University of Southern California Dissertations and Theses

CEO reputation: who benefits -- the firm and the CEO?

(USC Thesis Other)

CEO reputation: who benefits -- the firm and the CEO?

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

CEO REPUTATION:
WHO BENEFITS – THE FIRM AND THE CEO?

by

Sung-Han Lee

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(BUSINESS ADMINISTRATION)

August 2007

Copyright 2007 Sung-Han Lee

ii
Acknowledgements

I would like to thank my dissertation chair, S. Mark Young for his continuous
guidance and encouragement in the development of this dissertation. I also thank the
other members of my dissertation committee: Wim Van der Stede, Tatiana Sandino,
and Geert Ridder (outside member) for their guidance and support. From the
beginning of my Ph.D. program, they have been great mentors to encourage my
academic work and be considerate to my personal life. They shared joy together and
encouraged me to get through tough times.
I appreciate the valuable comments and suggestions from Sarah E. Bonner,
Clara Chen, Melissa Martin, Michal Matejka, Ken A. Merchant, and seminar
participants at Arizona State University, Singapore Management University,
Southern Methodist University, University of Illinois at Chicago, University of
Southern California, and University of Utah. I am grateful to the Leventhal School of
Accounting and Marshall School of Business at the University of Southern
California for financial support.
I would like to express my deep appreciation to my lovely wife, Yoon Koh,
who has supported me whole-heartedly to complete this doctoral dissertation. I also
thank my parents, Hoon-Koo Lee and Chun-Ja Lee, and my parents-in-law, Sung-
Sam Koh and Yeoung-Heui Kim for their constant encouragement and endless love.
Last but not least, I thank my precious gem and son, Nathan Ron Lee, for
giving me the joy of a life.

iii
Table of Contents

Acknowledgements ii
List of Tables iv
List of Figures v
Abstract vi
Chapter 1. Introduction 1
Chapter 2. CEO Reputation and Firm Performance: Theory 10
2.1. Agency Theory 10
2.2. Mass Communication Research 17
Chapter 3. CEO Reputation and Firm Performance: Empirical Evidence 22
3.1. Hypotheses Development 22
3.2. Sample 27
3.3. Measurement and Research Design 28
3.4. Empirical Results 37
3.5. Discussion 44
Chapter 4. CEO Reputation and Personal Benefits: Theory 45
4.1. Agency Theory 45
4.2. Attribution Theory 48
4.3. Expectancy-Disconfirmation Theory 55
Chapter 5. CEO Reputation and Personal Benefits: Empirical Evidence 63
5.1. Hypotheses Development 63
5.2. Sample 69
5.3. Measurement and Research Design 71
5.4. Empirical Results 77
5.5. Discussion 88
Chapter 6. Conclusion and Limitations 90
Bibliography 94
Appendices 105
Appendix A 105
Appendix B 107

iv
List of Tables

Table 1: Awards from the Business Press 28
Table 2: Validity Tests of CEO Reputation Proxy 32
Table 3: Descriptive Statistics (Performance Test Sample) 38
Table 4: Pearson Correlations (Performance Test Sample) 38
Table 5: CEO Reputation and Firm Performance 40
Table 6: Reasons for CEO Turnover 73
Table 7: Descriptive Statistics (Turnover Sample) 78
Table 8: Descriptive Differences by CEO Reputation 79
Table 9: Pearson Correlations (Turnover Sample) 80
Table 10: CEO Reputation and Forced Turnover 84
Table 11: CEO Reputation and Total Compensation 85

v
List of Figures

Figure 1: Conceptual Framework 2
Figure 2: Framework for Reputation in Agency Theory 12
Figure 3: Performance Persistence and Reversals 35
Figure 4: Framework for Attribution Theory 49
Figure 5: Framework for Expectancy-Disconfirmation Theory 55
Figure 6: Disconfirmation of Expectations 57
Figure 7: Framework for Expectations and Affective Behaviors 61

vi
Abstract
In this dissertation, I examine the potential economic value of CEO
reputation: performance improvement at the firm level and personal benefits to the
CEO such as compensation and job retention. Two perspectives on CEO reputation
offer different predictions regarding the benefits of CEO reputation. The ability
perspective in the agency literature advocates the economic benefits of CEO
reputation. The symbolic image perspective from recent CEO reputation studies,
however, argues that CEO reputation does not necessarily improve firm performance
or CEO job retention. I investigate which perspective is more consistent with
empirical evidence. The results of firm performance tests show that CEOs with well-
established reputations are able to sustain good firm performance but do not turn
around poor performance. These results imply that stakeholders might have to
consider replacing the CEO – no matter how highly regarded – with a turnaround
specialist when a firm suffers financially. Another finding from job security tests
shows that CEOs with high reputation are more likely to be dismissed than CEOs
with low reputation when they perform poorly. These results suggest that the
reputations of CEOs through promoting their own images in the media do not
necessarily secure their job titles. Finally, the results of compensation tests show that
CEO reputation increases pay-for-performance sensitivity.

1
Chapter 1. Introduction
The value of intangible assets has been prominently recognized in the
accounting and other (management and marketing) literatures (Keller 1993; Hansson
2004; Wyatt 2005). Studies have examined the economic value of intangible assets
such as corporate reputation, human capital, brand name, marketing savvy, and
relationships with customers (Chatterjee et al. 1992; Simon and Sullivan 1993;
Riahi-Belkaoui 2004). The reputations of chief executive officers (CEOs) have been
argued to be the most important facet of human capital affecting firm value in both
practice and theory (Nguyen-Dang 2005; Gaines-Ross 2002). A practitioner survey
from Burson-Marsteller shows that most financial analysts would recommend a
particular company’s stock to their clients based, at least in part, on the reputation of
the company’s CEO (Gaines-Ross 2002). A large body of agency literature argues
that CEOs try to improve their reputations as good leaders with high ability for their
future careers (Fama 1980; MacLeod and Malcomson 1988; Gibbons and Murphy
1992).
Recognizing the importance of CEO reputation as an intangible asset, I
examine the potential economic value of CEO reputation: performance improvement
at the firm level and personal benefits to the CEO such as compensation and job
retention as shown in Figure 1.
1
Firm performance improvement is concerned with
whether CEO reputation provides tangible economic value to the firm, while

1
I examine job retention benefits to CEOs in this study, but firms may benefit from retaining
good CEOs through improved performance in the future. I indirectly address job retention

2
Figure 1.
Conceptual Framework

personal benefits such as financial incentives and job security
2
are related to CEOs’
motivation to build their public reputations (Dewatripont et al. 1999). Assuming
these two motivations for CEO reputation, many studies have examined whether the
reputational motivations of CEOs affect their decision-making behaviors such as
investment and reporting decisions (Hirshleifer 1993; Sridar 1994). Very few studies,
however, have empirically examined whether the reputations of CEOs provide the

benefits to firms by examining whether retained CEOs turn around poor performance in the
near future.
2
Job security includes promotion to another company as well as retention in the current firm.
This study examines the retention of CEOs in the current firms.

3
sought-after benefits and whether the expectation of superior performance of highly
reputed CEOs, in fact, is delivered.
3
One of the reasons for the small number of
empirical studies related to CEO reputation is the difficulty of measuring the
reputations of CEOs. To overcome this concern, I measure CEO reputation using
two proxies which are most often employed by previous studies: CEO awards from
business journals and media exposure (Johnson et al. 1993; Milbourn 2003; Francis
et al. 2004; Malmendier and Tate 2005).
Empirical investigation of the economic benefits of CEO reputation is
important because two perspectives on CEO reputation offer different predictions
regarding the consequences of CEO reputation. The ability perspective found in the
agency literature advocates the economic benefits of CEO reputation (MacLeod and
Malcomson 1988; Gibbons and Murphy 1992). The symbolic image perspective
from recent CEO reputation studies, however, argues that CEO reputation does not
improve either firm performance or the CEO’s job retention (Francis et al. 2004;
Malmendier and Tate 2005). Previous studies adopt one perspective and develop/test
hypotheses based on the perspective.
4
Many studies in the accounting literature
support the ability perspective and take the economic benefits of CEO reputation for
granted. Studies in the management literature, however, recognize the psychological

3
Performance benefits are related to turnover and compensation decisions, because the
superior performance of highly reputed CEOs would result in higher job retention and
compensation.
4
The two perspectives have been developed and supported in separate research streams
without controversies. Some recent CEO reputation studies, however, criticize the ability
perspective and support the symbolic image perspective. This study investigates CEO
reputation from an ignostic point of view.

4
aspect of CEO reputation – perception (Fombrun and Shanley 1990; Kilduff and
Krackhardt 1994; Rindova et al. 2005). This study acknowledges that both the
perspectives are based on valid arguments and that the dual aspects of CEO
reputation (ability and perception) influence economic consequences simultaneously.
This study further tests whether one aspect of CEO reputation dominates the other
empirically.
The ability perspective, based on agency theory, argues that the reputation of
a CEO is the revealed (Bayesian updated) ability of the CEO as shown in Figure 1
(Gibbons and Murphy 1992; Johnson et al. 1993; Milbourn 2003).
5
If positive CEO
reputation reflects superior ability, CEO reputation improves firm performance.
Analysts recommend a company’s stock based on CEO reputation because a CEO
with a well-established reputation, believed to have high ability, will sustain good
performance or turn around poor performance (Gaines-Ross 2002). A CEO with a
lesser reputation, on the other hand, will reduce good performance or perpetuate poor
performance. A board of directors will give a highly regarded CEO more
opportunities to make up for his or her mistakes, because they believe that the CEO
will reverse current poor performance in the near term (Gaines-Ross 2002). Thus, the
ability perspective suggests that the reputation of the CEO will enhance his or her
likelihood of retaining the top job in the firm.

5
Many agency studies define CEO reputation as Bayesian updating about the CEO’s ability.
They argue that the reputation of a CEO is the adjusted estimate of the CEO’s ability using
past and current performance information. Johnson et al. (1993) find that good firm
performance help CEOs achieve reputation, which validates the argument of CEO reputation
as the ability of the CEO.

5
The symbolic image perspective, however, argues that the reputation of a CEO
is mainly the perceived image of the CEO via the media as shown in Figure 1
(Malmendier and Tate 2005).
6
If CEO reputation mainly reflects the symbolic image
of a CEO instead of ability, a well-known CEO will not necessarily maintain good
firm performance or reverse poor performance. Celebrity CEOs are often media-
created and do not live up to the heightened expectations of stakeholders,
7
a
phenomenon called the “CEO Disease” in the business press (Byrne, Symonds, and
Siler 1991). This CEO disease phenomenon lends support to the prediction that a well-
known CEO may serve as a scapegoat for poor performance as argued by Khurana
(2002) in his article “The curse of the superstar CEO.”
8
Positive CEO reputation often
raises the expectations of stakeholders, which increases gaps between expectations and
actual performance. Increased gaps between elevated expectations and poor
performance are mostly attributed to a high-profile CEO because people believe that a
symbolic leader determines the success or failure of an organization (Pfeffer 1977).
These increased gaps due to elevated expectations and the attribution of poor
performance to the well-known CEO decrease the satisfaction of stakeholders (Oliver

6
The perceived image of the CEO may come from past performance, but the media
considerably affects the perception of the CEO by the business community (shareholders,
boards of directors, analysts, and potential investors). The symbolic image perspective does
not deny the importance of past performance or ability rather puts more emphasis on the
possibility of misperception due to media exposures.
7
Malmendier and Tate (2005) show that celebrity CEOs often get distracted by non-value-
added activities (assuming board seats and writing books) to promote their own images in
the media.
8
By dismissing a well-known CEO, the board of directors signals to the business community
that better strategies will be taken to improve firm performance (Gamson and Scotch, 1964).
Dismissing a more-reputed CEO will provide more conviction to stakeholders for strategic
changes than dismissing a less-reputed CEO.

6
1980) and result in more CEO turnovers. Thus, the symbolic image perspective
suggests that the reputation of a CEO will impair his or her job retention under the
circumstances of poor performance.
The overall results of this study show that CEO reputation increases
compensation but not necessarily firm performance and job retention. First, the tests of
firm performance improvement partially support both perspectives: the ability
perspective for maintaining good performance (performance persistence) and the
symbolic image perspective for turning around poor performance (performance
reversals). I find that CEOs with well-established reputations sustain good firm
performance from one year to the next better than do CEOs with lesser reputations.
Well-known CEOs, however, do not necessarily turn around poor performance in the
near term. The results imply that stakeholders might have to consider replacing a well-
known CEO with a turnaround specialist when a firm suffers financially, because the
CEO may not turn around poor performance. The results also suggest that investors
had better be careful about investing in a company solely based on a celebrity CEO,
especially once firm performance slides.
Second, the tests of job retention show that more-reputed CEOs are more likely
to be dismissed under the circumstances of poor performance than less-reputed CEOs,
which supports the symbolic image perspective. These results might be related to the
results of firm performance reversal tests. Because well-known CEOs do not
necessarily turn around poor performance, even if they are leaders with strong
reputations, boards of directors would replace high-profile CEOs with turnaround

7
specialists. The results suggest that CEOs can secure their job positions by improving
firm performance rather than by promoting their images to the media. Well-known
CEOs risk serving as scapegoats because of the attribution of poor performance to
high-profile CEOs and heightened expectations from their positive reputations.
Finally, the tests of CEO compensation show that CEO reputation increases
pay-for-performance sensitivity. The results support the ability perspective and the
symbolic image perspective, because both these perspectives predict more sensitive
compensation to firm performance for highly regarded CEOs. The ability perspective
argues that CEOs with high reputation (ability) pressure firms to adopt performance-
based compensation, which increases pay-for-performance sensitivity. And the
symbolic image perspective argues that the attribution of performance to a well-known
CEO stimulates not only significant financial incentives for good performance, but
also significant financial penalties for poor performance. The results are consistent
with the evidence that CEO reputation is positively associated with equity-based pay
sensitivity (Milbourn 2003). This study, however, shows that CEO reputation
increases accounting performance pay sensitivity as well as market performance pay
sensitivity.
This study has three important contributions. First, this dissertation
contributes to the understanding of a more complete picture of the economic benefits
of CEO reputation. A large body of literature emphasizes the importance of CEO
reputation (Fama 1980; Gibbons and Murphy 1992) and uses CEO reputation
arguments to develop predictions about CEO behaviors (Hirshleifer 1993; Sridar

8
1994). However, scarce literature empirically investigates whether the reputation of a
CEO provides economic benefits to the firm and to the CEO himself or herself. This
is, to my best knowledge, the first study to examine whether CEO reputation affects
forced turnover decisions. This study also examines the effects of CEO reputation on
accounting performance pay sensitivity as well as stock performance pay sensitivity
(Milbourn 2003).
Second, this study tests which perspective of viewing CEO reputation is
empirically supported. Two perspectives emphasize different aspects of CEO
reputation which offer different predictions about the economic benefits of CEO
reputation. The results show that the reputation of a CEO reflects ability to maintain
good performance, which supports the ability perspective. CEO reputation, however,
does not necessarily provide performance benefits for a struggling firm to turn
around in the near future, which supports the symbolic image perspective. The
reputation of a CEO also impairs his or her job retention under the circumstances of
poor performance, which supports the symbolic image perspective.
Finally, this study investigates whether CEO reputation is valuable to firms as
an intangible asset. Many studies have examined whether corporate reputation and
brand name are associated with firm values (Riahi-Belkaoui 2004; Simon and
Sullivan 1993). Very few empirical studies have examined whether CEO reputation
is value-relevant and can be recognized as an intangible asset. This is the first study
to examine whether CEO reputation improves performance persistence and reversal
rather than performance level (Malmendier and Tate 2005).

9
The remainder of this dissertation is organized as follows. Chapter 2 reviews
the literature on the relation between CEO reputation and future firm performance.
Chapter 3 empirically investigates whether CEO reputation improves future firm
performance. Chapter 4 provides the literature review on whether CEO reputation
provides personal benefits such as job retention and compensation to CEOs. Chapter
5 presents evidence on the effect of CEO reputation on job retention and
compensation. Finally, Chapter 6 concludes and addresses the limitations of the
study.

10
Chapter 2. CEO Reputation and Firm Performance: Theory
2.1. Agency Theory
Theoretical agency studies view the reputation of a CEO as a learning
process about the CEO’s ability (Johnson et al. 1993; Milbourn 2003). Labor market
participants in business community estimate unobservable true ability using prior
information and form beliefs over the CEO’s ability using updated information.
Thus, as more performance information becomes known to these stakeholders, the
CEO’s ability is further revealed and estimated ability approaches to true ability
(MacLeod and Malcomson 1988; Gibbons and Murphy 1992; Johnson et al. 1993).
9

Agency theory explains how to best design contracts between the party (the
principal) who owns the capital and delegates tasks and the party (the agent) who
undertakes the tasks (Baiman 1982, Laffont and Martimort 2002). Agency research
has examined how the owner and the manager pursue their welfares using the firm’s
information systems and employment contracts (Baiman 1982). Agency theory plays
an important role in economics because of its applicability to modern business
environments, especially under the circumstance of the separation of ownership and
control (Jensen and Meckling 1976). The modern capital market facilitates the
diffuse ownership structure of firms, which in turn results in the separation of
ownership from management. The owner of a firm with diffused ownership

9
It takes some time for the board of directors to estimate the CEO’s ability reasonably
because firm performance is determined by many factors other than the CEO’s ability
(Gibbons and Murphy 1992). Thus, it is difficult to isolate the CEO’s influence from other
business factors. Moreover, a change in the firm’s environment makes this isolation of the
CEO’s contribution more difficult.

11
structures should delegate management tasks to the manager of the firm, which fits a
pure agency relationship (Baiman 1982).
Conflicting objectives between the owner and the manager and imperfect
information about the agent (information asymmetry) result in agency costs which
prevent the two parties from achieving optimal the first-best contract as shown in
Figure 2.
10
The objectives of the principal and the agent are different and even
conflicting because the utility of the owner is a decreasing function of payment to the
employee whereas that of the manager is an increasing function of compensation
from the owner (Laffont and Martimort 2002). Since the agent pursues his or her
own self-interest different from that of the principal, the manager attempts to
maximize his or her personal benefits at the cost of the owner.
11

The delegation of management tasks to the manager who has a different
objective than that of the owner becomes problematic when information about the
manager’s effort and ability is limited. If the principal has perfect information about
the agent, the owner intervenes to prevent the manager from sacrificing the
principal’s utility for the agent’s benefits. It is, however, difficult for the principal to
control or prevent this dysfunctional behavior because information about the agent’s
actions and ability is not available.
Agency costs are classified by the two types of private information: hidden
actions (effort) and hidden knowledge (ability). If the actions of the agent are

10
The first-best contract refers to an optimal contract given the condition that the principal
has perfect information about the agent.
11
This assumption of pursuing a private interest has its limitations, because psychological
aspects and social norms influence the decision making of individuals in many cases.
Agency theory, however, argues that the assumption of self-interest driven behaviors
explains many aspects of business decisions (Laffont and Martimort 2002).

12
Figure 2.
Framework of Reputation in Agency Theory

observed by the principal, the amount of the agent’s effort can be used as the basis
for the compensation contract (Baiman 1982). The owner can implement an
employment contract to make the manager exert the first-best level of effort. The
unobservability of the agent’s effort, however, results in moral hazard under which
the agent reduces his or her inputs (effort) and the principal cannot be ascertain
whether the agent exerts his or her best effort. If the ability of the agent is observed
by the principal, the owner can assign an appropriate task to the manager to operate
the firm efficiently and maximize outputs. The unobservability of the agent’s ability,
however, induces adverse selection under which an agent with low ability pretends to

13
have high ability and the principal cannot be sure whether the agent represents an
appropriate productivity type (Laffont and Martimort 2002). Agency costs such as
moral hazard and adverse selection prevent society from reaching the first-best
solution that could be obtained if all private information about the agent is known to
public or the principal.
The agency literature suggests two approaches to reduce or minimize agency
costs: contract designs and monitoring as shown in Figure 2. Traditional agency
studies have mainly stressed the primacy of incentive contracts to minimize agency
costs. Agency theory requires that the principal chooses an incentive contract, which
maximizes his or her own expected utility after taking into consideration of the
agent's utility-maximizing behaviors (Laffont and Martimort 2002). Once the
compensation contract is given, the agent will allocate effort to maximize his or her
utility. In other words, the employment contract is designed to align different self-
interests between the owner and the manager. To align different self-interests,
shareholders either reward managers based on the output of the firm or provide
ownership to managers as shown in Figure 2.
Performance-based compensation using observed outputs as a substitute of
unobserved inputs (effort and ability) motivates the agent by aligning the interest of
the agent to that of the principal (Laffont and Martimort 2002). In a typical agency
framework, the principal designs a contract to motivate the risk- and effort-averse
agent to exert unobservable effort in a production process. The principal receives the
residual after compensating the agent. In this formulation, the optimal contract bases
compensation for the agent on observed output which the principal uses as an

14
indicator of the effort expended by the agent (Baiman 1982). To best facilitate
output-based compensation, research has pursued which performance measure
motivates managers to exert the optimal level of effort. Many previous studies have
focused on the comparison between accounting-based measures and market-based
measures (Sloan, 1993; Barclay et al., 2000). Recent studies, inspired by balanced
scorecard (Kaplan 1992), admit that most firms employ multiple performance
measures including market-based measures, accounting-based measures, and
nonfinancial measures (Kaplan and Norton 2001). Research has shown that each
measure has its own advantages and disadvantages (Ittner and Larcker, 1998).
Convergence-of-interest hypothesis advocates the benefit of increased
management ownership (Jensen & Meckling 1976). If a manager owns 100 percent
of the firm’s equity, he or she will make operating decisions by considering the
marginal benefits of increasing outputs and the marginal costs of exerting effort,
which achieves the first-best solution (Jensen & Meckling 1976). As the owner-
manager sells equity claims, however, (minority) shareholders incur agency costs
due to management’s shirking and perquisite consumption (Jensen & Meckling
1976, Ang et al. 2000). Many studies argue that increasing the equity ownership of
management by granting restricted stocks and stock options reduces agency costs
(Core and Guay 1999 & 2001). Ang et al. (2000) find that agency costs significantly
decrease (i) when an insider instead of an outsider manages the firm; (ii) when
managers’ ownership share increases; and (iii) when the number of non-management
shareholders decreases.

15
Traditional agency research does not put emphasis on monitoring but
suggests the benefit of monitoring when it is difficult to align incentives using
employment contracts. Monitoring is defined as observation of an agent’s actions,
ability, or outcomes that is achieved by auditing, supervision by boards of directors,
formal control systems, budget restrictions, recruiting process, and other devices
(Jensen and Meckling 1976, Tosi et al. 1997). Monitoring can resolve or prevent
agency costs by capturing information about the manager’s effort and ability or
reducing information asymmetry between the principal and the agent (Laffont and
Martimort 2002). Monitoring as a vehicle of extracting information from the
manager is classified into direct supervision and selection process using the
manager’s reputation as shown in Figure 2.
Direct supervision is more direct and binding control systems preventing the
manager from behaving opportunistically than incentive contracts. Shavell (1979)
and Holmstrom (1979) argue that monitoring without any cost provides benefits to
the principal when the unobservability of the agent’s actions (information
asymmetry) has negative effects on the principal’s utility. Direct supervision,
however, has its limitation to observe the agent’s ability or effort perfectly and
becomes very costly to overcome this limitation, which makes complete monitoring
virtually impossible. Regardless of this limitation of direct supervision, strong
monitoring is beneficial and economical (the benefits outweigh the costs) when
managerial incentives gaps between the principal and the agent is great (Beatty and
Zajac 1994). Many empirical studies show the benefit of monitoring (Morck et al.
1988, Wruck 1989, Pagano and Roell 1998).

16
Selection process using the reputation of a manager can be regarded as broad
monitoring to find out the ability of the manager. Recruiting employees with high
ability and matching the employees with appropriate tasks require complex selection
procedures and significant costs to employers. Even after these complex recruiting
procedures, however, the true ability of the manager is not known with certainty to
shareholders, analysts, and the board of directors. Stakeholders often rely on the
reputation of the manager to estimate his or her ability. MacLeod and Malcomson
(1988) argue the dual roles of reputation: information about the ability of the
manager and motivation for the manager to achieve satisfactory performance.
Theoretical agency studies explain that the reputation of a CEO is derived
from a learning process (Bayesian assessment
12
) about the CEO’s ability (Johnson et
al. 1993; Milbourn 2003). Agency theory often assumes the unobservable ability
(denoted a) of an employee with shared beliefs about the distribution of ability
between the employer and the employee. The owner has to use the estimate of the
manager’s ability (denoted â) to assign a task and design a compensation contract.
The best estimate of the manager’s true ability in the beginning of the first year is the
population mean (denoted ā). Even though the manager’s true ability is not
observable, we can make inferences about the ability of the manager by observing
performance over time. The estimate of the manager’s ability at the end of period t
(denoted â
t
) is a weighted average of current performance (denoted y
t
) and estimated
ability at the previous period (denoted â
t-1
). The manager’s estimated ability at the

12
Bayesian inference involves collecting evidence to evaluate a hypothesis or estimate a true
value. The degree of belief in the hypothesis changes as evidence accumulates. Evidence
consistent with the hypothesis increases a degree of belief, whereas evidence inconsistent
with the hypothesis decreases a degree of belief.

17
end of period t approaches true ability over time with monotonically increasing
precision (DeGroot 1970). Appendix A explains detailed processes to show
monotonically increasing precision or decreasing variance over time. Johnson et al.
(1993) find that CEO awards are positively associated with past firm performance,
which shows that the reputations of CEOs reflect their true ability.

2.2. Mass Communication Research
Recent CEO reputation studies argue that the reputation of a CEO is greatly
influenced by the media exposure of the CEO as a superstar (Malmendier and Tate
2005). Mass communication research has shown the powerful effects of the media on
constructing audiences’ opinions and images of CEOs (Ball-Rokeach and Cantor
1986, Smith 1995, Deephouse 2000). This stream of research suggests the two main
roles of the media: recording public knowledge (a surveillance role) and influencing
public opinions (an agenda-setting role). The media serve not only as vehicles for
portraying reality and recording public events but also as active agents shaping
public opinions and images through editorials and feature articles (Fombrun and
Shanley 1990).
Communication research has shown that the media record public knowledge
and provide information about environments to audiences (Deephouse 2000).
Schramm (1949) views the function of news as reporting “the essential framework of
an event,” which suggests the surveillance role of the media (Lasswell 1949). A
critical norm among journalists is to report events, issues, and opinions thoroughly
and unbiasedly (Weaver and Wilhoit 1986). The Code of Ethics of the Society of
Professional Journalists (1996) states “The duty of the journalist is to further those

18
ends by seeking truth and providing a fair and comprehensive account of events and
issues. Conscientious journalists from all media and specialties strive to serve the
public with thoroughness and honesty.” Reporters and their employing organizations
attempt to avoid being accused of bias and embarrassed by inaccurate stories (Hallin
1986). If a publishing organization is known to convey inaccurate news consistently,
the credibility of the organization is undermined, which in turn results in decreased
circulation. In sum, an industry norm, journalist ethics, social or legal responsibility,
and profitability concerns constantly put pressures for the media to record public
knowledge thoroughly and unbiasedly.
Research, however, suggests that the media not only report reality to the
public but also influence public opinions (Deephouse 2000). McCombs and Shaw
(1972) argue that the media coverage of certain issues attracts attention from the
public and increases the salience of these issues, which implies the agenda-setting
role of the media. The media provide data that may become important and
meaningful information for the business community. A good example of how the
media can influence public opinions is the effects of Nazi and Communist
propaganda on social revolutions (George 1959, Lasswell et al. 1965). The news
press disseminates private information across wider audiences (Hayward and
Hamrick 1997). Some members of the public may have direct experience and
knowledge about an event or issue. The media collect the knowledge, publicize the
event or issue, and influence the opinions of the members without direct knowledge.
Behr and Iyengar (1985) show that CBS news stories on inflation and energy
increase the public awareness of these issues in a two-month period. Ader (1995)

19
finds that the increased coverage of pollution in The New York Times results in
increased concerns about pollution in the general public.
Recent research moves beyond an agenda-setting role and examines the
effects of the media on attitudes, behaviors, images, and the social construction of
reality (Gamson et al. 1992). The role of the media shaping public opinions becomes
more significant if media bias is considered (Gentzkow and Shapiro 2006).
Gentzkow and Shapiro (2006) show that slanted information by the media through
selective omission, choice of words, and the credibility of primary information
source provide different impression about what happened.
The roles of the media to record and influence public knowledge and
opinions suggest media exposure as one of the important sources of CEO reputation.
Media exposure is critical for a CEO’s reputation development, since the media not
only solidify the attribution of the firm’s success to the CEO but also diffuse the
CEO’s prestige across wider audiences (Cameron and Whetten 1983, Hayward and
Hambrick 1997). Dyck and Zingales (2002) argue that individuals obtain much of
their information from the mass media, which implies the importance of the media
coverage of CEOs in building their credibility and reputations. The media coverage
of CEOs removes some uncertainties, brings more transparency, and puts emphasis
on the viability of CEO vision, which helps the development of CEO reputation
(Nguyen-Dang 2005).
Chen and Meindl (1991) examine how the media’s portrayals of Donald
Burr, a CEO of People Express Airlines, influence the business community’s
perceptions of him. Once the business community constructs the favorable image of
the CEO, the mass media remain faithful to that image even if firm performance

20
decreases significantly (Chen and Meindl 1991). Meindl et al. (1985) find that the
media have a propensity to attribute successful companies to individual leaders and
that celebrity CEOs are often portrayed as “heroic” in the media. This impression
management perspective suggests that misperceptions about some celebrity CEOs
caused by media exposure possibly increase their reputations without the appropriate
evaluation of their ability.
Principles from social cognition theory help explain how and why only some
information from the media is incorporated into reputation formulation (Wartick
1992). Social cognition theory starts with two assumptions regarding individuals’
information processes (Fiske and Taylor 1984). First, individuals have only limited
capacities to process information for their judgments and decisions. Next, the
cognitive processes of perceiving, storing, retrieving and inferring lie along a
continuum from automatic (effortless) to “effortful” processes (Kiesler and Sproul
1982). Because stakeholders process information with limited capacities, they have
to pick and choose information from the media by combining automatic and effortful
processes. These selection processes due to limited capacities imply the influence of
the media on the images of CEOs.
Social cognition theory has spawned three bodies of theories: information
processing theory, social perception theory, and social motivation theory.
Information processing theory argues that salience, discrepancy, recency, and
stereotypes are important factors in organizing information in memory. Wartick
(1992) argues that information processing theory is the most relevant theory to
explain a relationship between media exposure and reputation formulation. He

21
suggests that stakeholders’ information processing to formulate CEO reputation is
influenced by the amount, tone, and recency of media coverage.
Social perception theory addresses how information is encoded and retrieved
depending on the power attached with information. Augmentation of “powerful”
sources and discounting of “weaker” forces due to limited processing capacities
induce bias, which in turn results in misperceptions about CEOs. Social motivation
theory explains how individuals selectively perceive their environments by
overvaluing or undervaluing certain information and creating preferences toward
some information. The theory explains how stakeholders selectively scan and
interpret information regarding their CEOs, which implies the image formulation of
CEOs by the media.
Knowing the importance of the media in reputation development, CEOs
attempt to command attention from analysts, investors, and the press so that they
become superstars or celebrities. McQuail (1985) suggests that managers
strategically attempt to influence their images or the images of their firms using press
articles and mass media presentations.
13
Khurana (2003) argues that a critical skill
that CEOs need to gain credibility and confidence from the business community is to
promote their bright future to the media and attract attention from the business press.
Malmendier and Tate (2005) argue that CEOs with celebrity status strive to maintain
their reputations by promoting their images to the media, which causes CEOs
distracted from performance improvement.

13
Given the biases of human judgments due to information availability (Tversky and
Kahneman 1974), media visibility (information availability) shapes the assessment of
audiences (McQuail 1985).

22
Chapter 3. CEO Reputation and Firm Performance: Empirical Evidence
3.1. Hypotheses Development
Reputation
The media’s heightened interest in CEOs’ actions and the visibility of CEOs
as superstars reflect the importance of CEO reputation (Gaines-Ross 2002). The
attention of the media and investors has been increasingly focused on CEOs to the
point where many CEOs are so well-known in the labor market and the business
press that they seem to have achieved celebrity status (Malmendier and Tate 2005).
Business leaders are as well known as movie or TV stars and exposed to the general
public as heroes or superstars (Hamilton and Zeckhauser 2004). For example, three
business leaders – Ted Turner of Turner Communications, Andrew Grove of Intel,
and Jeff Bezos of Amazon.com – were each awarded as Time’s Person of the Year.
CEO reputation which is closely interrelated to the well-being of a company
has been well recognized in practice. A survey by Burson-Marsteller shows that
more than 90% of financial analysts would recommend stock to their clients based
on the reputation of a CEO (Gaines-Ross 2002). A joint survey by Hill &
Knowlton/Korn/Ferry also finds that a CEO is the most important factor in building
corporate reputation (Stock 2003).
The importance of CEO reputation in practice calls for investigating the
consequences of CEO reputation. But before I investigate the economic benefits of

23
CEO reputation, reputation must first be defined.
14
The Merriam-Webster Dictionary
defines reputation as “recognition by other people of some characteristic or
ability.”
15
There seem to be two major aspects in this definition of reputation:
recognition/perception and characteristic/ability.
16
The dual aspects of CEO
reputation are clearly demonstrated in an examination of the two reasons why CEOs
become well-known in the business community. Some CEOs become famous
because they have superior ability, which translates into impressive performance as
leaders, whereas some CEOs become widely recognized because they are exposed to
the media favorably.
17

In this chapter, I investigate whether current CEO reputation improves future
firm performance. Johnson et al. (1993) show that current CEO reputation is the ex
post consequence of good past performance, which supports the argument that
positive reputation is a signal of superior ability. Yet, few studies have examined
whether current CEO reputation is an ex ante indicator of future firm performance
and whether the expected superior performance of highly reputed CEOs is delivered.
The investigation of the effects of CEO reputation on future performance tests

14
I mainly consider positively well-known reputation instead of negatively (notoriously)
widely-known reputation.
15
Note that the definition of reputation from the Webster Dictionary and that in the agency
literature (footnote 5) have a common factor of ability. The definition of reputation in many
agency studies, however, mainly pays attention to objective knowledge about ability whereas
that in this study recognizes psychological perceptions as well as rational expectations.
16
Hall (1992) defines corporate reputation as the combination of subjective perceptions held
by stakeholders and objective knowledge. Schwaiger (2004) argues that “corporate
reputation” does not differ from “corporate images” in several publications (Bromley 1993).
17
The dual aspects of reputation are not mutually exclusive, because some CEOs have
superior ability and great media exposure. Unfortunately, I cannot disentangle the dual
aspects of reputation clearly, since I cannot observe CEOs’ true ability.

24
whether this argument is empirically supported and whether CEO reputation has
economic value as an important facet of human capital.
The dual aspects of reputation induce two perspectives predicting different
consequences of CEO reputation: an ability perspective and a symbolic image
perspective.
18
The ability perspective argues that CEO reputation reflects the
superior ability of CEOs and thus improves firm performance (MacLeod and
Malcomson 1988; Gibbons and Murphy 1992). The symbolic image perspective,
however, maintains that CEO reputation reflects the media-created images of CEOs
and thus does not necessarily improve performance (Malmendier and Tate 2005). I
test which aspect of CEO reputation dominantly affects firm performance and which
perspective is more consistent with empirical evidence.
19

Ability Perspective
The ability perspective found in the agency literature maintains that the future
firm performance of more-reputed CEOs is likely to exceed that of less-reputed
CEOs (MacLeod and Malcomson 1988; Gibbons and Murphy 1992). The agency
literature describes reputation formation as a learning process (Bayesian assessment)
about the CEO’s ability (Johnson et al. 1993; Milbourn 2003). CEOs with high

18
The two perspectives focus on the different aspects of CEO reputation. The ability
perspective puts more emphasis on ability and views reputation as rational expectation of
ability using a Bayesian process. The symbolic image perspective, however, pays more
attention to the misperception of the business community and considers reputation inflated
ability.
19
This study does not test which perspective is valid, but acknowledges that both the
perspectives are based on valid arguments. I test which perspective is dominant empirically
when the dual aspects of CEO reputation have countervailing forces influencing economic
outcomes.

25
ability will be more likely to sustain good firm performance, whereas CEOs with low
ability will be more likely to sabotage good performance. Skilled CEOs will be more
likely to reverse poor performance, whereas unskilled CEOs will be more likely to
continue poor performance. If the reputations of CEOs reflect their ability, CEOs
with well-established reputations are more likely to maintain good performance or
reverse poor performance than CEOs with lesser reputations.

Symbolic Image Perspective
The symbolic image perspective, however, suggests that CEO reputation does
not necessarily improve firm performance (Francis et al. 2004; Malmendier and Tate
2005). Recent CEO reputation studies view the reputation of a CEO mainly as the
symbolic image
20
of the CEO created by the media (Malmendier and Tate 2005).
21

The media has a great influence on the image of a CEO as a superstar (Malmendier
and Tate 2005). Celebrity CEOs often do not live up to the expectations of the
business community once they have reached superstar status (Byrne, Symonds, and
Siler 1991). Malmendier and Tate (2005) view the celebrity status of CEOs as a
media-induced symbolic image. They show that CEOs with celebrity status
underperform relative to comparable CEOs who lack the visibility associated with
awards and other recognition. They find that pursuing celebrity status induces the
dysfunctional behavior of spending time on non-value-added activities (writing

20
Schwaiger (2004) argues that “corporate reputation” does not differ from “corporate
image” in several publications (Bromley 1993).
21
There is a stream of research examining the power of the press on compensation and other
economic consequences (Core et al. 2005). Those studies investigate whether the contents of
news articles affect economic decisions directly whereas this study examines the
consequences of CEO reputation which is affected by media exposures.

26
books and assuming board seats of other companies). High-profile CEOs get
distracted from business operation to get exposed to the media, and they may waste
their time and effort that could otherwise be used on improving firm performance.
Thus, if the reputations of CEOs reflect their media-created symbolic images, high-
profile CEOs do not necessarily maintain good firm performance or reverse poor
performance.
I posit the following competing hypotheses to test whether CEO reputation
improves future firm performance and reflects the ability of the CEO either in
maintaining good firm performance or reversing poor performance.
22

Hypothesis 1a (the ability hypothesis): CEOs with high reputation are more
likely to maintain good firm performance than are CEOs with low reputation.
Hypothesis 1b (the symbolic image hypothesis): CEOs with high reputation
are not more likely to maintain good firm performance than are CEOs with
low reputation.

Hypothesis 1c (the ability hypothesis): CEOs with high reputation are more
likely to reverse poor firm performance than are CEOs with low reputation.
Hypothesis 1d (the symbolic image hypothesis): CEOs with high reputation
are not more likely to reverse poor firm performance than are CEOs with low
reputation.

22
I develop separate hypotheses for the ability of maintaining good performance
(performance persistence) and that of turning around poor performance (performance
reversals), since Daine et al. (2005) find that a CEO does not necessarily have both sets of
ability. This measure of CEO ability (performance persistence) introduced by Daine et al.
(2005) has often been used to evaluate the ability of mutual fund managers and investment
analysts (Brown et al. 1999; Mikhail et al. 2004).

27
3.2. Sample
The sample of CEOs used in this study is collected from the ExecuComp
database and matched to accounting and market performance from the Compustat
and CRSP database. I first identify 4,612 unique CEOs from the ExecuComp
database for the years 1993 – 2004. I exclude 271 CEOs whose tenure is less than
one year,
23
which leaves the total of 4,341 CEOs in the sample. I collect CEO award
data from the leading business press such as Businees Week and Financial World
(Johnson et al. 1993; Malmendier and Tate 2005). I identify 661 CEOs with one or
more awards such as CEOs of the Year and 3,680 CEOs without any record of award
recognitions I also collect media exposure measures of all 4,341 CEOs from the
Factiva (previously Dow Jones Interactive) database (Milbourn 2003; Francis et al.
2004).
I identify a sample of 21,950 CEO years from the ExecuComp database for
the years 1993 – 2004. Next, I exclude CEO years if there is a change of a CEO so
that firm performance is solely attributed to one CEO. The tests of performance
persistence require the performance of the current and prior years. Thus, I exclude
the first, second, and last year as a CEO (t = 1, 2, and n), which leaves 14,917 CEO
years from the ExecuComp database. After merging this data set with Compustat for
accounting earnings and CRSP for stock returns, the final sample consists of 13,931
observations.

23
CEOs who stay less than one year are mostly interim CEOs. Also if a CEO stays less than
one year, it is difficult to attribute annual firm performance solely to the CEO.

28
3.3. Measurement and Research Design
Reputation
As per the definition of reputation from the Merriam-Webster Dictionary, a
CEO’s reputation is the recognition of his/her ability by the business community
(other executives, the board of directors, analysts, and investors). CEO reputation,
however, is difficult to be measured empirically. In this study, I employ the two most
often employed proxies of CEO reputation: CEO awards from the prestigious
business press (Johnson et al. 1993; Malmendier and Tate 2005) and media exposure
(Milbourn 2003; Francis et al. 2004).

Table 1.
Awards from the Business Press

Magazine Awards Period Number of awards
Business Week Best Managers 1988 –
current
1988 – 1991: 6
1992 – 1995: more than 6
1996 – current: 25
Financial World CEOs of the Year 1975 –
1997
Gold Award: 1
Silver Award: 10
Bronze and certificates of
distinction: 150 – 180
Forbes Best Performing
CEOs
2001 -
current
2001: 5
2002 – current: 10
Industry Week CEOs of the Year 1986 -
current
1986 - 1992: 12
1993 – current: 1 or more
Chief Executive CEO of the Year 1987 -
current
1987 – current: 1
Electronic Business CEO of the Year 1997 -
current
1997 – current: 1

29
First, I collect CEO award data such as “CEOs of the Year” or “Best
Managers” from various business journals: Business Week, Financial World, Forbes,
Industry Week, Chief Executive, and Electronic Business.
24
The winners of CEO
awards from these business magazines have become superstars and highly reputed in
the business community. I tabulate magazines, the period of awards, and the number
of awards bestowed each year in Table 1. Detailed nomination processes and
selection criteria for each magazine are described in Appendix B. AWARD is an
indicator variable set equal to 1 if the CEO was given any award for the last five
years or 0 otherwise.
25

Second, I collect the press coverage of CEOs by searching news articles from
all publications in the Factiva database. Given the agenda-setting role of the media
by attracting attention from the public, media exposure is likely to contribute to
reputation development. McQuail (1985) argues that the public are more likely to
perceive a manager as important and famous if the manager receives greater media
attention. Considering the information availability biases of human judgments
(Tversky and Kahneman 1974), measuring CEO reputation using media visibility has
its own validity. The amount of information disseminated through the business press

24
Johnson et al. (1993) collect data only from Financial World, because their sample period
only covers until 1987 (most business journals started bestowing the CEOs of the year after
1988). Malmendier and Tate (2005) collect the most comprehensive CEO award data from
Business Week, Financial World, Chief Executive, Forbes, Industry Week, Moningstar.com,
Time, Time/CNN, Electronic Business, and Ernst & Young.
25
I set five-year as a threshold criterion, since five-year seems to be a reasonable period for
the business community to remember honorees. I run robustness tests by varying this period
to three-year and permanently once a CEO receive any award. I set three-year as a threshold
criterion for media exposure, and I run sensitivity tests by varying this period to five- and
two-year.

30
and the mass media influences the manager’s reputation development (McQuail
1985). Kotha et al. (2001) argue that contrary to popular beliefs, media exposure
itself, rather than the tone (either positive or negative) of news articles, influences the
process of constructing opinions, images, and judgments about CEOs.
The media coverage of CEOs refers to the news reports relating to a specific
CEO of the firm within a defined period (Wartick 1992). I employ the number of
news articles given the name of a CEO and the company of the CEO following
Milbourn (2003), Francis et al. (2004), and other studies measuring reputation.
26
I
use the sum of multiple years’ news counts, because a CEO develops his/her
reputation over career years. More specifically I sum the last three years of media
counts to measure CEO reputation. After I collect the news coverage of all CEOs, I
derive MEDIA that is set equal to 1 if the CEO’s media count for the last three years
is more than the top 20% of the industry or 0 otherwise.
27

Given the controversy of measuring CEO reputation, I run two tests to
validate the proxies of CEO reputation. First, the sample of this study indicates the
Pearson correlation of 0.239 (p < 0.01, two-tailed) between AWARD and MEDIA.
28

26
I consider the multiple names of companies if the CEO serves multiple companies over
his/her career years.
27
I convert media coverage counts (a continuous variable) to MEDIA (an indicator variable),
since I need both CEO reputation measures, AWARD and MEDIA, to be comparable. I set
the threshold value of top 20% of the industry because only 15% of CEOs was bestowed any
award from the business press. An indicator variable also has advantages of interpreting
results and increasing the power of tests.
28
Pearson correlation of 0.239 is reasonably high compared to prior studies (Milbourn
2003). Milbourn (2003) employs four proxies of CEO reputation. The greatest correlation
among four proxies is 0.083 (p < 0.01, two-tailed) and even some proxies are not
significantly correlated.

31
I also examine whether AWARD is positively associated with MEDIA using logistic
regressions to check whether the correlation holds even after controlling for firm size
(market value) and firm performance (return on assets and market returns). Panel A
of Table 2 shows that both the indicator and continuous variables of press coverage
are positively related to award recognitions (p < 0.01 for both, two-tailed). Panel A
also shows that AWARD is positively associated with firm profitability (p < 0.01 for
both, two-tailed), which is consistent with Johnson et al. (1993).
Second, I also check the contents/tones of news articles to validate the use of
news article counts as a proxy of CEO reputation. Most prior studies employ the
number of articles searched in the Dow Jones Interactive database assuming that
CEOs with more news exposure have higher positive reputation. But there have been
concerns that news counts do not consider the tones of news articles (positive,
neutral, or negative news). I randomly select 100 CEOs in the sample and randomly
choose 10 news articles for each CEO to check whether CEOs with more media
exposure have more positive reputations. Table 2, Panel B reports that more than
94% of the articles portray CEOs in non-negative (neutral to positive) tones. CEOs
frequently and nonnegatively cited by the media develop better reputations than
other CEOs since the media allocate the audiences’ attention to touted CEOs
(Fombrun and Shanley 1990). The results of Panel B validate media counts as a
reasonable proxy of reputation without considering the contents of news articles,
which is consistent with Francis et al. (2004).

32
Table 2.
Validity Tests of CEO Reputation Proxy

Panel A: Logistic regressions of AWARD (award recognition) on MEDIA (media
exposure)
Expected
Sign
AWARD
t

Coefficient (Chi-Square)
Intercept ? -3.399***
(3863.77)
-3.480***
(4952.38)
MEDIA
t
+ 0.897***
(90.32)

MEDIA COUNT
t
+ 0.0005***
(13.01)
SIZE
t-1
+ 0.010***
(89.51)
0.013***
(130.61)
ROA
t-1
+ 1.359***
(16.65)
1.278***
(15.03)
RET
t-1
+ 0.416***
(24.50)
0.401***
(22.95)
Likelihood Ratio Chi-Squared 315.12 246.61
N 15,108 15,108

AWARD
it
is the CEO’s awards from business magazine (1= if the CEO was bestowed any award at
year t or 0 = otherwise), MEDIA
it
is the press coverage count of the CEO (1= if the CEO was in the
top 20% media exposure of the industry at year t or 0 = otherwise), MEDIA COUNT
it
is the press
coverage count of the CEO at year t, SIZE
t-1
is the firm’s market value of equity at the prior year,
ROA
it-1
is the firm’s returns of assets of the prior year adjusted by two-digit industry median, and
RET
it-1
is the firm’s stock returns of the prior year adjusted by two-digit industry median. *** denotes
significance at the 0.01 level (2-tailed).

Panel B: Descriptive information on the tone of media coverage
Subject % favorable %neutral
%
unfavorable
%
nonnegative % total
CEO 8.1% 25.2% 0.9% 33.3% 34.2%
Company 11.9% 49.7% 4.2% 61.6% 65.8%
Overall 20.0% 74.9% 5.1% 94.9% 100.0%

33
This study incorporates the two mostly employed proxies of CEO reputation:
award recognitions and media exposure. Previous studies use either recognition
awards (Johnson et al. 1993, Malmendier and Tate 2005) or media counts (Milbourn
2003, Francis et al. 2004) as a proxy of CEO reputation.
29
This study, however,
shows that the two proxies provide similar empirical results regarding the economic
benefits of CEO reputation. This study also shows the robustness of CEO reputation
proxies using the variations of awards and media coverage.

Performance
I measure firm performance using both accounting and market returns
following most previous studies (Murphy 1985; Coughlan and Schmidt 1985;
Lambert and Larcker 1987; Weisbach 1988; Engel et al. 2003). Stock returns
incorporate future expectations whereas accounting earnings reflect firms’ past
operating performance. Accounting earnings have an advantage in measuring short-
term profitability, whereas stock returns have an advantage of incorporating timely
and relevant information available in public (Weisbach 1988). Noticing the
advantages of each performance measure, Daine et al. (2005) employ both
accounting earnings and market returns to test performance persistence.
I use two-digit industry-adjusted returns on assets, ROA
t
(net income before
extraordinary items divided by total assets minus industry median), as my earnings
measure. I employ stock returns as a market performance measure. Abnormal stock

29
Francis et al. (2004) show that media coverage is positively associated with award
recognitions.

34
returns are calculated using the three-factor model (Fama and French 1992). I
cumulate monthly excess returns over the firm’s fiscal year and calculate two-digit
industry-adjusted abnormal returns, RET
t
.

Research Design
To examine the performance benefits of CEO reputation I employ regression
models using an interaction variable of REPUTATION * PERFORM. I examine
whether CEO reputation improves firm performance using performance persistence
rather than performance level (Malmendier and Tate 2005). Performance persistence
tests address both the ability to maintain good performance and the ability to reverse
poor performance whereas performance level tests only investigate the general
ability to outperform competitors. The ability to turn around poor performance is
important in this study, because it is closely related to a CEO turnover decision. To
estimate firm performance persistence and reversals moderated by CEO reputation, I
set the following regression model:
30

PERFORM
it
= β
0
+ β
1
PERFORM
it-1
* I
GOOD
+ β
2
PERFORM
it-1
* I
BAD
+ β
3
REPUTATION
it-1
+ P * REPUTATION
it-1
* PERFORM
it-1
* I
GOOD

+ R * REPUTATION
it-1
* PERFORM
it-1
* I
BAD
+ ∑ρ
k
Z
kit
+ ε
it,
(1)
where PERFORM
it-1
is the firm’s performance measured by ROA
it-1
(the firm’s
returns of assets) or RET
it-1
(the stock returns of the firm), I
GOOD
(I
BAD
) is an

30
Equation (1) is adopted from the regression model introduced by Daine et al. (2005). They
examine the CEO ability of “performing better” using the persistence of good performance
and the reversals of poor performance.

35
Figure 3.
Performance Persistence and Reversals

indicator variable set equal to one if the firm’s performance of previous year
(PERFORM
it-1
) was better (worse) than the industry median of that year or zero
otherwise,
31
REPUTATION
it-1
is the CEO’s reputation measured by AWARD
it-1
(the
CEO’s award from business magazines set equal to one if the CEO was bestowed
any award for the last five years or zero otherwise) or MEDIA
it-1
(the press coverage
count of the CEO set equal to one if the CEO was in the top 20% media exposure of
the industry for the last three years or zero otherwise), and ∑ Z
kit
is the vector of

31
If PERFORM
it-1
≥ 0 then I
GOOD
= 1, and if PERFORM
it-1
< 0 then I
BAD
= 1, since
PERFORM
it-1
is industry-adjusted firm performance of previous year.

36
control variables such as CEO total pay, firm growth, year dummy, and industry
dummy.
32

I am mainly interested in whether CEOs with high reputation are more likely
to continue prior good performance (performance persistence; P) or reverse prior bad
performance (performance reversals; R) than CEOs with low reputation. Figure 3
illustrates what the coefficients in equation (1) indicate. The coefficients of β
1
and β
2

indicate a time-series relationship between prior performance and current
performance. Specifically, the coefficient of β
1
( β
2
) indicates performance
persistence (performance reversals), when prior performance is better (worse) than
the industry median. The coefficient of P (R) indicates whether CEO reputation
moderates performance persistence (performance reversals) as shown in Figure 4.
The ability hypothesis regarding performance persistence (H1a) predicts P to be
positive, because a highly regarded CEO will maintain good performance from one
year to the next. The ability hypothesis regarding performance reversals (H1c)
predicts R to be negative, because a CEO with a well-established reputation will be
more likely to turn around poor performance in the near future than a CEO with a
lesser reputation. The symbolic image hypotheses regarding firm performance (H1b
and H1d), however, predict that CEO reputation is not associated with performance
persistence and reversals (insignificant P and R or even negative P and positive R, if

32
PAY is included as a control variable, because Daine et al. (2005) argue that compensation
is capturing CEO ability to improve firm performance in the future. GROWH is also related
to performance persistence, since the performance of high growth firms might not be
persistent. Both CEO compensation and firm growth are associated with CEO reputation.

37
celebrity CEOs waste their effort that could be used on improving firm
performance), because a psychological symbolic image does not necessarily improve
performance.

3.4. Empirical Results
Descriptive Statistics
Descriptive statistics for the variables used to test the hypotheses are
presented in Table 3. The mean (median) of ROA is 2.2% (1.7%), which suggests
that the distribution of ROA is not skewed. The mean (median) of RET is 2.1%
(0.0%). Although the mean value of RET is similar to that of ROA, the standard
deviation of RET (0.426) is much greater than that of ROA (0.153), which suggests a
wider range of RET than that of ROA. The mean of AWARD is 0.129, which
indicates that 12.9% of CEOs have been bestowed one or more awards for the last
five years. The standard deviation of MEDIA (0.425) is similar to that of AWARD
(0.335). The mean and median of PAY are $ 4.05 million and $ 1.78 million
respectively, which shows that the distribution of CEO total pay is skewed. I use the
log transformation of PAY to minimize the skewness of total pay distribution. The
mean (median) of GROWTH measured by the market-to-book ratio is 4.07 (2.29).
Univariate analyses using Pearson correlations are presented in Table 4.
AWARD
t-1
and ROA
t
are positively associated (r=0.027, p<0.01, two-tailed), which
implies that CEO reputation improves future accounting performance. Both
AWARD
t-1
and MEDIA
t-1
, however, are inversely related to RET
t
(r=-0.016 and

38
Table 3.
Descriptive Statistics (Performance Test Sample)

N Mean Median Std. dev
ROA
t
13,876 0.022 0.017 0.153
RET
t
13,262 0.021 0.000 0.426
AWARD
t-1
13,931 0.129 0 0.335
MEDIA
t-1
13,931 0.237 0 0.425
PAY
t-1
13,247 4,049.73 1,783.67 12,096
GROWTH
t-1
13,931 4.073 2.290 56.725

ROA ≡ the return on assets of the firm defined as net income before
extraordinary item divided by total assets;
RET ≡ the stock returns of the firm defined as abnormal returns using the
three-factor model (Fama and French, 1993);
AWARD ≡ the CEO’s awards from business magazine that equals 1 if the CEO
was bestowed any award for the last five years, 0 otherwise;
MEDIA ≡ the press coverage count of the CEO that equals 1 if the CEO was in
the top 20% media exposure of the industry for the last three years;
PAY ≡ the total flow compensation of the CEO (in thousands of dollars);
GROWTH ≡ the market to book ratio of the firm.

Table 4.
Pearson Correlations (Performance Test Sample)

RET
t
ROA
t-1
RET
t-1
AWARD
t-1
MEDIA
t-1

ROA
t
0.149*** 0.345*** 0.112*** 0.027*** -0.005
RET
t
1.000 -0.005 -0.002 -0.016* -0.021**
ROA
t-1
1.000 0.135*** 0.017** -0.013
RET
t-1
1.000 -0.030*** -0.031***
AWARD
t-1
1.000 0.239***
MEDIA
t-1
1.000
*** Correlation is significant at the 0.01 level (2-tailed).
** Correlation is significant at the 0.05 level (2-tailed).
* Correlation is significant at the 0.10 level (2-tailed).

39
-0.021, p<0.10 and p<0.05, two-tailed), which implies that CEO reputation does not
improve but decreases future stock returns. These correlations show the conflicting
effect of CEO reputation of future performance, which needs further analyses using
multivariate regression models.
AWARD
t-1
and MEDIA
t-1
are positively correlated (r=0.239, p<0.01, two-
tailed), which shows the construct validity of the two reputation proxies. A
correlation between ROA
t
and RET
t
is significantly positive (r=0.149, p<0.01, two-
tailed), which shows a high association between accounting performance and market
performance. ROA
t
is positively related to ROA
t-1
(r=0.345, p<0.01, two-tailed),
which shows the autocorrelation of accounting performance. ROA
t
is also positively
related to RET
t-1
(r=0.112, p<0.01, two-tailed), which shows that stock price reflects
the future expectations of accounting performance. RET
t
, however, is not related to
any of past performance measure (ROA
t-1
or RET
t-1
).

Empirical Tests
I examine whether CEOs with high reputation perform better than CEOs with
low reputation as shown in Table 5. The regression models test whether CEO
reputation improves performance persistence and performance reversals. Table 5,
Panel A presents the regression results using accounting earnings and Panel B
reports those using market returns. The results show that both AWARD * ROA and
MEDIA * ROA have significantly positive coefficients (P = 0.601 and 0.235, p<0.01
for both, two-tailed) when the firm performance of the prior year is better than the

40
Table 5.
CEO Reputation and Firm Performance
ROA
it-1
is the firm’s returns of assets of the prior year adjusted by two-digit industry median (ROA
(Good) if the ROA of the firm is better than the industry median, ROA (Bad) if the ROA of the firm is
worse than the industry median), RET
it-1
is the firm’s stock returns of the prior year adjusted by two-
digit industry median, AWARD
it-1
is the CEO’s awards from business magazine (1= if the CEO was
bestowed any award for the last five years or 0 = otherwise), MEDIA
it-1
is the press coverage count of
the CEO (1= if the CEO was in the top 20% media exposure of the industry for the last five years or 0
= otherwise), PAY
it-1
is the total flow compensation of the CEO of the firm, and GROWTH
it-1
is the
market to book ratio of the firm. *** denotes significance at the 0.01 level (2-tailed), ** at the 0.05
level, and * at the 0.10 level.

Panel A: Accounting performance

ROA
t

Coefficient (t-value)
Intercept -0.030 (-1.31) -0.025 (-1.08)
ROA (Good)
t-1
0.197*** (15.52) 0.203*** (15.94)
ROA (Bad)
t-1
0.331*** (30.89) 0.323*** (29.37)
AWARD
t-1
-0.015*** (-3.20)
AWARD * ROA (Good)
t-1
0.601*** (10.09)
AWARD * ROA (Bad)
t-1
-0.094 (-0.74)
MEDIA
t-1
-0.011*** (-3.40)
MEDIA * ROA (Good)
t-1
0.235*** (6.85)
MEDIA * ROA (Bad)
t-1
0.051* (1.85)
PAY
t-1
0.0002* (1.69) 0.0003** (2.17)
PAY
t-1
* ROA (Good)
t-1
-0.0020*** (-2.59) -0.0018** (-2.29)
PAY
t-1
* ROA (Bad)
t-1
-0.0036*** (-3.14) -0.0049** (-3.69)
GROWTH
t-1
0.0003*** (2.83) 0.0010*** (6.97)
GROWTH
t-1
* ROA (Good)
t-1
-0.0018*** (-2.56) -0.0066*** (-6.81)
GROWTH
t-1
* ROA (Bad)
t-1
0.0023*** (9.08) 0.0032*** (11.38)
Year dummy Yes Yes
Industry dummy Yes Yes
F-Statistic 26.80 25.99
Adjusted R
2
0.144 0.140
N 13,193 13,193

41
Table 5 (Continued).
CEO Reputation and Firm Performance
RET
it-1
is the firm’s stock returns of the prior year adjusted by two-digit industry median (RET
(Good) if the RET of the firm is better than the industry median, RET (Bad) if the RET of the firm is
worse than the industry median), AWARD
it-1
is the CEO’s awards from business magazine (1= if the
CEO was bestowed any award for the last five years or 0 = otherwise), MEDIA
it-1
is the press
coverage count of the CEO (1= if the CEO was in the top 20% media exposure of the industry for the
last five years or 0 = otherwise), PAY
it-1
is the total flow compensation of the CEO of the firm, and
GROWTH
it-1
is the market to book ratio of the firm. *** denotes significance at the 0.01 level (2-
tailed), ** at the 0.05 level, and * at the 0.10 level.

Panel B: Market performance

RET
t

Coefficient (t-value)
Intercept 0.106 (1.39) 0.105 (1.37)
RET (Good)
t-1
0.191*** (2.61) 0.185** (2.52)
RET (Bad)
t-1
-0.033 (-0.28) -0.035 (-0.29)
AWARD
t-1
-0.024 (-1.54)
AWARD * RET (Good)
t-1
0.156*** (2.76)
AWARD * RET (Bad)
t-1
-0.018 (-0.27)
MEDIA
t-1
-0.015 (-1.20)
MEDIA * RET (Good)
t-1
0.065** (1.94)
MEDIA * RET (Bad)
t-1
-0.011 (-0.24)
PAY
t-1
-0.0062 (-1.29) -0.0057 (-1.14)
PAY
t-1
* RET (Good)
t-1
-0.0299*** (-3.12) -0.0294*** (-3.06)
PAY
t-1
* RET (Bad)
t-1
0.0023 (0.15) 0.0027 (0.16)
GROWTH
t-1
-0.0002 (-0.33) -0.0002 (-0.31)
GROWTH
t-1
* RET (Good)
t-1
0.0003 (0.25) 0.0002 (0.24)
GROWTH
t-1
* RET (Bad)
t-1
0.0023 (0.73) 0.0023 (0.73)
Year dummy Yes Yes
Industry dummy Yes Yes
F-Statistic 4.56 4.51
Adjusted R
2
0.023 0.023
N 12,666 12,666

42
industry median. AWARD * RET and MEDIA * RET also have significantly
positive coefficients (P = 0.156 and 0.065, p < 0.01 and p < 0.05, two-tailed). The
results suggest that CEOs with well-established reputation better sustain good
performance from one year to the next than CEOs with lesser reputations, which
supports the ability hypothesis (H1a).
On the other hand, when the firm performance of the prior year is worse than
the industry median, most interaction variables (AWARD * ROA, AWARD * RET,
and MEDIA * RET) do not have significant coefficients. The coefficient of MEDIA
* ROA, however, is significantly positive (R = 0.051, p < 0.10, two-tailed), which
implies that well-known CEOs might even aggravate poor accounting performance.
The results indicate that CEO reputation may not turn around struggling firms in the
near term, which supports the symbolic image hypothesis (H1d). It is worth noting
that the results of performance tests support the ability hypothesis and the symbolic
image hypothesis partially. Well-known CEOs often have the ability to maintain
good performance but not necessarily the ability to turn around poor performance.
The results, however, might be caused by the selection criteria of CEO awards
because most business journals bestow recognition awards to CEOs who maintain
good performance better than their peers. If I consider CEOs known as turnaround
specialists, the results of performance reversals might be different. The results also
might be caused by earnings management. Malmendier and Tate (2005) show that
celebrity CEOs often manage earnings to meet heightened expectations. They report

43
that once it has been more than five years since the last award, celebrity CEOs can
no longer manipulate earnings and report more negative earnings.

Robustness Tests
I run robustness tests to check whether varying the proxies of CEO reputation
(AWARD and MEDIA) provides similar results to those of the empirical tests. I set
five years of lagged period after CEOs receive recognition awards. I vary this lagged
value of five years to three years (AWARD is set to 1 if the CEO received any award
for the last three years) and permanence (AWARD is set to 1 throughout sample
period once the CEO was bestowed any award). I also change the summation period
of media counts from three years to five or two years. To check the validity of
converting a continuous variable (the sum of media counts) to an indicator variable
(MEDIA), I vary the threshold of top 20% of the industry to 10% or 30%.
Overall, the results of sensitivity tests using the various proxies of reputation
do not qualitatively alter the conclusions, but there are some variations in detailed
results. Performance persistence tests report that 11 out of 16 coefficients (P) are
significantly positive but that only 1 coefficient is significantly negative, which
suggests relatively consistent results supporting the ability hypothesis (H1a).
Performance reversal tests show that 12 out of 16 coefficients (R) are not significant
(3 coefficients are significantly positive and 1 coefficient is significantly negative).

44
The results suggest that well-known CEOs do not necessarily reverse poor
performance, which supports the symbolic image hypothesis (H1d).
33

3.5. Discussion
This chapter empirically examines whether current CEO reputation improves
future firm performance measured by accounting earnings and market returns.
Empirical investigation about whether positive reputation helps sustain good
performance and turn around poor performance is needed to show whether CEO
reputation is valuable to firms as an intangible asset. The issue of the performance
benefits of CEO reputation (Chapter 3) is important to determine whether
stakeholders and boards of directors should provide personal benefits such as job
retention and compensation to highly regarded CEOs (Chapter 5).
The results of firm performance tests show that CEOs with well-established
reputations maintain good performance. Well-known CEOs, however, do not
necessarily reverse poor performance in the near future. The results imply that
shareholders and a board of directors are better off by keeping a highly regarded
CEO when a firm prospers, because the highly regarded CEO has the ability to
sustain good performance. Stakeholders, however, may have to consider replacing a
high-profile CEO with a turnaround specialist when the firm suffers financially,
since the well-known CEO may not turn around poor performance. Investors had
better be careful about investing in a company solely based on a superstar CEO once
firm performance decreases.

33
I run robustness tests to control firm-specific effects or CEO-specific effects. The results
after considering firm-specific effects provide qualitatively the same conclusion.

45
Chapter 4. CEO Reputation and Personal Benefits: Theory
4.1. Agency Theory
Traditional agency studies put emphasis on formal or explicit compensation
contracts to minimize agency costs such as moral hazard and adverse selection
(Baiman 1982). A large body of literature has examined whether financial incentive
schemes facilitate optimal solutions by overcoming interest conflicts and information
asymmetry between the principal and the agent (Laffont and Martimort 2002).
Promotion, retention, and the stream of future compensation, however, are more
important to managers in the long term than current financial rewards. Yet, implicit
incentive schemes of career concerns or reputation are not frequently examined by
previous agency studies (Dewatripont et al. 1999).
Career concerns are important whenever the labor market (internally a
current employer and externally potential employers) uses managers’ current
performance to assess their ability. Managers have incentives to increase this
assessment, since stakeholders often determine the future wages, promotion, or
retention of executives based on updated beliefs about their ability. Fama (1980) first
discusses the importance of implicit career concerns as well as explicit compensation
contracts. Career concerns were then formally modeled by Holmstrom (1982) and
empirically tested by Gibbons and Murphy (1992). MacLeod and Malcomson (1988)
examine career concerns for promotion instead of those for future compensation
(Fama 1980, Holmstrom 1982). Chevalier and Ellison (1999) empirically show that
the CEO’s age indicating reputation concerns affects job retention.
Fama (1980) argues that implicit career concerns discipline managers
because managers are motivated to improve the stream of their future wages.

46
Assuming that the market uses information rationally to assess the talent of
managers, their current good performance will generate high wages in the future via
their increased reputations or beliefs about their productivity. Thus, managers have
strong incentives to enhance their reputations or the market values of their human
capital.
Holmstrom (1982) shows how implicit career concerns affect a manager’s
behavior. In a two-period setting, the principal uses the performance at the first
period to determine the salary of the agent at the second period. The performance at
the first period may be the best available information to estimate the manager’s
ability, because the principal does not know the ability or effort level of the agent
(DeGroot 1970). Thus, the manager has motivation to improve performance to boost
the market’s assessment of his or her productivity. This conclusion in a two-period
setting is further generalized in a multi-period setting which is a more realistic
assumption for external validity (Holmstrom 1982). The results imply that even
wage incentive schemes implicitly motivate managers through reputation building
incentives or career concerns which were ignored in previous agency studies
focusing on explicit output-based compensation. Holmstrom (1982), however, shows
that implicit career concerns are not a perfect substitute for explicit contracts, since
reputation concerns distort effort allocations. Career concerns without incentive
contracts induce the manager to work too hard in early years but not hard enough in
later years.
Gibbons and Murphy (1992) show that career concerns create important
incentives on top of compensation contracts. The principal designs a compensation
contract to optimize total incentives combining implicit incentives from career

47
concerns and explicit incentives from the compensation contract. Thus, the principal
sets explicit incentives from the compensation contract strongest for the agent close
to retirement, because implicit career concerns are weakest for that agent. The
principal, however, puts much less emphasis on explicit incentives from the
compensation contract for the agent in early years of his or her career, since implicit
career concerns provide enough incentives to that agent. Gibbons and Murphy
(1992) find that pay-performance sensitivity increases as CEOs near retirement.
34

MacLeod and Malcomson (1988) argue that promotion in hierarchical
organizations is important motivation for a manager to build good reputation through
superior performance. The firm with hierarchy cannot assign new employees in
appropriate ranks because it does not know their ability. New employees will not
voluntarily choose appropriate ranks, because even employees with low ability will
choose the highest rank for the highest wages. This mismatch between the ability of
employees and their ranks distorts a sorting mechanism and decreases productivity.
Thus, the firm has to design a selection mechanism to assign all new employees to
the lowest rank and promote employees based on productivity (ability).
Reputational concerns or career concerns are also applied to job retention as
well as promotion. Some might argue that CEOs might not be motivated for
promotion because they are in the top position of the firm. Job retention, however, is

34
Most agency models explaining career concerns assume uncertainty about the manager’s
ability (Holmstrom 1982, MacLeod and Malcomson 1988, Gibbons and Murphy 1992).
There is an argument against the applicability of career concerns to CEO settings. Most
CEOs were long-time employees of firms and internally promoted, so shareholders and a
board of directors know the ability of a new CEO. Shareholders and the board of directors,
however, might not be certain about the ability of the new CEO because skills to lead whole
organization and initiate strategy as a CEO are different from those to implement strategy as
an executive (Gibbons and Murphy 1992).

48
an important concern for a CEO, since the board of directors can replace the
incumbent CEO with an internal heir or an outsider who is known better than the
current CEO. Thus, even CEOs have motivation to build and maintain their
reputations as executives with high ability. Chevalier and Ellison (1999) examine
whether the association between the likelihood of turnover (being terminated) and
past performance is moderated by reputational concerns. They show that termination
is more performance-sensitive for younger managers with more reputational
concerns.

4.2. Attribution Theory
Attribution theory is a theory about how people explain things and make
causal inferences of phenomena (Weiner 1986). Human beings keep asking
ourselves “why?” People have a strong need to understand and explain why
something happen (Heider 1958). Attribution theory describes the processes of
causal attribution to satisfy this need to explain phenomena. It examines how
individuals collect information and put it together to assess causality about human
behaviors or events (Kelley 1967).
Heider is a pioneer to raise these issues of psychological attribution in his
book, The Psychology of Interpersonal Relationships (1958). He suggests scientific
studies examining how individuals make causal inferences about people and
phenomena around them by discovering the naïve epistemology of a social perceiver.
Heider (1958) assumes that a naïve perceiver searches for the reason of things
around him or her, because he or she feels a strong need to find the cause of an event

49
Figure 4.
Framework of Attribution Theory

or behavior. Thus, the naïve perceiver attempts to develop the perceptions of persons
or things through collecting information about actions, intentions, and environments.
Heider (1958) proposes a set of rules by which an ordinary/lay person makes causal
inferences of an event or attributes responsibility to another person. He, however,
notes that the lay attributor is prone to certain biases in this attribution processes. The
naïve attributor puts more weight on salient information than on other information in
the function of causality assessment.
Attribution theory emerged from Heider’s (1958) naïve psychology was
reformulated by Kelley (1967) and Jones and Davis (1965), the two general lines of

50
research as shown in Figure 4. Kelley (1972) argues that a lay attributor “generally
acts like a good scientist, examining the covariation between a given effect and
various possible causes.” The naïve perceiver reaches an explanation for an event by
investigating the extent to which the possible causes and the effect covariate across
time and person. Kelley (1972) maintains that the lay attributor, with some
exceptions, mostly makes causal inferences in “a reasonable and unbiased manner.”
Kelley (1967) suggests the three factors of causal inferences: consensus,
consistency, and distinctivenss. For example, a perceiver’s favorable judgment about
a restaurant is attributed to the restaurant if the perceiver knows that other people
like the restaurant, the perceiver enjoys the restaurant every time he or she goes
there, and the perceiver seldom enjoys restaurants. But the perceiver’s favorable
judgment about the restaurant is attributed to the perceiver’s personality if the
perceiver finds that most people do not like the restaurant, the perceiver enjoys the
restaurant most times, and the perceiver enjoys most restaurants.
Jones and Davis (1965) examine how a perceiver infers another’s intention or
personality from his or her behaviors. They argue that individuals employ
incomplete information processes to reach a salient and satisfactory explanation for
phenomenon. According to cognitive psychology, individuals search for information
to explain an event or behavior, but often stop data search when they come across
satisfactory and salient information (Tversky and Kahneman 1974). People often use
salient and available evidence to make causal inferences instead of reviewing all
relevant information. Social psychology examines how salient and distinct cues are

51
used for causal inferences (Taylor and Fiske 1978). Theorists in social psychology
develop and generalize this principle of salience that people tend to attend salient
stimuli and perceive salient factors causing an observed effect. Jones and Davis
(1965) state the following paragraph:
The perceiver seeks to find sufficient reason why the person acted and why
the act took a particular form. Instead of the potentially infinite regress of
causes and effect which characterizes an impersonal, scientific analysis, the
perceiver’s explanation comes to a stop when an intention or motive has the
quality of being reason enough.

Hamilton (1980) explains the difference between the two general lines of
research, Kelley (1967 & 1972) and Jones and Davis (1965), by refining the
definition of attribution. He acknowledges that attribution theories lack the direct
definition of attribution. Hamilton (1980) suggests that attribution covers two distinct
inquiries: an explanatory inquiry and a sanctioning inquiry as shown in Figure 4. A
main difference between the two inquiries lies in starting points for inquiries. An
explanatory inquiry starts with a known effect with an attempt to understand or
discover unknown causes. A sanctioning inquiry, however, begins with a known act
as well as a known effect with an attempt to find whether the act causes the effect.
The task in explanation is to search for potential causes and infer causality whereas
that in sanctioning is to determine whether the wrongful act causes the harm or loss.
The objective of explanation is understanding whereas that of sanctioning is
responsibility judgment. Kelley’s model (1967 & 1971) using a “covariation
principle” is the model of explanatory inquiry which puts emphasis on scientific and
unbiased approaches. Jones and Davis’s (1965) model adopting a rule of “could have

52
done otherwise,” however, is the model of sanctioning inquiry which focuses on
biased and incomplete processes.
The advantage of distinguishing between explanation and sanctioning is to
identify clearly which model should be applied under specific situations. When
managers analyze why an industry leader is successful using benchmark analyses,
they use the model of explanatory inquiry to find the critical success factors of the
industry leader. When the effectiveness of an organization is in doubt, however, the
stakeholders of the organization employ a sanctioning approach to judge
responsibility. The key issue in sanctioning processes is to determine whether the
actor could have brought different results by choosing different actions. Leaders
would be responsible for poor organization performance if they could have produced
better outcomes by behaving differently (Hamilton 1980, Lord and Smith 1983).
Pfeffer (1977) argues that organizational performance is mostly attributed to a
leader’s actions instead of situations because people desire or believe that the leader
determines the success of the organization. According to the model of sanctioning
inquiry, individuals tend to assign responsibility to persons instead of situations
using “could have done otherwise” tests (Jones and Davis 1965). Social
psychologists have shown that human beings tend to over-attribute causality to
personal factors and under-attribute causality to situational factors, which is called
“fundamental attribution error (Hamilton 1980).”
People’s perceptions or beliefs about whether the leader’s actions influence
organizational effectiveness play a more important role in attribution processes than

53
the actual impact of the leader’s actions on organization performance, which shows
biased and incomplete processes in responsibility judgment.
35
Social Psychologists
further develop the incomplete processes of causal inference based on salient stimuli
(Jones and Nisbett 1972, Taylor and Fiske 1975 & 1978, Taylor et al. 1977, Pryor
and Kriss 1977). Jones and Nisbett (1972) develop how perceptual salience affects
perceived causality in assigning responsibility. If a perceiver attends to a part of
his/her environments, the information from the part becomes perceptually salient to
the perceiver. The perceiver puts more weights on this information perceived salient
and adopts the salient part as a potential cause of outcomes. In other words, attention
or a point of view determines perceptual salience which in turn leads to perceived
causality.
Taylor and Fiske (1978) argue that individuals make causal inferences with
little thought behind it and that people respond to the most salient stimuli in their
environments, which is called “top of the head” phenomena. People spend very little
time to assess potential alternatives causing an effect and collect little data beyond
the first thing that comes to mind (available in that situation), yet respond with
thoughtless guess. Contrary to the argument of Kelley (1972) “top of the head”

35
Khanna and Poulsen (1995) find that management actions between bankrupt firms and
control (matched) firms are not significantly different and that neither set of managers takes
value-reduction actions. The results suggest that perceptions or beliefs about who is
responsible for poor firm performance are more important in sanctioning than the
effectiveness of executives’ decisions. The results also imply that executives often serve as
scapegoats.

54
phenomena are far more common than he admits.
36
Research has shown that
attribution bias caused by salient stimuli reliably accounts for a great number of
experimental findings and that this bias affects our causal inferences, learning,
evaluations, and the imputations of personal dispositions (personality, attitude, and
so on).
Taylor and Fiske (1975) examine whether the salience of actors affects the
causal attribution of observers. They manipulate perceptual salience using the seating
arrangement of actors (physical attention) and the instruction of attention to a
specific actor (mental attention). They find that observers perceive a salient
individual as causal in an event, which suggests that people often attend to the most
salient cue in causal attribution. Taylor et al. (1977) investigate whether the novelty
of actors affects the causality perception of observers. They manipulate the salience
of an actor using a solo African American (race), woman, or man (gender). They find
that a solo black, woman, or man is perceived as being more influential or talkative
than is the same person in a non-solo situation, which strongly supports the salience-
causality link. Taylor et al. (1977) also find that attention or salience increases
perceived representativeness which in turn increases perceived attribution.
37

36
Kelley (1972) argues that a naïve attributor mostly makes causal inference in an unbiased
manner with some exceptions.
37
Research has examined why attention or salience induces such a great effect on causal
inferences. Pryor and Kriss (1977) investigate whether the availability of information recall
mediates the link between salience (attention) and biased attribution. They show that salient
information is more quickly retrieved and identified in a reaction time task.

55
4.3. Expectancy-Disconfirmation Theory
Expectancy disconfirmation theory is a theory explaining that customer
satisfaction is based on expectations as well as the actual quality of product
(Anderson 1973, Oliver 1980). According to expectancy disconfirmation theory
customer satisfaction is a function of pre-purchase expectations, perceived product
performance, and expectancy disconfirmation (Oliver 1980, Zwick et al. 1995).
38

Previous studies show that expectations provide an anchor or a frame of reference in
comparative processes (Anderson 1973, Santos and Boote 2003). End results worse
than expectations (a negative disconfirmation) are evaluated below the reference
point, whereas outcomes better than standards (a positive disconfirmation) are rated
Figure 5.
Framework of Expectancy-Disconfirmation Theory

38
Disconfirmation indicates discrepancy between standards (expectations) and the actual
performance of product.

56
above the reference point (Oliver 1980). The expectancy disconfirmation model
shows that high expectations are more likely to result in a negative disconfirmation
which in turn decreases satisfaction (Cadotte et al. 1980). In other words, elevated
expectations decrease satisfaction via a negative disconfirmation as shown in Figure
5 (Anderson 1973, Oliver 1980, Cadotte et al. 1980).
39

Anderson (1973) argues that the effect of disconfirmation between
expectations and actual product performance on customers’ product evaluation can
be explained by four psychological theories: contrast theory, cognitive dissonance
theory, generalized negativity thesis, and assimilation-contrast approach. These four
psychological theories offer different predictions regarding the effect of high
expectations on satisfaction as shown in Figure 6.
First, contrast theory suggests that customers who experience a negative
disconfirmation (a product worse than expectations) will magnify discrepancy
between actual product quality and product expectations (Hovland et al. 1957).
Contrast theory assumes that a negative surprise effect or contrast between
expectations and outcomes will make customers exaggerate the negative
disconfirmation (Anderson 1973). If the actual quality of product is less valuable
than expectations, customers will evaluate the product less favorably than if they had
no prior expectations (Sherif and Hovlan 1961, Freedman 1964, Whittaker 1965).
For example, suppose a customer watches a movie with high expectations but finds
the movie less interesting than he or she expected. The disappointment at the movie
is exaggerated by a negative disconfirmation due to high expectations, which

39
Zwick et al. (1995) explain that expectations are thought to have direct positive effects on
satisfaction and indirect negative effects on satisfaction through a perceived negative
disconfirmation.

57
Figure 6.
Disconfirmation of Expectations

makes the customer complain that the movie is the worst movie he or she has ever
watched. Spector (1956) examines whether a negative disconfirmation affects the
evaluation of a financial reward. He finds that subjects whose expectations are higher
than the actual amount of the reward are less satisfied with the reward than are those
whose expectations are the same as the reward. Contrast theory predicts the effect of
expectations on evaluation as shown by the dashed line in Figure 6 (Anderson 1973).
Second, cognitive dissonance theory presumes that disparity between
expectations and actual performance induces a state of dissonance or psychological
discomfort (Chapanis and Chapanis 1964, Rosenberg 1965, Feldman 1966,
Oshikawa 1968). This dissonance in turn puts pressure for customers to reconcile a

58
difference between expectations and product quality, since they want product
information from their own experience and advertisement consistent (Holloway
1963). One potential approach to resolve the perceived disparity is to raise the
evaluation of the product as an assimilation strategy (Festinger 1957). For example, a
consumer confronting the disconfirmation of a less interesting movie than
expectations resolves the discomfort by re-evaluating the movie not really bad as it
first appeared. Cognitive dissonance theory predicts the effect of expectations on
satisfaction as shown by the dotted line in Figure 6, which provides an opposite
prediction to that of contrast theory (Anderson 1973).
Third, generalized negativity theory argues that any disparity between the
product expected and the product received induces a generalized negative hedonic
state (Carlsmith and Aronson 1963). This generalized negativity state makes a
perceiver less satisfied and less pleasant, which in turn decreases the ratings of the
product. Carlsmith and Aronson (1963) examine whether the manipulation of
subjects’ expectations regarding the tastes of bitter and sweet solutions affects the
ratings of sweet tastes. When subjects expecting the sweet solution taste the bitter
solution, they rate the bitter solution more bitter, which supports contrast theory.
When subjects expecting the bitter solution taste the sweet solution, however, they
evaluate the sweet solution less sweet, which supports assimilation theory. Carlsmith
and Aronson (1963) explain these seemingly conflicting results by arguing that any
disconfirmation (either positive or negative) induces a generalized negative hedonic
state which decreases product ratings. Generalized negativity theory predicts that any
absolute level of discrepancy decreases satisfaction as shown by the line of
alternating dots and dashes in Figure 6.

59
Finally, assimilation-contrast approach combines contrast theory and
cognitive dissonance theory as its name indicates. Hovland et al. (1957) argue that
individuals have the ranges of acceptance, rejection, and neutrality. If product quality
is slightly different from expected quality (within an acceptance range), individuals
tend to reduce perceptual dissonance by increasing the evaluation of product
performance as shown in assimilation processes. If discrepancy between product
quality and expectations is greater than a certain threshold, however, the disparity
tends to be exaggerated as shown in contrast processes. Thus, assimilation-contrast
theory predicts that the effect of disconfirmation on evaluation represents the S-
shaped curve as shown in Figure 6.
Anderson (1973) manipulates product expectations and examines how the
product ratings of subjects change over different levels of expectations given fixed
product quality. He finds that product evaluations are assimilated toward
expectations within an acceptable range. The ratings at a “very high” expectation
condition, however, plummet, which best supports assimilation-contrast theory. The
results suggest that unrealistically high expectations created by excessive advertising
negatively affect customer satisfaction. The results also imply that elevated
expectations due to the reputations of CEOs might impair their job retention via a
negative disconfirmation.
Santos and Boote (2003) note that consumers do not have one expectation of
product quality rather they have a set of expectations at various levels. They review
the customer satisfaction literature and develop nine groups of expectations in the
form of a hierarchy from 56 different definitions of ‘expectation’ (Miller 1977,
Olson and Dover 1979, Swan and Trawick 1980, Woodruff et al. 1983, Tse and

60
Wilton 1988, Zeithaml et al. 1993, Spreng et al. 1996). Santos and Boote (2003)
label nine groups of expectations as the ideal, the should, the desired, the predicted,
the deserved, the adequate, the minimum tolerable, the intolerable, and the worst
imaginable from highest to lowest as shown in Figure 7. Among nine groups of
expectations, the predicted (will be) expectation has been widely employed in
previous studies. Santos and Boote (2003) indicate that the predicted is considered
the ‘core expectation’, whereas all other expectations are regarded as ‘peripheral
expectations.’
The advantage of enumerating nine groups of expectations is to classify
customer satisfaction into detailed post-purchase affective states. The customer
satisfaction literature argues that satisfaction may not be a simple affective state in
one scale or dimension (Erevelles and Leavitt 1992, Kumar and Olshavsky 1997).
Santos and Boote (2003) classify traditional satisfaction into four affective states:
delight, satisfaction, acceptance, and dissatisfaction. They consider different levels of
expectations, disconfirmation (positive or negative), and the zone of indifference to
develop more useful post-purchase affective states than a traditional satisfaction
construct. The zone of indifference lies when performance is less valuable than the
desired and more valuable than the minimum tolerable as shown in Figure 7
(Woodruff et al. 1983, Erevelles and Leavitt 1992).
40

Satisfaction or acceptance is induced if disparity between expectations and
performance lies within the zone of indifference. If actual performance is greater

40
The customer satisfaction literature argues that the disconfirmation of expectations does
not incur, if actual performance is slightly greater than or less than expectations (Woodruff
et al. 1983). Simple confirmation (where performance is similar to expectations) results in
neither satisfaction nor dissatisfaction but indifference (Buttle 1996).

61
Figure 7.
Conceptual Framework of Expectations and Affective Behaviors

than the predicted expectation but discrepancy lies within the zone of indifference
(lower than the desired), satisfaction may occur. If actual quality is lower than the
predicted expectation but disparity does not lie beyond the zone of indifference,
acceptance incurs. Delight and dissatisfaction arise only when disparity exists
beyond this zone of indifference as shown in Figure 7. Delight is a temporary and
extreme positive state compared with satisfaction. Delight may occur if perceived
performance is over and above the desired level of expectation. Dissatisfaction
incurs if a negative disconfirmation falls below the minimum tolerable level of
expectation.

62
Four affective states based on the zone of difference and disconfirmation
induce different actions of customers as shown in Figure 7 (Bearden and Teel 1983,
Santos and Boote 2003). Delighted consumers compliment a product whereas
dissatisfied customers complain the product after they consume it (Robinson and
Berl 1979). Consumers with satisfaction or acceptance (within the zone of
indifference) do not take any action but adjust expectations for next purchases
(Bearden and Teel 1983, Boote 1998). Difference between acceptance and
dissatisfaction is critical in this study, since only dissatisfied stakeholders take
actions to punish CEOs with poor performance. The reputations of CEOs raise the
minimum tolerable level of firm performance as well as the predicted expectation,
which in turn increases the likelihood that sliding performance triggers forced
turnovers. However, boards of directors have the lower minimum tolerable level for
less reputed CEOs and accept poor firm performance without taking any action. Thus,
the poor performance of a well-known CEO might trigger CEO dismissal decision
by the board of directors, whereas that of a less-known CEO might be accepted by
the board of directors.

63
Chapter 5. CEO Reputation and Personal Benefits: Empirical Evidence
5.1. Hypotheses Development
I investigate whether CEO reputation provides personal benefits to CEOs. As
explained in Chapter 2, reputation has two major aspects: ability and perception.
These dual aspects of CEO reputation induce two perspectives predicting the
different consequences of CEO reputation. The ability perspective argues that CEO
reputation reflects the superior ability of CEOs and thus improves their job retention
and compensation (MacLeod and Malcomson 1988; Gibbons and Murphy 1992).
The symbolic image perspective, however, maintains that CEO reputation reflects
the media-created images of CEOs and thus does not improve but impairs job
retention (Malmendier and Tate 2005).

CEO Reputation and Job Retention
I examine whether CEO reputation moderates a known inverse relationship
between firm performance and CEO turnover. Research has examined the effects of
firm performance on CEO turnover and finds that poor performance results in more
CEO dismissals (Couglan and Schmidt 1985; Warner et al. 1988; Weisbach 1988;
Puffer and Weintrop 1991; Fee and Headlock 2002; Dahya et al. 2002, Farrell and
Whidbee 2003). CEO dismissals based on firm performance sort out CEOs with high
ability from CEOs with low ability, discipline CEOs to exert their best efforts, and
eventually improve future performance (Weisbach 1988; Dahya et al. 2002; Goyal
and Park 2002; Denis and Denis 1995; Khurana and Nohria 2000).

64
There is conflicting anecdotal evidence about whether the reputations of
CEOs improve or impair their job retention under the circumstances of poor
performance. In the first case, the excellent reputation of a CEO allowed him to keep
his job and gave him the opportunity to make up for his mistakes that had hurt the
firm’s past performance. In April 1985, Coca-Cola CEO Roberto Goizueta
introduced “New Coke,” but consumers reacted negatively disparaging the taste of
“New Coke.” In response to this turmoil, Goizueta had to resume manufacturing
“Classic Coke.” Despite what was widely considered the largest marketing blunder
of the 1980s, Coca-Cola survived and continued to be a successful company, and
Goizueta did not lose his job. There were many factors contributing to the smooth
resolution of this crisis, but the well-respected reputation of Roberto Goizueta is
worth noting. Goizueta had introduced “Diet Coke” successfully in 1982, which
changed the soft-drink market dramatically. Because he had earned credibility from
shareholders and the board of directors, they were willing to accept the risk of
potential failures in the future. Stakeholders believed that Coca-Cola would recover
from the marketing misstep and continue to be prosperous under Goizueta’s
leadership.
In the second case, however, the reputation of a CEO put pressure on him to
step down. Eckhard Pfeiffer made Compaq the world’s largest maker of personal
computers by increasing the annual sales of Compaq from $ 3.27 billion in 1991 to $
31.2 billion in 1998. Pfeiffer, however, failed to respond to plummeting PC prices
and growing internet sales in the late 1990s. The company’s stumbling performance

65
in the first quarter of 1999 raised doubts that Compaq would achieve its annual sales
goal of $ 50 billion by the year 2000 and led to the ousting of Eckhard Pfeiffer in
April 1999. In contrast to the previous example of Roberto Goizueta, Pfeiffer’s
success and reputation did not help him survive his encounter with severe
competition. A gap between an elevated goal ($ 50 billion of sales by 2000) and
actual performance ($38 billion of sales in 1999) was mostly attributed to Eckhard
Pfeiffer because he had been closely connected with Compaq symbolically - he was
identified very closely with Compaq by many stakeholders. The attention of the
media and investors was focused on Pfeiffer, a single human being, instead of on
Compaq. Eckhard Pfeiffer had become a surrogate for Compaq and was expected to
take the hit for the poor performance of the company. Pfeiffer ultimately served as a
scapegoat.
The conflicting anecdotal cases above might explain two perspectives
offering different predictions regarding the job retention benefits of CEO reputation.
The ability perspective predicts that the reputations of CEOs will help them survive
hard times. The reputations of CEOs mainly reflect their ability to reverse current
poor performance in the near term according to the ability perspective. A more-
reputed CEO will be more likely to turn around poor performance, whereas a less-
reputed CEO will be more likely to perpetuate poor performance (Daine et al. 2005).
A board of directors and shareholders are willing to provide more chances to a
highly regarded CEO despite current poor performance because they have
confidence in the flourishing future of the firm under his or her leadership. The

66
ability perspective predicts that CEO reputation will help him/her to survive through
hard times. Thus, CEOs with well-established reputations are less likely to be
dismissed than are CEOs with lesser reputations as shown in the case of Coca-Cola
CEO Roberto Goizueta.
The symbolic image perspective, on the other hand, predicts that the
reputations of CEOs will impair their job retention. Poor firm performance is more
often attributed to symbolic leaders according to attribution theory. As CEOs
develop their reputations, they become more symbolically connected with the
companies they run and many stakeholders do not separate the CEOs per se from the
companies (Gaines-Ross 2002). For example, Microsoft reminds many people of Bill
Gates, Apple of Steve Jobs, GE of Jack Welch, and Boeing of Phil Condit. As CEOs
achieve mutual identity with their companies, they get adulation when firms prosper
but accusations when firms slide. A well-known CEO as the symbol of the company
provides a target for blame when things go wrong. Stakeholders blame CEO
sanctioning poor firm performance to a leader instead of explaining why it happened.
As a leader becomes well-known, poor performance is more attributed to the
reputed leader. Taylor and Fiske (1975) show that any salient person in a group is
perceived as mostly causal in their experiments. Considering the fact that a leader of
a group is the most salient and novel individual, the leader is highly likely perceived
as responsible for poor performance in many cases. Thus, more distinct and reputed
CEOs are more blamed for poor performance than less reputed CEOs. Gamson and
Scotch (1964) show, for example, that the manager of a baseball team serves as a

67
scapegoat, when the team’s performance plummets. It is difficult to dismiss the
whole team, yet the firing of the manager signals to the world that the failure of the
team was the result of the manager’s misjudgments and that newer and better steps
will be taken to improve team performance.
A high-profile CEO is at greater risk of serving as a scapegoat because the
positive reputation of the CEO raises the expectations of stakeholders.
41
Expectancy
disconfirmation theory argues that CEOs feel more pressured as they develop
reputations due to these raised expectations of stakeholders. The elevated
expectations of stakeholders increase gaps between expectations and actual
performance. These increased gaps eventually decrease the satisfaction of
stakeholders and increase the likelihood of the CEO’s turnover (Anderson 1973;
Oliver 1980; Cadotte et al. 1980). Firm performance poorer than stakeholders’
expectations (a negative disconfirmation) decreases their satisfaction, which leads to
more turnover of CEOs. Thus, more-reputed CEOs are more likely to be replaced for
poor performance than are less-reputed CEOs as shown in the case of Compaq CEO
Eckhard Pfeiffer. If actual performance is lower than expected performance but
disparities lie within the zone of indifference, consumers do not take any action
(acceptance). However, if negative disconfirmations fall below minimum tolerable
levels, consumers will take complaining responses (dissatisfaction). The reputations
of CEOs raise minimum tolerable levels of firm performance and performance slides

41
Malmendier and Tate (2005) argue that heightened expectations because of the celebrity
status of CEOs induce CEOs to manipulate earnings to meet elevated goals.

68
might trigger forced turnovers. However, the boards with less reputed CEOs have
low expectations and minimum tolerable levels and accept poor firm performance
without taking any action.
I posit the following competing hypotheses to test the moderating effects of
CEO reputation on an inverse relationship between firm performance and forced
turnover.
Hypothesis 2a (the ability hypothesis): CEOs with high reputation are less
likely to be forced out under the circumstances of poor performance than are
CEOs with low reputation.
Hypothesis 2b (the symbolic image hypothesis): CEOs with high reputation
are more likely to be forced out under the circumstances of poor performance
than are CEOs with low reputation.

CEO Reputation and Compensation
The agency literature advocates that performance-based compensation
minimizes agency costs by motivating CEOs’ efforts and attracting CEOs with high
ability (Holmstrom 1979; Demski 1979; Lazear and Rosen 1981). Many empirical
studies have shown that both accounting and market performance are positively
associated with CEO compensation (Murphy 1985; Coughlan and Schmidt 1985;
Lambert and Larcker 1987; Jensen and Murphy 1990).
This study examines whether CEO reputation affects pay-for-performance
sensitivity. The ability perspective based on agency theory argues that compared to a
flat salary contract, an output-based incentive contract will attract CEOs with high
ability (Demski and Feltham, 1978). CEOs with high reputation (ability) are more
likely to choose firms adopting outcome-based compensation than CEOs with low

69
reputation. CEOs with high reputation also pressure the firms they lead to put more
weight on performance-based bonus or stock compensation, which increases pay-for-
performance sensitivity (Milbourn 2003).
42

The symbolic image perspective argues that a symbolic connection between a
well-known CEO and a firm will make stakeholders attribute most of firm
performance to the CEO (Taylor et al. 1977). And this attribution induces significant
financial incentives for good performance but significant penalties (negligible bonus)
for poor performance. Thus, more-reputed CEOs will have higher pay-for-
performance sensitivity than less-reputed CEOs.
I posit the following hypothesis to test the effects of CEO reputation on pay-
performance sensitivity. Notice that both the ability and symbolic image perspectives
predict that CEO reputation positively affects pay-performance sensitivity whereas
the hypotheses regarding CEO turnover (H2a vs. H2b) offer conflicting predictions.
Hypothesis 3 (the ability/symbolic image hypothesis): CEOs with high
reputation will have higher pay-performance sensitivity than CEOs with low
reputation.

5.2. Sample
The sample of CEOs used in this study is collected from the ExecuComp
database. To collect CEO reputation data I first identify 4,612 unique CEOs from the
ExecuComp database for the years 1993 – 2004. The final sample for CEO
reputation measures has the total of 4,341 CEOs after excluding 271 CEOs whose

42
An optimal compensation contract requires that pay-for-performance sensitivity is an
increasing function of a CEO’s (perceived) ability.

70
tenure is less than one year. I collect CEO award data from the leading business
press, which identifies 661 CEOs with one or more awards and 3,680 CEOs without
any record of award recognitions (Johnson et al. 1993; Malmendier and Tate 2005). I
also collect media count data of all 4,341 CEOs from the Factiva database (Milbourn
2003; Francis et al. 2004).
I first identify a sample of 21,950 CEO years from the ExecuComp database
for the years 1993 – 2004. I then exclude CEO years if there is a change of a CEO to
eliminate a noise from using the firm performance of multiple CEOs. I exclude the
first and second year as a CEO for turnover tests (t = 1 and 2), since the tests require
the performance of the prior year. And I exclude the first and last year as a CEO for
compensation tests (t = 1 and n), because the tests require the performance of the
current year. This exclusion leaves 16,527 CEO years for turnover tests and 17,149
CEO years for compensation tests. After merging these data sets with ExecuComp
(compensation, age,
43
and ownership percentage), Compustat (accounting earnings
and market-to-book ratio), and CRSP (stock return and market value), the final
samples consist of 14,063 and 15,035 observations for turnover and compensation
tests respectively.

43
I first collect the age data of CEOs from the ExecuComp database, but the ExecuComp
has age information for only 20% of the sample. So I search age information from Forbes,
10-K, proxy statements, and the Factiva database.

71
5.3. Measurement and Research Design
Reputation
I employ the two most often employed proxies of CEO reputation: CEO
award recognitions (Johnson et al. 1993; Malmendier and Tate 2005) and media
exposure (Milbourn 2003; Francis et al. 2004) as explained in Chapter 2. First, I
collect CEO awards such as “CEOs of the Year” or “Best Managers” from various
business journals. AWARD is an indicator variable set equal to 1 if the CEO was
given any award for the last five years or 0 otherwise.
Second, I collect the media exposure of CEOs by searching news articles
from the Factiva database. Following previous studies measuring reputation
(Milbourn 2003, Francis et al. 2004), I employ the number of news articles given the
name of a CEO and the company of the CEO. I use the sum of the last three years of
media counts. I derive MEDIA that is set equal to 1 if the CEO’s media count for the
last three years is more than the top 20% of the industry or 0 otherwise.

Turnover
Another challenging task of this study is to identify forced turnover events. I
first identify 1,987 CEO turnover events using the dates of CEOs’ departures
(LEFTOFC) from the ExecuComp database. I eliminate the turnovers of interim
CEOs and CEO changes related to mergers and acquisitions, spin-offs, liquidations,
and ownership control changes (proxy fights), which leaves the total of 1,616
turnover observations. Next, I verify the detailed reason of turnover by reading news

72
articles from the Factiva database and further categorize turnover into 426 forced
turnovers (poor performance, strategic change, scandal, moving to a smaller
company, pursuing other interests, and personal issues) and 1,190 voluntary
turnovers (retirement, death, health reason, remaining as a chairman, moving to a
larger company, and leadership succession for transition). Table 6 presents the
frequency and percentage of the detailed reason of CEO turnover. The percentages of
turnover reasons are similar to previous studies (Warner et al. 1988; Weisbach
1988). TURN is an indicator variable set equal to 1 if the CEO stepped down
involuntarily or 0 otherwise.
It is challenging to determine the actual reasons of CEO departures because
of the following two reasons (DeFond and Park 1999; Engel et al. 2003). First, many
news articles often pay attention to the fact of a CEO change rather than the reason
for CEO turnover. Second, the media gets its most information from the company’s
spokesperson or the CEO leaving the firm who hesitates to reveal underlying reasons
for turnover.
44
Regardless of the challenging task of classifying an ambiguous
turnover event as either forced turnover or voluntary turnover, this classification
provides more reliable results. Since voluntary turnover is not associated with
performance in most cases, using general turnover without detailed classification
decreases the power of a turnover test.

44
For example, the spokesperson of the company often attributes a CEO change to pursuing
other interests and personal issues. But those reasons are interpreted as forced turnover
among financial analysts and the business community in many cases. Thus, I classify the
reason of pursuing other interests and personal issues as forced turnover as most previous
studies did (Engel et al, 2003).

73
Table 6.
Reasons for CEO Turnover

Number Percentage
Retirement or remain as chairman 927 57.37%
Death 26 1.61%
Health 29 1.79%
Leadership succession 91 5.63%
Promoted to another company 46 2.85%
No reason given 42 2.60%
No article 29 1.79%
Non-forced 1,190 73.64%

Poor performance 164 10.15%
Legal/scandal 36 2.23%
Demoted within firm 37 2.29%
Demoted to another company 11 0.68%
Pursue other interests 123 7.61%
Disagreement with board/policy 55 3.40%
Forced 426 26.36%

Total 1,616 100.00%
Source: Factiva.

Compensation
I measure CEO compensation using CEO total compensation defined as the
sum of total flow compensation and the change in the value of the CEO’s equity
(stock and options) portfolio (Milbourn 2003, Core et al. 2003). The value of the
CEO’s stock holdings is calculated using the number of stock shares held by the
CEO multiplied by stock price at year end from the ExecuComp database.
45
The
value of stock option holdings is estimated as the sum of two ExecuComp items:

45
Milbourn (2003) uses the percentage of total shares outstanding held by the CEO to
calculate the value of the CEO’s stock holdings. But I use the number of stock shares
because the number of stock shares (SHROWN) has fewer missing values than the
percentage of total shares held by the CEO (SHROWNPC).

74
INMONEX (the value of exercisable in-the-money options) and INMONUN (the
value of unexercisable in-the-money options) which are consistent with Milbourn
(2003).
Previous studies often use CEO total pay defined as the sum of salary, cash
bonus, long-term incentive plan payouts, the value of restricted stocks granted, the
value of options granted, and any other annual pay (Sloan 1993, Baker et al. 1998).
The change in the value of the CEO’s equity portfolio, however, constitutes the
majority of the variability in pay-sensitivities, though annual total pay reflects
significant portions of total compensation (Murphy 1999, Milbourn 2003). Baker
(1987) and Hall and Liebman (1998) argue that the change in the value of a CEO’s
equity portfolio should be included in computing the CEO’s compensation to
measure monetary incentives. They explain that CEOs make decisions considering
not only annual flow compensation but also the effect of the decisions on his firm-
specific portfolio. Core et al. (2003) show that CEO total compensation is a better
measure of monetary incentives than total pay.

Performance
I measure firm performance using both accounting earnings and market
returns as explained in Chapter 2 (Murphy 1985; Coughlan and Schmidt 1985;
Lambert and Larcker 1987; Weisbach 1988; Engel et al. 2003). Previous studies
have established an association between accounting performance and
turnover/compensation. Boards of directors and firms often attribute CEO dismissals
to poor accounting performance as well as market performance (Weisbach 1988;

75
Engel et al., 2003). Compensation contracts often rely on accounting earnings as a
performance measure (Murphy, 1999). I use two-digit industry-adjusted returns on
assets, ROA
t
(net income before extraordinary items divided by total assets minus
industry median), as my earnings measure. I employ stock returns as a market
performance measure. Abnormal stock returns are calculated using the three-factor
model (Fama and French 1992). I cumulate monthly excess returns over the firm’s
fiscal year and calculate two-digit industry-adjusted abnormal returns, RET
t
.

Research Design
To investigate the job retention and compensation benefits of CEO reputation
I employ regression models using an interaction variable between reputation and
firm performance. I first examine the moderating effects of CEO reputation on the
relationship between firm performance and CEO turnover. To test the second set of
hypotheses, I estimate the following logit regression model:
46

Logit (TURN
it
) = β
0
+ β
1
ROA
it-1
+ β
2
RET
it-1
+ β
3
REPUTATION
it-1

+ γ
1
REPUTATION
it-1
* ROA
it-1
+ γ
2
REPUTATION
it-1
* RET
it-1
+ ∑ρ
k
Z
kit
+ ε
it
, (2)
where TURN
it
is the turnover decision of the CEO (1 = if forced turnover or 0 =
otherwise), REPUTATION
it-1
is the CEO’s reputation measured by AWARD
it-1
(the
CEO’s award from business magazines) or MEDIA
it-1
(the press coverage count of

46
The logit regression of testing CEO turnover has the problem of unbalanced sample by
nature, because only 3% of the sample are forced turnover observations. Cramer (1999)
indicates the problem of low predictability but Maddala (1991) shows that coefficients are
unbiased.

76
the CEO), ROA
it
is the firm’s returns of assets, RET
it-1
is the stock returns of the
firm, and ∑ Z
kit
is the vector of control variables such as firm size, CEO age, tenure,
ownership percentage, CEO total pay, year dummy, and industry dummy.
47

Many studies find an inverse relationship between firm performance and
CEO turnover, which would result in β
1
and β
2
to be negative in equation (2)
(Couglan and Schmidt 1985; Warner, Watts, and Wruck 1988; Weisbach 1988). The
coefficients of γ
1
and γ
2
indicate whether CEO reputation moderates the inverse
relationship between performance and turnover. The ability hypothesis regarding
turnover (H2a) predicts γ
1
and γ
2
to be positive, since the reputations of CEOs help
them survive through hard times. According to the symbolic image hypothesis
regarding turnover (H2b), however, I expect γ
1
and γ
2
to be negative, since the
reputations of CEOs impair their job retention under the circumstances of poor
performance.
Next, I investigate the effects of CEO reputation on pay-performance
sensitivity. To test the third hypothesis, I estimate the following OLS regression
model:
COMP
it
= β
0
+ β
1
ROA
it
+ β
2
RET
it
+ β
3
REPUTATION
it-1

+ γ
1
REPUTATION
it-1
* ROA
it
+ γ
2
REPUTATION
it-1
* RET
it

+ ∑ρ
k
Z
kit
+ ε
it
, (3)

47
Previous studies show that firm size, CEO age, tenure, and ownership percentage are
associated with CEO turnover (Barro and Barro 1990; Murphy 1999; Engel et al. 2003;
Huson et al. 2004). Firm size, CEO age, tenure, and the ownership are also related to CEO
reputation (Johnson et al. 1993; Milbourn 2003).

77
where COMP
it
is the total compensation of the CEO, REPUTATION
it-1
is the CEO’s
reputation measured by AWARD
it-1
(the CEO’s award from business magazines) or
MEDIA
it-1
(the press coverage count of the CEO), ROA
it
is the firm’s returns of
assets, RET
it
is the stock returns of the firm, and ∑ Z
kit
is the vector of control
variables such as firm size, ownership percentage, firm growth, year dummy, and
industry dummy.
48

Many empirical studies have shown that both accounting and market
performance are positively associated with CEO compensation, which suggests β
1

and β
2
to be positive in equation (3) (Murphy 1985; Lambert and Larcker 1987;
Jensen and Murphy 1990). The hypothesis regarding pay-performance sensitivity
(the ability and symbolic image hypothesis; H3) predicts γ
1
and γ
2
to be positive,
since CEO reputation increases pay-performance sensitivity.

5.4. Empirical Results
Descriptive Statistics
Descriptive statistics for the variables used to test the hypotheses are
presented in Table 7. The mean (median) of ROA is 2.4% (1.7%), which suggests
that the distribution of ROA is not skewed. The mean (median) of RET is 3.8%
(0.8%). Although the mean value of RET is similar to that of ROA, the standard
deviation of RET (0.432) is much greater than that of ROA (0.145), which suggests a

48
Previous studies show that firm size, ownership percentage, and firm growth are related to
CEO compensation (Ciscel and Caroll 1980; Murphy 1985; Kostuik 1990; Hall and Liebman
1998).

78
Table 7.
Descriptive Statistics (Turnover Sample)

N Mean Median Std. dev
ROA
t-1
14,063 0.024 0.017 0.145
RET
t-1
14,063 0.038 0.008 0.432
AWARD
t-1
14,063 0.140 0 0.347
MEDIA
t-1
14,063 0.239 0 0.427
TURN
t
14,063 0.023 0 0.149
AGE
t
14,063 55.580 56 7.627
TENURE
t
14,063 8.980 7 7.470
SHROWN
t
14,063 2.810 0.400 6.476
SIZE
t
14,063 5,159.62 1,004.46 18,452
PAY
t-1
13,525 4,028.73 1,826.10 11,549

ROA ≡ the return on assets of the firm defined as net income before
extraordinary item divided by total assets;
RET ≡ the stock returns of the firm defined as abnormal returns using the
three-factor model (Fama and French, 1993);
AWARD ≡ the CEO’s awards from business magazine that equals 1 if the CEO
was bestowed any award for the last five years, 0 otherwise;
MEDIA ≡ the press coverage count of the CEO that equals 1 if the CEO was in
the top 20% media exposure of the industry for the last three years, 0
otherwise;
TURN ≡ the turnover decision of the CEO that equals 1 if forced turnover, 0
otherwise;
AGE ≡ the age of the firm’s CEO;
TENURE ≡ the tenure of the CEO;
SHROWN ≡ the percentage of the CEO’s ownership;
SIZE ≡ the market value of equity of the firm (in millions of dollars);
PAY ≡ the total flow compensation of the CEO of the firm (in thousands of
dollars).

79
Table 8.
Descriptive Differences by CEO Reputation

Panel A: Turnover ratio differences by CEO reputation

ROA
Top Quintile
Bottom
Quintile Difference
Yes 1.12% 7.04% 5.92%
AWARD
No 1.59% 4.94% 3.35%
Yes 2.31% 7.04% 4.73%
MEDIA
No 1.28% 4.49% 3.21%
RET

Top Quintile
Bottom
Quintile Difference
Yes 1.21% 5.31% 4.10%
AWARD
No 1.08% 4.43% 3.35%
Yes 1.98% 7.78% 5.80%
MEDIA
No 0.85% 3.52% 2.67%

Panel B: Compensation differences by CEO reputation
(in thousands of dollars)
ROA
Top Quintile
Bottom
Quintile Difference
Yes 9,245.1 3,174.7 6,070.3
AWARD
No 4,281.9 2,906.1 1,375.8
Yes 9,515.7 5,263.7 4,252.0
MEDIA
No 3,619.6 2,201.6 1,418.0
RET

Top Quintile
Bottom
Quintile Difference
Yes 9,764.4 7,661.2 2,103.1
AWARD
No 3,758.9 3,038.9 720.1
Yes 8,393.9 7,107.4 1,286.5
MEDIA
No 3,212.5 2,443.8 768.7

80

Table 9.
Pearson Correlations (Turnover Sample)

Panel A: Basic Variables
RET AWARD MEDIA TURN COMP SIZE
ROA 0.137*** 0.031*** 0.002 -0.039*** 0.107*** 0.125***
RET 1.000 -0.025*** -0.032*** -0.079*** 0.368*** -0.114***
AWARD 1.000 0.244*** -0.015* 0.039*** 0.304***
MEDIA 1.000 0.046*** 0.101*** 0.418***
TURN 1.000 -0.007 0.008
COMP 1.000 0.133***
SIZE 1.000
*** Correlation is significant at the 0.01 level (2-tailed).
** Correlation is significant at the 0.05 level (2-tailed).
Panel B: Interaction Variables
RET *
AWARD
ROA *
MEDIA
RET *
MEDIA
TURN COMP
ROA * AWARD 0.286*** 0.341*** 0.111*** -0.029*** 0.061***
RET * AWARD 1.000 0.103*** 0.340*** -0.046*** 0.127***
ROA * MEDIA 1.000 0.219*** -0.026*** 0.095***
RET * MEDIA 1.000 -0.069*** 0.182***
TURN 1.000 -0.007
COMP 1.000
*** Correlation is significant at the 0.01 level (2-tailed).
** Correlation is significant at the 0.05 level (2-tailed).

81
wider range of RET than that of ROA. The mean of AWARD is 0.140, which
indicates that 14.0% of CEOs have been bestowed one or more awards for the last
five years. The standard deviation of MEDIA (0.427) is similar to that of
AWARD (0.347). The mean and median of PAY are $ 4.03 million and $ 1.83
million respectively. The mean (median) of COMP is $ 16.13 billion ($ 3.22 billion).
The mean and median of SIZE are $ 5.16 billion and $ 1.00 billion. These results
show that the distributions of CEO flow pay, CEO total compensation, and firm size
are skewed, which suggests the need of normalization via log transformations.
Table 8 presents descriptive information showing the differences of turnover
and compensation by CEO reputation.
49
Panel A of Table 8 reports the differences of
turnover ratio between CEO reputation groups (2 x 2 design: CEO reputation x firm
performance). I divide performance group using performance quintile and compare
turnover ratios between top quintile and bottom quintile. Panel A of Table 8 shows
that turnover ratio increases as firm performance decreases, which implies an inverse
relationship between performance and turnover. The results show that the turnover
ratios of more-reputed CEOs in bottom performance quintile (7.04%, 7.04%, 5.31%,
and 7.78%) are always greater than those of less-reputed CEOs (4.94%, 4.49%,
4.43%, and 3.52%). The results also report that the differences of turnover ratio
between top performance quintile and bottom performance quintile for more-reputed
CEOs (5.92%, 4,73%, 4.10%, and 5.80%) are greater than those for less-reputed

49
Table 8 does not provide statistical analyses but simply illustrates group differences by
CEO reputation.

82
CEOs (3.35%, 3.21%, 3.35%, and 2.67%). The results suggest that CEOs with high
reputation are more likely to be dismissed as performance slides than CEOs with low
reputation.
Panel B of Table 8 presents the differences of CEO compensation using the
same group comparison scheme as the one used for turnover ratio differences. The
results show that CEO compensation increases as firm performance increases, which
implies a positive association between performance and compensation. The results
also report that the differences of compensation between top performance quintile
and bottom performance quintile for more-reputed CEOs (6.07 million, 4.252
million, 2.103 million, and 1.286 million) are always greater than those for less-
reputed CEOs (1.375 million, 1.418 million, 0.720 million, and 0.768 million). The
results suggest that CEOs with high reputation have higher pay-performance
sensitivity than CEOs with low reputation.
Univariate analyses using Pearson correlations are presented in Table 9.
Panel A of Table 9 presents Pearson correlations among basic variables such as
AWARD, MEDIA, ROA, RET, TURN, and COMP whereas Panel B reports
correlations among interaction variables (AWARD * ROA, AWARD * RET,
MEDIA * ROA, and MEDIA * RET) and the two dependent variables (TURN and
COMP). Both AWARD and MEDIA are positively associated with COMP (r=0.039
and 0.101, p < 0.01 for both, two-tailed), which shows the rent extractions of highly
regarded CEOs (Malmendier and Tate 2005). MEDIA is positively related to TURN
(r=0.046, p < 0.01, two-tailed), whereas AWARD is negatively associated with

83
TURN (r=-0.015, p < 0.10, two-tailed). These correlations show the conflicting
effects of CEO reputation on forced turnover, which needs further analyses in
multivariate tests.
Both ROA and RET are inversely related to TURN (r=-0.039 and -0.079, p <
0.01 for both, two-tailed) and positively associated with COMP (r=0.107 and 0.368,
p < 0.01 for both, two-tailed), which is consistent with previous studies (Couglan and
Schmidt 1985; Murphy 1985; Lambert and Larcker 1987; Warner et al. 1988;
Weisbach 1988). AWARD and MEDIA are positively correlated (r=0.244, p<0.01,
two-tailed). A correlation between ROA and RET is significantly positive (r=0.137,
p<0.01, two-tailed), which shows a high association between accounting
performance and market performance. Both AWARD and MEDIA are positively
associated with SIZE (r=0.304 and 0.418, p < 0.01 for both, two-tailed), which
indicates the need of including firm size as a control variable.
Panel B of Table 9 shows that all of four interaction variables between CEO
reputation and firm performance are negatively correlated to TURN (r=-0.029, -
0.046, -0.026, and -0.069, p < 0.01 for all four, two-tailed), which supports the
symbolic hypothesis (H2b) instead of the ability hypothesis (H2a). Panel B also
indicates that all interaction variables are positively associated with COMP (r=0.061,
0.127, 0.095, and 0.182, p < 0.01 for all four, two-tailed), which supports the
hypothesis regarding pay-performance sensitivity (H3).

84
Table 10.
CEO Reputation and Forced Turnover
TURN
it
is the turnover decision of the CEO (1 = if forced turnover or 0 = otherwise), ROA
it-1
is the
firm’s returns of assets of the prior year adjusted by two-digit industry median, RET
it-1
is the firm’s
stock returns of the prior year adjusted by two-digit industry median, AWARD
it-1
is the CEO’s awards
from business magazine (1= if the CEO was bestowed any award for the last five years or 0 =
otherwise), MEDIA
it-1
is the press coverage count of the CEO (1= if the CEO was in the top 20%
media exposure of the industry for the last five years or 0 = otherwise), SIZE
it
is the firm’s market
value of equity, AGE
it
is the age of the firm’s CEO, SHROWN
it
is the percentage of the CEO’s
ownership, TENURE
it
is the tenure of the CEO, and PAY
it-1
is the total flow compensation of the
CEO of the firm. *** denotes significance at the 0.01 level (2-tailed), ** at the 0.05 level, and * at the
0.10 level.

Forced turnover (TURN
t
)
Coefficient (Chi-Square)
Intercept -2.675** (4.59) -2.242* (3.14)
ROA
t-1
-0.349 (2.26) -0.293 (1.86)
RET
t-1
-1.141*** (57.23) -1.092*** (40.11)
AWARD
t-1
-0.617** (5.57)
AWARD * ROA
t-1
-6.842** (6.90)
AWARD * RET
t-1
-1.990*** (11.20)
MEDIA
t-1
0.667*** (18.94)
MEDIA * ROA
t-1
-1.001 (1.90)
MEDIA * RET
t-1
-0.552* (3.08)
SIZE
t
0.117** (4.97) -0.015 (0.08)
AGE
t
-0.007 (0.56) -0.007 (0.54)
SHROWN
t
-0.153*** (20.26) -0.153*** (20.74)
TENURE
t
-0.030** (5.625) -0.033*** (6.76)
PAY
t-1
-0.126* (3.62) -0.153*** (6.73)
Year dummy Yes Yes
Industry dummy Yes Yes
Likelihood Ratio Chi-Squared 292.45 304.88
N 13,525 13,525

85
Table 11.
CEO Reputation and Total Compensation
COMP
it
is the total compensation of the CEO, ROA
it
is the firm’s returns of assets of the prior year
adjusted by two-digit industry median, RET
it
is the firm’s stock returns of the prior year adjusted by
two-digit industry median, AWARD
it-1
is the CEO’s awards from business magazine (1= if the CEO
was bestowed any award for the last five years or 0 = otherwise), MEDIA
it-1
is the press coverage
count of the CEO (1= if the CEO was in the top 20% media exposure of the industry for the last five
years or 0 = otherwise), SIZE
it
is the firm’s market value of equity, SHROWN
it
is the percentage of
the CEO’s ownership, and GROWTH
it
is the market to book ratio of the firm.*** denotes significance
at the 0.01 level (2-tailed), ** at the 0.05 level, and * at the 0.10 level.

Total compensation (COMP
t
)
Coefficient (t-value)
Intercept 5.573*** (11.37) 5.824*** (11.90)
ROA
t
0.899*** (5.07) 0.863*** (4.6)
RET
t
2.948*** (49.39) 2.934*** (46.09)
AWARD
t-1
0.176** (2.02)
AWARD * ROA
t
-0.005 (-0.01)
AWARD * RET
t
0.828*** (3.41)
MEDIA
t-1
0.491*** (7.74)
MEDIA * ROA
t
1.177** (2.37)
MEDIA * RET
t
0.263* (1.77)
SIZE
t
0.247*** (13.84) 0.190*** (10.07)
GROWTH
t
-0.0003 (-0.72) -0.0003 (-0.71)
SHROWN
t
-0.031*** (-7.60) -0.314*** (-7.72)
Year dummy Yes Yes
Industry dummy Yes Yes
F-Statistic 55.18 56.15
Adjusted R
2
0.240 0.244
N 14,050 14,050

86
Empirical Tests
I first investigate whether CEO reputation provides job security benefits to
CEOs. The results of job retention verify the findings of the descriptive analyses
shown in Table 8, Panel A. Table 10 presents the logit regression results of testing
whether CEO reputation moderates an inverse relationship between performance and
turnover. The results show that AWARD * ROA has a significantly negative
coefficient ( γ
1
= -6.842, p < 0.05, two-tailed). The results also report that both
AWARD * RET and MEDIA * RET have significantly negative coefficients ( γ
2
= -
1.990 and -0.552, p < 0.01 and p < 0.10, two-tailed). MEDIA * ROA, however, is
not significantly associated with forced turnover. The overall results of turnover tests
suggest that CEO reputation does not improve but impairs job retention because
high-profile CEOs serve as scapegoats when firms suffer from poor performance,
which supports the symbolic image hypothesis (H2b).
The results of turnover tests might be related to the results of firm
performance reversal tests. Because well-known CEOs do not necessarily turn
around poor performance, boards of directors would replace well-known CEOs with
turnaround specialists. But, the results of performance reversals alone cannot explain
why CEOs with high reputation are more likely to be dismissed than CEOs with low
reputation, since more reputed CEOs and less reputed CEOs have the similar ability
to reverse poor performance. The attribution of poor performance to high-profile
CEOs and increased gaps between heightened expectations and actual performance
would play a critical role in dismissing well-known CEOs in struggling companies.

87
Next, I examine whether the reputations of CEOs increase their
compensation. The overall results of compensation tests are similar to the findings of
the descriptive information shown in Panel B of Table 8. Table 11 presents the
regression results testing whether CEO reputation affects pay-performance
sensitivity. The results indicate that MEDIA * ROA has a significantly positive
coefficient ( γ
1
= 1.177, p < 0.05, two-tailed). The results also show that both
AWARD * RET and MEDIA * RET have significantly positive coefficients ( γ
2
=
0.828 and 0.263, p < 0.01 and p < 0.10, two-tailed). AWARD * ROA, however, is
not significantly associated with total compensation (COMP). Overall results suggest
that CEO reputation significantly increases pay-performance sensitivity for both
accounting and market performance, which supports the hypothesis regarding pay-
performance sensitivity (H3). This study shows that CEO reputation positively affect
accounting-based pay-performance sensitivity as well as equity-based pay-
performance sensitivity (Milbourn 2003). Moreover, CEO reputation affects pay-
performance sensitivity through accounting performance more than market
performance.

Robustness Tests
I run robustness tests to check whether varying the proxies of CEO reputation
(AWARD and MEDIA) provides similar results to those of the empirical tests. I set
five years of lagged period after CEOs receive recognition awards. I vary this lagged
value of five years to three years (AWARD is set to 1 if the CEO received any award

88
for the last three years) and permanence (AWARD is set to 1 throughout sample
period once the CEO was bestowed any award). I also change the summation period
of media counts from three years to five or two years. To check the validity of
converting a continuous variable (the sum of media counts) to an indicator variable
(MEDIA), I vary the threshold of top 20% of the industry to 10% or 30%.
Overall, the results of sensitivity tests using the various proxies of reputation
do not qualitatively alter the conclusions, but there are some variations in detailed
results. Turnover tests report that 11 out of 16 coefficients ( γ
1
or γ
2
) are significantly
negative and 5 coefficients are not significant. The results indicate that CEO
reputation impairs job retention under the circumstances of poor performance, which
supports the symbolic image hypothesis (H2b). Compensation tests show that 11 out
of 16 coefficients ( γ
1
or γ
2
) are significantly positive and that 4 coefficients are not
significant. The results suggest that CEO reputation positively affect pay-
performance sensitivity, which supports the hypothesis regarding pay-performance
sensitivity (H3).

5.5. Discussion
This chapter investigates whether CEO reputation provides better job
retention and compensation. The personal benefits of CEO reputation is related to the
performance benefits of CEO reputation, since the benefits of job retention and
compensation is justified if CEO reputation improves future performance.

89
Shareholders and boards of directors determine the job retention and compensation
of CEOs based on beliefs about the superior ability of highly regarded CEOs.
The results of job retention tests report that well-known CEOs are more
likely to be dismissed than CEOs with low reputation as performance slides. The
results of compensation tests show that CEO reputation increases pay-performance
sensitivity. The results imply that CEOs had better put more effort towards
improving firm performance than towards promoting their own images in the media
for their job retention. The celebrity images of CEOs alone do not help and may even
impair their job retention, even though well-known CEOs are rewarded with more
compensation. High-profile CEOs risk serving as scapegoats when firms struggle
financially due to the attribution of poor performance to high-profile CEOs and
heightened expectations.

90
Chapter 6. Conclusion and Limitations
In this dissertation I test empirically whether CEO reputation provides three
economic benefits: firm performance, job retention, and compensation. Empirical
investigation of the three economic benefits of CEO reputation is needed, since there
are two perspectives that provide different predictions. The ability perspective,
which views reputation as ability, argues that CEO reputation will improve firm
performance, job retention, and compensation. On the other hand, the symbolic
image perspective suggests that CEO reputation will not improve firm performance
and job retention. The reputations of CEOs might even impair their job retention
because high-profile CEOs may serve as scapegoats for poor performance. I derive
competing hypotheses based on the two arguments and empirically test the two
perspectives.
Overall, the results indicate that the reputations of CEOs increase their
compensation but not necessarily firm performance and job retention. The results of
firm performance tests suggest that reputed CEOs sustain good performance but may
not turn around poor performance. The results of job retention tests show that the
reputations of CEOs impair their job retention under the circumstances of poor
performance. The results of compensation tests report that CEO reputation increases
pay-performance sensitivity.
This study has the following practical implications. First, promoting CEO’s
own images in the media does not necessarily secure their job positions. As CEOs
achieve superstar status, they are rewarded with more compensation. Well-known

91
CEOs, however, risk serving as scapegoats when firms struggle financially. The
celebrity images of CEOs alone do not help and may even impair their job retention
due to the attribution of poor performance to high-profile CEOs and heightened
expectation. Second, shareholders and a board of directors are better off by keeping a
well-known CEO when a firm prospers, but may have to consider replacing a
celebrity CEO with a turnaround specialist when the firm suffers financially. A
highly regarded CEO has the ability of performance persistence (maintaining good
performance) but not necessarily the ability of performance reversals (turning around
poor performance). The results also suggest that investors had better be careful about
investing in a company solely based on a celebrity CEO once firm performance
slides.
There are several limitations to this study. First, both award recognition and
media coverage counts might not be good proxies of CEO reputation. Both proxies
may be affected by firm size and political processes that are more related to the
symbolic images of CEOs than their ability. These two proxies, however, are the
most widely accepted/used measures of reputation in the previous literature (Johnson
et al. 1993; Milbourn 2003; Francis et al. 2004; Malmendier and Tate 2005). This
study and previous studies provide validity tests to check whether award and media
coverage are reasonable proxies of CEO reputation. Milbourn (2003) uses CEO
tenure, outside-hired CEOs, and past performance as the proxies of CEO reputation.
CEO tenure, however, might be one of the consequences of CEO reputation and past
performance might be one of the determinants of CEO reputation (Johnson et al.

92
1993). Thus, CEO tenure and past performance are not reasonable proxies to
examine an association between CEO reputation and future performance/job
retention.
Second, it is very difficult to classify CEO departures into voluntary or forced
turnover by reading news articles. News articles seldom specify the reasons of CEO
departures, since those reasons are confidential in many cases. Categorizing
ambiguous turnover into forced or voluntary turnover, however, provides more
reliable results as shown in the previous literature (Couglan and Schmidt 1985;
Warner, Watts, and Wruck 1988; Weisbach 1988; Fee and Headlock 2002; Farrell
and Whidbee 2003).
Despite the above limitations of measurement, this study contributes to the
understanding of a more complete picture of the economic benefits of CEO
reputation. Many studies in the accounting literature based on agency theory argue
that CEOs with positive reputations have superior ability. The economic benefits of
CEO reputation were simply taken for granted. CEO reputation, however, is affected
by the media considerably and the business community might have misperception or
over-confidence regarding CEOs’ ability. This psychological aspect of CEO
reputation sometimes has a countervailing force against the ability aspect of CEO
reputation as shown in conflicting anecdotal evidence of Coca-Cola CEO Roberto
Goizueta and Compaq CEO Eckhard Pfeiffer.
This study investigates which aspect of CEO reputation dominantly affects
economic benefits empirically. Empirical results support that the psychological

93
aspect of CEO reputation as the perceptions of CEOs dominates the ability aspect of
CEO reputation in job retention decisions. The results of performance benefit tests
give us a clue as to explain conflicting anecdotal evidence. The results imply that
Compaq dismissed Eckhard Pfeiffer, because it was struggling financially due to
severe competition and needed to turn around in the near future. The results also
suggest that Coca-Cola kept Roberto Goizueta, since Coca-Cola only had to resume
producing its traditional product to resolve the marketing crisis. Coca-Cola mainly
needed to maintain good performance instead of turning around poor performance in
the near future.
This study calls for more future research regarding CEO reputation. The
association between CEO reputation and corporate reputation is an interesting issue
to pursue in the future. It is worthwhile to examine whether CEO reputation affects
the likelihood of getting recruited in other firms. The effects of CEO reputation on
the expectations of analysts/investors need to be investigated.

94
Bibliography
Aaker, D.A. 1991. Managing Brand Equity - Capitalizing on the Value of a Brand
Name. New York: The Free Press.
Ader, C.R. 1995. A longitudinal study of agenda setting for the issue of
environmental pollution. Journalism and Mass Communication Quarterly 72:
300 – 311.
Anderson, R.E. 1973. Consumer Dissatisfaction: The effect of disconfirmed
expectancy on perceived product performance. Journal of Marketing
Research 10: 38 – 44.
Ang, J.S., R.A. COLE, and J.W. LIN. 2000. Agency Costs and Ownership Structure.
Journal of Finance 55: 81 – 106.
Baber, W.R., S.H. Kang, and K.R. Kumar. 1999. The explanatory power of earnings
levels vs. earnings changes in the context of executive compensation. The
Accounting Review 74: 459 – 472.
Baiman, S. 1982. Agency research in managerial accounting: A survey. Journal of
Accounting Literature 1: 154 – 213.
Ball-Rokeach, S.J. and M.G. Cantor. 1986. Media, audience, and social structure.
Newbury Park: Sage.
Barclay, M.J., D. Gode, and S.P. Kothari, 2000. The advantages of using earnings for
compensation: Matching delivered performance. Working paper: University
of Rochester, New York University, and Massachusetts Institute of
Technology.
Bearden, W.O. and J.E. Teel. 1983. Selected determinants of consumer satisfaction
and complaint reports. Journal of Marketing Research 20: 21 – 28.
Beatty, R.P. and E.J. Zajac. 1994. Managerial incentives, monitoring, and risk
bearing: A study of executive compensation, ownership, and board structure
in initial public offerings. Administrative Science Quarterly 39: 313 – 335.
Behr, R.L. and S. Iyengar. 1985. Television news, real-world cues, and changes in
the public agenda. Public Opinion Quarterly 49: 38 – 57.
Boote, J. 1998. Towards a comprehensive taxonomy and model of consumer
complaining behaviour. Journal of Consumer Satisfaction, Dissatisfaction
and Complaining Behavior 11: 140 – 151.

95
Buttle, F.A. 1996. SERVQUAL: Review, critique, research agenda. European
Journal of Marketing 30: 8 – 32.
Byrne, J.A., W.C. Symonds, and H.F. Siler. 1991. CEO Disease. Business Week
(April 1): 52.
Cadotte, E.R., R.B. Woodruff, and R.L. Jenkins. 1987. Expectations and norms in
models of consumer satisfaction. Journal of Marketing Research 24: 305 –
314.
Cameron, K.S. and D.A. Whetten. 1983. Organizational effectiveness: A comparison
of multiple models. New York: Academic Press.
Cardozo, R.N. 1965. An experimental study of customer effort, expectation, and
satisfaction. Journal of Marketing Research 2: 244 – 249.
Carlsmith, J.M. and E. Aronson. 1963. Some hedonic consequences of the
confirmation and disconfirmation of expectancies. Journal of Abnormal and
Social Psychology 66: 151 – 156.
Chapanis, N.P. and A. Chapanis. 1964. Cognitive dissonance: Five years later.
Psychological Bulletin 61: 1 – 22.
Chatterjee, S., Lubatkin, M.H., Schweiger, D.M., and Weber, Y. 1992. Cultural
differences and shareholder value in related mergers: Linking equity and
human capital. Strategic Management Journal 13: 319 – 335.
Chen, C.C. and J.R. Meindl. 1991. The construction of leadership images in the
popular press: The case of Donald Burr and People Express. Administrative
Science Quarterly 36: 521 – 551.
Chevalier, J. and G. Ellison. 1999. Career concerns of mutual fund managers. The
Quarterly Journal of Economics 389 – 432.
Coughlan, A.T. and R.M. Schmidt. 1985. Executive compensation, management
turnover, and firm performance: An empirical investigation. Journal of
Accounting and Economics 7: 43 – 66.
Dahya, J., J.J. McConnell, and N.G. Travlos. 2002. The Cadbury Committee,
corporate performance, and top management turnover. The Journal of
Finance 57: 461 – 483.
Daines, R., V.B. Nair, and L. Kornhauser. 2005. The good, the bad, and the lucky:
CEO pay and skill. Working paper, Stanford University, University of
Pennsylvania, and New York University.

96
Deephouse, D.L. 2000. Media reputation as a strategic resource: An integration of
mass communication and resource-based theories. Journal of Management
26: 1091 – 1112.
DeFond, M.L. and C.W. Park. 1999. The effect of competition on CEO turnover.
Journal of Accounting Economics 27: 33 – 56.
DeGrrot, M.H. 1970. Optimal Statistical Decisions. New York: McGraw-Hill Inc.
Demski, J.S. 1976. Uncertainty and evaluation based on controllable performance,
Journal of Accounting Research 14: 230 – 245.
Denis, D.J. and D.K. Denis. 1995. Performance changes following top management
dismissals. The Journal of Finance 50: 1029 – 1057.
Dewatripoint M., I. Jewitt, and J. Tirole. 1999. The economics of career concerns,
Part I: Comparing information structures. The Review of Economic Studies
66: 183 – 198.
Dyck, A. and L. Zingales. 2002. The corporate governance role of the media.
Working paper: NBER.
Engel, E., R.M. Hayes, and X. Wang. 2003. CEO turnover and properties of
accounting information. Journal of Accounting Economics 36: 197 – 226.
Erevelles, S. and C. Leavitt. 1992. A comparison of current models of consumer
satisfaction/ dissatisfaction. Journal of Consumer Satisfaction,
Dissatisfaction and Complaining Behavior 5: 104 – 114.
Fama, E.F. 1980. Agency problems and the theory of the firm. The Journal of
Political Economy 88: 288 – 307.
Fama, E.F. and K.R. French. 1992. The Cross-Section of Expected Stock Returns.
The Journal of Finance 47: 427 – 465.
Farrell, K.A. and D.A. Whidbee. 2003. Impact of firm performance expectations on
CEO turnover and replacement decisions. Journal of Accounting and
Economics 36: 165 – 196.
Fee, C.E. and C.J. Hadlock. 2004. Management turnover across the corporate
hierarchy. Journal of Accounting and Economics 37: 3 – 38.
Feldman, S. 1966. Cognitive Consistency: Motivational antecedents and behavioral
consequences. New York: Academic Press.

97
Festinger, L. 1957. A theory of cognitive dissonance. New York: Harper & Row.
Fiske, S.T. and S.E. Taylor. 1984. Social Cognition. New York: Random House.
Fombrun C. and M. Shanley. 1990. What’s in a name? Reputation building and
corporate strategy. Academy of Management Journal 33: 233 – 258.
Francis, J., A.H. Huang, S. Rajgopal, and A.Y. Zang. 2004. CEO reputation and
earnings quality. Working paper, Duke University and University of
Washington.
Freedman, J.L. 1964. Involvement, discrepancy and change. Journal of Abnormal
and Social Psychology 69: 290 – 295.
Gaines-Ross, L. 2003. CEO capital: A guide to building CEO reputation and
company success. Hoboken: John Wiley & Sons.
Gamson, W., D. Croteau, W. Hoynes, and T. Sasson. 1992. Media images and social
construction of reality. Annual Review of Sociology 18: 373 – 393.
Gamson, W.A. and N.A. Scotch. 1964. Scapegoating in baseball. American Journal
of Sociology 70: 69 – 72.
Gentzkow, M. and J.M. Shapiro. 2006. Media bias and reputation. Journal of
Political Economy 114: 280 – 316.
George, A.L. 1959. Propaganda analysis: A study of inferences made from Nazi
propaganda in World War II. Evanston: Row, Peterson, and Company.
Gibbons, R. and K. J. Murphy. 1992. Optimal Incentive Contracts in the Presence of
Career Concerns: Theory and Evidence. The Journal of Political Economy
100: 468 – 505.
Goyal, V.K. and C.W. Park. 2002. Board leadership structure and CEO turnover.
Journal of Corporate Finance 8: 49 – 66.
Hallin, D. 1986. The uncensored war. Berkeley: University of California Press.
Hamilton, J.T. and R. Zeckhauser. 2004. Media coverage of CEOs: Who? What?
Where? When? Why? Working paper: Duke University and Harvard
University.
Hamilton, V.L. 1980. Intuitive psychologist or intuitive lawyer? Alternative models
of the attribution process. Journal of Personality and Social Psychology
39:767 – 772.

98
Hansson, B. 2004. Human capital and stock returns: Is the value premium an
approximation for return on human capital? Journal of Business Finance &
Accounting 31: 333 – 357.
Hayward, M.L.A. and D.C. Hambrick. 1997. Explaining the premiums paid for large
acquisitions: Evidience of CEO hubris. Administrative Science Quarterly 42:
103 – 127.
Heider, F. 1958. The psychology of interpersonal relations. New York: Wiley.
Hirshleifer, D. 1993. Managerial reputation and corporate investment decisions.
Financial Management 22: 145 – 160.
Holloway, R.J. 1967. An experiment on consumer dissonance. Journal of Marketing
31: 39 – 43.
Holmstr őm, B. 1979. Moral hazard and observability. Bell Journal of Economics 10:
74-91.
Holmstr őm, B. 1982. Managerial incentive problems: A dynamic perspective. In
Swedish School of Economics (Ed.), Essays in economics and management
in honor of Lars Wahlbeck. Helsinki: Swnska Handelshogkolan.
Hovland, C.I., O.J. Harvey, and M. Sherif. 1957. Assimilation and contrast effects in
reactions to communication and attitude change. Journal of Abnormal and
Social Psychology 55: 244 – 252.
Ittner, C.D. and D.F. Larcker, 1998. Are Nonfinancial Measures Leading Indicators
of Financial Performance? An Analysis of Customer Satisfaction. Journal of
Accounting Research 36 : 1 – 35.
Jensen, M.C. and W.H. Meckling. 1976. Theory of the firm: Managerial behavior,
agency costs and ownership structure. Journal of Financial Economics 3: 305
– 360.
Johnson, W.B., S.M. Young, and M. Welker. 1993. Managerial reputation and the
informativeness of accounting and market measures of performance.
Contemporary Accounting Research 10: 305 – 332.
Jones, E.E. and K.E. Davis. 1965. From acts to dispositions: The attribution process
in person perception. In L. Berkowitz (Ed.), Advances in experimental social
psychology. Vol. 2. New York: Academic Press.
Jones, E.E. and R.E. Nisbett. 1972. The actor and the observer: Divergent
perceptions of the causes of behavior. In E.E. Jones, D.E. Kanouse, H.H.

99
Kelley, R.E. Nisbett, S. Valins, and B. Weiner (Eds.), Attribution: Perceiving
the causes of behavior. Morristown: General Learning Press.
Kaplan, R. and D. Norton. 1992. The balanced-scorecard: Measures that drive
performance. Harvard Business Review, 71 – 79.
Kaplan, R.S. and D.P. Norton. 2001. Transforming the balanced scorecard from
performance measurement to strategic management: Part I. Accounting
Horizons 15: 87 – 104.
Keller, K.L. 1993. Conceptualizing, measuring, and managing customer-based brand
equity. Journal of Marketing 57: 1-22.
Kelley, H.H. 1967. Attribution theory in social psychology. In D. Levine (Ed.),
Nebraska Symposium on Motivation 15: 192 – 238.
Kelley, H.H. 1972. Attribution in social interaction. In E.E. Jones, D.E. Kanouse,
H.H. Kelley, R.E. Nisbett, S. Valins, and B. Weiner (Eds.), Attribution:
Perceiving the causes of behavior. Morristown: General Learning Press.
Kelley, H.H. 1973. The processes of causal attribution. American Psychologist 28:
107 – 128.
Khanna, N. and A.B. Poulsen. 1995. Managers of financially distressed firms:
Villains or scapegoats? The Journal of Finance 30: 919 – 940.
Khurana, R. 2002. The curse of the superstar CEO. Harvard Business Review 80: 60
– 65.
Khurana, R. and N. Nohria. 2000. The performance consequences of CEO turnover.
Working paper, Harvard University.
Kiesler, S. and L. Sproul. 1982. Managerial response to changing environments:
Perspectives on problem sensing from social cognition. Administrative
Science Quarterly 27: 548 – 570.
Laffont, J.J. and D. Martimort. 2002. The theory of incentives: The principal-agent
model. New Jersey: Princeton University Press.
Lambert, R. and D. Larcker. 1987. An analysis of the use of accounting and market
measures of performance in executive compensation contracts. Journal of
Accounting Research 25: 85 – 129.

100
Lasswell, H.D. 1949.The structure and function of communication in society. In W.
Schramm (Ed.), Mass communications: 102 – 115. Urbana: University of
Illinois Press.
Lasswell, H.D., N. Leites, and Associates. 1965. Language of politics. Cambridge:
MIT Press.
Lazear, E.P. and S. Rosen. 1981. Rank-order tournaments as optimum labor
contracts. The Journal of Political Economy 89: 841 – 864.
Lord, R.G. and J.E. Smith. 1983. Theoretical, information processing, and situational
factors affecting attribution theory models of organizational behavior.
Academy of Management Review 8: 50 – 60.
MacLeod, W.B. and J.L. Malcomson. 1988. Reputation and hierarchy in dynamic
models of employment. The Journal of Political Economy 96: 832 – 854.
Malmendier, U. and G. Tate. 2005. Superstar CEOs. Working paper, Stanford
University and University of Pennsylvania.
McCombs, M.E. and D.L. Shaw. 1972. The agenda setting function of the mass
media. Public Opinion Quarterly 69: 813 – 824.
McQuail, D. 1985. Sociology of mass communication. Annual Review of Sociology
11: 93 – 111.
Meindl, J.R., S.R. Ehrlich, and J.M. Dukerich. 1985. The romance of leadership.
Administrative Science Quarterly 30: 78-102.
Milbourn, T.T. 2003. CEO reputation and stock-based compensation. Journal of
Financial Economics 68: 233 – 262.
Miller, J.A. 1977. Studying satisfaction, modifying models, eliciting expectations,
posing problems, and making meaningful measurements. In Hunt, H.K. (Ed.)
Conceptualization and measurement of consumer satisfaction and
dissatisfaction. 72 – 91. Bloomington: Indiana University.
Morck, R., A. Shleifer, and R.W. Vishny. 1988. Management ownership and market
valuation: An empirical analysis. Journal of Financial Economics 20: 293–
315.
Murphy, K.J. 1985. Corporate performance and managerial remuneration: An
empirical investigation. Journal of Accounting and Economics 7: 11 – 42.
Murphy, K.J. 2003. Lecture note for FBE 599. University of Southern California.

101
Nguyen-Dang, B. 2005. Investor recognition hypothesis, firm value, and corporate
governance: Evidence from media coverage of CEOs. Working paper, HEC
Paris School of Management.
Oliver, R.L. 1980. A cognitive model of the antecedents and consequences of
satisfaction decisions. Journal of Marketing Research 17: 460 – 469.
Olson J.C. and P. Dover. 1979. Disconfirmation of consumer expectations through
product trial. Journal of Applied Psychology 64: 179 – 189.
Oshikawa, S. 1968. The theory of cognitive dissonance and experimental research.
Journal of Marketing Research 5: 429 - 430.
Pagano, M. and A. Roell. 1998. The choice of stock ownership structure: Agency
costs, monitoring, and the decision to go public. Quarterly Journal of
Economics 187 – 225.
Pfeffer, J. 1977. The ambiguity of leadership. The Academy of Management Review
2: 104 – 112.
Pryor, J.B. and M. Kriss. 1977. The cognitive dynamics of salience in the attribution
process. Journal of Personality and Social Psychology 35: 49 – 55.
Puffer, S.M. and J.B. Weintrop. 1991. Corporate performance and CEO turnover:
The role of performance expectations. Administrative Science Quarterly 36:
1-19.
Riahi-Belkaoui, A. 2005. Corporate reputation, internalization and the market
valuation of multinational firms. Working paper, University of Illinois at
Chicago.
Robinson, L.M. and R.L. Berl. 1979. What about compliments: A followup study on
customer complaints and compliments. In Hunt, H.K. and R.L. Day (Eds.)
Refining concepts and measures of consumer satisfaction and complaining
behavior. Bloomington: Indiana University.
Rosenberg, M.J. 1965. When dissonance fails: On eliminating evaluation
apprehension from attitude measurement. Journal of Personality and Social
Psycholgy 1: 28 – 42.
Santos, J. and J. Boote. 2003. A theoretical exploration and model of consumer
expectations, post-purchase affective states and affective behaviour. Journal
of Consumer Behavior 3: 142 – 156.

102
Schramm, W. 1949. The nature of news. In W. Schramm (Ed.), Mass
communications: 288 – 303. Urbana: University of Illinois Press.
Schudson, M. 1978. Discovering the news. New York: Basic Books.
Shavell, S. 1979. Risk sharing and incentives in the principal and agent relationship
model. Bell Journal of Economics 10: 55-73.
Sherif, M. and C.I. Hovland. 1961. Social Judgment: Assimilation and contrast
effects in communication and attitude change. New Haven: Yale University
Press.
Shleifer, A. and R.W. Vishny. 1986. Large shareholders and corporate control.
Journal of Political Economy 94: 461–488.
Shoemaker, P.J. 1996. Hardwired for news: Using biological and cultural evolution
to explain the surveillance function. Journal of communication 46: 32 – 47.
Simon, C.J. and M.W. Sullivan. 1993. The measurement and determinants of brand
equity: A financial approach. Marketing Science 12: 28 – 52.
Sloan, R.G. 1993. Accounting earnings and top executive compensation. Journal of
Accounting and Economics 17: 55 – 100.
Smith, J. 1995. Understanding the media: A sociology of mass communications.
Cresskill: Hampton Press.
Society of Professional Journalists. 1996. Code of ethics.
http://www.spj.org/ethics.asp.
Spector, A.J. 1956. Expectations, fulfillment, and morale. Journal of Abnormal and
Social Psychology 52: 51 – 56.
Spreng, R., S.B. MacKenzie, and R.W. Olshavsky. 1996. A re-examination of the
determinants of consumer satisfaction. Journal of Marketing 60: 15 – 32.
Sridhar, S.S. 1994. Managerial reputation and internal reporting. The Accounting
Review 69:343 – 363.
Stock, H. 2003. CEOs Take Responsibility For Reputation Role. Investor Relations
Business (Nov 10): 1.
Swan, J.E. and F.I. Trawick. 1980. Satisfaction related to predictive vs. desired
expectations: A field study. In Day, R.L. and Hunt H.K. (Eds.) New findings

103
on consumer satisfaction and complaining. 15 – 22. Bloomington: Indiana
University.
Taylor, S.E. and S.T. Fiske. 1975. Point of view and perceptions of causality.
Journal of Personality and Social Psychology 32: 439 – 445.
Taylor, S.E. and S.T. Fiske. 1978. Salience, attention, and attribution: Top of the
head phenomena. Advances in Experimental Social Psychology 11: 249 –
288.
Taylor, S.E., S.T. Fiske, M. Close, C. Anderson, and A. Ruderman. 1977. A solo
status as a psychological variable: The poser of being distinctive, Working
paper, Harvard University.
Tosi, H.L., J.P. Katz, and L.R. Gomez-Mejia. 1997. Disaggregating the agency
contract: The effects of monitoring, incentive alignment, and term in office
on agent decision making. Academy of Management Journal 40: 584 – 602.
Tse, D.K. and P.C. Wilton. 1988. Models of consumer satisfaction formation: An
extension. Journal of Marketing Research 25: 204 – 212.
Tversky, A. and D. Kahneman. Judgment under uncertainty: Heuristics and biases.
Science 185: 1124 – 1131.
Warner, J.B., R.L. Watts, and K.H. Wruck. 1988. Stock prices and top management
changes. Journal of Financial Economics 20: 461 – 492.
Wartick, S.L. 1992. The relationship between intense media exposure and change in
corporate reputation. Business Society 31: 34 – 49.
Weaver, D.H. and G.C. Wilhoit. 1986. The American journalists: A portrait of U.S.
newspeople and their work. Bloomington: Indiana University Press.
Weiner, B. 1986. An attributional theory of motivation and emotion. New York:
Springer-Varlag.
Weisbach, M.S. 1988. Outside directors and CEO turnover. Journal of Financial
Economics 20: 431 – 460.
Whittaker, J.O. Attitude change and communication – attitude discrepancy. Journal
of Social Psychology 65: 141 – 147.
Woodruff, R.B. E.R. Cadotte, and R.L. Jenkins. 1983. Modeling consumer
satisfaction processes using experience-based norms. Journal of Marketing
Research 20: 296 – 304.

104
Wruck, K.H. 1989. Equity ownership concentration and firm value: evidence from
private equity financings. Journal of Financial Economics 23: 3–28.
Wyatt, A. 2005. Accounting recognition of intangible assets: Theory and evidence
on economic determinants. The Accounting Review 80: 967 – 1003.
Zeithaml, V.A., L.A. Berry, and A. Parasuraman. 1993. The nature and determinants
of customer expectations of service. Journal of the Academy of Marketing
Science 21: 1 – 12.
Zwick, R., R. Pieters, and H. Baumgartner. 1995. On the practical significance of
hindsight bias: The case of the expectancy-disconfirmation model of
consumer satisfaction. Organizational Behavior and Human Decision
Processes 64: 103 – 117.

105
Appendix A. Learning Model
DeGroot (1970) shows how estimates approach closer to a true value based
on Bayesian updating as information about the true value is known (Murphy 2003).
Assume that both the principal and the agent know the distribution of the
agent’s ability but only the agent knows his true ability. If true ability (denoted a) is
comes from the normal distribution with a mean of ã and a variance of σ
2
, the we
have a prior belief about the true ability the agent as follows: a ~ N(ã, σ
2
).
In the first year of a new employee, the expected value (best estimate) of a is
as follows: E
0
(a) ≡ â = ã.
Then output x at t is given by x
t
= a + u
t
, where u
t
~ N (0, σ
u
2
).
We can learn about true ability (a) over time, since x
t
~ N (a, σ
u
2
).
After we observe x
1
, x
2
, …, x
n
, we can update the estimate of ability using
both prior information and output information as follows:
Define
i
∑
≡ x
n
1
x n .
â
t
= E [ a | Φ
t
] =
2 2
2 2
t
x t a
~
σ σ
σ σ
+
+
u
t
u
, where Φ
t
≡ { ã, σ
2
, σ
u
2
, x
1
, x
2
, …, x
t
}.
Note that â = ã when t = 0, and â = n x when t → ∞.
The variance of the estimate of ability is as follows:
2
t
a
σ

≡ E [ a | Φ
t
]

≡
2 2
2 2
t σ σ
σ σ
+
u
u

106
Note that → 0 when t → ∞. In other words, the variance of â decreases or the
precision of â increases as more output information is known to the principal. The above
procedure shows the estimated ability â will approach a (true ability) with monotonically
decreasing variance.
2
t
a
σ

107
Appendix B. The Detailed Procedures of CEO Awards
Business Week has bestowed the “Best Managers” in the beginning of every
year since 1988
50
. The magazine surveys a panel of about 150 writers and editors
around the world to determine whose performance should be lauded and renown.
Business Week chose 6 best CEOs until 1991 but gradually increased the number of
honorees starting 1992. It finally selected the top 25 managers of the year in 1996
and has continued to choose 25 best managers to the present
51
. Financial World had
awarded the annual “CEOs of the Year” from 1975 to 1997. The magazine ceased
publication, but it was a highly respected magazine in the business community.
About 40 securities analysts rated the performance of CEOs and the editorial staff of
the magazine compiled the list of the best CEOs in March or April of each year. The
editorial staff first chose three CEOs in each industry and gave the Bronze Award to
the top-rated CEOs and certificates of distinction to the other two CEOs. A panel of
experts again chose a Gold Award winner and 10 Silver Awards winners. I include
all four kinds of awards (Gold, Silver, Bronze, and certificates of distinction) to
measure CEO reputation. There are about 150 to 180 honorees (three CEOs in 50 to
60 industries) every year, which makes Financial World the most dominant source of
CEO awards.

50
Business Week also bestows the “Best Entrepreneurs” at the same time it picks the Best
Managers. However, I exclude the Best Entrepreneurs from my sample, since most
entrepreneurs manage small companies which are not included in Compustat and CRSP.
51
The total number of the Best Managers chosen from 1988 to 2004 is 221. There are more
than 221 CEOs awarded, but many CEOs of foreign firms are excluded in the ExecuComp
database.

108
Forbes has chosen the “Best Performing CEOs” since 2001. The editorial
staff of the magazine rates CEO rankings based on efficiency (performance vs.
compensation). The magazine chose 5 winners in 2001 and 10 winners since 2002.
Industry Week used to choose the “CEOs of the Year” in categories such as
consumer goods, finance, high-tech, heavy industry, industrial sector, or services
sector until 1992. The magazine stopped categorizing awards and switched the
format to select a single CEO or multiple CEOs of the Year since 1993. Chief
Executive magazine has chosen a “Chief Executive of the Year” since 1987. The
magazine accepts nominations for CEO of the year from CEOs, chairmen,
presidents, other executives (COOs and CFOs), and board of directors. The selection
committee makes the final decision for a “Chief Executive CEO of the Year” among
top nominees. Electronic Business magazine has annually chosen a “CEO of the
Year” since 1997. The editorial staff of the magazine accepts nominations from
electronics companies (mostly large, publicly held, and US-based firms). The
editorial staff first analyzes the performance of nominators’ companies and the
Editor-in-Chief lastly considers qualitative factors such as integrity, innovation, and
business ethics. Both Chief Executive and Electronic Business have bestowed one
CEO each year since their first nomination.

Abstract (if available)

Abstract In this dissertation, I examine the potential economic value of CEO reputation: performance improvement at the firm level and personal benefits to the CEO such as compensation and job retention. Two perspectives on CEO reputation offer different predictions regarding the benefits of CEO reputation. The ability perspective in the agency literature advocates the economic benefits of CEO reputation. The symbolic image perspective from recent CEO reputation studies, however, argues that CEO reputation does not necessarily improve firm performance or CEO job retention. I investigate which perspective is more consistent with empirical evidence. The results of firm performance tests show that CEOs with well-established reputations are able to sustain good firm performance but do not turn around poor performance. These results imply that stakeholders might have to consider replacing the CEO -- no matter how highly regarded -- with a turnaround specialist when a firm suffers financially. Another finding from job security tests shows that CEOs with high reputation are more likely to be dismissed than CEOs with low reputation when they perform poorly. These results suggest that the reputations of CEOs through promoting their own images in the media do not necessarily secure their job titles. Finally, the results of compensation tests show that CEO reputation increases pay-for-performance sensitivity.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Empirical analysis of factors driving stock options grants and firms volatility

PDF

CEO selection performance: does board experience matter?

PDF

Political connections and access to bond capital: Reputation or collusion?

PDF

Leveling the playing field: unbiased tests of the relative information content of book income and taxable income

PDF

Does mandatory adoption of international financial reporting standards decrease the voting premium for dual-class shares: theory and evidence

PDF

Incorporating social preferences into incentive system design

PDF

Who manages the firm matters: the incremental effect of individual managers on accounting quality

PDF

Accrual quality and expected returns: the importance of controlling for cash flow shocks

PDF

Corporate reputation crisis in the digital age: a comparative study on Abercrombie & Fitch’s reputation crisis in the U.S., China and Taiwan

PDF

Behavioral approaches to industrial organization

PDF

For the love of the game? ownership and control in the NBA

PDF

Essays on the firm and stakeholders relationships: evidence from mergers & acquisitions and labor negotiations

PDF

Essays on development and health economics: social media and education policy

PDF

The 21st-century principal: the recruitment, mentoring, and retention of principals

PDF

A "second chance" program for vulnerable young adults: a program evaluation and a randomized controlled trial of adjunctive motivational interviewing to improve retention

PDF

Essays on competition between multiproduct firms

PDF

The benefits and costs of accreditation of undergraduate medical education programs leading to the MD degree in the United States and its territories

PDF

How product designs and presentations influence consumers’ post-acquisition decisions

PDF

The voices of teacher attrition: Perceptions of retention and turnover at an international school in Thailand

PDF

Determining industry values to strengthen attorney-client relationships and performance

Asset Metadata

Creator Lee, Sung-Han (author)

Core Title CEO reputation: who benefits -- the firm and the CEO?

School Leventhal School of Accounting

Degree Doctor of Philosophy

Degree Program Business Administration

Degree Conferral Date 2007-08

Publication Date 07/30/2007

Defense Date 06/25/2007

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag CEO reputation,compensation,firm performance,job retention,OAI-PMH Harvest

Language English

Advisor Young, S. Mark (committee chair), Ridder, Geert (committee member), Sandino, Tatiana (committee member), Van der Stede, Wim A. (committee member)

Creator Email sunghanl@marshall.usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m727

Unique identifier UC1220740

Identifier etd-Lee-20070730 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-535037 (legacy record id),usctheses-m727 (legacy record id)

Legacy Identifier etd-Lee-20070730.pdf

Dmrecord 535037

Document Type Dissertation

Rights Lee, Sung-Han

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu