Internal Governance and Real Earnings Management

This research paper examines the impact of internal governance on the extent of real earnings management in U.S. corporations. The study finds that the extent of real earnings management decreases with key subordinate executives' horizon and influence. The results are robust to alternative measures of internal governance and various approaches to address potential endogeneity. The paper contributes to the literature by shedding light on how the members of the management team work together in shaping financial reporting quality.

Singapore Management University
Institutional Knowledge at Singapore Management U
Research Collection School Of Accountancy School of Accountancy
7-2016
Internal Governance and Real Earnings
Management
Qiang CHENG
Singapore Management University, [e-mail]
Jimmy LEE
Singapore Management University, [e-mail]
Terry J. SHEVLIN
University of California-Irvine
Follow this and additional works at: http://ink.library.smu.edu.sg/soa_research
Part of the Accounting Commons, Business Law, Public Responsibility, and Ethics Com
and the Corporate Finance Commons
This Journal Article is brought to you for free and open access by the School of Accountancy at Institutional Knowledge at Singapore
University. It has been accepted for inclusion in Research Collection School Of Accountancy by an authorized administrator of Institu
Knowledge at Singapore Management University. For more information, please email [e-mail].
Citation
CHENG, Qiang; LEE, Jimmy; and SHEVLIN, Terry J.. Internal Governance and Real Earnings Management. (2016). Acco
Review. 91, (4), 1051-1085. Research Collection School Of Accountancy.
Available at: http://ink.library.smu.edu.sg/soa_research/987

Internal governance and real earnings management
Qiang Cheng
[e-mail]
Singapore Management University
Jimmy Lee
[e-mail]
Singapore Management University
Terry Shevlin
[e-mail]
University of California at Irvine
August 2015
ABSTRACT
We examine whether internal governance affects the extent of real earnings management in U.S.
corporations. Internal governance refers to the process through which key subordinate executives
provide checks and balances in the organization and affect corporate decisions. Using the
number of years to retirement to capture key subordinate executives’ horizon incentives and
using their compensation relative to CEO compensation to capture their influence within the
firm, we find that the extent of real earnings management decreases with key subordinate
executives’ horizon and influence. The results are robust to alternative measures of internal
governance and to various approaches used to address potential endogeneity including a
difference-in-differences approach. In cross-sectional analyses, we find that the effect of internal
governance is stronger for firms with more complex operations where key subordinate
executives’ contribution is higher, is enhanced when CEOs are less powerful, is weaker when the
capital markets benefit of meeting or beating earnings benchmarks is higher, and is stronger in
the post-SOX period. This paper contributes to the literature by examining how internal
governance affects the extent of real earnings management and by shedding light on how the
members of the management team work together in shaping financial reporting quality.
Key words: internal governance, real earnings management, top management team
JEL codes: G32, M40
 We thank Xia Chen, Richard Frankel, Weili Ge, Frank Hodge, Bin Ke, Jim Naughton, Dan Segal, Holly Skaife,
Terry Warfield, Anne Wyatt, Huai Zhang, and workshop and conference participants at the 2012 Singapore three-
school research conference, the 2013 Financial Accounting and Reporting Section mid-year conference, the 2013
University of Technology, Sydney Conference, the 2013 AAA annual meeting, Southern Methodist University, and
the University of Wisconsin – Madison for helpful comments. Cheng and Lee thank the School of Accountancy
Research Center (SOAR) at Singapore Management University for financial support. Shevlin acknowledges
financial support from the Paul Merage School of Business at the University of California-Irvine.
Published in Accounting Review, In-Press, 2015 August
http://dx.doi.org/10.2308/accr-51275

1
I. INTRODUCTION
We examine whether internal governance affects the extent of real earnings management. 1
Internal governance refers to the process through which key subordinate executives provide
checks and balances in the organization and affect corporate decisions.2 We focus on key
subordinate executives, or specifically the top four executives with the highest compensation
other than the CEO, because we hypothesize that they are the most likely group of employees
that have both the incentive and the ability to influence the CEO in corporate decisions. As
argued in Acharya, Myers, and Rajan (2011), key subordinate executives have strong incentives
not to take actions that increase short-term performance at the expense of long-term firm value.
This tradeoff between current and future firm value is particularly salient in the case of real
earnings management because overproduction and cutting of R&D expenditures are costly and
can reduce the long-term value of the firm (e.g., Graham, Harvey, and Rajgopal 2005; Bhojraj,
Hribar, Picconi, and McInnis 2009; Cohen and Zarowin 2010). In addition, we expect these key
subordinate executives to have more direct impact on corporate decisions, such as research and
development, production, and other activities that affect operating cash flows, and as a result, the
extent of real earnings management. In contrast, these executives, with the exception of the CFO,
have little direct influence on the accrual process. Thus, we focus on real earnings management
in this paper.
The motivation for the research question is two-fold. First, the majority of the papers in the
literature explicitly or implicitly assume that the CEO is the sole decision maker for financial
1 Following Roychowdhury (2006, 336), we define real earnings management as “management actions that deviate
from normal business practices, undertaken with the primary objective of meeting certain earnings thresholds.”
Some papers in the literature refer to “real earnings management” as “real activities management.”
2 We use the term “internal governance” to be consistent with some of the closely related studies (e.g., Acharya et al.
2011). We refer to governance mechanisms other than the monitoring by the key subordinate executives broadly as
“other governance mechanisms.”

2
reporting quality, which includes both accrual and real earnings management.3 Focusing only on
the CEO does not provide a complete picture because firm management is typically a shared
effort of all top executives (Finkelstein 1992). Recent literature starts to examine how CFOs
affect the quality of financial reporting (e.g., Jiang, Petroni, and Wang 2010; Feng, Ge, Luo, and
Shevlin 2011). However, the impact of other executives has been largely overlooked. As
discussed briefly below and in detail in Section II, recent studies argue that subordinate
executives usually have longer decision horizons and they can influence corporate decisions
through various means. We hypothesize that differential preferences arising from differential
horizons can affect the extent of real earnings management.
Second, while there are studies focusing on the impact of various corporate governance
mechanisms on corporate decisions (e.g., board independence and institutional ownership), little
is known about whether there are checks and balances within the management team. This lack of
knowledge is an important omission because control is not just imposed from the top-down or
from the outside, but also from the bottom-up (Fama 1980).
Key subordinate executives usually care more about the long-term firm value than the CEO
for several important reasons. First, as argued in Acharya et al. (2011), some of these executives
desire to become the CEO in the future. As candidates for the CEO position in the future, key
subordinate executives care about cash flows that the firm can generate in the future, which are
in turn a function of the firm’s current investments. As a result, these executives are less likely to
sacrifice long-term investments to meet short-term earnings targets. Second, key subordinate
executives have more to lose relative to their total wealth from corporate underperformance than
the CEO. They are usually younger and have more remaining years of employment. As such, the
3 Some papers pool all top five executives covered in the ExecuComp database together and examine their collective
influence on financial reporting (e.g., Cheng and Warfield 2005). The distinct impact of other executives is not
identified in such analyses.

3
potential loss of income for failing to find a comparable job in the future is high for younger
executives and increases with horizon. Third, Fama (1980) argues that in general, a manager’s
outside opportunity wage depends on other managers’, including the CEO’s, actions and firm
performance. This effect can motivate the key subordinate executives to be more long-term
oriented and to exert monitoring on the CEO.
Not only do key subordinate executives have incentives to increase long-term firm value,
they also have the means to influence corporate decisions toward their preferences. Prior
research argues that because key subordinate executives’ effort is an important determinant of
current cash flows and the CEO’s welfare, the CEO will consider key subordinate executives’
preferences when making important corporate decisions; otherwise, subordinate executives
might not work hard, hence reducing current and future cash flows and the CEO’s welfare (Allen
and Gale 2000; Acharya et al. 2011).
The above discussion implies that the effectiveness of internal governance depends on the
decision horizon of key subordinate executives and the influence they have on the CEO. In this
paper, we use the number of years until retirement age (assumed to be 65) to capture these
executives’ decision horizon and we use the level of their compensation relative to the CEO’s to
capture their influence. We expect that the longer the horizon and the higher the relative
compensation, the more effective is internal governance, and the lower the extent of real earnings
management. Of course, subordinate executives might have the same incentives as the CEO to
increase short-term performance at the expense of long-term value. Or, subordinate executives
might be afraid of the consequences of disobeying the CEO (e.g., being demoted or fired) and
hence do not exert monitoring on the CEO.4 In addition, it is possible that the key subordinate
4 See Feng et al. (2011) for evidence on the role of powerful CEOs in influencing CFOs to undertake material
accounting manipulations.

4
executives are in a tournament or competition for the CEO’s job with external candidates; as a
result, they could undertake real earnings management to increase short-term earnings and/or to
curry favor with the CEO who likely plays an important role in selecting his/her successor. These
possibilities introduce tension to our research question and thus whether internal governance can
effectively reduce the extent of real earnings management is an empirical question.
We test our hypothesis using 11,994 firm-year observations from the S&P 1500 firms in
the period 1993-2011. The empirical results are consistent with our prediction. We find that the
extent of real earnings management decreases with subordinate executives’ horizon and relative
compensation. The results hold after we control for CEO and firm characteristics that might
affect the extent of real earnings management (e.g., CEO horizon, CEO compensation structure,
firm age, analyst coverage, firm size, firm performance, leverage, firm growth opportunities, and
other governance mechanisms). When we split the sample firms into suspect firms – the
subsample of firms that meet or just beat analysts’ forecasts – and other firms, we find that the
results only hold for the suspect firms, where CEOs have incentives to engage in upward real
earnings management. We do not find results for the other firms. The remaining analyses are
thus based on the sample of suspect firms.
In the main analyses, we use the relative compensation of the key subordinate executives to
capture their ability to influence the CEO on key corporate decisions. An alternative
interpretation of our results is that this proxy captures CEO entrenchment, not internal
governance per se, and entrenched CEOs engage in more real earnings management. In an
additional analysis, we use two alternative measures to investigate the robustness of our results
and to address this alternative explanation. More specifically, we use the abnormal compensation
of subordinate executives and whether the subordinate executives sit on other companies’ boards

5
as alternative proxies for their influence. Our inferences based on these two alternative proxies
remain the same.
As with many corporate governance studies, we recognize that our analyses might be
subject to endogeneity concerns because firms’ internal governance is arguably endogenously
determined. The factors that affect the strength of internal governance might also affect the
extent of real earnings management. We address this endogeneity concern using a number of
approaches. First, we use the lagged values of internal governance in all our analyses and include
a comprehensive list of control variables that are likely correlated with both internal governance
and the extent of real earnings management. Second, we use an instrumental variable approach to
further control for potential endogeneity concerns. Specifically, following related prior studies
(e.g., Coles, Daniel, and Naveen 2006; Boone, Field, Karpoff, and Raheja 2007; Kale, Reis, and
Venkateswaran 2009; Bebchuk, Cremers, and Peyer 2011), we use the two-year lagged value of
internal governance, the industry-year median of internal governance, the number of named
executives in the proxy statement, and an indicator for outside CEOs as instruments. Our
inferences remain the same. Third, we adopt a difference-in-differences design by examining the
impact on the extent of real earnings management of the appointment of a subordinate executive
as an independent director of another company (one of our alternative proxies for internal
governance). We find that before such appointments, firms that later on have subordinate
executives serving as independent directors of other companies do not differ in the extent of real
earnings management from the firms without such subordinate executives. However, after such
appointments, firms with subordinate executives serving as independent directors experience a
significant decrease in the extent of real earnings management compared to other firms. These

6
tests indicate that our results are not driven by the potential endogeneity concern.5
To corroborate the inference from the main analyses, we conduct a series of cross-sectional
analyses. First, key subordinate executives’ ability to influence the CEO’s decision hinges on
their contribution to firm performance and we argue that their contribution is greater when the
firm’s operations are more complex. Accordingly, we expect that the impact of internal
governance is higher when operation complexity is higher. We use industry R&D intensity and a
common factor based on the number of geographical segments, geographical sales concentration,
and foreign sales to capture the complexity of a firm’s operations. The results are consistent with
our prediction that the impact of internal governance is stronger when operation complexity is
higher. Second, we find that the effect of internal governance is stronger when the CEO is more
effectively monitored and less powerful, proxied for by higher board independence, higher
institutional ownership, and an indicator for newly appointed outside CEOs. This result also
indicates that other governance mechanisms can enhance subordinate executives’ ability to
influence the CEO’s decisions. Third, we find that the effect of internal governance is attenuated
for firms in financial distress, for firms that routinely meet or beat earnings targets, and for firms
with upcoming financing activities, presumably because subordinate executives have weaker
incentives to constrain real earnings management when the capital markets benefit of meeting or
beating earnings benchmarks is higher.
We also conduct a series of additional tests to ensure the robustness of our results and to
provide additional insights. First, the Sarbanes-Oxley (SOX) Act exerts a shock to firms’
governance (e.g., requiring higher board independence) and the extent of real earnings
management (Cohen, Dey, and Lys 2008). As such, we expect internal governance to be more
5 Our cross-sectional analyses also mitigate the endogeneity concern because it is arguably harder for an omitted
variable to explain both our main and cross-sectional findings.

7
effective in constraining the extent of real earnings management in the post-SOX period.
Consistent with our prediction, we find that our results are stronger in the post-SOX period than
in the pre-SOX period.
Second, we find that internal governance is more effective in constraining real earnings
management for firms in more homogeneous and competitive industries, where CEOs
presumably have greater career concerns and thus have stronger incentives to manage earnings to
report better financial performance (Parrino 1997; DeFond and Park 1999). Lastly, we find that
internal governance is less effective in constraining real earnings management for firms with
large forthcoming fixed-date option grants, where CEOs presumably have incentives to engage
in downward earnings management in order to reduce the exercise price of option grants (e.g.,
McAnally, Srivastava, and Weaver 2008).
This paper contributes to the literature in two important ways. First, this paper is the first to
examine the association between internal governance and the extent of real earnings
management. This examination is important as it sheds light on how the members of the
management team work together and shape financial reporting. This paper differs from and
complements studies on the impact of CFOs’ characteristics on accrual quality or the likelihood
of earnings restatements/frauds by focusing on all key subordinate executives and by focusing on
real earnings management.
Second, our examination of internal governance helps provide a more complete picture of
how firms work. Unlike prior research which generally views top executives as a unified team,
this paper provides evidence that key subordinate executives can serve an important monitoring
role and that effective internal governance can reduce the extent of CEOs’ myopic behavior. Our
study answers Fama’s (1980, 293) call for research on internal governance. He argues that while

8
each manager is concerned with the performance of others in the firm and as a consequence,
undertakes certain monitoring of other managers, both above or below, “less well appreciated,
however, is the monitoring that takes place from bottom to top.”
The remainder of the paper is organized as follows. Section II provides a summary of prior
research and develops hypotheses. Section III describes the sample and data and presents the
research design. Section IV reports the main analysis of the impact of internal governance on the
extent of real earnings management, the analysis based on alternative proxies for internal
governance, and analyses addressing the potential endogeneity concerns. Section V reports the
cross-sectional analyses and Section VI additional analyses. Section VII concludes.
II. LITERATURE REVIEW AND HYPOTHESIS DEVELOPMENT
Literature Review
We rely and build on two steams of the earnings management literature: the impact of
individual executives on financial reporting quality and real earnings management.
One of the fundamental drivers of earnings management is the pressure on managers to
deliver short-term performance that is used in contracting and firm valuation. See, for examples,
DeFond and Park (1997) on the pressure related to job security, Matsunaga and Park (2001) on
the pressure related to meeting earnings benchmarks, and Bartov, Givoly, and Hayn (2002) and
Kasznik and McNichols (2002) on the capital market pressure to deliver short-term performance.
A recent survey study, Dichev, Graham, Harvey, and Rajgopal (2013), concludes that “about 20
percent of firms manage earnings to misrepresent economic performance, and for such firms 10
percent of EPS is typically managed.” Using a different research methodology, Dyck, Morse,
and Zingales (2013) also conclude that earnings management and accounting frauds are

9
prevalent. Given the vast literature on earnings management, we do not provide a detailed
literature review here and we refer readers to the review papers that discuss in greater detail the
demand for earnings management and how managers benefit from this activity (e.g., Schipper
1989; Healy and Wahlen 1999; Dechow and Skinner 2000; Fields, Lys, and Vincent 2001;
Dechow, Ge, and Schrand 2010).6
Most prior studies tend to focus on the management team as a whole or solely on the CEO
as the person(s) held primarily responsible for earnings management within the firm. Recently,
the literature starts to examine the effect of CFOs on earnings quality. For example, Geiger and
North (2006) find that the appointment of new CFOs is associated with a decrease in
discretionary accruals and that the result is largely driven by new CFOs who are hired from
outside. Focusing on CFO directors, Bedard, Hoitash, and Hoitash (2014) find that firms with
CFOs who sit on their own board exhibit higher reporting quality (e.g., lower likelihood of
internal control weaknesses, lower likelihood of restatements, and higher accruals quality). Ge,
Matsumoto, and Zhang (2011) find that CFOs matter for various accounting choices, such as
discretionary accruals, the likelihood of meeting or just beating earnings expectations, and the
likelihood of restatements.7
There are also studies contrasting the impact of CFOs’ incentives with that of CEOs’ on
earnings management. Jiang et al. (2010) find that the magnitude of accruals and the likelihood
of meeting or just beating analysts’ forecast are more sensitive to CFOs’ than to CEOs’ equity
incentives in the pre-SOX period. In contrast, Feng et al. (2011) find that while CEOs of firms
6 While the literature focusses primarily on accrual-based earnings management, the argument on the demand for
and the benefit (to managers) of earnings management apply to real earnings management as well. Indeed, the recent
development of the real earnings management literature builds on prior studies of accrual-based earnings
management.
7 Ge et al. (2011) capture the effect of CFO style by using a fixed effect model to track CFOs who work in multiple
companies over their sample period.

10
that are involved in material accounting manipulations (manipulations that violate GAAP) have
higher equity incentives than their counterparts in other firms, CFOs of these accounting
manipulation firms have similar levels of equity incentives as their counterparts in other firms.
Despite their lack of incentives, CFOs who are involved in material accounting manipulations
suffer substantial losses. Feng et al. conclude that the direct financial gain is not the main
motivation for CFOs to be involved in earnings manipulation. Rather, CFOs likely succumb to
powerful CEOs’ pressure to manipulate financial statements.
We extend this line of research by focusing on a broader set of key subordinate executives,
including not only CFOs but also other top executives, and examine their impact on the extent of
real earnings management. We focus on real earnings management for two reasons. 8 First, the
tradeoff between increasing short-term performance and increasing long-term firm value is
important for real earnings management. For example, cutting R&D expenditures now to meet
current year’s earnings targets will lead to lower long-term investment and likely reduce the
company’s ability to compete in the product markets and to generate profits in the future.
Consistent with this notion, Bhojraj et al. (2009), Leggett, Parsons, and Reitenga (2009), and
Mizik (2010) report that firms that reduce discretionary spending to beat earnings benchmark
exhibit long-term underperformance. Cohen and Zarowin (2010) and Mizik and Jacobson (2008)
document that firms engaging in real earnings management prior to seasoned equity offerings
have poorer operating performance in the future. Graham et al. (2005) also provide supporting
8 In an untabulated analysis, we examine the impact of internal governance on accrual earnings management. Ex-
ante, whether non-CFO subordinate executives can influence the extent of accrual earnings management is unclear.
On one hand, key subordinate executives do not play a direct role in accrual management because unlike the CFO,
they are not directly involved in the financial reporting process. On the other hand, they likely have an important
influence over the corporate culture and the overall corporate attitude toward earnings management. If the key
subordinate executives focus on the long-term value of the firm, their preference might manifest in less accrual-
based earnings management. After considering the interrelationship between real and accrual earnings management,
we do not find robust evidence that subordinate executives have a significant impact on the extent of accrual
earnings management, consistent with these executives playing a more limited role in the financial reporting
process.

11
evidence based on their surveys of CFOs.9 Second and importantly, key subordinate executives
have more direct control and influence over real activities, such as R&D expenditures,
production volumes, and sales decisions, than over accrual-based earnings management.
To our knowledge, ours is the first study that explicitly examines the impact of subordinate
executives on the extent of real earnings management. The extant literature on real earnings
management has focused primarily on documenting the existence of real earnings management.
For example, Graham et al. (2005) report that 80 percent of surveyed CFOs stated that, in order
to deliver earnings, they would decrease research and development (R&D), advertising, and
maintenance expenditures, while 55 percent said they would postpone a new project, both of
which are real activities manipulation. Roychowdhury (2006) documents the existence of real
earnings management in firms that meet or just beat earnings benchmarks. Cohen et al. (2008)
find that the extent of real earnings management is higher in the post-SOX period than in the pre-
SOX period. We extend this line of research by examining how internal governance affects the
extent of real earnings management, complementing studies that examine the impact on real
earnings management of other governance mechanisms, such as institutional ownership, board
independence, and employment agreement (e.g., Bushee 1998; Carcello, Hollingsworth, Klein,
and Neal 2006; Zhao 2011; Chen, Cheng, Lo, and Wang 2015).
Hypothesis Development
Main Hypothesis
In this section, we discuss why key subordinate executives have both the incentive and
9 In contrast, Gunny (2010) finds that firms engaging in real earnings management to report small positive earnings
exhibit better subsequent performance and she attributes this result to the signaling role of real earnings
management. In light of this contradictory evidence, in an untabulated analysis we examine the association between
our real earnings management proxies and future performance in our sample. Unlike Gunny (2010), we find that our
measures of real earnings management are associated with significantly lower one-year-ahead returns on assets and
cash flow from operations.

12
ability to provide monitoring and reduce the extent of real earnings management.
As discussed above, one of the fundamental drivers of earnings management is the pressure
on CEOs to deliver short-term performance. While it is possible that key subordinate executives
are under similar or even greater pressure to deliver short-term performance, yet as compared to
CEOs, key subordinate executives have longer horizons that induce them to care more about
long-term firm value for three reasons. First, one of the career objectives of many key
subordinate executives is to become the next CEO when the current CEO retires or resigns. As
documented in Cremers and Grinstein (2011), 68.6 percent of CEOs are promoted within the
firm; in other words, in 68.6 percent of the CEO turnover cases, one of the key subordinate
executives becomes the next CEO.10 As the potential CEO in the future, these subordinate
executives care about the cash flows that the firm can generate in the future. Since a company’s
performance depends critically on the capital stock (i.e., value enhancing assets), how the
company performs when the subordinate manager becomes the CEO depends on current
investment. Thus, subordinate executives are hypothesized to care more about long-term
investment and therefore less likely to support activities that sacrifice long-term positive NPV
investments to meet short-term earnings targets.
Second, subordinate executives have more to lose in the event of corporate
underperformance and operational failures. Key subordinate executives are usually younger than
the CEO. In our sample, they are three years younger at the sample median, and this difference
represents a 30 percent increase in the number of years of remaining employment (i.e., the
number of years until the assumed retirement age of 65). Their future compensation likely
10 Based on data from ExecuComp, we find that 59.7 percent of the CEOs in our sample were promoted within the
company. Within this group of CEOs, 36.0 percent were the Chief Operating Officer, 40.8 percent were the
President, and 7.5 percent were the Vice President. These statistics confirm that inside-CEOs are generally selected
from the set of key subordinate executives we study.

13
represents a larger proportion of their overall income and wealth. While the CEO might also
suffer from poor firm performance, the concept of diminishing marginal utility suggests that the
relative impact, i.e., the impact of the potential loss related to the individual’s total wealth, is
important. As such, lower compensation due to poor firm performance in the future or loss of
income due to the difficulty of finding a comparable job is higher for younger executives and
increases with their horizon. This is the same idea underlying the horizon problem discussed in
Dechow and Sloan (1991).
Third, Fama (1980) argues that in general, a manager’s outside opportunity wage depends
on other managers’, including the CEO’s, actions and firm performance. This effect can motivate
the key subordinate executives to be more long-term oriented and to exert monitoring on the
CEO.
The above discussion implies that subordinate executives have longer horizons than the
CEO. The longer the subordinate executives’ horizon, the stronger their incentives not to
sacrifice long-term value for short-term goals.
Not only do subordinate executives have incentives, they also have the means to influence
the CEO’s decision. The current CEO’s welfare depends on the cash flow in the current period,
which is affected by the key subordinate executives’ effort levels.11 If the CEO does not consider
the subordinate executives’ interests, subordinate executives can work less diligently, hence
reducing the current cash flow and the CEO’s welfare (Allen and Gale 2000; Acharya et al.
2011).12 Anticipating this, it is in the best interest of the CEO to consider subordinate executives’
11 The importance of these key subordinate executives is self-evident. In a recent study, Graham, Harvey, and Puri
(2013) find that only about 15 percent of the surveyed CEOs and CFOs indicate that the CEO is the sole-decision
maker in their firms regarding important corporate decisions, such as M&A, capital allocation and investments.
12 This argument is plausible because an individual’s effort level is usually unobservable and subordinate executives
have the best information to decide the appropriate effort level. (This is the same reason why top executives are
given performance-based bonus and stock-based compensation, not just a fixed salary).

14
interests, motivating subordinate executives to work harder (Landier, Sraer, and Thesmar 2009).
Applying the above general discussion to the real earnings management setting, if the CEO
chooses real activity manipulation that decreases long-term firm value, key subordinate
executives will choose a lower effort level. Anticipating this, the CEO will be less likely to
engage in real earnings management. In other words, if the CEO does not engage in real earnings
management, then the subordinate executives’ interest is aligned and they will work harder to
improve both current and future firm performance.
In addition, the CEO needs the subordinate executives’ cooperation to engage in real
earnings management because subordinate executives are usually more informed than the CEO
in their own functional areas. For example, the president in charge of production likely has more
information about the supply of raw materials and the demand from customers. Hence, he or she
will play an important role if the firm decides to overproduce in order to increase the current
period’s earnings. Similarly, the executive in charge of R&D is better informed and can
influence whether and how much the firm can reduce the current period’s R&D. That is, while
the CEO has the formal authority to make the decision, subordinate executives have the real
authority, e.g., effective control over the decisions, due to their information advantage (Aghion
and Tirole 1997). As such, the CEO will have to take the subordinate executives’ preferences
into consideration.
Overall, the effectiveness of key subordinate executives’ influence in curbing myopic
behavior depends on their horizon and their ability to influence CEOs’ decisions. The longer the
horizon and the more influence the key subordinate executives have, the more effective the
internal governance, and the less likely that the company will engage in real earnings
management. Thus, our first hypothesis (in alternative form) is as follows:

15
H1: The extent of upward real earnings management is negatively associated with the
effectiveness of internal governance.
As discussed below, we use key subordinate executives’ horizon (i.e., the number of years until
retirement) and their relative pay (i.e., the average pay of subordinate executives divided by CEO
pay) to capture the effectiveness of internal governance.
There are two critical assumptions underlying H1. First, we rely on prior research to argue
that the CEO has incentives to increase short-term performance at the expense of long-term
value, such as to increase job security (DeFond and Park 1997) and to increase compensation
(e.g., Healy 1985; Cheng and Warfield 2005). One might argue that subordinate executives
might be as myopic as the CEO. In addition, it is possible that the key subordinate executives are
in a tournament or competition for the CEO’s position with external candidates. As a result, they
might undertake real earnings management to increase short-term earnings and/or to curry favor
with the CEO who likely plays a role in selecting his/her successor. If this is the case, we will not
find results consistent with H1. Second, while prior research indicates that key subordinate
executives have the ability to influence CEOs’ decisions, CEOs have the power to demote or fire
these subordinate executives. Job security concerns can motivate subordinates to cooperate with
CEOs in carrying out myopic behavior (Feng et al. 2011). Of course, firing key subordinate
executives because they do not cooperate in myopic behavior can backfire. Having nothing to
lose after being fired, subordinate executives can become “whistle-blowers” and reveal the
inappropriate behavior to the board, investors, and the press, or seek legal action against the firm
for inappropriate dismissal. This potential outcome will deter CEOs from freely firing
subordinate executives who choose not to engage in myopic behavior. Again, if subordinate
executives have no influence on CEOs’ myopic behavior or if CEOs have no incentive to engage
in earnings management, we will not find results consistent with H1. Thus, whether we find

16
results consistent with H1 is an empirical question.
Cross-sectional Analyses
To corroborate our theory and main hypothesis that key subordinate executives have the
ability and incentive to influence the extent of real earnings management, we propose several
cross-sectional predictions that exploit the variation in subordinate executives’ ability and
incentive. These cross-sectional tests also help rule out competing explanations for the main
result.
Variation in subordinate executives’ contribution
One key assumption underlying H1 is that subordinate executives can influence corporate
decisions to reflect their preferences. Because one of the fundamental reasons why key
subordinate executives can influence CEOs’ decisions is their contribution to firm performance,
the greater the subordinate executives’ contribution, the greater is their potential influence on
CEOs (Finkelstein 1992; Acharya et al. 2011). Prior research indicates that complex firms are
more difficult to manage and requires the collective efforts of all executives (e.g., Graham et al.
2013). We thus expect the impact of internal governance to be stronger in more complex firms
than in other firms. Our next hypothesis (in alternative form) is as follows:
H2: The effectiveness of internal governance in reducing the extent of upward real
earnings management is stronger in more complex firms than in other firms.
We discuss the proxy for firm complexity in the empirical section.
Variation in CEO power
Subordinate executives’ ability to influence CEOs’ decision is likely affected by how
powerful the CEOs are. According to Adams, Almeida, and Ferreira (2005), powerful CEOs are
those who can consistently influence key decisions in their firms, despite the potential opposition
from other executives. In firms where CEOs are powerful, decision making authority is usually

17
centralized in the hands of the CEO and thus these CEOs are able to push their agenda even if
the decision may be viewed as sub-optimal. Consistent with this reasoning, Feng et al. (2011)
find that CFOs likely succumb to powerful CEOs’ pressure to manipulate financial statements.
Therefore, we expect subordinate executives to have lower ability to influence CEOs’ decision
when CEOs hold substantial power and authority within the firm. Conversely, we expect internal
governance to be more effective in constraining the extent of real earnings management when
CEOs are less powerful, and thus our third hypothesis (in alternative form) is as follows:
H3: The effectiveness of internal governance in reducing the extent of upward real
earnings management is stronger for firms with less powerful CEOs than for other
firms.
We discuss the proxy for CEOs’ power in the empirical analysis section.
Capital markets benefits of meeting or beating earnings expectations
In developing the main hypothesis, we argue that subordinate executives have incentives to
reduce the extent of real earnings management because such activities can reduce firm value in
the long run. However, if such activities can increase firm value in the short run that also benefit
subordinate executives, they will have weaker incentives to constrain real earnings management.
Prior research documents the capital markets benefit of meeting or beating earnings expectations
(e.g., Bartov et al. 2002; Kasznik and McNichols 2002). If the benefit is high enough to
outweigh the cost of real earnings management, then the effectiveness of internal governance is
expected to be lower. Firms in financial distress, such as those with poor credit rating, benefit
more from meeting or beating earnings benchmarks because missing earnings expectations could
lead to credit rating downgrades, inhibiting the firm’s ability in obtaining future financing and
thus perpetuating financial distress. Consistent with this reasoning, Jiang (2008) finds that the
reduction in the cost of debt from beating earnings benchmark is more pronounced for firms with

18
high default risk. Firms which habitually meet or beat earnings expectations also benefit more
from meeting earnings benchmarks because of the additional market premium from consistently
meeting earnings targets (Kaznik and McNichols 2002). Finally, firms that have a forthcoming
debt or equity issuance benefit more from beating earnings benchmarks, which can increase the
proceeds from debt/equity financing. Therefore, in instances where the capital markets benefit of
reporting higher earnings is high, we expect subordinate executives to have weaker incentives to
constrain real earnings management, and thus our last hypothesis (in alternative form) is as
follows:
H4: The effectiveness of internal governance in reducing the extent of upward real
earnings management is weaker for firms with higher capital markets benefit of
meeting or beating earnings expectations than for other firms.
III. RESEARCH DESIGN
Sample
We obtain our initial sample of firms from Compustat ExecuComp in the period from 1993
to 2011. We limit our examination to firms with compensation details of the top five executives
and require at least five executives (including the CEO) to be reported in the annual proxy
statement.13 To ensure that we have an appropriate measure of CEO’s influence within the firm,
we require the CEO to be in office for the entire year. We exclude firms in financial (2-digit
SICs between 60 and 69) and utility (2-digit SICs of 49) industries because firms in regulated
industries have different financial reporting incentives from other firms. We then merge the
sample of executive-level data with Compustat, CRSP, and I/B/E/S to obtain the data for the
other variables required for the analyses and we drop the observations that have missing values
13 About 9 percent of firm-year observations in ExecuComp report compensation information for fewer than five
executives. Following Bebchuk et al. (2011), we exclude these observations from our sample to ensure that our
measure of key subordinate executives’ influence is comparable across firms.

19
for these variables. Our final sample consists of 11,994 firm-years, and Panel A of Table 1
reports the sample selection process.
Table 1, Panel B reports the job titles of the key subordinate executives in our sample
firms.14 In our empirical tests, key subordinate executives refer to the top four non-CEO
executives included in the ExecuComp database.15 The CFO is usually included in the top four
executives, with an increased frequency in recent years, possibly because of the increasing
influence of CFOs in the post-SOX era. Other key executives reported in the proxy statements
usually hold job titles such as Chief Operating Officer (COO), President, Executive or Senior
Vice President, and Vice President. These titles suggest that the key subordinate executives in
our sample usually hold very important positions and thus have significant responsibilities within
the firm, leading to their ability to monitor the CEO and to influence real earnings management.
Measure of Internal Governance
In this paper, we posit that the effectiveness of internal governance increases with key
subordinate executives’ incentives and ability to monitor the CEO. We measure key subordinate
executives’ monitoring incentives based on their decision horizon, which we proxy for using the
number of years until the age of retirement (assumed to be 65):16,17
14 Ideally, we would like to categorize the job title of the key subordinate executives based on their job function,
such as sales, production, and R&D. However, the job titles in ExecuComp do not indicate the job scope of the key
executives, and many firms categorize their job titles by business segments (e.g. subsidiaries), geographical
segments or product segments rather than by function. As such, we can only provide generic job titles.
15 We limit our scope of subordinate executives to the top four executives other than the CEO because most firms
only disclose the compensation details of the top five executives (including the CEO) in their proxy statements.
16 We use the horizon of key subordinate executives, not their horizon relative to the CEO’s, because it is the
horizon itself that leads subordinate executives to care about long-term firm value. The difference in horizon does
not necessarily capture executives’ incentives to increase long-term firm value. For example, in firm A, subordinate
executives are on average 50 years old and the CEO is 55 years old; in firm B, subordinate executives are on
average 60 years old and the CEO is 65 years old. While the difference in horizon between subordinate executives
and the CEO is the same for the two firms, firm A’s subordinate executives have longer horizon, arguably care more
about the firm’s long-term value, than their counterparts in Company B. In the empirical analyses, we include CEO
horizon to control for its impact on the extent of real earnings management. Nevertheless, we obtain qualitatively
similar results when using the difference in horizon.

20
Exec_Horizon = 65 – the average age of key subordinate executives
Next, we measure key subordinate executives’ ability to monitor the CEO based on their
influence within the organization. We posit that competitive labor markets dictate the
compensation of top executives and hence their compensation reflects their contribution to, and
their influence within, the firm.18 In addition, Finkelstein (1992) argues that an executive’s
compensation reflects her power derived from her structural position in the firm. Therefore, our
measure of key subordinate executives’ ability to monitor the CEO is defined as follows:19
_ Average annualcompensation of key subordinate executives
CEO’s annualcompensation
We scale the average compensation of key subordinate executives by CEO’s annual
compensation because we want to capture key subordinate executives’ influence within the firm.
The level of key executives’ compensation varies across firms and does not capture how much
influence the executives have within the firm. For example, subordinate executives in a large
firm might erroneously be regarded as having more influence than their counterparts in a small
firm if one uses the unscaled level of compensation as the proxy for their influence within the
firm. In an additional analysis, we use the unscaled abnormal compensation of key subordinate
executives as an alternative proxy and the inferences remain the same; see Section IV for details.
Finally, we derive an aggregate measure of a firm’s overall internal governance
effectiveness based on both the incentive and ability of key subordinate executives to monitor the
CEO. Specifically, we standardize both Exec_Horizon and Exec_PayRatio and sum the
17 Assuming a different retirement age, such as 70, does not change the regression results (except the intercept)
because the retirement age is assumed to be a cross-sectional constant and is thus just a scalar.
18 Executives’ compensation is also closely related to their outside opportunity wage, which is then related to their
influence within the firm. An executive with a higher outside opportunity wage is more likely to stand by his or her
position and is less concerned with the CEO’s reaction (e.g., being demoted or fired).
19 Some prior studies use variations of the inverse of this measure, or pay slice, to capture tournament incentives
(Kale et al. 2009) or CEO entrenchment (Bebchuk et al. 2011; Feng et al. 2011). We explore alternative proxies
below to address the concern that our inferences are confounded by these alternative interpretations.

21
standardized measures as our proxy for the firm’s overall internal governance effectiveness
(Int_Governance).20,21
Measure of Real Earnings Management
We derive our measure of real earnings management following prior studies
(Roychowdhury 2006; Cohen and Zarowin 2010). In particular, we use three individual metrics,
abnormal levels of cash flow from operations (RM_CFO), production costs (RM_PROD) and
discretionary expenses (RM_DISX), and two aggregate metrics (RM1 and RM2) to measure the
level of real earnings management. These individual measures are the residuals from the
corresponding estimation model, as described in the Appendix. Executives can artificially inflate
reported earnings by: 1) accelerating sales using aggressive price discounts and/or more lenient
credit terms which results in abnormally low cash flow from operations (CFO); 2) reducing the
costs of sales by increasing production so as to spread the fixed costs of production over a larger
number of units, thereby resulting in abnormally high production cost (PROD); 3) reducing the
amount of discretionary research and development (R&D), advertising, and selling, general and
administrative (SG&A) expenses which result in abnormally low discretionary expenses (DISX).
Therefore, higher abnormal PROD, lower abnormal CFO, and lower abnormal DISX are
consistent with income-increasing real earnings management. For ease of interpretation, all
measures (RM_CFO, RM_PROD, and RM_DISX) are defined to be increasing in reported
20 Specifically, for each of the two variables, we deduct the sample mean and then divide the difference by the
sample standard deviation of the variable. We also explore an alternative aggregate measure based on a non-linear
combination of Exec_Horizon and Exec_PayRatio and obtain quantitatively similar results (untabulated). In
particular, we form tercile ranks on Exec_Horizon and Exec_PayRatio, sum the tercile rank of both variables, and
then rescale the aggregate measure to lie within zero and one.
21 Note that Int_Governance is not a common factor of Exec_Horizon and Exec_PayRatio; we are not assuming that
these two variables are highly correlated and capture the common underlying construct. Instead, we argue that these
two variables capture different dimensions of internal governance; executives with long horizon and high pay
relative to the CEO (a high value of Int_Governance) have both the incentive and the ability to monitor CEOs
compared to their counterparts with short horizon and low pay relative to the CEO (a low value of Int_Governance).

22
earnings.22
Following Cohen and Zarowin (2010), we define two aggregate measures of real earnings
management, RM1 and RM2, to capture the total amount of real earnings management engaged
by the firm in a particular fiscal year:23
RM1 = RM_DISX +RM_ PROD
RM2 = RM_CFO + RM_DISX
Empirical Model
Below we describe the research design for the main test of H1. The design for other tests is
described in the corresponding empirical analysis sections. To test H1, we estimate the following
regression:
ti,
ti,1-ti,1-ti,ti,
+Year_FE+EIndustry_F+
olsFirm_Contr+lsCEO_Contro+anceInt_Govern+=RM

 ψγ , (1)
where RMi,t is the measure of real earnings management and Int_Governancei,t-1 is the measure
of a firm’s internal governance strength, as discussed above. Hypothesis H1 predicts a negative
coefficient on Int_Governance. CEO_Controlsi,t-1 are the CEO characteristics that are included to
control for the CEO’s incentives and power in the prior fiscal year; Firm_Controlsi,t are
contemporaneous firm-level control variables; Industry_FE and Year_FE are industry and year
22 Prior research on accruals-based earnings management suggests that discretionary accrual models might be mis-
specified when applied to firms with extreme financial performance (e.g., Dechow, Sloan, and Sweeney 1995;
Kothari, Leone, and Wasley 2005). A similar concern may apply to the real earnings management measures. In an
untabulated analysis, we use a similar research design as proposed in Kothari et al. (2005) and estimate
performance-matched real earnings management proxies. We use two alternative performance measures, earnings
(ROA) and operating cash flow (CFO), because unlike accrual earnings management, real earnings management
affects both earnings and cash flow. Our inferences remain the same. We do not use this approach in the main
analyses because of smaller sample size due to the requirement of matching firms in the same industry-year with
ROA/CFO within a narrow bandwidth. Cohen, Pandit, Wasley, and Zach. (2011) also note that performance-
matched real earnings management measures can provide conservative tests that under-reject the null hypotheses
relating to income-increasing real earnings management, which is our economic phenomenon of interest.
23 We do not use an aggregate measure based on all three real earnings management proxies because, as suggested in
Roychowdhury (2006) and Cohen and Zarowin (2010), some activities that lead to abnormally high production costs
might also lead to abnormally low CFO. Therefore, combining these two measures can result in double counting. In
addition, we note that the three individual measures capture different types of real earnings management. As a result,
we do not use a common factor based on these three measures in the analyses.

23
fixed-effects, respectively.24 We use the lagged value of all variables relating to internal
governance and CEO’s characteristics to alleviate the potential endogeneity concern. We also
utilize an instrumental variable approach and a difference-in-differences analysis to further
mitigate this concern, as discussed in Section IV. Appendix A includes the detailed definition of
all variables. To mitigate the influence of extreme values, all continuous variables are winsorized
at the 1 percent and 99 percent levels. Because we use a pooled sample, we use firm and year
clustered standard errors to control for cross-sectional and time-series dependence in the data
(Petersen 2009; Gow, Ormazabal, and Taylor 2010).
We include CEO control variables to mitigate the concern that our proxies for key
subordinate executives’ incentives and ability to monitor the CEO merely capture the effect of
CEO’s incentives and power on real earnings management. Specifically, we include the CEO’s
decision horizon (CEO_Horizon), proxied for by the number of years until the age of retirement
(assumed to be 65), the CEO’s annual compensation (CEO_Comp), and CEO’s pay-for-
performance sensitivity (CEO_PPS), measured as the sensitivity of the CEO’s equity portfolio to
the firm’s stock performance (Core and Guay 2002).
Following prior studies, we include several firm-level control variables to capture the
impact of firm characteristics on the extent of real earnings management. The inclusion of these
variables can also help alleviate the omitted correlated variable concern arising from potential
endogeneity of internal governance. Firm age (Firm_Age) is included because younger firms,
which are usually high-growth firms and are expected to obtain additional financing in the
future, likely face greater capital markets pressure to deliver and hence are more likely to engage
in real earnings management to meet earnings targets (Skinner and Sloan 2002; Erickson,
24 Because of the inclusion of industry and year fixed effects, the intercept (α) captures the extent of real earnings
management for firms in the industry and year that do not have corresponding indicators in the regression and when
all independent variables have values of zero. As such, we do not present the estimates of the intercept in the tables.

24
Hanlon, and Maydew 2006; Armstrong, Larcker, Ormazabal, and Taylor 2013). We include the
number of analysts following (N_Analyst) because the monitoring by financial analysts is likely
to constrain real earnings management (Cohen and Zarowin 2010). Lastly, firm performance
(ROA), firm size (Size), the book-to-market ratio (B/M), and leverage (Leverage) are included as
controls for other firm-specific characteristics such as capital structure and growth opportunities
that likely affect real earnings management (Roychowdhury 2006; Cohen and Zarowin 2010).25
Descriptive Statistics
Table 1, Panel C reports descriptive statistics on the regression variables. Because the
model for real earnings management is estimated using the ExecuComp universe and our sample
is similarly obtained from ExecuComp, the means and medians of the individual real earning
management proxies are close to zero. The mean (median) decision horizon of key subordinate
executives (Exec_Horizon) is 12.70 (13.00) years, which is longer than that of the CEO’s mean
(median) decision horizon (CEO_Horizon) of 9.50 (10.00) years by 33.7 percent (30.0 percent).
This comparison supports the notion that key subordinate executives have longer decision
horizons than the CEO. The mean (median) annual compensation of the key subordinate
executive relative to that of the CEO is 0.558 (0.436). By construction, the summary measure of
internal governance, Int_Governance, has a mean of zero. As our sample firms are from
ExecuComp which only includes firms from the S&P1500, our sample firms are significantly
more mature (mean Firm_Age of 22.9 years), have more analysts following (mean N_Analyst of
25 We also control for other variables that might affect the extent of real earnings management, such as the G-index,
an indicator for CEO-Chairman duality, the pay-for-performance sensitivity of key subordinate executives, and the
squared term of internal governance measures. Similarly, we control for several variables that have been used to
proxy for the cost of real earnings management: market share, Z-score, institutional ownership, and marginal tax rate
(Zang 2012). The untabulated analyses indicate that the results on the variables of interest are qualitatively similar.
The G-index and the CEO-Chairman duality exhibit marginally significant coefficients in some specifications,
consistent with the extent of real earnings management being higher when there are more anti-takeover measures
and when the CEO is more likely to be entrenched. The other aforementioned variables are insignificant in most
specifications. We omit these controls in our main analyses in favor of a more parsimonious empirical model and a
larger and more generalizable sample.

25
11.1 analysts), have better performance (mean ROA of 5.5%), and are larger (mean Size, ln(Total
assets), of 7.3), as compared to the firms covered in the Compustat universe in the same time
period.26 The average book-to-market ratio is 0.505 and the average leverage is 0.512.
Table 2 reports the Pearson correlation table of the variables in our main analysis. The
three measures of real earnings management (RM_CFO, RM_PROD, and RM_DISX) are highly
positively correlated with each other except for the correlation between RM_CFO and
RM_DISX. These high correlations suggest that firms manage one real activity in tandem with
other real activities. By construction, RM1 and RM2 are highly correlated with individual
components and with each other. The correlation between Exec_Horizon and Exec_PayRatio is
positive, but the relatively low correlation coefficient (0.08) suggests that key subordinate
executives’ decision horizon and influence capture different aspects of firms’ internal
governance. Consistent with H1, almost all real earnings management measures are negatively
associated with the proxies of internal governance. None of the correlations between control
variables are high enough to impose a multicollinearity problem.27
IV. MAIN ANALYSES – TESTS OF H1
Full sample analysis
In this section, we report our main tests of H1. We first analyze the separate impact of
26 The average firm in the Compustat universe in the same period is 13.1 years old, is followed by 4.4 analysts, has
ROA of -0.8% and Size of 5.8.
27 While internal governance is negatively associated with firm performance and size, we do not believe that
performance and size drive the documented results. First, we control for both firm performance and size in the
multiple regression analyses. Second, as mentioned above, the inferences remain the same when we use
performance-matched real earnings management measures. Third, we separately examine the association between
internal governance and both firm performance and size in the suspect firms and non-suspect firms subsample
(samples defined below). We find that internal governance is negatively associated with firm performance and size
for both subsamples. Given that we do not find consistent results in the hypothesized direction in the non-suspect
sample, as discussed in Section IV, the negative association between internal governance and both firm performance
and size is unlikely to drive the results in suspect firms.

26
executive horizon and pay ratio on the extent of real earnings management, and then the impact
of the combined internal governance measure. Table 3 presents the results. For ease of
exposition, all measures of real earnings management are multiplied by 100.
Table 3, Panel A presents the separate impact of subordinate executives’ decision horizon
and pay ratio on real earnings management. We find that as predicted in H1, both executives’
decision horizon and influence are significantly negatively associated with the extent of real
earnings management, whether proxied for by the three individual measures (with the exception
of the association between Exec_Horizon and RM_CFO) or by the two summary measures.28
The results on control variables are generally consistent with prior studies. We find some
evidence that firms with CEOs that have longer horizon are less likely to engage in real earnings
management. CEOs with higher compensation (which also signifies their ability in the
competitive labor market) are less likely to engage in real earnings management, suggesting that
better-ability CEOs are associated with better earnings quality (Demerjian, Lev, and McVay
2013). We also find that firms with more analysts following and better performance are less
likely to engage in real earnings management and that larger firms and firms with higher book-
to-market and leverage are more likely to engage in real earnings management. Finally, there is
also evidence that younger firms are more likely to engage in real earnings management.
Table 3, Panel B reports the analysis of the impact of the overall internal governance on
real earnings management. Consistent with the results reported above, the overall internal
governance (Int_Governance) is significantly associated with a lower extent of real earnings
28 In an untabulated analysis, we explore the potential non-linearity in the impact of executives’ decision horizon on
real earnings management by constructing three piece-wise linear terms in Exec_horizon, following the approach
used in Himmelberg, Hubbard, and Palia (1999). We find that when subordinate executives’ horizon is short – less
than 5 years, there is no impact on the extent of real earnings management. The impact occurs when executive
horizon is between 5 and 15 years and executive horizon beyond 15 years has no incremental effect. This result
indicates that executives’ incentive to monitor the CEO is low when their horizon is too short (less than 5 years) and
does not increase further after 15 years.

27
management. The effect of internal governance on real earnings management is also
economically significant. A one standard deviation increase in Int_Governance is associated with
a decrease in RM1 and RM2 of 3.0 percent and 1.7 percent of total assets, respectively. 29
We conduct a series of additional analyses to ensure the robustness of the results and we do
not tabulate the results to save space. First, we examine whether our results are driven by CFOs’
characteristics. For this purpose, we exclude CFOs from our measurement of internal
governance. The untabulated results are quantitatively similar, suggesting that other key
subordinate executives do influence real earnings management. Second, in the main analyses, we
use the average of executive horizon and pay ratio to construct internal governance measures.
We find similar results (1) when we use the median of key executives’ decision horizon and pay
ratio in order to mitigate the concern that our results are driven by extreme values in the internal
governance variables, and (2) when we use the maximum value of key executives’ decision
horizon and pay ratio (internal governance can arguably be exerted by the executive who has the
greatest incentive and ability to monitor the CEO).
Overall, the results reported above are consistent with H1 which predicts that the extent of
real earnings management is negatively associated with the effectiveness of internal governance.
Suspect firms versus non-suspect firms
One drawback of using the full sample to test H1 is that CEOs’ incentives to engage in
earnings management are not salient. To increase our ability to detect real earnings management,
we focus on firm-years where there is a greater likelihood of earnings management – when firms
meet or just beat important earnings benchmarks (e.g., Burgstahler and Dichev 1997; Degeorge,
29 The impact on RM1 expressed as a percentage of total assets is computed as -2.021 (the coefficient on
Int_Governance) × 1.468 (the sample standard deviation of Int_Governance). Note that all measures of real earnings
management are already multiplied by 100 and hence presented as a percentage of total assets. The impact on RM2
is computed analogously.

28
Patel, and Zeckhauser 1999). For this purpose, we follow prior research (e.g., Roychowdhury
2006) and limit our sample to firm-years with earnings surprise between zero and one percent of
share price, where earnings surprise is calculated as actual earnings minus the most recent
consensus analyst forecast before the earnings announcement. We then test H1 using this sample
of “suspect-firms.”
Panel A of Table 4 presents the results. We find that internal governance is negatively
correlated with both summary measures of real earnings management (t-statistic = -3.83 and -
4.37 for RM1 and RM2, respectively). The inferences are the same when we examine individual
real earnings management measures or individual internal governance measures. We do not
tabulate the results to preserve space.
While Panel A presents the negative effect of internal governance on real earnings
management, one might wonder whether suspect firms with ineffective internal governance
indeed manage earnings upward. For this purpose, we split the suspect sample into five groups
based on the quintile of internal governance. We find that for the subsample with internal
governance in the bottom quintile, the average RM1 (RM2) is 0.017 (0.007), significantly
different from zero at the 0.01 (0.08) level (untabulated). This test indicates that suspect firms
with less effective internal governance indeed engage in upward real earnings management.
The arguments underlying H1 imply that we will not find a negative association between
internal governance and the extent of real earnings management in a sample where CEOs have
low incentives to engage in upward earnings management. Therefore, as a falsification test, we
re-run our main analyses on a sample where we do not expect earnings management and hence
internal governance is less likely to matter. Specifically, we construct a sample of firm-years
with earnings surprises less than -0.5 percent of stock price (big miss) and larger than 1 percent

29
of stock price (big beat). We exclude the sample of firm-years with earnings surprises between -
0.5 percent and 0 percent of stock price for two reasons. First, given the potential stock price
penalty associated with missing analyst forecast, it is possible that managers engaged in upward
earnings management but still failed to meet the benchmark. Second, managers near the
important earnings benchmark may still manage earnings upwards to meet other internal and
unobservable targets (Roychowdhury 2006; Zang 2012). Panel B of Table 4 reports the results
using this sample of “non-suspect” firms. We do not find a significant coefficient on the internal
governance variable (t-statistic = -1.26 and -0.81 for RM1 and RM2, respectively). This result
reinforces our inference that internal governance plays a more important role in constraining real
earnings management when the incentives to meet or beat earnings target is high.
Given that we find the predicted results only for the suspect firms, we focus on this sub-
sample in the remaining analyses.
Alternative Measures of Key Subordinate Executives’ Influence
In our earlier analyses, we use the subordinate executives’ compensation relative to the
CEO’s as a proxy for their influence within the firm. However, this measure might also capture
other constructs such as agency problems (Bebchuk et al. 2011; Feng et al. 2011): a lower ratio
of subordinate executive pay to CEO pay implies entrenched CEOs. If so, our results could be
interpreted as less entrenched CEOs (with high executive to CEO pay ratio) engaging in less real
earnings management. We do not think this alternative explanation is valid because our results
are robust to controlling for corporate governance variables, as discussed above. To further
refute this alternative interpretation, we utilize an alternative proxy for key subordinate
executives’ influence – their abnormal compensation. To do so, we follow the compensation
model used by Core, Guay, and Larcker (2008) and regress the logged total compensation of the

30
subordinate executives on economic determinants, including prior year’s logged firm sales, S&P
500 membership dummy, prior year’s book-to-market ratio, current and prior year’s stock
returns, current and prior year’s return on assets, and industry and year fixed effects. We use the
residual from this regression as a proxy for subordinate executives’ abnormal compensation
(Exec_AbComp). This variable is not based on a comparison with CEO compensation and is thus
not subject to the alternative interpretation, and as usually interpreted in the compensation
literature, the abnormal compensation captures the executive’s market value. Executives with
higher abnormal compensation are more influential and thus better able to constrain the extent of
earnings management.
Table 5, Panel A reports the regression results based on this variable in instead of
Exec_Payratio in Equation (1). As observed from columns (1) to (4), our results are robust to
using this alternative proxy for subordinate executives’ influence, either as a stand-alone
measure or in the combined internal governance measure with executive horizon.30
In addition, throughout the paper, we use the compensation-based measure to capture key
executives’ influence within the firm. In an additional analysis, we explore an alternative
measure of key subordinate executives’ influence: the number of directorships in other firms
held by these executives (Other_Director). Finkelstein (1992) argue that sitting on other firms’
boards reflects an executive’s power. Masulis and Mobbs (2011) also argue that these executives
are more influential and are more likely to be the CEO in the future. Based on these findings, we
expect that the key subordinate executives who have directorships in other firms to exert greater
influence over the current CEO and that the more directorships they have, the stronger their
influence. To test this prediction, we add this alternative measure of key executives’ power to
30 In another untabulated sensitivity test, we use firms’ total asset as the deflator when calculating executive pay
ratio which avoids using CEO pay as the denominator of the pay ratio. Our inferences also remain the same.

31
Equation (1) and report the results in Table 5, Panel B. In our sample, 8.0 percent of firm-year
observations have at least one key executive holding directorship(s) in other firms.31
In columns (1) and (2), we use Other_Director in place of Exec_PayRatio as an alternative
proxy for key subordinate executives’ influence. As predicted, we find that Other_Director is
negatively associated with RM1 and RM2, significant at the 0.01 level in both models (t-statistic
= -2.44 and -2.55, respectively). This result suggests that key subordinate executives with outside
directorships exert greater influence in constraining real earnings management. In columns (3)
and (4), we explore whether Other_Director captures a different dimension of executives’
influence than Exec_PayRatio by including both variables in the same regression. We find that
both variables have significantly negative coefficients, suggesting that Other_Director represents
a different aspect of executives’ influence within the firm.
In sum, our results reported above are not due to the alternative explanation based on CEO
entrenchment or agency problems, and our results hold when using alternative measures of
subordinate executives' influence.
Addressing Endogeneity Concerns
We recognize that our analyses might be subject to endogeneity concerns because firms’
internal governance is arguably endogenously determined and the determinants of the internal
governance might also affect the extent of real earnings management. For example, some of the
firms that are conscientious about real earnings management might select young and powerful
subordinate executives to balance the influence of the CEO and these firms might also have
lower extent of earnings management for other reasons (such as having a strong board of
directors). For another example, unobservable subordinate executive talent might affect both
31 Within the group of firms with key subordinate executives serving as directors in other firms, 66 percent (23
percent, 11 percent) of the firms have one (two, three or more) key subordinate executive serving as directors in
other firms.

32
measures; talented subordinate executives are likely paid more, leading to higher value of the
measured executive pay ratio, and are given more discretion in undertaking discretionary
investments. We do not believe that these alternative arguments can explain our results.
First, theoretically, the potential omitted correlated variable, such as talent, likely affect
subordinate executives and CEOs similarly. For example, under the alternative argument based
on executive talent, if the board is willing to award more talented subordinate executives higher
compensation and more discretion, they should be willing to do the same for the CEO. That is,
the board will give more talented CEOs higher compensation and more discretion. This implies
that in the absence of CEO entrenchment, firms with highly paid CEOs will be less likely to cut
discretionary expenditures, leading to lower executive pay ratio and lower extent of real earnings
management, i.e., a positive association between the two. This prediction is opposite to what H1
predicts and what we find above. Second, as highlighted earlier, we mitigate this concern by
using the lagged values of internal governance and include a comprehensive list of control
variables that are likely correlated with both internal governance and the extent of real earnings
management in the main analyses or robustness checks, including corporate governance
variables. Third, some of our cross-sectional analyses also mitigate this concern because it is
arguably harder for an omitted correlated variable to explain both our main and cross-sectional
findings.32 Lastly, as discussed above, we do not find consistent results for non-suspect firms. If
omitted correlated variables drive the results, we should expect to find similar results in non-
suspect firms as well.
Nevertheless, in this section we use two approaches, an instrumental variable approach and
32 For example, it is difficult to argue why executive talent plays a less important role for firms that are expected to
benefit more from meeting or just beating earnings targets (see the development of H4 in Section II). There is no
compelling reason to believe that the capital market benefit of meeting or just beating earnings targets to the firm
should vary with the talent of the subordinate executives.

33
a difference-in-differences analysis, to further address the endogeneity concerns.
An Instrumental Variable Approach
In this section, we employ a two-stage least square instrumental variable approach to
further address endogeneity concerns, as commonly used in the literature. In the first stage
regression, we regress internal governance on the instrument variables and in the second stage,
we use the predicted internal governance to explain the extent of real earnings management. For
this purpose, we utilize four instruments: 1) the one-year lagged value of internal governance
(Lagged_Int_Governance);33 2) the industry-year median value of internal governance (Ind-
Year-Median_Int_Governance); 3) an indicator variable, Outside_CEO, that equals one if the
current CEO is recruited from outside, and zero otherwise; and 4) the number of named
executives in the annual proxy statement besides the CEO (Named_Exec). For the first two
instruments, we follow prior studies (e.g., Coles et al. 2006; Boone et al. 2007; Kale et al. 2009)
and use the lagged endogenous variable and the industry-year median endogenous variable as
instruments, based on the reasoning that firm-specific governance practice that persists over time
is more likely to be exogenous to the current year’s decision (e.g., the extent of real earnings
management engaged to meet the current year’s short-term earnings targets) and that industry-
specific governance characteristic are more likely exogenous because they are not under the
firm’s control in any particular year. The choice of the latter two instruments is based on related
prior studies that utilize similar instruments (e.g., Kale et al. 2009; Bebchuk et al. 2011). Based
on these studies, we argue that when the CEO is recruited from outside, the CEO is less likely to
possess as much firm-specific knowledge as an inside-CEO and the influence of other executives
is likely higher, improving the effectiveness of internal governance. In a similar vein, having a
33 Recall that our measures of internal governance are lagged one year in all our empirical specifications. Therefore,
this instrument is lagged two years relative to our outcome variable.

34
higher number of named executives in the annual proxy statement implies a greater number of
highly-paid executives and a stronger presence of divisional managers who can arguably increase
the effectiveness of internal governance. However, we are not aware of any prior research
suggesting that having an outside CEO or the number of divisional managers is associated with
the extent of real earnings management. As discussed below, we conduct the tests suggested by
Larcker and Rusticus (2010) and find that these four instruments are relevant and valid.
We report the first stage regression results in Column (1) of Table 6, where we regress
Int_Governance on all four instruments as well as the controls used in the second stage
regression. As predicted, we find that the instrument variables are significantly positively
associated with Int_Governance with the exception of Named_Exec (t-statistic = 27.46, 9.70,
2.75, and 0.90, respectively). The weak identification test suggests that these four instruments are
relevant and powerful: the F statistic for the joint explanatory power of the instrument variables
is 303.29, significantly higher than the critical value of 13.96, as suggested in Stock, Wright, and
Yogo (2002).
Columns (2) and (3) of Table 6 report the second stage regression results. We find that
predicted internal governance estimated from the first-stage regression is significant and
negatively associated with RM1 and RM2 (t-statistic = -3.35 and -4.32, respectively). The result
from the over-identification test of all instruments is insignificant (J-statistic = 5.197 and 5.499
for the two columns, respectively), suggesting that the instruments are valid in the second stage
regression.34
34 In an additional robustness test, we follow Larcker and Rusticus’s (2010) recommendation and conduct sensitivity
analyses on the choice of instruments, and all the 2SLS test statistics are robust to using various subsets of the four
instruments such as: 1) lagged and industry-year median internal governance and Outside_CEO; 2) lagged and
industry-year median internal governance; 3) Outside_CEO and Named_Exec.

35
A Difference-in-differences Analysis
As an alternative approach to address endogeneity concerns, we conduct a difference-in-
differences (DID) analysis. If the omitted correlated variables that affect both internal
governance and the extent of real earnings management are time-invariant, they are controlled
for in the DID analysis. Because the year-on-year change in executive horizon is 1 by
construction and executive pay ratio is relatively sticky over time, we examine the change in one
of the alternative measures of internal governance – key subordinate executives serving on other
firms’ boards. In particular, we utilize a DID research design, as in Bertrand and Mullainathan
(2003) and Chan, Chen, Chen, and Yu (2012), and examine the impact on real earnings
management of the new appointment of key executives as independent directors on other firm’s
boards. We construct two variables: 1) an indicator variable (CID_FIRM ) that equals one if the
firm has at least one key executive who holds independent directorships in other firms during the
sample period, and zero otherwise; 2) an indicator variable (POST_CID_FIRM ) that equals one
for firm-years after the first key executive is appointed as an independent director in other firms,
and zero otherwise.35 The coefficient on CID_FIRM captures the difference in real earnings
management between firms with key subordinate executives being externally appointed as
independent directors (i.e., treatment firm) and the other firms in the pre-appointment period.
The coefficient on POST_CID_FIRM captures the incremental effect of CID_FIRM on real
earnings management after the appointment of key executives as external independent director. 36
We do not include a separate variable for the post-appointment period as it is subsumed by the
35 There are only very few firms with two or more subordinate executives concurrently serving as independent
directors on other firms’ boards.
36 We exclude firm-years where the treatment firm becomes a non-treatment firm to have a cleaner set of treatment
firms. In unreported analyses, our results are similar if we include these excluded firm-years in the analyses. In
another robustness test, we restrict the treatment sample to firms with at least two pre-appointment years and at least
two post-appointment years. Our inferences remain the same.

36
year fixed effects. (POST_CID_FIRM is essentially the interaction between the CID_FIRM and
an indicator for the post-appointment period.)
Table 7 presents the result from this analysis. We observe a significant decrease in the
extent of real earnings management after the appointment of the first key executive as
independent directors in other firms; the coefficient on POST_CID_FIRM is significantly
different from zero (t-statistic = -2.42 and -2.75, respectively).37 This result indicates that key
subordinate executives have a causal effect on the extent of real earnings management after their
appointment as independent directors in other firms, presumably as a result of the increase in
their standing and influence within their own firms.
Overall, the robust results based on the instrumental variable approach and the DID
analysis strengthen our earlier inference that key executives exercise significant influence over
real earnings management, mitigating concerns that our results are driven by omitted correlated
variables.
V. CROSS-SECTIONAL ANALYSES
Research Design
To test H2-H4, we estimate the following regression:
ti,ti,
1-ti,ti,1-ti,
ti,1-ti,ti,
+Year_FE+EIndustry_F+olsFirm_Contr+
lsCEO_Contro+ng_VARConditionianceInt_Govern+
ng_VARConditionianceInt_Govern+=RM



ψ
γ

, (2)
where Conditioning_VARi,t is a conditioning variable that moderates the association between a
firm’s internal governance effectiveness and real earnings management. All other variables are
defined as above. To preserve space, we focus on the two aggregate measures of real earnings
37 We also find that the coefficient on CID_FIRM is insignificantly different from zero; that is, there is no significant
difference in the extent of real earnings management in the pre-appointment period between firms that have
executives being appointed as external independent directors and those without.

37
management (RM1 and RM2) and the aggregate measure of the firm’s internal governance
(Int_Governance). The estimation of regression (2) is similar to that of regression (1). To test
H2, H3, and H4, Conditioning_VARi,t refers to proxies for key subordinate executives’
contribution to the firm’s performance, proxies for CEO power, and proxies for the benefit of
meeting or beating earnings benchmarks, respectively. We explain the proxies below in the
corresponding sections.
The Conditioning Effect of Firm Complexity – Test of H2
To test H2, we examine whether the effectiveness of internal governance in constraining
real earnings management is stronger in firms where key subordinate executives’ contribution to
the firm’s performance is expected to be higher. We expect key subordinate executives’
contribution to the firm’s performance to be more important when the firm operates in an R&D
intensive industry where technological complexity is high and when the complexity surrounding
operating in diverse geographical locations is high (e.g., Finkelstein 1992; Graham et al. 2013).
We proxy for operation complexity using the following two measures: (1) an indicator for high
R&D intensity (IND_RD), which equals one (zero) if the average R&D intensity in the industry-
year is above (below) the sample median; and (2) an indicator for high geographical complexity
(GEO_Complexy), which equals one (zero) for firm-year observations with above (below) the
median first principle component of the following three variables: the number of geographical
segments, geographical sales concentration, and the percentage of foreign sales. 38 To test H2, we
replace Conditioning_VARi,t in Equation (2) with each of the two measures and we expect a
negative coefficient on the interaction term.
Table 8 reports the regression results. We find the association between internal governance
38 We do not combine IND_RD and GEO_Complexy into one common factor because unreported factor analysis
results in two principle components with an eigenvalue greater than one, suggesting that these two measures appear
to capture different constructs.

38
and the extent of real earnings management is significantly more negative for firms in industries
with higher R&D intensity (Panel A, t-statistic = -2.58 and -2.90, respectively) and for firms
with more diverse geographical operations (Panel B, t-statistic = -1.63 and -2.00, respectively).
Overall, the results in Table 8 are consistent with hypothesis H2 that the impact of internal
governance is stronger in more complex firms where key subordinate executives are expected to
play a more important role in the firm’s operations.
The Conditioning Effect of CEO Power – Test of H3
H3 predicts that the effectiveness of internal governance is higher when CEOs are less
powerful. We measure CEO power using three proxies. The first two measures are based on two
commonly studied governance mechanisms: the monitoring by the board of directors and by
institutional shareholders. We expect CEOs to be less powerful when other strong governance
mechanisms are in place. Prior research documents that the effectiveness of board monitoring
increases with board independence (e.g., Weisbach 1988; Klein 2002) and that institutional
investors are better monitors than other shareholders (e.g., Bushee 1998; Parrino, Sias, and
Starks 2003; Chen, Harford, and Li 2007). Thus, we predict that the effectiveness of internal
governance increases with board independence and institutional ownership, and we construct
indicator variables that equal one (zero) if board independence (BD_IND) and institutional
ownership (Inst_Own) are above (below) the corresponding sample median.39 The third proxy is
based on CEO’s tenure and whether he is recruited from outside. We expect a CEO who is
recently recruited from outside to be less experienced on his new position and thus less powerful.
Hence, we create an indicator variable (New_OutsideCEO) that equals one if the CEO is
39 While the monitoring by the board of directors and institutional investors is probably the most commonly
examined dimensions of corporate governance, there are other dimensions of corporate governance. Examining all
possible dimensions of corporate governance is beyond the scope of this paper.

39
recruited from outside and the CEO’s tenure is less than three years, and zero otherwise. 40, 41 To
test H3, we replace Conditioning_VARi,t with one of these three proxies and we expect a negative
coefficient on the interaction term in Equation (2).
Table 9 presents the regression results. We find that the effectiveness of internal
governance in constraining the extent of real earnings management is higher in firms with higher
board independence (Panel A, t-statistic = -1.55 and -2.38, respectively), in firms with higher
institutional ownership (Panel B, t-statistic = -1.58 and -1.52, respectively), and in firms with
newly appointed outside CEOs (Panel C, t-statistic = -3.21 and -2.50, respectively). These results
are consistent with hypothesis H3 that internal governance is more effective when CEOs are less
powerful and that effective board oversight and higher institutional ownership can enhance key
subordinate executives’ ability to monitor the CEO.42
The Conditioning Effect of the Capital Markets Benefit of Meeting or Beating Earnings
Targets – Test of H4
Finally, we examine whether subordinate executives’ incentives to constrain real earnings
management vary with the capital markets benefit of meeting or beating earnings targets. We
expect subordinate executives to have weaker incentives to constrain real earnings management
when the capital markets benefit of reporting higher earnings is high because they will enjoy the
benefit as well. We proxy for the benefits of reporting higher earnings using three measures: (1)
an indicator variable (Distress) that equals one if the Z-score of the firm is lower than 1.81 and
40 We do not focus on CEO tenure solely because a newly-appointed CEO could have worked within the firm for
many years and thus would be very experienced and knowledgeable about his new position. Therefore these CEOs
are arguably powerful.
41 Results are qualitatively similar when we use different tenure cutoffs around 3, such as 2.5 years or 3.5 years.
42 This finding is also consistent with Acharya et al. (2011, 691) who analytically show that “a combination of
internal governance and a rudimentary form of outside governance by shareholders can improve the efficiency of the
firm dramatically.” By “rudimentary form of outside governance by shareholders,” they refer to shareholders’ ability
to take over the firm and replace the CEO if necessary. That is, the effectiveness of internal governance can be
enhanced by the monitoring by shareholders who care about long-term value and have the ability to discipline the
CEO if needed.

40
the bond rating of the firm is below the investment grade, and zero otherwise; (2) an indicator
variable (Hab_Beater) that equals one if the firm is a habitual beater (i.e., meeting or beating at
least three out of the last four quarters, and at least six out of the last eight quarters), and zero
otherwise; and (3) an indicator variable (Capital_Issue) that equals one if the firm has significant
financing activities (i.e., issuing debt or equity greater than or equal to three percent of market
value) in the following fiscal year, and zero otherwise.43 We expect the benefits of reporting
higher earnings to be higher for firms with poor credit rating, for firms that are habitual
benchmark beaters, and for firms with forthcoming financing activities, and hence we expect a
positive coefficient on the interaction term in Equation (2).
Table 10 presents the regression results. Consistent with H4, we find that internal
governance is less effective in constraining real earnings management for firms in financial
distress (Panel A, t-statistic = 1.90 and 2.22, respectively), for firms that are habitual beaters
(Panel B, t-statistic = 1.95 and 1.58, respectively), and for firms with future financing activities
(Panel C, t-statistic = 1.07 and 1.42, respectively). Overall, the results are consistent with
hypothesis H4 that subordinate executives have weaker incentives to constrain real earnings
management when the capital markets benefit to reporting higher earnings is greater.
VI. ADDITIONAL ANALYSES AND SENSITIVITY CHECKS
The Effectiveness of Internal Governance: Pre- versus Post-SOX Period
The Sarbanes-Oxley Act (hereafter, “SOX”), passed on July 30, 2002, aims at
strengthening corporate governance and mitigating managerial incentives to manipulate earnings
43 We use a relatively high cutoff of three percent of market value to classify debt or equity issuance so that we can
focus on instances where the benefits of reporting higher earnings are greater as well as to prevent misclassification
of debt and equity issuance (e.g., issuing equity for employee stock options plans, debt conversion). Our results are
robust to using other cutoffs between two percent to five percent of market value.

41
via accruals. Prior research (e.g., Graham et al. 2005; Cohen et al. 2008) finds that the passage of
SOX and the increased regulatory scrutiny on accrual-based earnings management led many
firms to switch from accrual to real earnings management. When CEOs switch to value-
decreasing real activities manipulations, we expect key subordinate executives to exert more
influence over real earnings management in the post-SOX period than in the pre-SOX period. In
addition, the passage of SOX increases the overall emphasis on corporate governance. Hence,
key subordinate executives are likely to obtain greater support from other governance
mechanisms, such as the board of directors, in the monitoring of the CEO, also leading to more
effective internal governance. Note that as shown in Section V, internal governance and other
governance mechanisms work as complements, rather than substitutes. As such, SOX represents
an exogenous shock that affects the effectiveness of internal governance in reducing the extent of
real earnings management: both the effectiveness of internal governance and the extent of real
earnings management are increased, and we should observe stronger results in the post-SOX
period. In contrast, if the results documented above are driven by endogenous (or optimal)
decision of the firm, we should not observe any change in the effectiveness of internal
governance.
To test this prediction, we create an indicator variable (Post_SOX) that equals one if the
fiscal year is after 2003, and zero otherwise, and replace Conditioning_VAR in Equation (2) with
Post_SOX.44 Because of the inclusion of the Post_SOX variable, we cannot include year fixed-
effects; instead we include a time trend variable (Time), which is fiscal year minus 1993, the first
fiscal year of the sample. The results are presented in Table 11. Consistent with our predictions,
the coefficient on the interaction term is significantly negative (t-statistic = -2.29 and -2.76,
44 We do not include observations in 2002 and 2003 in the post-SOX period because these two years are regarded as
a transition period when many sections of SOX were not yet fully effective. The results are quantitatively similar if
we include 2002 and 2003 in the post-SOX period.

42
respectively), implying that the effectiveness of internal governance in constraining real earnings
management is strengthened in the post-SOX period.
Self-Serving CEOs
An implicit assumption in our hypothesis is that the CEO has private incentives to increase
short-term performance at the expense of long-term value. This assumption is based on the
findings in prior research. However, not all CEOs are the same and our results should be stronger
in cases where the CEOs are particularly self-serving because subordinate executives have
stronger incentives to constrain real earnings management when the CEOs are perceived to be
more self-serving. In this section, we explore a setting where CEOs are more likely to be self-
serving. For this purpose, we rely on prior literature to identify instances where CEOs have
greater career concerns and thus stronger incentives to manage earnings to report better financial
performance. Parrino (1997) shows that it is easier to identify and replace poorly performing
CEOs in homogeneous industries. Similarly, DeFond and Park (1999) show that market
competition is likely to enhance the importance of accounting earnings in relative performance
evaluation, and indeed they find that accounting-based measures are more associated with CEO
turnover in industries with high competition. Following these arguments, we predict that CEOs
in homogenous and competitive industries are more self-serving and have greater incentives to
manage earnings for job security consideration.45 As such, internal governance is more effective
for firms in such industries. To test our prediction, we create an indicator variable
(Self_Serving_CEO) that equals one (zero) if the firm is in the more homogenous and
competitive industries, which are classified based on the sample median of the first principle
component of the industry homogeneity measure used in Parrino (1997) and the inverse of
45 On the other hand, Giroud and Mueller (2010) find that managerial slack is lower in competitive industries, which
suggests that market competition improves firm governance and thus mitigates managerial self-serving behavior.

43
industry sales concentration ratio.
Table 12 presents the regression results. As reported in this table, we find that consistent
with our prediction, the negative effect of internal governance on real earning management is
stronger in more homogenous and competitive industries (t-statistic = -4.09 and -3.94,
respectively).
The impact of internal governance when CEOs have incentives to engage in downward
earnings management
To triangulate our results, we identify a situation where CEOs have incentives to engage in
downward earnings management and then test whether internal governance plays a less
important role in constraining earnings management. The argument underlying H1 is that strong
internal governance reduces upward real earnings management because such manipulation
reduces long-term firm value. Presumably, if downward real earnings management does not have
an adverse impact on long-term firm value, subordinate executives will not restrain the extent of
real earnings management. It thus follows that internal governance plays a less important role in
situation where CEOs have incentives to report lower earnings. Following prior research (e.g.,
McAnally et al. 2008), we use forthcoming fixed-date option grants to capture CEOs’ incentives
to engage in downward earnings management. McAnally et al. (2008) find that CEOs have
incentives to miss earnings targets prior to fixed-date option grants, because CEOs profit from a
reduced option strike price if the firm’s stock price decreases after missing earnings targets. 46
Following McAnally et al. (2008), we create an indicator variable (Future_Option_Grant) that
equals one if the one-year ahead fixed-date option grants scaled by salary after the earnings
announcement is greater than the sample median and the firm misses analyst forecast by a small
46 We focus on fixed-date option grants because the grant dates of these options are known and thus managers
cannot time or backdate the options (McAnally et al. 2008).

44
margin (less than 0.5 percent of stock price) or a really large margin (more than 10 percent of
stock price), and zero otherwise.47 We predict the negative effect of internal governance to be
weaker, or the coefficient on the interaction term of internal governance and
Future_Option_Grant to be positive.
Table 13 presents the regression results. As predicted, we find that the negative effect of
internal governance on real earning management is significantly attenuated when CEOs have
large forthcoming fixed-date option grants (t-statistic = 2.05 and 2.12, respectively). Moreover,
the F-test indicates that the net effect of internal governance (β1 + β2) is insignificant; that is,
internal governance is not associated with the extent of real earnings management for the firms
with large forthcoming fixed-date option grants. These results corroborate our evidence that
internal governance only plays an important role in constraining upward earnings management.
VII. CONCLUSION
In this paper, we examine whether key subordinate executives have the incentive and
ability to constrain the extent of real earnings management. Compared to the CEO, key
subordinate executives are usually younger, have longer horizon, and care more about future
performance. Also, key subordinate executives have the ability to influence CEOs’ decisions
because of their significant involvement in the firm’s operations as well as their contribution to
the firm’s current performance, which are important to the CEO. Using the number of years to
retirement to capture key subordinate executives’ incentives and their compensation relative to
the CEO’s to capture their influence within the firm, we find that the extent of real earnings
management decreases with key subordinate executives’ horizon and influence. Our results are
47 Because we are interested in the firms where managers have the incentives to manage earnings downward and
benefit from missing earnings targets (i.e., reduction in option strike price prior to option grants), we examine the
full sample in this set of analysis.

45
robust to alternative measures of key subordinate executives’ ability to influence corporate
decisions: the abnormal compensation of subordinate executives and the number of directorships
in other firms held by these executives. Our inferences also remain the same after we control for
potential endogeneity concerns using an instrumental variable approach and a difference-in-
differences approach.
We then examine whether the impact of internal governance varies with proxies for key
subordinate executives’ contribution, proxies for CEO power, and proxies for capital markets
benefit of meeting earnings benchmarks. We find that the effect of internal governance is
stronger in more complex firms where key subordinate executives play a more important role,
stronger in firms where the CEO is less powerful, and weaker in firms where the capital markets
benefit of meeting earnings benchmark is higher. We conduct a series of additional tests to
ensure the robustness of our results and to provide additional insights. First, we find that our
results are stronger in the post-SOX period when real earnings management is likely more
prevalent than in the pre-SOX period. Second, we find that internal governance is more effective
in constraining real earnings management for firms where CEOs presumably have greater career
concerns and thus have more incentives to manage earnings to report a better financial
performance. Lastly, we find that the effect of internal governance is weaker for firms with large
forthcoming fixed-date option grants, where CEOs presumably have incentives to manage
earnings downward to reduce the exercise price of the option grants.
We contribute to the literature by examining the impact of internal governance on the
extent of real earnings management. This examination is important because it sheds light on how
the members of the management team work together to shape financial reporting. Unlike prior
research that generally views executives as a unified team, this paper provides evidence that

46
subordinate executives can provide an important monitoring role on the CEOs from the bottom
up and that effective internal governance can reduce the extent of real earnings management.
This paper differs from and complements studies on the impact of CFO characteristics on accrual
quality or the likelihood of earnings restatements/frauds by focusing on all subordinate
executives and by focusing on real earnings management.

47
REFERENCES
Acharya, V., S. C. Myers, and R. G. Rajan. 2011. The internal governance of firms. The Journal of
Finance 66, 689-720.
Adams, R. B., H. Almeida, and D. Ferreira. 2005. Powerful CEOs and their impact on corporate
performance. Review of Financial Studies 18 (4), 1403-1432.
Aghion, P., and J. Tirole. 1997. Formal and real authority in organizations. Journal of Political Economy
105 (1), 1-29.
Allen, F., and D. Gale. 2000. Comparing financial systems. Cambridge, MA: MIT Press.
Armstrong, C., D. F. Larcker, G. Ormazabal, and D. J. Taylor. 2013. The relation between equity
incentives and misreporting: The role of risk-taking incentives. Journal of Financial Economics
109, 327-350.
Bartov, E., D. Givoly, and C. Hayn. 2002. The rewards to meeting or beating earnings expectations.
Journal of Accounting and Economics 33 (2), 173-204.
Bebchuk, L. A., M. Cremers, and U. Peyer. 2011. The CEO pay slice. Journal of Financial Economics
102, 199-221.
Bedard, J. C., R. Hoitash, and U. Hoitash. 2014. Chief financial officers as inside directors.
Contemporary Accounting Research 31(3), 787-817.
Bertrand, M., and S. Mullainathan. 2003. Enjoying the quiet life? Corporate governance and managerial
preferences. Journal of Political Economy 111, 1043-1075.
Bhojraj, S., P. Hribar, M. Picconi, and J. McInnis. 2009. Making sense of cents: An examination of firms
that marginally miss or beat analyst forecasts. The Journal of Finance 64 (5), 2361-2388.
Boone, A. L., L. C. Field, J. M. Karpoff, and C.G. Raheja. 2007. The determinants of board size and
composition: An empirical analysis. Journal of Financial Economics 85, 66-101.
Burgstahler, D. C., and I. D. Dichev. 1997. Earnings management to avoid earnings decreases and losses.
Journal of Accounting and Economics 24 (1), 99-126.
Bushee, B. J. 1998. The influence of institutional investors on myopic R&D investment behavior. The
Accounting Review 73, 305-333.
Carcello, J. V., C. W. Hollingsworth, A. Klein, and T. L. Neal. 2006. Audit committee financial expertise,
competing corporate governance mechanisms, and earnings management. Working paper,
University of Tennessee and New York University.
Chan, L., K. Chen, T-Y. Chen, and Y. Yu. 2012. The effects of firm-initiated clawback provisions on
earnings quality and auditor behavior. Journal of Accounting and Economics 54, 180-196.
Chen, X., Q. Cheng, A. Lo, and X. Wang. 2015. CEO contractual protection and managerial short-
termism. The Accounting Review, forthcoming in the September 2015 issue.
Chen, X., J. Harford, and K. Li. 2007. Monitoring: Which institutions matter? Journal of Financial
Economics 86, 279-305.
Cheng, Q., and T. D. Warfield. 2005. Equity incentives and earnings management. The Accounting
Review 80, 441-476.
Cohen, D. A., A. Dey, and T. Z. Lys. 2008. Real and accrual-based earnings management in the pre- and
post-Sarbanes-Oxley periods. The Accounting Review 83, 757-787.

48
Cohen, D., S. Pandit, C. Wasley, and T. Zach. 2011. Measuring real earnings management. Working
paper, University of Texas at Dallas, University of Illinois at Chicago, University of Rochester and
The Ohio State University.
Cohen, D. A., and P. Zarowin. 2010. Accrual-based and real earnings management activities around
seasoned equity offerings. Journal of Accounting and Economics 50, 2-19.
Coles, J. L., N. D. Daniel, and L. Naveen. 2006. Managerial incentives and risk-taking. Journal of
Financial Economics 79, 431-468.
Core, J. E., and W. Guay. 2002. Estimating the value of employee stock option portfolios and their
sensitivity to price and volatility. Journal of Accounting Research 40, 613-630.
Core, J. E., W. Guay, and D. F. Larcker. 2008. The power of the pen and executive compensation.
Journal of Financial Economics 88, 1-25.
Cremers, M., and Y. Grinstein. 2011. Does the market for CEO talent explain controversial CEO pay
practices? Working paper, Yale University and Cornell University.
Dechow, P., W. Ge, and C. Schrand. 2010. Understanding earnings quality: A review of the proxies, their
determinants and their consequences. Journal of Accounting and Economics 50, 344-401.
Dechow, P. M., and D. Skinner. 2000. Earnings management: reconciling the views of accounting
academics, practitioners, and regulators. Accounting Horizons 14, 235-250.
Dechow, P. M., and R. G. Sloan. 1991. Executive incentives and the horizon problem. Journal of
Accounting and Economics 14(1), 51-89.
Dechow, P., R. Sloan, and A. Sweeney. 1995. Detecting earnings management. The Accounting Review
70, 193-225.
DeFond, M. L., and C. W. Park. 1997. Smoothing income in anticipation of future earnings. Journal of
Accounting and Economics 23, 115-139.
DeFond, M. L., and C. W. Park. 1999. The effect of competition on CEO turnover. Journal of Accounting
and Economics 27, 35-56.
Degeorge, F., J. Patel, and R. Zeckhauser. 1999. Earnings management to exceed thresholds. Journal of
Business 72 (1), 1-35.
Demerjian, P. R., B. Lev, M. F. Lewis, and S. E. McVay. 2013. Managerial ability and earnings quality.
The Accounting Review 88(2), 463-498.
Dichev, I., J. Graham, C. R. Harvey, and S. Rajgopal. 2013. Earnings quality: Evidence from the field.
Journal of Accounting and Economics 56, 1-33.
Dyck, A., A. Morse, and L. Zingales. 2013. How pervasive is corporate fraud? Working paper,
University of Toronto and University of Chicago.
Erickson, M., M. Hanlon, and E. L. Maydew. 2006. Is there a link between executive equity incentives
and accounting fraud? Journal of Accounting Research 44, 113-143.
Fama, E. F. 1980. Agency problems and the theory of the firm. Journal of Political Economy 88, 288-
307.
Feng, M., W. Ge, S. Luo, and T. Shevlin. 2011. Why do CFOs become involved in material accounting
manipulations? Journal of Accounting and Economics 51, 21-36.
Fields, T. D., T. Z. Lys, and L. Vincent. 2001. Empirical research on accounting choice. Journal of
Accounting and Economics 31, 255-307.

49
Finkelstein, S. 1992. Power in top management teams: Dimensions, measurement, and validation.
Academy of Management Journal 35 (3), 505-538.
Ge, W., D. Matsumoto, and J. L. Zhang. 2011. Do CFOs have style? An empirical investigation of the
effect of individual CFOs on accounting practices. Contemporary Accounting Research 28, 1141-
1179.
Geiger, M. A., and D. S. North. 2006. Does hiring a new CFO change things? An investigation of
changes in discretionary accruals. The Accounting Review 81, 781-809.
Giroud, X., and H. M. Mueller. 2010. Does corporate governance matter in competitive industries?
Journal of Financial Economics 95, 312-331.
Gow, I. D., G. Ormazabal, and D. J. Taylor. 2010. Correcting for cross-sectional and time-series
dependence in accounting research. The Accounting Review 85, 483-512.
Graham, J. R., C. R. Harvey, and M. Puri. 2013. Capital allocation and delegation of decision-making
authority within firms. Working paper, Duke University.
Graham, J. R., C. R. Harvey, and S. Rajgopal. 2005. The economic implications of corporate financial
reporting. Journal of Accounting and Economics 40, 3-73.
Gunny, K. A. 2010. The relation between earnings management using real activities manipulation and
future performance: Evidence from meeting earnings benchmarks. Contemporary Accounting
Research 27 (3), 855-888.
Healy, P. M. 1985. The effect of bonus schemes on accounting decisions. Journal of Accounting and
Economics 7, 85-107.
Healy, P. M., and J. M. Wahlen. 1999. A review of the earnings management literature and its
implications for standard setting. Accounting Horizons 13 (4), 365-383.
Himmelberg, C. P., R. G. Hubbard, and D. Palia. 1999. Understanding the determinants of managerial
ownership and the link between ownership and performance. Journal of Financial Economics 53
(3), 353-384.
Jiang, J. 2008. Beating earnings benchmarks and the cost of debt. The Accounting Review 83 (2), 377-416
Jiang, J., K. Petroni, and I. Wang. 2010. CFOs and CEOs: who has the most influence on earnings
management. Journal of Financial Economics 96, 513-526.
Kale, J., E. Reis, and A. Venkateswaran. 2009. Rank order tournaments and incentive alignment: The
effect on firm performance. The Journal of Finance 64, 1479-1512.
Kasznik, R., and M. F. McNichols. 2002. Does meeting earnings expectations matter? Evidence from
analyst forecast revisions and share prices. Journal of Accounting Research 40 (3), 727-759.
Klein, A. 2002. Audit committee, board of directors characteristics, and earnings management. Journal of
Accounting and Economics 33, 375-400.
Kothari, S. P., A. J. Leone, and C. E. Wasley. 2005. Performance matched discretionary accrual
measures. Journal of Accounting and Economics 39, 163-197.
Landier, A., D. Sraer, and D. Thesmar. 2009. Optimal dissent in organizations. Review of Economic
Studies 76, 761-794.
Larcker, D. F., and T. O. Rusticus. 2010. On the use of instrumental variables in accounting research.
Journal of Accounting and Economics 49, 186-205.
Leggett, D., L. Parsons, and A. Reitenga. 2009. Real earnings management and subsequent operating
performance. Working paper, University of Alabama.

50
Masulis, R. W., and S. Mobbs. 2011. Are all inside directors the same? Evidence from the external
directorship market. The Journal of Finance 66, 823-872.
Matsunaga, S. R., and C. W. Park. 2001. The effect of missing a quarterly earnings benchmark on the
CEO’s annual bonus. The Accounting Review 76 (3), 313-332.
McAnally, M., A. Srivastava, and C. Weaver. 2008. Executive stock options, missed earnings targets, and
earnings management. The Accounting Review 83 (1), 185-216.
Mizik, N. 2010. The theory and practice of myopic management. Journal of Marketing Research 47, 594-
611.
Mizik, N., and R. Jacobson. 2008. Earnings inflation through accruals and real activity manipulation:
Its prevalence at the time of an SEO and the financial market consequences. Working paper,
Columbia University and University of Washington.
Parrino, R. 1997. CEO turnover and outside succession: A cross-sectional analysis. Journal of Financial
Economics 46, 165-197.
Parrino, R., R. W. Sias, and L. T. Starks. 2003. Voting with their feet: institutional ownership changes
surrounding forced CEO turnover. Journal of Financial Economics 68, 3-46.
Petersen, M. A. 2009. Estimating standard errors in finance panel data sets: Comparing approaches.
Review of Financial Studies 22, 435-480.
Roychowdhury, S. 2006. Earnings management through real activities manipulation. Journal of
Accounting and Economics 42, 335-370.
Schipper, K. 1989. Commentary: Earnings Management. Accounting Horizons 3 (4), 91-102.
Skinner, D. J., and R. G. Sloan. 2002. Earnings surprises, growth expectations, and stock returns or don’t
let an earnings torpedo sink your portfolio. Review of Accounting Studies 7, 289-312.
Stock, J. H., J. H. Wright, and M. Yogo. 2002. A Survey of Weak Instruments and Weak Identification in
Generalized Method of Moments. Journal of Business and Economic Statistics 20, 518-529.
Weisbach, M. 1988. Outside directors and CEO turnover. Journal of Financial Economics 20, 431-460.
Zang, A. Y. 2012. Evidence on the trade-off between real activities manipulation and accrual-based
earnings management. The Accounting Review 87 (2), 675-703.
Zhao, J. D. 2011. The association between corporate governance and the earnings surprises games.
Working paper, The University of Melbourne.

51
APPENDIX
Variables Definition
BD_IND An indicator variable that equals one (zero) if the firm-year observation is
above (below) the median percentage of independent director.
B/M The book to market ratio in the current fiscal year, defined as book value
of equity (CEQ) divided by the market value of equity (CSHO*PRCC_F).
Capital_Issue An indicator variable that equals one if the firm issues debt or equity
greater than or equals three percent of market value in the following fiscal
year, and zero otherwise.
CEO_Comp The CEO’s logged total compensation in the prior fiscal year.
CEO_Horizon CEO’s decision horizon, defined as retirement age of 65 minus the age of
the CEO.
CEO_PPS The normalized pay-for-performance sensitivity of the CEO’s portfolio of
equity in the prior fiscal year, measured similarly to Core and Guay
(2002).
CID_FIRM An indicator variable that equals one if the firm has at least one key
executive that serves as an independent director on other firms’ boards
during the sample period, and zero otherwise.
POST_CID_FIRM An indicator variable that equals one for firm-years after the key executive
is appointed as an independent director in other firms, and zero otherwise.
Distress An indicator variable that equals one if the Z-score of the firm is less than
1.81 and the bond rating of the firm is below investment grade, and zero
otherwise.
Exec_Abcomp Subordinate executives’ abnormal compensation, calculated as the logged
(1 + abnormal compensation + sample minimum abnormal compensation),
where abnormal compensation is the residual from a regression of
executives’ mean total compensation on known determinants of CEO Pay
(logged sales, S&P500 membership, book-to-market, returns and lagged
returns, ROA and lagged ROA, and industry and year fixed effects).
Exec_Horizon Subordinate executives’ decision horizon, defined as retirement age of 65
minus the average age of other executives.
Exec_PayRatio Subordinate executives’ pay ratio, calculated as the average total
compensation of subordinate executives scaled by the CEO’s total
compensation, measured in the prior fiscal year.
Firm_Age The age of the firm, defined as the number of years since the firm’s stock
returns is first reported in the monthly stock files of CRSP.
Future_Option_ An indicator that equals one if the one-year ahead fixed-date option grant
Grant scaled by salary after the earnings announcement is greater than the
sample median and the firm misses analyst forecast by a small margin
(less than 0.5 percent of stock price) or a large margin (more than 10
percent of stock price), and zero otherwise.
GEO_Complexy An indicator variable that equals one (zero) if the firm-year observation is
above (below) the median first principle component of the following three
variables: 1) number of geographical segments; 2) geographical sales
concentration and; 3) percentage of foreign sales.

52
Hab_Beater An indicator variable that equals one if the firm meets or beats earnings
targets at least three out of the last four quarters, and at least six out of the
last eight quarters, and zero otherwise.
IND_RD An indicator variable that equals one (zero) if the average R&D intensity
in the industry-year is above (below) the sample median.
Ind-Year-Median_ The industry-year median value of internal governance.
Int_Governance
Inst_Own An indicator variable that equals one (zero) if the firm-year observation is
above (below) the median institutional ownership.
Int_Governance Firm’s overall internal governance, measured as the sum of the
standardized value of Exec_Horizon and Exec_PayRatio.
Lagged_Int_ The one-year lagged value of internal governance.
Governance
Leverage The leverage ratio in the current fiscal year, defined as total liabilities (AT
– CEQ) divided by total assets (AT).
N_Analyst The number of analysts following the firm in the current fiscal year,
obtained from I/B/E/S.
Named_Exec The number of named executives in the annual proxy statement besides
the CEO in the prior fiscal year.
New_OutsideCEO An indicator equals one if the CEO is recruited from outside and the
CEO’s tenure is less than three years, zero otherwise.
Other_Director The number of independent directorships in other firms held by key
subordinate executives.
Outside_CEO An indicator variable that equals one if the current CEO is recruited from
outside, and zero otherwise.
Post_SOX An indicator variable that equals one if fiscal year is 2002 and onward,
and zero otherwise.
RM_CFO Negative of the residual from the cash flow from operations (CFO) model:
1 Δ
The model is estimated by industry (at the Fama-French 48 industry level)
and year and requires at least ten observations for each industry-year
combination, using firms from the ExecuComp universe.
RM_DISX Negative of the residual from the discretionary expenses (DISX) model:
1 Δ
The model is estimated by industry (at the Fama-French 48 industry level)
and year and requires at least ten observations for each industry-year
combination, using firms from the ExecuComp universe.
RM_PROD The residual from production Costs (PROD) model:
1 Δ Δ
PROD is defined as the sum of the cost of goods sold (COGS) and the
change in inventory (ΔINVT). The model is estimated by industry (at the
Fama-French 48 industry level) and year and requires at least ten

53
observations for each industry-year combination, using firms from the
ExecuComp universe.
RM1 An aggregate measure of real earnings management, defined as the sum of
RM_PROD and RM_DISX.
RM2 An aggregate measure of real earnings management, defined as the sum of
RM_CFO and RM_DISX.
ROA Return on assets in the current fiscal year, defined as earnings before
extraordinary items (IB), scaled by beginning total assets (AT).
Self_Serving_CEO An indicator variable that equals one (zero) if the firm-year observation is
above (below) the median first principle component of the following two
variables: 1) industry homogeneity based on Parrino (1997) and; 2)
industry competition based on the inverse of industry sales concentration
ratio.
Size Firm size, calculated as the logged value of total assets (AT) in the current
fiscal year.
Time A time trend variable which equals the difference between the current
fiscal year and 1993.

54
TABLE 1
Sample Selection and Descriptive Statistics
Panel A: Sample Selection
Obs.
Total number of firm-year observations from 1993-2011 with Compustat,
Execucomp and I/B/E/S data 23,647
Less: financials and utilities firms (5,133)
Less: missing values for variables used in the regressions (6,520)
Final sample 11,994
Number of unique firms 2,005
Panel B: Titles of Key Subordinate Executives
Title Obs. %
Chief Financial Officer (CFO) 9,556 19.92
Chief Operating Officer (COO) 5,245 10.93
President 6,888 14.36
Executive Vice President 7,361 15.34
Senior Vice President 7,347 15.31
Vice President 6,543 13.64
Others 5,036 10.50
Total 47,976 100.00

55
TABLE 1 (Cont’d)
Panel C: Sample and Descriptive Statistics
Variables Obs. Mean Median Std. Dev. Q1 Q3
RM_CFO 11,994 -0.002 -0.002 0.084 -0.047 0.041
RM_PROD 11,994 -0.003 0.002 0.173 -0.092 0.087
RM_DISX 11,994 -0.002 0.009 0.184 -0.083 0.094
RM1 11,994 -0.006 0.010 0.336 -0.162 0.168
RM2 11,994 -0.004 0.005 0.202 -0.101 0.102
Exec_Horizon 11,994 12.697 13.000 6.462 9.000 17.000
Exec_PayRatio 11,994 0.558 0.436 0.514 0.324 0.596
Int_Governance 11,994 0.000 -0.132 1.468 -0.782 0.578
CEO_Horizon 11,994 9.496 10.000 7.800 5.000 15.000
CEO_Comp 11,994 7.867 7.848 1.075 7.095 8.610
CEO_PPS 11,994 0.285 0.211 0.236 0.106 0.393
Firm_Age 11,994 22.941 17.000 18.796 9.000 31.000
N_Analyst 11,994 11.070 9.000 7.833 5.000 15.000
ROA 11,994 0.055 0.061 0.105 0.022 0.105
Size 11,994 7.345 7.180 1.518 6.244 8.324
B/M 11,994 0.505 0.424 0.382 0.261 0.649
Leverage 11,994 0.512 0.514 0.218 0.357 0.645
Notes to Table 1:
RM_CFO is a real earnings management proxy that negatively affects cash flow from operations. RM_PROD is a
real earnings management proxy that negatively affects production. RM_DISX is a real earnings management proxy
that negatively affects discretionary expenses. RM1 and RM2 are aggregate measures of real earnings management.
Exec_Horizon is the subordinate executives’ decision horizon. Exec_PayRatio is the subordinate executives’ ability
to influence the CEO. Int_Governance is the firm’s overall internal governance, measured as the sum of the
standardized value of Exec_Horizon and Exec_PayRatio. CEO_Horizon is the CEO’s decision horizon.
CEO_Comp is the CEO’s logged total compensation. CEO_PPS is the pay-for-performance sensitivity of the CEO’s
portfolio of equity. Firm_Age is the age of the firm. N_Analyst is the number of analysts following the firm. ROA is
the return on assets in the current fiscal year. Size is the logged value of total assets in the current fiscal year. B/M is
the book-to-market ratio in the current fiscal year. Leverage is the leverage ratio in the current fiscal year.

TABLE 2
Pearson Correlation Table
1 2 3 4 5 6 7 8 9 10 11 12 13 14
1 RM_CFO 1.00
2 RM_PROD 0.43 1.00
3 RM_DISX 0.00 0.75 1.00
4 RM1 0.22 0.93 0.94 1.00
5 RM2 0.42 0.87 0.90 0.95 1.00
6 Exec_Horizon 0.00 -0.05 -0.08 -0.07 -0.07 1.00
7 Exec_PayRatio 0.01 -0.01 -0.03 -0.02 -0.02 0.08 1.00
8 Int_Governance 0.00 -0.04 -0.07 -0.06 -0.06 0.73 0.73 1.00
9 CEO_Horizon 0.02 -0.03 -0.06 -0.05 -0.05 0.16 0.05 0.15 1.00
10 CEO_Comp -0.07 -0.03 0.00 -0.02 -0.03 -0.05 -0.45 -0.34 -0.03 1.00
11 CEO_PPS -0.13 -0.06 -0.02 -0.04 -0.07 0.12 0.28 0.28 -0.06 0.05 1.00
12 Firm_Age 0.01 0.00 0.01 0.01 0.02 -0.26 -0.13 -0.27 -0.18 0.29 -0.13 1.00
13 N_Analyst -0.19 -0.12 -0.04 -0.09 -0.12 -0.02 0.02 0.00 -0.01 0.43 0.25 0.15 1.00
14 ROA -0.49 -0.29 0.06 -0.12 -0.16 -0.05 -0.05 -0.07 -0.06 0.06 0.15 0.04 0.20 1.00
15 Size -0.04 0.05 0.09 0.08 0.06 -0.17 -0.07 -0.16 -0.11 0.66 0.14 0.46 0.62 0.06
16 B/M 0.23 0.21 0.09 0.16 0.18 -0.02 -0.04 -0.04 -0.02 -0.10 -0.20 -0.01 -0.27 -0.31
17 Leverage 0.20 0.13 0.04 0.09 0.12 -0.09 -0.07 -0.10 -0.03 0.21 -0.17 0.23 0.02 -0.26
Notes to Table 2:
RM_CFO is a real earnings management proxy that negatively affects cash flow from operations. RM_PROD is a real earnings manage
negatively affects production. RM_DISX is a real earnings management proxy that negatively affects discretionary expenses. RM1 and
measures of real earnings management. Exec_Horizon is the subordinate executives’ decision horizon. Exec_PayRatio is the subordina
influence the CEO. Int_Governance is the firm’s overall internal governance, measured as the sum of the standardized value of Exec_H
Exec_PayRatio. CEO_Horizon is the CEO’s decision horizon. CEO_Comp is the CEO’s logged total compensation. CEO_PPS is the p
sensitivity of the CEO’s portfolio of equity. Firm_Age is the age of the firm. N_Analyst is the number of analysts following the firm. R
in the current fiscal year. Size is the logged value of total assets in the current fiscal year. B/M is the book-to-market ratio in the curren
the leverage ratio in the current fiscal year. All correlations except those in shaded cells are statistically significant at the 0.05 level or b

TABLE 3
Internal Governance and Real Earnings Management
Panel A: Key Executives' Decision Horizon, Power and Real Earnings Management
Pred. (1) (2) (3) (4)
H1 RM_CFO RM_PROD RM_DISX RM1
Coef. t-stats Coef. t-stats Coef. t-stats Coef. t-stats
Exec_Horizon ― -0.010 -0.66 -0.108 -2.45 *** -0.142 -2.90 *** -0.248 -2.75 ***
Exec_PayRatio ― -0.308 -1.46 * -2.283 -3.92 *** -2.762 -3.81 *** -5.108 -4.06 ***
CEO_Horizon 0.006 0.45 -0.045 -1.19 -0.097 -2.04 ** -0.144 -1.74 *
CEO_Comp -0.351 -2.34 ** -2.120 -4.07 *** -2.536 -4.54 *** -4.726 -4.51 ***
CEO_PPS -0.008 -0.02 0.901 0.61 -0.212 -0.12 0.808 0.25
Firm_Age 0.006 1.13 -0.069 -3.43 *** -0.083 -3.67 *** -0.152 -3.65 ***
N_Analyst -0.132 -7.06 *** -0.344 -5.97 *** -0.367 -6.01 *** -0.716 -6.26 ***
ROA -33.550 -16.20 *** -36.110 -9.49 *** 21.710 5.58 *** -14.840 -2.12 **
Size 0.404 3.11 *** 3.169 7.16 *** 3.620 7.74 *** 6.858 7.82 ***
B/M 2.429 7.11 *** 6.357 6.16 *** 5.430 6.06 *** 11.810 6.27 ***
Leverage 4.724 7.31 *** 5.377 2.80 *** 4.461 2.16 ** 9.788 2.62 ***
Industry and Year FE YES YES YES YES
Adjusted R2 0.280 0.139 0.060 0.078
Observations 11,994 11,994 11,994 11,994
Panel B: Overall Internal Governance and Real Earnings Management
Pred. (1) (2) (3) (4)
H1 RM_CFO RM_PROD RM_DISX RM1
Coef. t-stats Coef. t-stats Coef. t-stats Coef. t-stats
Int_Governance ― -0.102 -1.36 * -0.893 -4.26 *** -1.124 -4.63 *** -2.021 -4.61 ***
CEO_Horizon 0.006 0.45 -0.045 -1.19 -0.096 -2.04 ** -0.143 -1.74 *
CEO_Comp -0.307 -2.06 ** -1.902 -4.22 *** -2.306 -4.94 *** -4.256 -4.80 ***
CEO_PPS -0.041 -0.10 0.736 0.49 -0.385 -0.21 0.455 0.14
Firm_Age 0.006 1.09 -0.070 -3.52 *** -0.085 -3.76 *** -0.155 -3.74 ***
N_Analyst -0.132 -7.03 *** -0.345 -5.98 *** -0.368 -6.02 *** -0.718 -6.27 ***
ROA -33.530 -16.16 *** -36.010 -9.46 *** 21.820 5.59 *** -14.610 -2.09 **
Size 0.384 3.11 *** 3.070 7.09 *** 3.516 7.79 *** 6.646 7.79 ***
B/M 2.438 7.22 *** 6.403 6.17 *** 5.479 6.05 *** 11.900 6.27 ***
Leverage 4.730 7.32 *** 5.405 2.81 *** 4.491 2.17 ** 9.848 2.63 ***
Industry and Year FE YES YES YES YES
Adjusted R2 0.280 0.139 0.060 0.078
Observations 11,994 11,994 11,994 11,994

TABLE 3 (Cont’d)
Notes to Table 3:
RM_CFO is a real earnings management proxy that negatively affects cash flow from operations. RM_PROD is a real earnings manage
negatively affects production. RM_DISX is a real earnings management proxy that negatively affects discretionary expenses. RM1 and
measures of real earnings management. All measures of real earnings management are multiplied by 100 for the ease of exposition. Ex
subordinate executives’ decision horizon. Exec_PayRatio is the subordinate executives’ ability to influence the CEO. Int_Governance i
internal governance, measured as the sum of the standardized value of Exec_Horizon and Exec_PayRatio. CEO_Horizon is the CEO’s
CEO_Comp is the CEO’s logged total compensation. CEO_PPS is the pay-for-performance sensitivity of the CEO’s portfolio of equity
the firm. N_Analyst is the number of analysts following the firm. ROA is the return on assets in the current fiscal year. Size is the logg
the current fiscal year. B/M is the book-to-market ratio in the current fiscal year. Leverage is the leverage ratio in the current fiscal yea
corrected for cross-sectional and time-series dependence (Petersen 2009; Gow et al. 2010). ***, **, and * indicate statistical significan
0.10 level or better, respectively (one-tailed test where there is a prediction, two-tailed test otherwise).

59
TABLE 4
Internal Governance and Real Earnings Management – Partitioned by Suspect and
Non-Suspect Firms
Panel A: Suspect Firms
(1) (2)
Pred. RM1 RM2
Coef. t-stats Coef. t-stats
Int_Governance ― -1.857 -3.83 *** -1.229 -4.37 ***
CEO_Horizon -0.125 -1.38 -0.086 -1.71 *
CEO_Comp -4.725 -4.54 *** -2.967 -5.24 ***
CEO_PPS -0.376 -0.10 -0.781 -0.34
Firm_Age -0.169 -3.61 *** -0.086 -3.28 ***
N_Analyst -0.653 -5.20 *** -0.471 -6.55 ***
ROA -26.550 -2.44 ** -19.240 -3.03 ***
Size 6.573 7.05 *** 3.892 7.43 ***
B/M 18.070 5.58 *** 11.590 6.22 ***
Leverage 14.050 2.86 *** 11.160 4.09 ***
Industry and Year FE YES YES
Adjusted R2
0.095 0.117
Observations 7,701 7,701
Panel B: Non-Suspect Firms
(1) (2)
Pred. RM1 RM2
Coef. t-stats Coef. t-stats
Int_Governance ― -0.994 -1.26 -0.329 -0.81
CEO_Horizon -0.123 -0.83 -0.068 -0.88
CEO_Comp -1.677 -1.19 -1.048 -1.35
CEO_PPS 5.025 0.78 1.712 0.46
Firm_Age -0.135 -2.66 *** -0.065 -2.32 **
N_Analyst -0.736 -3.31 *** -0.470 -3.88 ***
ROA 7.187 0.93 -3.661 -0.87
Size 4.481 3.09 *** 2.451 3.66 ***
B/M 7.076 4.84 *** 4.421 4.25 ***
Leverage 10.430 1.87 * 10.080 3.51 ***
Industry and Year FE YES YES
Adjusted R2
0.051 0.053
Observations 1,803 1,803
Notes to Table 4:
Suspect firms are firm-years with earnings surprise between 0 and 1 percent of stock price, while Non-Suspect
firms are firm-years with earnings surprise less than -0.5 percent of stock price or more than 1 percent of stock
price. RM1 and RM2 are aggregate measures of real earnings management. All measures of real earnings
management are multiplied by 100 for the ease of exposition. Int_Governance is the firm’s overall internal
governance, measured as the sum of the standardized value of Exec_Horizon and Exec_PayRatio.
CEO_Horizon is the CEO’s decision horizon. CEO_Comp is the CEO’s logged total compensation. CEO_PPS
is the pay-for-performance sensitivity of the CEO’s portfolio of equity. Firm_Age is the age of the firm.
N_Analyst is the number of analysts following the firm. ROA is the return on assets in the current fiscal year.
Size is the logged value of total assets in the current fiscal year. B/M is the book-to-market ratio in the current
fiscal year. Leverage is the leverage ratio in the current fiscal year. Standard errors are corrected for cross-
sectional and time-series dependence (Petersen 2009; Gow et al. 2010). ***, **, and * indicate statistical
significance at the 0.01, 0.05 and 0.10 level or better, respectively (one-tailed test where there is a prediction,
two-tailed test otherwise).

TABLE 5
Alternative Measure of Key Executives' Influence and Real Earnings Management
Panel A:Key Executives' Abnormal Compensation
(1) (2) (1)
Pred. RM1 RM2 RM1
Coef. t-stats Coef. t-stats Coef. t-stats
Exec_Horizon ― -0.219 -2.11 ** -0.150 -2.50 ***
Exec_AbComp ― -4.807 -1.77 ** -3.781 -2.26 **
Int_Governance ― -1.268 -2.73 ***
CEO_Horizon -0.165 -1.82 * -0.109 -2.18 ** -0.166 -1.83 *
CEO_Comp -2.508 -2.76 *** -1.453 -2.86 *** -2.433 -2.64 ***
CEO_PPS -1.326 -0.34 -1.305 -0.55 -1.317 -0.34
Firm_Age -0.165 -3.49 *** -0.085 -3.14 *** -0.165 -3.49 ***
N_Analyst -0.652 -5.09 *** -0.474 -6.38 *** -0.652 -5.09 ***
ROA -24.560 -2.14 ** -18.420 -2.70 *** -24.620 -2.15 ** -
Size 5.825 6.12 *** 3.397 6.41 *** 5.820 6.11 ***
B/M 19.190 5.86 *** 12.280 6.49 *** 19.220 5.85 ***
Leverage 15.560 3.22 *** 11.860 4.31 *** 15.490 3.21 ***
Industry and Year FE YES YES YES
Adjusted R2
0.094 0.118 0.094
Observations 7,441 7,441 7,441

TABLE 5 (Cont’d)
Panel B:Key Executives' Independent Directorships in Other Firms
(1) (2) (3)
Pred. RM1 RM2 RM1
Coef. t-stats Coef. t-stats Coef. t-stats
Exec_Horizon ― -0.224 -2.16 ** -0.154 -2.57 *** -0.207 -2.00 **
Other_Director ― -2.059 -2.44 *** -1.281 -2.55 *** -2.022 -2.44 ***
Exec_PayRatio ― -5.321 -3.86 ***
CEO_Horizon -0.143 -1.56 -0.097 -1.91 * -0.128 -1.40
CEO_Comp -3.371 -3.56 *** -2.070 -4.00 *** -5.392 -4.65 ***
CEO_PPS -2.270 -0.60 -2.011 -0.88 -0.067 -0.02
Firm_Age -0.158 -3.39 *** -0.080 -3.03 *** -0.162 -3.49 ***
N_Analyst -0.674 -5.39 *** -0.484 -6.76 *** -0.646 -5.21 ***
ROA -25.790 -2.35 ** -18.740 -2.93 *** -26.740 -2.47 **
Size 6.250 6.55 *** 3.668 6.72 *** 7.012 7.24 ***
B/M 18.330 5.61 *** 11.760 6.25 *** 17.770 5.49 ***
Leverage 14.200 2.90 *** 11.270 4.13 *** 13.780 2.81 ***
Industry and Year FE YES YES YES
Adjusted R2
0.093 0.115 0.097
Observations 7,701 7,701 7,701
Notes to Table 5:
RM1 and RM2 are aggregate measures of real earnings management. All measures of real earnings management are multiplied by 100 for the e
Exec_Horizon is the subordinate executives’ decision horizon. Exec_AbComp is the logged (1 + abnormal compensation + sample minimum a
abnormal compensation is defined as the residual from a regression of executives’ mean total compensation on known determinants of CEO Pa
membership, book-to-market, returns and lagged returns, ROA and lagged ROA, and industry and year fixed effects). Int_Governance is the fir
governance, measured as the sum of the standardized value of Exec_Horizon and Exec_AbComp. Other_Director is the number of independent
held by key subordinate executives. Exec_PayRatio is the subordinate executives’ ability to influence the CEO. CEO_Horizon is the CEO’s de
the CEO’s logged total compensation. CEO_PPS is the pay-for-performance sensitivity of the CEO’s portfolio of equity. Firm_Age is the age o
number of analysts following the firm. ROA is the return on assets in the current fiscal year. Size is the logged value of total assets in the curren
to-market ratio in the current fiscal year. Leverage is the leverage ratio in the current fiscal year. Standard errors are corrected for cross-sectiona
(Petersen 2009; Gow et al. 2010). ***, **, and * indicate statistical significance at the 0.01, 0.05 and 0.10 level or better, respectively (one-taile
prediction, two-tailed test otherwise).

62
TABLE 6
Internal Governance and Real Earnings Management - Instrumental Variables (2SLS)
Approach
(1) (2) (3)
Pred. Int_Governance RM1 RM2
Coef. t-stats Coef. t-stats Coef. t-stats
Predicted_Int_Governance ― -3.075 -3.35 *** -2.225 -4.32 ***
CEO_Horizon 0.000 -0.15 -0.154 -1.64 -0.080 -1.56
CEO_Comp -0.471 -14.66 *** -5.466 -4.41 *** -3.584 -5.16 ***
CEO_PPS 0.568 6.76 *** 0.325 0.07 0.041 0.02
Firm_Age -0.004 -3.42 *** -0.179 -3.53 *** -0.090 -3.17 ***
N_Analyst 0.009 2.70 *** -0.574 -4.03 *** -0.434 -5.24 ***
ROA -0.048 -0.19 -30.500 -2.33 ** -22.440 -2.84 ***
Size 0.137 5.46 *** 6.760 6.33 *** 4.082 6.93 ***
B/M -0.096 -1.22 17.930 5.49 *** 11.200 5.90 ***
Leverage -0.062 -0.45 15.250 3.00 *** 11.710 4.16 ***
Lagged_Int_Governance 0.559 27.46 ***
Ind-Year-Median_Int_Governance 0.464 9.70 ***
Outside_CEO 0.103 2.75 ***
Named_Exec 0.014 0.90
Industry and Year FE YES YES YES
Adjusted R2 0.571 0.109 0.136
Observations 5,611 5,611 5,611
Kleibergen-Paap rk Wald F statistic 303.29 *** 303.29 ***
(Weak identification test)
Hansen J-statistic 5.197 5.499
(Over-identification test of all instr.)
Notes to Table 6:
RM1 and RM2 are aggregate measures of real earnings management. All measures of real earnings management
are multiplied by 100 for the ease of exposition. Int_Governance is the firm’s overall internal governance,
measured as the sum of the standardized value of Exec_Horizon and Exec_PayRatio. CEO_Horizon is the
CEO’s decision horizon. CEO_Comp is the CEO’s logged total compensation. CEO_PPS is the pay-for-
performance sensitivity of the CEO’s portfolio of equity. Firm_Age is the age of the firm. N_Analyst is the
number of analysts following the firm. ROA is the return on assets in the current fiscal year. Size is the logged
value of total assets in the current fiscal year. B/M is the book-to-market ratio in the current fiscal year.
Leverage is the leverage ratio in the current fiscal year. Lagged_Int_Governance is the one-year lagged value of
internal governance. Ind-Year-Median_Int_Governance is the industry-year median value of internal
governance. Outside_CEO is an indicator equals one if the current CEO is recruited from outside, and zero
otherwise. Named_Exec is the number of named executives in the annual proxy statement besides the CEO.
Standard errors are corrected for cross-sectional and time-series dependence (Petersen 2009; Gow et al. 2010).
***, **, and * indicate statistical significance at the 0.01, 0.05 and 0.10 level or better, respectively (one-tailed
test where there is a prediction, two-tailed test otherwise).

63
TABLE 7
Real Earnings Management surrounding the New External Appointment of
Subordinate Executives as Independent Directors
(1) (2)
Pred. RM1 RM2
Coef. t-stats Coef. t-stats
Exec_Horizon ― -0.272 -2.73 *** -0.178 -3.11 ***
CID_Firm 1.925 0.75 1.433 1.01
Post_CID_Firm ― -5.079 -2.42 *** -3.243 -2.75 ***
CEO_Horizon -0.100 -1.07 -0.071 -1.36
CEO_Comp -3.560 -3.68 *** -2.199 -4.27 ***
CEO_PPS -2.662 -0.68 -2.192 -0.93
Firm_Age -0.130 -2.76 *** -0.063 -2.39 **
N_Analyst -0.759 -5.98 *** -0.534 -7.19 ***
ROA -20.560 -1.77 * -15.660 -2.34 **
Size 6.648 6.66 *** 3.893 6.72 ***
B/M 17.680 5.50 *** 11.310 6.41 ***
Leverage 12.420 2.38 ** 10.140 3.54 ***
Industry and Year FE YES YES
Adjusted R2
0.091 0.111
Observations 6,675 6,675
Notes to Table 7:
RM1 and RM2 are aggregate measures of real earnings management. All measures of real earnings management
are multiplied by 100 for the ease of exposition. Exec_Horizon is the subordinate executives’ decision horizon.
CID_Firm is an indicator equals one if the firm has at least one key executive who holds independent
directorships in other firms during the sample period, and zero otherwise. Post_CID_Firm is an indicator equals
one for firm-years after the key executive is appointed as an independent director in other firms, and zero
otherwise. CEO_Horizon is the CEO’s decision horizon. CEO_Comp is the CEO’s logged total compensation.
CEO_PPS is the pay-for-performance sensitivity of the CEO’s portfolio of equity. Firm_Age is the age of the
firm. N_Analyst is the number of analysts following the firm. ROA is the return on assets in the current fiscal
year. Size is the logged value of total assets in the current fiscal year. B/M is the book-to-market ratio in the
current fiscal year. Leverage is the leverage ratio in the current fiscal year. Standard errors are corrected for
cross-sectional and time-series dependence (Petersen 2009; Gow et al. 2010). ***, **, and * indicate statistical
significance at the 0.01, 0.05 and 0.10 level or better, respectively (one-tailed test where there is a prediction,
two-tailed test otherwise).

64
TABLE 8
Internal Governance and Real Earnings Management conditioning on Key Executives’
Contribution
Panel A: Industry Research and Development Intensity
Pred. (1) (2)
H2 RM1 RM2
Coef. t-stats Coef. t-stats
Int_Governance -0.449 -0.78 -0.395 -1.18
IND_RD -2.005 -2.16 ** -1.335 -2.26 **
Int_Governance × IND_RD ― -2.688 -2.58 *** -1.592 -2.90 ***
Controls YES YES
Industry and Year FE YES YES
Adjusted R2
0.098 0.120
Observations 7,700 7,700
Panel B: Factor Analysis of Geographical Operating Complexity
Pred. (1) (2)
H2 RM1 RM2
Coef. t-stats Coef. t-stats
Int_Governance -1.012 -1.79 * -0.638 -1.95 *
GEO_Complexy -7.971 -4.11 *** -4.654 -4.08 ***
Int_Governance × GEO_Complexy ― -1.599 -1.63 * -1.142 -2.00 **
Controls YES YES
Industry and Year FE YES YES
Adjusted R2
0.105 0.127
Observations 7,701 7,701
Notes to Table 8:
RM1 and RM2 are aggregate measures of real earnings management. All measures of real earnings management
are multiplied by 100 for the ease of exposition. Int_Governance is the firm’s overall internal governance,
measured as the sum of the standardized value of Exec_Horizon and Exec_PayRatio. IND_RD is indicator
equals one (zero) if the average R&D intensity in the industry year is above (below) the sample median.
GEO_Complexy is an indicator equals one (zero) if the firm-year observation is above (below) the median first
principle component of the following three variables: 1) number of geographical segments; 2) geographical
sales concentration and; 3) percentage of foreign sales. CEO_Horizon is the CEO’s decision horizon.
CEO_Comp is the CEO’s logged total compensation. CEO_PPS is the pay-for-performance sensitivity of the
CEO’s portfolio of equity. Firm_Age is the age of the firm. N_Analyst is the number of analysts following the
firm. ROA is the return on assets in the current fiscal year. Size is the logged value of total assets in the current
fiscal year. B/M is the book-to-market ratio in the current fiscal year. Leverage is the leverage ratio in the
current fiscal year. Standard errors are corrected for cross-sectional and time-series dependence (Petersen 2009;
Gow et al. 2010). ***, **, and * indicate statistical significance at the 0.01, 0.05 and 0.10 level or better,
respectively (one-tailed test where there is a prediction, two-tailed test otherwise).

65
TABLE 9
Internal Governance and Real Earnings Management conditioning on CEO Power
Panel A: Board Independence
Pred. (1) (2)
H3 RM1 RM2
Coef. t-stats Coef. t-stats
Int_Governance -1.419 -2.46 ** -0.902 -2.71 ***
BD_IND -5.782 -4.53 *** -3.493 -4.67 ***
Int_Governance × BD_IND ― -1.111 -1.55 * -0.953 -2.38 ***
Controls YES YES
Industry and Year FE YES YES
Adjusted R2
0.108 0.132
Observations 4,796 4,796
Panel B: Institutional Ownership
Pred. (1) (2)
H3 RM1 RM2
Coef. t-stats Coef. t-stats
Int_Governance -1.669 -2.49 ** -1.189 -3.31 ***
Inst_Own 1.080 0.71 1.083 1.26
Int_Governance × Inst_Own ― -1.398 -1.58 * -0.763 -1.52 *
Controls YES YES
Industry and Year FE YES YES
Adjusted R2
0.102 0.127
Observations 6,731 6,731
Panel C: New Outside CEO
Pred. (1) (2)
H3 RM1 RM2
Coef. t-stats Coef. t-stats
Int_Governance -1.217 -2.36 ** -0.871 -2.98 ***
New_OutsideCEO -2.906 -0.99 -1.400 -0.83
Int_Governance × New_OutsideCEO ― -3.703 -3.21 *** -1.858 -2.50 ***
Controls YES YES
Industry and Year FE YES YES
Adjusted R2
0.095 0.118
Observations 7,181 7,181

66
TABLE 9 (Cont’d)
Notes to Table 9:
RM1 and RM2 are aggregate measures of real earnings management. All measures of real earnings management
are multiplied by 100 for the ease of exposition. Int_Governance is the firm’s overall internal governance,
measured as the sum of the standardized value of Exec_Horizon and Exec_PayRatio. BD_IND is an indicator
equals one (zero) if the firm-year observation is above (below) the median percentage of independent director.
Inst_Own is an indicator equals one (zero) if the firm-year observation is above (below) the median institutional
ownership. New_OutsideCEO is an indicator equals one if the CEO is recruited from outside and the CEO’s
tenure is less than three years, zero otherwise. CEO_Horizon is the CEO’s decision horizon. CEO_Comp is the
CEO’s logged total compensation. CEO_PPS is the pay-for-performance sensitivity of the CEO’s portfolio of
equity. Firm_Age is the age of the firm. N_Analyst is the number of analysts following the firm. ROA is the
return on assets in the current fiscal year. Size is the logged value of total assets in the current fiscal year. B/M is
the book-to-market ratio in the current fiscal year. Leverage is the leverage ratio in the current fiscal year.
Standard errors are corrected for cross-sectional and time-series dependence (Petersen 2009; Gow et al. 2010).
***, **, and * indicate statistical significance at the 0.01, 0.05 and 0.10 level or better, respectively (one-tailed
test where there is a prediction, two-tailed test otherwise).

67
TABLE 10
Internal Governance and Real Earnings Management conditioning on the Benefits of
Meeting or Beating Earnings Expectations
Panel A: Financial distress
Pred. (1) (2)
H4 RM1 RM2
Coef. t-stats Coef. t-stats
Int_Governance -1.840 -3.57 *** -1.232 -4.13 ***
Distress -2.166 -0.97 -0.717 -0.51
Int_Governance × Distress + 2.747 1.90 ** 1.786 2.22 **
Controls YES YES
Industry and Year FE YES YES
Adjusted R2
0.092 0.115
Observations 7,465 7,465
Panel B: Habitual Beater
Pred. (1) (2)
H4 RM1 RM2
Coef. t-stats Coef. t-stats
Int_Governance -3.003 -3.93 *** -1.768 -3.98 ***
Hab_Beater 2.194 1.82 * 1.218 1.87 *
Int_Governance × Hab_Beater + 1.519 1.95 ** 0.742 1.58 *
Controls YES YES
Industry and Year FE YES YES
Adjusted R2
0.097 0.118
Observations 7,234 7,234
Panel C: Debt or Equity Issuance
Pred. (1) (2)
H4 RM1 RM2
Coef. t-stats Coef. t-stats
Int_Governance -2.047 -3.82 *** -1.374 -4.50 ***
Capital_Issue 3.784 3.58 *** 2.204 3.59 ***
Int_Governance × Capital_Issue + 0.717 1.07 0.561 1.42 *
Controls YES YES
Industry and Year FE YES YES
Adjusted R2
0.097 0.119
Observations 7,701 7,701

68
TABLE 10 (Cont’d)
Notes to Table 10:
RM1 and RM2 are aggregate measures of real earnings management. All measures of real earnings management
are multiplied by 100 for the ease of exposition. Int_Governance is the firm’s overall internal governance,
measured as the sum of the standardized value of Exec_Horizon and Exec_PayRatio. Distress is an indicator
equals one if the Z-score of the firm is less than 1.81 and the bond rating of the firm is below investment grade,
and zero otherwise. Hab_Beater is an indicator equals one if the firm has meet or beat at least three out of the
last four quarters, and at least six out of the last eight quarters, and zero otherwise. Capital_Issue is an indicator
equals one if the firm issues debt or equity greater than or equals three percent of market value in the following
fiscal year, and zero otherwise. CEO_Horizon is the CEO’s decision horizon. CEO_Comp is the CEO’s logged
total compensation. CEO_PPS is the pay-for-performance sensitivity of the CEO’s portfolio of equity.
Firm_Age is the age of the firm. N_Analyst is the number of analysts following the firm. ROA is the return on
assets in the current fiscal year. Size is the logged value of total assets in the current fiscal year. B/M is the
book-to-market ratio in the current fiscal year. Leverage is the leverage ratio in the current fiscal year. Standard
errors are corrected for cross-sectional and time-series dependence (Petersen 2009; Gow et al. 2010). ***, **,
and * indicate statistical significance at the 0.01, 0.05 and 0.10 level or better, respectively (one-tailed test
where there is a prediction, two-tailed test otherwise).

69
TABLE 11
Internal Governance and Real Earnings Management conditioning on the passage of
SOX
(1) (2)
Pred. RM1 RM2
Coef. t-stats Coef. t-stats
Int_Governance -1.171 -2.38 ** -0.775 -2.91 ***
Post_SOX -2.721 -1.02 -1.519 -0.98
Int_Governance × Post_SOX ― -1.547 -2.29 ** -1.114 -2.76 ***
CEO_Horizon -0.126 -1.33 -0.083 -1.61
CEO_Comp -4.990 -4.35 *** -3.153 -5.01 ***
CEO_PPS -1.306 -0.32 -1.449 -0.59
Firm_Age -0.175 -3.77 *** -0.092 -3.49 ***
N_Analyst -0.657 -5.19 *** -0.470 -6.41 ***
ROA -33.350 -3.49 *** -22.170 -3.74 ***
Size 6.748 6.93 *** 4.035 7.37 ***
B/M 15.830 4.76 *** 10.350 5.33 ***
Leverage 13.690 2.74 *** 10.890 3.91 ***
Time 0.062 0.29 0.018 0.14
Industry FE YES YES
Adjusted R2
0.097 0.118
Observations 6,929 6,929
Notes to Table 11:
RM1 and RM2 are aggregate measures of real earnings management. All measures of real earnings management
are multiplied by 100 for the ease of exposition. Int_Governance is the firm’s overall internal governance,
measured as the sum of the standardized value of Exec_Horizon and Exec_PayRatio. Post_SOX is an indicator
equals one if fiscal year is on or after 2002, and zero otherwise. CEO_Horizon is the CEO’s decision horizon.
CEO_Comp is the CEO’s logged total compensation. CEO_PPS is the pay-for-performance sensitivity of the
CEO’s portfolio of equity. Firm_Age is the age of the firm. N_Analyst is the number of analysts following the
firm. ROA is the return on assets in the current fiscal year. Size is the logged value of total assets in the current
fiscal year. B/M is the book-to-market ratio in the current fiscal year. Leverage is the leverage ratio in the
current fiscal year. Time is a time trend variable which equals to the difference between the current fiscal year
and 1993. Standard errors are corrected for cross-sectional and time-series dependence (Petersen 2009; Gow et
al. 2010). ***, **, and * indicate statistical significance at the 0.01, 0.05 and 0.10 level or better, respectively
(one-tailed test where there is a prediction, two-tailed test otherwise).

70
TABLE 12
Internal Governance and Real Earnings Management conditioning on Self-serving
CEOs
(1) (2)
Pred. RM1 RM2
Coef. t-stats Coef. t-stats
Int_Governance 0.263 0.41 -0.023 -0.06
Self_Serving_CEO 3.607 1.05 1.809 0.95
Int_Governance × Self_Serving_CEO ― -3.835 -4.09 *** -2.182 -3.94 ***
CEO_Horizon -0.150 -1.63 -0.100 -2.00 **
CEO_Comp -4.806 -4.64 *** -3.021 -5.36 ***
CEO_PPS -0.367 -0.10 -0.747 -0.33
Firm_Age -0.165 -3.42 *** -0.084 -3.11 ***
N_Analyst -0.689 -5.44 *** -0.492 -6.74 ***
ROA -27.800 -2.58 *** -19.870 -3.14 ***
Size 6.838 7.31 *** 4.033 7.71 ***
B/M 17.740 5.57 *** 11.420 6.15 ***
Leverage 13.560 2.73 *** 11.020 3.97 ***
Industry and Year FE YES YES
Adjusted R2
0.103 0.124
Observations 7,601 7,601
Notes to Table 12:
RM1 and RM2 are aggregate measures of real earnings management. All measures of real earnings management
are multiplied by 100 for the ease of exposition. Int_Governance is the firm’s overall internal governance,
measured as the sum of the standardized value of Exec_Horizon and Exec_PayRatio. Self_Serving_CEO is an
indicator equals one (zero) if the firm-year observation is above (below) the median first principle component of
the following two variables: 1) industry homogeneity based on Parrino (1997) and; 2) industry competition
based on the inverse of industry sales concentration ratio. CEO_Horizon is the CEO’s decision horizon.
CEO_Comp is the CEO’s logged total compensation. CEO_PPS is the pay-for-performance sensitivity of the
CEO’s portfolio of equity. Firm_Age is the age of the firm. N_Analyst is the number of analysts following the
firm. ROA is the return on assets in the current fiscal year. Size is the logged value of total assets in the current
fiscal year. B/M is the book-to-market ratio in the current fiscal year. Leverage is the leverage ratio in the
current fiscal year. Standard errors are corrected for cross-sectional and time-series dependence (Petersen 2009;
Gow et al. 2010). ***, **, and * indicate statistical significance at the 0.01, 0.05 and 0.10 level or better,
respectively (one-tailed test where there is a prediction, two-tailed test otherwise).

71
TABLE 13
Internal Governance and Real Earnings Management conditioning on Future Option
Grants
(1) (2)
Pred. RM1 RM2
Coef. t-stats Coef. t-stats
Int_Governance (β1) -2.081 -4.75 *** -1.206 -4.61 ***
Future_Option_Grant 0.518 0.25 0.270 0.22
Int_Governance × Future_Option_Grant (β2) + 2.939 2.05 ** 1.571 2.12 **
CEO_Horizon -0.141 -1.71 * -0.091 -1.95 *
CEO_Comp -4.275 -4.81 *** -2.625 -5.38 ***
CEO_PPS 0.458 0.14 -0.458 -0.23
Firm_Age -0.155 -3.74 *** -0.077 -3.29 ***
N_Analyst -0.716 -6.29 *** -0.497 -7.72 ***
ROA -14.640 -2.09 ** -14.480 -3.33 ***
Size 6.662 7.77 *** 3.878 8.27 ***
B/M 11.890 6.25 *** 7.539 7.28 ***
Leverage 9.846 2.63 *** 8.900 4.39 ***
Industry and Year FE YES YES
Adjusted R2
0.078 0.100
Observations 11,994 11,994
F-test of β1 + β2 = 0 0.35 0.23
Notes to Table 13:
RM1 and RM2 are aggregate measures of real earnings management. All measures of real earnings management
are multiplied by 100 for the ease of exposition. Int_Governance is the firm’s overall internal governance,
measured as the sum of the standardized value of Exec_Horizon and Exec_PayRatio. Future_Option_Grant is an
indicator variable that equals one if the one-year ahead fixed-date option grant scaled by salary after the
earnings announcement is greater than the sample median and the firm misses analyst forecast by a small margin
(less than 0.5 percent of stock price) or a large margin (more than 10 percent of stock price), and zero otherwise.
CEO_Horizon is the CEO’s decision horizon. CEO_Comp is the CEO’s logged total compensation. CEO_PPS
is the pay-for-performance sensitivity of the CEO’s portfolio of equity. Firm_Age is the age of the firm.
N_Analyst is the number of analysts following the firm. ROA is the return on assets in the current fiscal year.
Size is the logged value of total assets in the current fiscal year. B/M is the book-to-market ratio in the current
fiscal year. Leverage is the leverage ratio in the current fiscal year. Standard errors are corrected for cross-
sectional and time-series dependence (Petersen 2009; Gow et al. 2010). ***, **, and * indicate statistical
significance at the 0.01, 0.05 and 0.10 level or better, respectively (one-tailed test where there is a prediction,
two-tailed test otherwise).