)	−0.139 (−1.46)	0.589** (2.91)	0.437** (2.80)
RS_t-1	0.183 (0.47)	0.641 (1.40)	0.693 (1.24)	0.257 (1.38)	0.813 (0.76)
UNEX2_t-1	−0.186* (−2.48)	−0.698 (−1.41)	−0.197 (−1.23)	−0.056 (−1.14)	−0.225 (−1.76)
EX2_t-1	0.004 (1.36)	0.107 (1.39)	−0.007 (−0.72)	−0.109 (−1.35)	−0.014 (1.04)
UNEX_t-1 * OC	0.045 (0.96)	0.030 (0.66)	0.733 (1.25)	0.612 (1.31)	1.396** (3.33)
EX_t-1 * OC	0.045 (0.34)	0.627 (1.51)	0.135 (1.72)	0.396* (2.12)	0.576 (1.61)
RS_t-1 * OC	−0.516 (−1.24)	−0.012 (−1.02)	−0.824 (−1.24)	−0.247 (−1.20)	−2.069 (−1.49)
UNEX2_t-1 * OC	0.002 (1.51)	0.113 (0.41)	0.212 (0.23)	0.156* (2.31)	0.284** (2.72)
EX2_t-1 * OC	0.002 (1.22)	−0.149 (−1.23)	−0.051 (−0.75)	−0.013 (−1.30)	−0.029 (−1.73)
FRAN_t-1		1.739 (1.51)	0.902 (0.60)	1.921* (1.89)	3.431*** (3.29)
SIZE_t-1	1.007*** (13.43)	1.376*** (4.20)	1.319*** (9.55)	1.119*** (23.31)	1.043*** (10.49)
Q_t-1	−0.127 (−1.79)	−0.144 (−0.74)	−0.560** (−2.98)	−0.365*** (−3.70)	−0.105 (−1.23)
TEN_t-1	−0.002 (−0.14)	−0.066 (−1.72)	−0.018 (−0.27)	−0.009 (−0.83)	0.070* (−2.34)
AGE_t	−0.033** (−3.30)	0.002 (0.05)	−0.045** (−3.00)	−0.044* (−2.62)	−0.340* (−2.41)
GEN			−0.473 (−1.03)	−0.774* (−2.49)	−0.423 (−1.15)
N	220	43	130	131	130
R ²	0.55	0.76	0.61	0.89	0.82

Significant at .05. **Significant at .01. ***Significant at .001.

The findings of the quintile regression in the loss domain using Equation 2 are shown in Table 6. The estimated effect of franchising was similar to the estimation in the gain domain: CEOs in firms with high levels of franchising were likely to take greater risks (Quintiles 4 and 5). Such a correlation was statistically significant and strongest in the most franchised firms (t = 3.33; p < .01). Therefore, Hypothesis 3b was not supported. In summary, our findings showed there was a positive relationship between franchising and CEO’s propensity for risk-taking: CEOs in highly franchised firms were more likely to take risks than other CEOs in both gain and loss domains.

Table 6

Results of Quintile Regression: Franchising and SRT in the Loss Domain

	Least Franchised				Most Franchised
	Quintile 1	Quintile 2	Quintile 3	Quintile 4	Quintile 5
OC	0.052 (0.55)	0.412 (1.04)	0.735** (2.80)	0.880** (4.33)	1.719*** (5.18)
PUNEX_t-1	0.298 (1.71)	0.152 (1.44)	0.211 (0.58)	1.023 (1.99)	0.462 (1.77)
PEX_t-1	−0.502 (−1.16)	−0.325 (−1.74)	−0.225 (−1.62)	−1.086 (−1.96)	−0.250 (−1.40)
PRS_t-1	0.130 (1.30)	−0.147 (–1.27)	0.104 (1.11)	0.446 (1.91)	1.150 (1.33)
PUNEX_t-1 * OC	−0.104 (–1.46)	0.802* (2.67)	0.077 (1.14)	0.010* (2.17)	0.378* (2.18)
PEX_t-1 * OC	0.282 (1.06)	−0.341 (–1.34)	−0.278 (–1.44)	−0.825 (–1.29)	0.519 (0.69)
PRS_t-1 * OC	−0.278 (–1.29)	−0.041 (–1.13)	0.771 (0.70)	−0.493 (–1.62)	−0.849 (–1.73)
FRAN_t-1		2.064 (1.36)	−0.860 (–0.54)	1.677* (2.17)	2.665** (3.33)
SIZE_t-1	0.998*** (12.40)	1.491*** (5.22)	1.215*** (9.05)	1.168*** (22.52)	0.974*** (10.71)
Q_t-1	−0.098 (–1.27)	−0.254 (–1.94)	−0.529** (–2.98)	−0.120 (–1.27)	−0.105 (–1.00)
TEN_t-1	−0.003 (–0.25)	−0.088* (–2.39)	−0.016 (–0.76)	−0.027 (–1.71)	0.448 (0.68)
AGE_t	−0.029** (–2.75)	−0.047 (–1.59)	−0.044** (–2.78)	0.028 (1.57)	0.034* (2.46)
GEN		−0.255 (–1.48)	−0.706 (–1.56)	−1.07** (–3.43)	−1.198*** (–3.78)
N	203	41	122	119	123
R ²	0.50	0.76	0.59	0.79	0.81

Note: The dependent variable in the regressions is strategic risk taking, SRT_t-1, constructed by the sum of the log of R&D investment, capital investment, acquisition investment, and long-term debt. OC is CEO overconfidence (=1 if overconfident, =0 otherwise). PUNEX_t-1 is the perceived loss in the accumulated value of unexercisable stock options. PEX_t-1 is the perceived loss in the accumulated value of exercisable stock options. PRS_t-1 is the perceived loss in the accumulated value of restricted stock. The perceived loss variables are presented in percentage as a difference between the value in the current year (t) and the value in the previous year (t − 1) divided by the value in the previous year (t − 1). PUNEX_t-1 * OC is the product of the accumulated value of unexercisable stock options and overconfidence. PEX_t-1 * OC is the product of the accumulated value of exercisable stock options and overconfidence. PRS_t-1 * OC is the product of the accumulated value of restricted stock and overconfidence. FRAN_t-1 is the degree of franchising, measured by the ratio of the number of franchise units to total units. SIZE_t-1 is the natural log of assets at the beginning of the year. Q_t-1 is the market value of assets over the book value of assets. TEN_t-1 is the number of years employed as CEO. AGE_t is CEO’s age. GEN is the gender of a CEO (=0 if male, =1 female).

Significant at .05. **Significant at .01. ***Significant at .001.

Discussion

A large body of compensation literature has been devoted to fundamental controversies about whether or not EBC is an efficient incentive mechanism for SRT. This study provided some important insights on this question by incorporating the behavioral bias and industry-specific characteristic into the analysis of the effect of EBC on SRT. We believe that empirical evidence discovered in this study could shed light on unanswered questions on the association between EBC and SRT. In particular, we found that CEOs’ willingness to take risky actions is associated with: (a) the value for different forms of EBC incorporated into their calculations of their personal wealth, (b) CEO overconfidence, and (c) the level of franchising. These findings suggest that restaurant firms should understand the role of both behavioral and industry-specific characteristics to effectively and efficiently develop, implement, and administer executive EBC plans.

Moderation Effects of Overconfidence

First, we found that in the gain domain, overconfidence positively moderated the relationship between the accumulated value of unexercisable stock options and SRT. As shown in Figure 1, the moderation effect was most pronounced at higher levels of value in unexercisable stock options. In particular, SRT dropped more rapidly for non-overconfident CEOs than for overconfident CEOs at higher levels of value in unexercisable stock options. The effect of overconfidence was a facilitation of SRT at high levels, which led to a less prominent curvilinear relationship. Moreover, the maximum of the curve for overconfident CEOs was placed to the right of that for non-overconfident CEOs, indicating less sensitivity to prospective losses stemming from high levels of value in unexercisable options. For example, we might expect overconfident CEOs to be more tolerant to risk aversion bias when personal wealth associated with their stock option holdings increases. Given that the negative influence of unexercisable stock options on SRT decreases in magnitude for overconfident CEOs, restaurant firms should carefully consider CEOs’ behavioral aspects when designing executive compensation plans. Granting stock options to overconfident CEOs could be effective in preventing suboptimal investment problems especially when the firm expects strong future prospects in the long term. Board of directors and shareholders can also extend option vesting periods to encourage overconfident CEO to engage in risk-taking investments.

Figure 1

Overconfidence Moderation Effect in the Gain Domain

On the other hand, overconfidence failed to positively moderate the effect of exercisable stock options on SRT. This could occur because the concern for losses is stronger for exercisable stock options than unexercisable stock options. Prospect theory expects CEOs to be risk-averse when they are heavily endowed with the value of stock options and become concerned about losses (Kahneman, Knetsch, & Thaler, 1990). The value of exercisable stock options may be instantly realized whereas the value of unexercisable stock options may not be endowed until they become excisable (Devers, Cannella, et al., 2007; Larraza-Kintana et al., 2007). This implies that when holding exercisable stock options, the concern for losses may outweigh the positive influence of overconfidence, suppressing risky behavior. We maintain that even overconfident CEOs could develop a greater motivation to reduce risks when they perceive prospective downside in exercisable options. For those CEOs who received large amounts of EBC in the past, firms may reduce proportions of EBC in their pay plans as it no longer motivates risk-taking. Instead, cash-based pay such as bonuses can be used as an alternative driver for risk-taking (Devers et al., 2008).

Consistent with loss aversion arguments, our findings showed that CEOs will take more risks when they perceive loss in the accumulated value of unexercisable stock options. We further found that overconfidence leads to a stronger positive relationship between the perceived loss in the accumulated unexercisable stock options and SRT. However, this positive moderating effect of overconfidence was not found with exercisable stock options and restricted stock. Although not statistically significant, both main effects and interaction effects for exercisable stock options and restricted stock were negative. These results imply that CEOs may perceive their personal wealth to be at considerable risk of loss when holding exercisable options and restricted stock but may not when holding unexercisable options (Devers, Wiseman, et al., 2007; Larraza-Kintana et al., 2007). Losses associated with unexercisable options may seem unreal at the time of loss due to the long-term horizon of these options. As a result, they may choose to pursue risky investments, expecting their stock options to recover from losses and generate gains.

Taken together, these findings suggest that understanding the behavioral aspects of CEOs could help restaurant firms design more efficient pay-to-perform incentive contracts. For example, firms can save significantly in the long-term by reducing large stock option grants to CEOs when unnecessary. Stock options are costly because new shares are often issued to back those stock options, which eventually harm their earnings per share. Furthermore, money saved on stock options can be deployed to increase operating income and thereby, create shareholder value.

The Impact of Franchising on CEO’s Risk-Taking Behavior

Given various industry-specific characteristics, CEOs in different industries may display different degrees of risk-taking and different perceptions of risks. We found partial support for a positive relationship between franchising and SRT. These results highlighted the benefits of franchising in the context of EBC and SRT. As franchising promotes increased levels of risk-taking, it may not be required to pay expensive EBC to encourage additional risk-taking when CEOs pursue expansion through franchising. One possible explanation is that cash flow generated from successful franchise operations may provide them with more room (insurance) to take risky investment projects. For example, firms will collect greater royalty and advertising fees as they are charged as a percentage of revenues. Franchisors can aggressively expand investments when their financial resources are affluent. Therefore, we maintain that franchising can help firms engage in more risky investments as it provides additional protection or comfort against risk involved in investment (Alon, 2001; Combs & Ketchen, 1999). This indicates that various industry-specific characteristics need to be carefully considered rather than paying expensive EBC systematically based on short-term outcomes.

Contributions

This study makes several important theoretical and practical contributions. First, this study added to the compensation literature by empirically testing some of the theoretical arguments of the BAM. Based on prospect theory, our study effectively assessed the effects of different forms of EBC on SRT in gain and loss domains (Kahneman & Tversky, 1979). Most prior compensation studies have focused on the gain domain to examine the association between EBC and SRT (Devers et al., 2008; Larraza-Kintana et al., 2007; Sanders & Hambrick, 2007). To empirically test varying risk behaviors, we developed a theoretical framework that could effectively represent a loss frame in the context of EBC. Considering loss domains that engender risk-seeking behaviors could provide theoretical and empirical insights on the association between EBC and SRT.

Another contribution to the literature is incorporating behavioral aspects into the analysis of the link between EBC and SRT. Prior research has not empirically examined the behavioral influence of CEOs in the context of an EBC–SRT relationship. Our findings suggest that overconfidence may help align risk preferences even when EBC fails to do so. Therefore, study findings could hold important implications for practitioners seeking to improve shareholder value in the long term by efficiently designing and implementing executive compensation packages. For example, shareholders and boards of directors should carefully examine the current practice of compensation plans in line with behavioral aspects of CEOs especially when the proportion of EBC increases in total compensation.

Finally, this study offered several findings that contribute to our understanding of the strategic implications of franchising in the U.S. restaurant industry. Using a sample of publicly traded restaurant firms, we found a positive relationship between franchising and risk-taking behavior. These results indicated that strategic choices made by CEOs could also have a strong influence on risk-taking behavior. Previous studies did not consider exogenous factors that might explain risk-taking behavior. Therefore, our contribution is the recognition that strategic choices and industry conditions could play an important role in determining executive compensation plans. Our findings could provide restaurant firms with helpful guidance to create efficient compensation mechanisms especially when they are trying to expand through franchising.

Limitations and Future Research

Our study has several limitations. First, the proxy for overconfidence used in this study could be improved in future studies. Using 6 years of Chief Financial Officer surveys, Ben-David et al. (2007) developed a proxy for overconfidence. Future studies could use various alternative methods, such as surveys and/or interviews, to measure managerial overconfidence at different levels and improve our findings. For example, our sample represents a relatively small group of CEOs in the U.S. restaurant industry because it was selected based on CEO EBC information. Data collection on managers at various levels may help create more representative samples.

Second, our examination was limited to the valuation of in-the-money EBC. Based on prospect theory, gain frames were created when CEOs perceive gain in the value of EBC. On the other hand, loss frames were created by assuming that CEOs perceive decreased value of in-the-money EBC as losses. In this regard, two separate regression analyses were conducted in gain and loss frames. Although not presented, the full model analyses also found qualitatively similar results. Prospect theory implies that the reference point chosen determines how the outcomes are framed. That is, CEOs using one reference point may frame stock option value as a gain whereas using a different reference point they may frame the same value as a loss. Prior studies tested prospect theory by manipulating the reference points using different wording for identical information (Bazerman, 1984; Kühberger, 1998; Levin, Schneider, & Gaeth, 1998; Tversky & Kahneman, 1981). Although the literature found some support for this approach, it remains a challenge to correctly identify the reference point (Devers, Cannella, et al., 2007; Holmes et al., 2011). Future studies could more fully explore the effects of the reference point.

Finally, future studies can further investigate the association between franchising and investment. Unlike our expectations, we found a positive relationship between franchising and SRT. This relation needs to be examined more carefully by considering firms’ financial conditions. For example, financial distress or excessive use of leverage can reduce a firm’s incentive to invest (Maksimovic & Titman, 1991). Controlling for the differences in firms’ financial conditions, future research can better explain the impact of franchising on corporate investment.

Footnotes

Appendix A

Notes

Kwanglim Seo (e-mail: kwanglim@hawaii.edu) is an assistant professor in the School of Travel Industry Management at University of Hawaii, Manoa. Amit Sharma (e-mail: aus22@psu.edu) is a professor in the School of Hospitality Management at The Pennsylvania State University, University Park, PA.

References

Aggrawal

R. K.

Samwick

A. A.

(2003). Why do managers diversify their firms? Agency reconsidered. Journal of Finance, 58, 71-118.

Aiken

L. S.

West

S. G.

(1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage.

Alon

(1999). The internationalization of US franchising systems. New York, NY: Garland.

Alon

Drtina

Gilbert

J. P.

(2004, March). The impact of franchising on the returns on equity chains in the restaurant sector. Paper presented at the proceedings of the Annual International Society of Franchising Conference, Las Vegas, NV.

Baker

Ruback

R. S.

Wurgler

(2007). Behavioral corporate finance. In Espeneckbo

(Ed.), Handbook of corporate finance: empirical corporate finance (pp. 145-188). Amsterdam, Netherlands: North-Holland.

Barber

Ghiselli

Deale

(2006). Assessing the relationship of CEO compensation and company financial performance in the restaurant segment of the industry. Journal of Food Service Business Research, 9(4), 65-82.

Bazerman

M. H.

(1984). The relevance of Kahneman and Tversky’s concept of framing to organizational behavior. Journal of Management, 10, 333-343.

Beckman

C. M.

Haunschild

P. R.

(2002). Network learning: The effects of partners’ heterogeneity of experience on corporate acquisitions. Administrative Science Quarterly, 47, 92-124.

Belsley

D. A.

Kuh

Welsch

R. E.

(1980). Regression diagnostics: Identifying influential data and source of collinearity. New York, NY: Wiley.

10.

Ben-David

Graham

J. R.

Harvey

C. R.

(2007). Managerial overconfidence and corporate policies (Working Paper). Chicago, IL: University of Chicago and Duke University.

11.

Berle

Means

(1932). The modern corporation and private property. Chicago, IL: Commerce Clearing House.

12.

Bertrand

Schoar

(2003). Managing with style: The effect of managers on firm policies. Quarterly Journal of Economics, 68, 1169-1208.

13.

Bliss

R. T.

Rosen

R. J.

(2001). CEO compensation and bank mergers. Journal of Financial Economics, 61, 107-138.

14.

Bolton

Chen

Wang

(2011). A unified theory of Tobin’s q, corporate investment, financing, and risk management. Journal of Finance, 66, 1545-1578.

15.

Brown

Sarma

(2007). CEO overconfidence, CEO dominance and corporate acquisitions. Journal of Economics and Business, 59, 358-379.

16.

Bryan

S. L.

Hwang

Lilen

(2000). CEO stock-based compensation: an empirical analysis of incentive-intensity, relative mix, and economic determinants. Journal of Business, 73, 661-693.

17.

Camerer

Lovallo

(1999). Overconfidence and excess entry: An experimental approach. American Economic Review, 89, 306-318.

18.

Combs

J. G.

Castrogiovanni

G. J.

(1994). Franchisor strategy: A proposed model and empirical test of franchise versus company ownership. Journal of Small Business Management, 32, 37-47.

19.

Combs

J. G.

Ketchen

D. J.

(1997). The effects of resource scarcities and agency costs on choice of organizational form in the US restaurant industry. Journal of Hospitality & Tourism Research, 21(2), 27-43.

20.

Combs

J. G.

Ketchen

D. J.

(1999). Explaining inter-firm cooperation and performance: Toward a reconciliation of predictions from the resource-based view and organizational economics. Strategic Management Journal, 20, 867-888.

21.

Coffee

J. C.

(1988). Shareholders versus managers: The strain in the corporate web. Knights, raiders, and targets: The impact of the hostile takeover. New York, NY: Oxford University Press.

22.

Coles

J. L.

Daniel

N. D.

Naveen

(2006). Managerial incentives and risk-taking. Journal of Financial Economics, 79, 431-468.

23.

Core

J. E.

Guay

W. R.

Larcker

D. F.

(2003). Executive equity compensation and incentives: A survey. Economic Policy Review, 9(1), 27-50.

24.

Dalbor

Oak

Rowe

(2010). How closely CEO compensation tied to performance? An examination of the US restaurant industry. Journal of Hospitality Financial Management, 18(2), 1-17.

25.

Dant

R. P.

Paswan

A. K.

Stanworth

(1996). Ownership redirection trends in franchising: A cross-sectional investigation. International Journal of Entrepreneurial Behavior & Research, 2(3), 48-67.

26.

Datta

Iskandar-Datta

Raman

(2001). Executive compensation and corporate acquisition decisions. Journal of Finance, 56, 2299-2336.

27.

Dawson

J. F.

Richter

A. W.

(2006). Probing the three-way interactions in moderated multiple regression: development and application of a slope difference test. Journal of Applied Psychology, 91, 917-926.

28.

De Noble

A. F.

Olsen

M. D.

(1986). The food service industry environment: Market volatility analysis. FIU Hospitality Review, 4(2), 89-100.

29.

Devers

C. E.

Cannella

A. A.

Reilly

G. P.

Yoder

M. E.

(2007). Executive compensation: a multidisciplinary review of recent developments. Journal of Management, 33, 1017-1072.

30.

Devers

C. E.

McNamara

Wiseman

R. M.

Arrfelt

(2008). Moving closer to the action: Examining compensation design effects on firm risk. Organization Science, 19, 548-566.

31.

Devers

C. E.

Wiseman

R. M.

Holmes

R. M.

(2007). The effects of endowment and loss aversion in managerial stock option valuation. Academy of Management Journal, 50(1), 1-18.

32.

Eisenhardt

K. M.

(1989). Building theories from case study research. Academy of Management Review, 14, 532-550.

33.

Fama

E. F.

(1976). Efficient capital markets: Reply. Journal of Finance, 31, 143-145.

34.

Gervais

(2010). Capital budgeting and other investment decisions, In Kent Baker

Nofsinger

John R.

(Eds.), Behavioral finance: Investors, corporations, and markets (pp.413-434). Hoboken, NJ: Wiley.

35.

Gervais

Heaton

J. B.

Odean

(2011). Overconfidence, compensation contracts, and capital budgeting. Journal of Finance, 66, 1735-1777.

36.

Gervais

Odean

(2001). Learning to be overconfident. Review of Financial Studies, 14(1), 1-27.

37.

Goel

A. M.

Thakor

A. V.

(2008). Overconfidence, CEO selection and corporate governance. Journal of Finance, 63, 2737-2784.

38.

Gow

I. D.

Ormazabal

Taylor

T. J.

(2010). Correcting for cross-Sectional and time-series dependence in accounting research. Accounting Review, 85, 483-512.

39.

Choi

Y. H.

(2004). CEO compensation in the casino industry. Journal of Hospitality & Tourism Research, 28, 143-155.

40.

Guillet

B. D.

Kucukusta

Xiao

(2011). An examination of executive compensation in the restaurant industry. International Journal of Hospitality Management, 31, 86-95.

41.

Hackbarth

(2008). Managerial traits and capital structure decision. Journal of Financial and Quantitative Analysis, 43, 843-882.

42.

Hall

Murphy

(2002). Stock options for underdiversified executives. Journal of Accounting and Economics, 33, 3-42.

43.

Harrington

R. J.

(2005). The how and who of strategy making: Models and appropriateness for firms in hospitality and tourism industries. Journal of Hospitality & Tourism Research, 29, 372-395.

44.

Hastie

Dawes

R. M.

(2001). Rational choice in an uncertain world: The psychology of judgment and decision making. Thousand Oaks, CA: Sage.

45.

Hausman

J. A.

(1978). Specification tests in econometrics. Econometrica, 46, 1251-1271.

46.

Hayward

Hambrick

D. C.

(1997). Explaining the premiums paid for large acquisitions: Evidence of CEO hubris. Administrative Science Quarterly, 42, 103-127.

47.

Heaton

J. B.

(2002). Managerial optimism and corporate finance. Financial Management, 31, 33-45.

48.

Holmes

R. M.

Bromiley

Devers

C. E.

Holcomb

T. R.

McGuire

J. B.

(2011). Management theory applications of prospect theory: Accomplishments, challenges, and opportunities. Journal of Management, 37, 1069-1107.

49.

Hoskisson

R. E.

Hitt

M. A.

Hill

C. W.

(1993). Managerial incentives and investment in R&D in large multiproduct firms. Organization Science, 4, 325-341.

50.

Hsu

Jang

Canter

D.D.

(2010). Factors affecting franchise decisions in the restaurant industry. Journal of Hospitality & Tourism Research, 34, 440-454.

51.

Huo

Y. H.

Kwansa

(1994). Effect of operating and financial leverage on firm’s risk. Journal of the International Academy of Hospitality Research, 8(3), 1-10.

52.

IHS Global Insight. (2012). Franchise business economic outlook for 2013. Washington, DC: International Franchise Association Educational Foundation.

53.

Jensen

M. C.

Meckling

W. H.

(1976). Theory of the firm: Managerial behavior, agency costs, and ownership structure. Journal of Financial Economics, 3, 305-336.

54.

Jensen

M. C.

Murphy

K. J.

(1991). Performance pay and top-management incentives. Journal of Political Economy, 98, 225-264.

55.

Kahneman

(2003). A perspective on judgment and choice: Mapping bounded rationality. American Psychologist, 58, 697-720.

56.

Kahneman

Knetsch

J. L.

Thaler

R. H.

(1990). Experimental tests of the endowment effect and the Coase theorem. Journal of Political Economy, 98, 1325-1348.

57.

Kahneman

Tversky

(1979). Prospect theory: An analysis of decision under risk. Econometrica, 47, 313-327.

58.

Khan

M. A.

(1999). Restaurant franchising (2nd ed.). New York, NY: Wiley.

59.

Kim

(2005). A preliminary examination of determinants of CEO cash compensation in the U.S. restaurant industry from an agency theory perspective. Journal of Hospitality & Tourism Research, 29, 341-355.

60.

Koh

Lee

Boo

(2009). Does franchising help restaurant firm value? International Journal of Hospitality Management, 28, 289-296.

61.

Kühberger

(1998). The influence of framing on risky decisions: A meta-analysis. Organizational Behavior and Human Decision Processes, 72, 23-55.

62.

Lafontaine

Bhattacharyya

(1995). The role of risk in franchising. Journal of Corporate Finance, 2, 39-74.

63.

Langer

(1975). The illusion of control. Journal of Personality and Social Psychology, 32, 311-328.

64.

Larcker

D. F.

(1983). The association between performance plan adoption and corporate capital investment. Journal of Accounting and Economics, 5, 3-30.

65.

Larraza-Kintana

Wiseman

R. M.

Gomez-Mejia

L. R.

Welbourne

(2007). Disentangling compensation and employment risks using the behavioral agency model. Strategic Management Journal, 28, 1001-1019.

66.

Levin

I. P.

Schneider

S. L.

Gaeth

G. J.

(1998). All frames are not created equal: A typology and critical analysis of framing effects. Organizational Behavior and Human Decision Processes, 76, 149-188.

67.

Lim

(2011). Loss aversion or CEO confidence? Unpacking the accumulation effects of CEO stock-based incentives on strategic risk-taking (Working paper). Atlanta, GA: Georgia State University.

68.

Madanoglu

Karadag

(2006). Alignment of CEO compensation and firm performance in the quick-service restaurants. Paper presented at the 2006 CHRIE Conference Proceedings, Arlington, VA.

69.

Madanoglu

Lee

Castrogiovanni

G. J.

(2011). Franchising and firm financial performance among U.S. restaurants. Journal of Retailing, 87, 406-417.

70.

Makri

Lane

P. J.

Gomez-Mejia

L. R.

(2006). CEO incentives, innovation, and performance in technology-intensive firms: A reconciliation of outcome and behavior- based incentive schemes. Strategic Management Journal, 27, 1057-1080.

71.

Maksimovic

Titman

(1991). Financial policy and reputation for product quality. Review of Financial Studies, 4, 175-200.

72.

Malmendier

Tate

(2005a). CEO overconfidence and corporate investment. Journal of Finance, 60, 2661-2700.

73.

Malmendier

Tate

(2005b). Does overconfidence affect corporate investment? CEO overconfidence measures revisited. European Financial Management, 11, 649-659.

74.

Malmendier

Tate

(2008). Who makes acquisitions? CEO overconfidence and the market’s reaction. Journal of Financial Economics, 89, 20-43.

75.

March

J. G.

Shapira

(1987). Managerial perspectives on risk and risk taking. Management Science, 33, 1404-1418.

76.

Martin

R. E.

(1988). Franchising and risk management. American Economic Review, 78, 954-968.

77.

Martin

G. P.

Wiseman

R. M.

Gomez-Mejia

L. R.

(2013). Executive stock options as mixed gambles: Re-visiting the behavioral agency model (Working Paper). Retrieved from http://works.bepress.com/geoffrey_martin/2

78.

Milgrom

Roberts

(1992). Economics, organization, & management. Englewood Cliffs, NJ: Prentice Hall.

79.

Morrison

(2000). Entrepreneurship, intrepreneurship and franchising. In Lashley

Morrison

(Eds.), Franchising hospitality services (pp. 69-71). Woburn, MA: Butterworth-Heinemann.

80.

Morrison

Macmillan

(2000). Financial fundamentals. In Lashley

Morrison

(Eds.), Franchising hospitality services (pp. 114-144). Woburn, MA: Butterworth-Heinemann.

81.

Ofek

Yermack

(2000). Taking stock: equity-based compensation and the evolution of managerial ownership. Journal of Finance, 55, 1367-1384.

82.

Oxenfeldt

A. R.

Kelly

A. O.

(1968). Will successful franchise chains ultimately become wholly-owned chains? Journal of Retailing, 44(4), 69-83.

83.

Ozdemir

Kizildag

Upneja

(2013). Does risk matter in CEO compensation contracting? Evidence from US restaurant industry. International Journal of Hospitality Management, 34, 372-383.

84.

Parrino

R. A.

Poteshman

Weinbach

M. S.

(2005). Measuring investment distortions when risk-averse managers decide whether to undertake risky projects. Financial Management, 34(1), 21-60.

85.

Parsa

H. G.

Self

J. T.

Njite

King

(2005). Why restaurants fail. Cornell Hospitality Quarterly, 46, 304-323.

86.

PriceWaterhouseCoopers. (2011). Economic impact of franchised businesses. A study for the International Franchise Association Educational Foundation. Washington, DC: Author.

87.

Rajgopal

Shevlin

(2002). Empirical evidence on the relation between stock option compensation and risk taking. Journal of Accounting and Economics, 33, 145-171.

88.

Roh

Y. S.

(1998). The effect of franchisor brand name capital and competency on the cost of fast-food franchise. Journal of Hospitality & Tourism Research, 22, 188-200.

89.

Roh

Y. S.

(2002). Size, growth rate and risk sharing as the determinants of propensity to franchise in chain restaurants. International Journal of Hospitality Management, 21(1), 43-56.

90.

Roh

Y. S.

Kwag

H. M.

(1997). The ownership structure of property rights: Theory and empirical evidence of restaurant franchising. Journal of Hospitality & Tourism Research, 21(2), 75-85.

91.

Roll

(1986). The hubris hypothesis of corporate takeovers. Journal of Business, 59, 197-216.

92.

Rubin

P. H.

(1978). The theory of the firm and the structure of the franchise contract. Journal of Law and Economics, 21, 223-233.

93.

Sanders

W. G.

(2001). Behavioral responses of CEOs to stock ownership and stock option pay. Academy of Management Journal, 44, 477-492.

94.

Sanders

W. G.

Hambrick

D. C.

(2007). Swing for the fences: The effects of CEO stock option on company risk-taking and performance. Academy of Management Journal, 50, 1055-1078.

95.

Sawers

Wright

Zamora

(2006, November). Loss aversion, stock-based compensation and managerial risk-seeking behavior (Working paper; SSRN Abstract ID 864224). Retrieved from http://papers.ssrn.com/sol3/papers.cfm?abstract_id=864224

96.

Seo

Sharma

(2013). Does equity-based compensation encourage strategic risk-taking? Exploring the effects of equity-based compensation in the U.S. restaurant industry. International Journal of Tourism Sciences, 13(2), 76-101.

97.

Sharpe

W. F.

(1970). Portfolio theory and capital markets. New York, NY: McGraw Hill.

98.

Sul

Khan

M. A.

(2006). Restaurant franchisees’ view of the relationships among the determinants of franchisors’ entrepreneurial strategy processes. Journal of Hospitality & Tourism Research, 30, 427-451.

99.

Tabachnick

B. G.

Fidell

L. S.

(1996). Using multivariate statistics. New York, NY: HarperCollins.

100.

Thaler

R. H.

(1980). Toward a positive theory of consumer choice. Journal of Economic Behavior & Organization, 1, 39-60.

101.

Thompson

S. B.

(2011). Simple formulas for standard errors that cluster by both firm and time. Journal of Financial Economics, 99, 1-10.

102.

Tosi

H. L.

Katz

Gomez-Mejia

L. R.

(1997). Disaggregating the agency contract: The effects of monitoring, incentive alignment, and CEO tenure on managerial decision making. Academy of Management Journal, 40, 584-602.

103.

Tosi

H. L.

Werner

Katz

J. P.

Gomez-Mejia

L. R.

(2000). How much does performance matter? A meta-analysis of CEO pay studies. Journal of Management, 26, 301-339.

104.

Tvede

(1999). The psychology of finance. Chichester, England: Wiley.

105.

Tversky

Kahneman

(1974). Judgment under uncertainty: Heuristics and biases. Science, 185, 1124-1131.

106.

Tversky

Kahneman

(1981). The framing of decisions and the psychology of choice. Science, 211, 453-458.

107.

Tversky

Kahneman

(1986). Rational choice and the framing of decisions. Journal of Business, 59, 251-278.

108.

Tversky

Kahneman

(1992). Advances in prospect theory: Cumulative representation of uncertainty. Journal of Risk and Uncertainty, 5, 297-323.

109.

Welbourne

T. M.

Andrews

A. O.

(1996). Predicting the performance of initial public offerings: Should human resource management be in the equation? Academy of Management Journal, 39, 891-919.

110.

Wiseman

R. M.

Bromiley

(1996). Toward a model of risk in declining organizations: An empirical examination of risk, performance and decline. Organization Science, 7, 524-543.

111.

Wiseman

R. M.

Gomez-Mejia

L. R.

(1998). A behavioral agency model of managerial risk taking. Academy of Management Review, 23, 133-153.

112.

Wright

Kroll

Krugg

J. A.

Pettus

(2007). Influences of top management team incentives on firm risk taking. Strategic Management Journal, 28, 81-89.

113.

Yermack

(1997). Good timing: CEO stock option awards and company news announcements. Journal of Finance, 52, 449-476.

114.

Zhang

X. M.

Bartol

K. M.

Smith

K. G.

Pfarrer

M. D.

Khanin

D. M.

(2008). CEOs on the edge: Earnings manipulation and stock-based incentive misalignment. Academy of Management Journal, 51, 241-258.