Board Committees, CEO Compensation, and Earnings Management

Author(s): Christian Laux and Volker Laux

Source: The Accounting Review , MAY 2009, Vol. 84, No. 3 (MAY 2009), pp. 869-891
Published by: American Accounting Association

Stable URL: https://www.jstor.org/stable/27784197

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

American Accounting Association is collaborating with JSTOR to digitize, preserve and extend
access to The Accounting Review

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
 THE ACCOUNTING REVIEW American Accounting Association
Vol. 84, No. 3 DOI: 10.2308/accr.2009.84.3.869
2009
pp. 869-891

Board Committees, CEO Compensation,

and Earnings Management
Christian Laux
Goethe-University Frankfurt am Main

Volker Laux
The University of Texas at Austin

ABSTRACT: We analyze the board of directors' equilibrium strategies for setting CEO
incentive pay and overseeing financial reporting and their effects on the level of earn
ings management. We show that an increase in CEO equity incentives does not nec
essarily increase earnings management because directors adjust their oversight effort
in response to a change in CEO incentives. If the board's responsibilities for setting
CEO pay and monitoring are separated through the formation of committees, then the
compensation committee will increase the use of stock-based CEO pay, as the in
creased cost of oversight is borne by the audit committee. Our model generates
predictions relating the board committee structure to the pay-performance sensitivity
of CEO compensation, the quality of board oversight, and the level of earnings
management.
Keywords: corporate governance; executive compensation; earnings management;
board oversight.

I. INTRODUCTION
There is a widespread concern that the surge in CEO pay and recent spectacular cases
of accounting fraud reflect a failure of corporate governance. A better understanding
of the relation between corporate governance, CEO pay, and earnings manipulation
is of great interest to both empirical and theoretical researchers, as well as regulators. A
key element in this relation is the role of the board of directors. In order to understand
how boards carry out their duties, it is important to recognize that most board functions
are performed by committees. As a director interviewed by Lorsch and Maclver (1989, 59)

We are grateful to John Core (editor) and an anonymous reviewers for their very helpful comments and sugges
tions. We also thank Harry Evans, Roman Inderst, Dirk Jenter, Bill Kinney, Christian Leuz, Paul Newman, Jason
Schloetzer, Raghu Venugopalan, and workshop participants at the Humboldt University Berlin, London School of
Economics, University of Bern, University of Frankfurt, University of Lausanne, University of Mainz, The Uni
versity of Texas at Austin, the 2006 EIASM Workshop on Accounting and Economics, the 2006 German Finance
Association Annual Meeting, the 2007 American Accounting Association FARS Meeting, and the 2007 European
Financial Management Association Meeting for their helpful comments.
Editor's note: Accepted by John Core.
Submitted: March 2007
Accepted: October 2008
Published Online: May 2009

869

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
 870 Laux and Laux

remarked: "The work of the board is done in committees" (see also Adams [2003] for
further discussion).
Delegating different board functions to different committees implies a separation of
tasks and functions on boards. We focus on the two board functions of designing the CEO
incentive pay scheme and overseeing the financial reporting process (monitoring) and study
how task separation affects corporate governance.
The compensation committee is responsible for setting the pay scheme for the CEO.
Due to the prevalence of stock-based executive compensation in corporations (Core et al.
2003), we focus on the share price as a performance measure for executive compensation.
Although higher-powered incentive schemes motivate the CEO to take value-enhancing
actions, ceteris paribus, they may also induce manipulative actions that boost the (short
term) stock price at the expense of long-term shareholder value. We model the interaction
between the productive and manipulative activities as a multi-task agency problem in the
spirit of Feltham and Xie (1994). In particular, the CEO in our setting has an incentive to
distort the earnings report upward in an attempt to mislead investors about future cash
flows. Although, as in Stein (1989), investors are not fooled in equilibrium,1 earnings man
agement is costly to current shareholders because it negatively affects the long-term value
of the firm.
One of the main objectives of the audit committee is to detect and prevent earnings
management. In particular, the audit committee is responsible for monitoring the integrity
of the firm's financial statements and the performance of the internal audit function. The
magnitude of earnings management in the firm therefore depends on both the CEO's level
of manipulation and the board's level of monitoring. Since an increase in stock-based
compensation enhances the CEO's direct incentive to manipulate earnings, the need for
board oversight increases with the size of the stock grant. Thus, there is a positive spillover
effect of the pay-performance sensitivity of CEO compensation on the audit committee's
incentive to engage in monitoring.
We assume that the board of directors is interested in the firm's long-term value. Mon
itoring is not contractible and not observable, which implies that the board cannot commit
to a certain level of oversight. However, the board is able to exploit the spillover effect of
CEO compensation to indirectly commit to oversight. That is, by increasing the link be
tween CEO pay and performance, the board can credibly signal to the CEO that it will
take its oversight function seriously. This indirect commitment to oversight is beneficial,
since it curbs the CEO's incentive to take manipulative actions in the first place.
To discuss the effects of task separation on boards, consider a situation without com
mittees where the whole board is responsible for both setting CEO pay and monitoring. In
this case, the board will under-exploit the positive spillover (indirect commitment) effect,
because it has to bear the personal cost of subsequent monitoring. That is, the anticipation
of burdensome monitoring lowers the board's desire to offer the CEO a powerful incentive
pay scheme. In such a situation, it is beneficial for shareholders to form committees and
to delegate the task of setting CEO pay to a compensation committee and the oversight
task to an audit committee. The formation of committees does not in itself guarantee a
clean separation of functions. Since it is not uncommon for board members to sit on more
than one committee, task separation on boards not only depends on the presence of com
mittees, but also on the degree of director overlap between the two committees. For this
study we define director overlap as the fraction of directors who serve on both committees.

1 See also Fudenberg and Tirole (1986) and Holmstrom (1999).

The Accounting Review May 2009

American Accounting Association

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
Board Committees, CEO Compensation, and Earnings Management 871

We show that the degree of director overlap determines the extent to which the compen
sation committee will take advantage of the positive spillover effect of CEO incentive pay
on board oversight. If director overlap decreases, i.e., if task separation is higher, then the
compensation committee is less involved in the monitoring task and hence less concerned
about the cost of monitoring. As a result, the compensation committee will increase the
pay-performance sensitivity of CEO compensation in order to induce a higher equilibrium
level of monitoring. Our model therefore predicts that an increase in task separation on
boards leads to more powerful incentive schemes for the CEO and an increase in board
oversight.
In our setting, an increase in stock-based CEO compensation is not necessarily asso
ciated with an increase in the level of earnings management. Clearly, a more powerful
incentive package increases the CEO's direct incentive to engage in manipulation, but, at
the same time, it also increases the audit committee's equilibrium choice of monitoring,
which in turn prevents and deters manipulation. Depending on which effect is stronger, the
magnitude of earnings management can increase or decrease with the pay-performance
sensitivity of the CEO's compensation plan.
We determine the optimal level of task separation on boards from the perspective of
shareholders and show that a positive level of task separation is always beneficial regardless
of whether this results in more or less earnings management.
In Section II we provide an overview of related studies. Section III introduces the
model. In Section IV we consider a benchmark case in which the board's monitoring choice
is observable and contractible. In Section V we analyze the equilibrium outcome for our
main setting, where monitoring is unobservable. We also show in this section how task
separation on boards influences the boards' decision-making, and we determine the optimal
degree of task separation from shareholders' perspective. Empirical implications are dis
cussed in Section VI. Section VII concludes. Appendix A provides a table of notation and
Appendix B contains the proofs.

II. RELATED STUDIES

There is a growing empirical literature that analyzes the relation between audit com
mittee or board characteristics, such as independence, size, and director background on the
one hand, and firm performance, CEO compensation, CEO turnover, and earnings man
agement on the other hand (e.g., Klein 2002; Farber 2005; Larcker et al. 2007; and for a
survey, Hermalin and Weisbach 2003). Our paper carves out another channel through which
the board structure has an impact: the separation of board functions. Our analysis provides
a number of testable predictions relating director overlap to the pay-performance sensitivity
of CEO compensation, earnings management, and board oversight.
Theoretical contributions on boards focus mainly on the question of how a change in
the board's power or independence affects CEO compensation and turnover (Hermalin and
Weisbach 1998; Almazan and Suarez 2003; Hermalin 2005). Adams and Ferreira (2007)
analyze the board's two roles of advising and monitoring the CEO. They show that if both
functions are performed by the same group of directors, then the CEO is unwilling to reveal
information that helps directors to provide advice, since directors can also use the infor
mation for monitoring purposes. A separation of functions is beneficial because it serves
as a substitute for a commitment not to use the revealed information against the CEO. In
contrast, in our setting, the separation of functions changes how the compensation com
mittee accounts for the cost of monitoring when choosing the pay-performance sensitivity
of the CEO's compensation contract. With separated responsibilities, the compensation

The Accounting Review May 2009

American Accounting Association

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
 872 Laux and Laux

committee is more willing to provide strong incentives to the CEO, which puts the
audit committee under greater pressure to diligently perform the oversight task.
Our paper is also related to the growing literature on earnings management (e.g.,
Fischer and Verrecchia 2000; Ewert and Wagenhofer 2005; Goldman and Slezak 2006).
Goldman and Slezak (2006) show that linking pay to the firm's share price provides the
CEO with incentives to manipulate accounting information. They analyze how an exoge
nous change in the level of monitoring influences the equilibrium levels of the pay
performance sensitivity and manipulation. In contrast, in our setting, the audit committee's
level of oversight is endogenous and unobservable, which creates a commitment problem
with respect to monitoring. We show that the board is able to exploit the pay-performance
sensitivity of CEO pay to indirectly commit to monitoring and analyze how the separation
of tasks on the board affects the extent to which the board will make use of this commitment
device. Goldman and Slezak (2006) also discuss the potential effects of recent regulatory
changes that require the separation of accounting services from consulting services. They
argue that this type of task separation eliminates "easy" ways to manipulate accounting
information and thereby makes manipulation more costly for the firm. Thus, in their model
task separation is assumed to have a direct effect on the cost of manipulation. In our
model, the effect of task separation is more in the spirit of the multi-task agency literature,
where the multi-task problem arises not only on the level of the CEO, but also on the
board. The board has to perform two functions, and the degree of task separation affects
how it performs these tasks. Thereby, task separation has an indirect effect on earnings
management.
The interaction between the productive and manipulative activities is related to the
multi-task agency literature introduced by Holmstrom and Milgrom (1991), Baker (1992),
and Feltham and Xie (1994). They show that the use of imperfect performance measures
can lead to distorted incentives in the sense that the agent allocates his effort inefficiently
among different tasks. In such a case, it can be optimal to employ a low-powered incentive
scheme in order to limit dysfunctional behavior. We extend this literature by introducing
endogenous monitoring by the board, which is assumed to reduce the agent's ability to
"game" the performance measure. Hence, in our setting the amount of effort distortion is
a function of the audit committee's equilibrium choice of monitoring, which in turn depends
on the pay-performance sensitivity of CEO pay. This modeling feature leads to a number
of new results. For example, if the effort distortion becomes more costly to the firm (i.e.,
the direct cost of earnings management to the firm increases), then the board does not
necessarily respond to this change by reducing the pay-performance sensitivity. Instead, it
may actually increase CEO incentives in order to induce a greater equilibrium level of
monitoring that prevents and deters earnings manipulation.

III. THE MODEL

Model Stages
Stage 1
In the first stage, the firm is established, and shareholders hire a CEO to run the firm.
Shareholders also employ a board of directors and decide on how the board will be organ
ized. The board of directors has to perform the two functions of setting CEO pay and
overseeing the financial reporting process. These functions can be delegated to the com
pensation committee and the audit committee. The formation and the composition of these
committees is commonly observable.
To ensure their participation, directors obtain a fixed salary for their services on the
board, wB. We normalize directors' personal costs of sitting on the compensation committee

The Accounting Review May 2009

American Accounting Association

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
Board Committees, CEO Compensation, and Earnings Management 873

to be zero, but performing the oversight function on the audit committee involves a personal
cost. Therefore, audit committee members must receive an additional compensation, which
we denote by wA.2

Stage 2
The compensation committee designs the incentive pay scheme for the CEO. In general,
the committee has several alternatives, e.g., it can use linear or nonlinear contracts and
choose between different performance measures. For the purposes of this paper, these dif
ferences are not essential. Since empirical evidence suggests that stock-based compensation
is the dominant source of CEOs' direct incentives (Core et al. 2003), we focus on executive
stock plans as an incentive tool. The compensation committee specifies the number of shares
awarded to the CEO, denoted b, and a fixed salary, denoted wc. We also refer to b as the
pay-performance sensitivity of the CEO's compensation scheme. The board's choice of
the pay parameters, wc and b, is publicly observable.
There are two ways to award the CEO the shares: the firm can either issue new shares
of stock, thereby diluting the claims of current shareholders, or it can use cash to buy the
required shares of stock in the open market. Both alternatives yield equivalent results. For
ease of exposition, we choose the second alternative.

Stage 3
The CEO chooses a productive effort, al9 which positively affects future cash flows. If
the manager does not engage in earnings manipulation, then the interim earnings report
issued in stage 3, R, is an undistorted signal of the firm's long-term cash flow, V, where R
= V = vxax + t|. The factor r\ is the state of the world, which is normally distributed with
mean zero and variance cr^.
Following the earnings management literature (Stein 1989; Feltham and Xie 1994;
Goldman and Slezak 2006), we assume that the CEO sells his stock prior to the realization
of the terminal firm value (stage 4). This creates an incentive for the CEO to manipulate
the interim earnings report R at the expense of long-term firm value, V. The task of the
audit committee is to oversee the financial reporting process in order to reduce the mag
nitude of earnings management. The level of oversight reflects the time and attention that
the audit committee devotes to its duties, such as preparing for meetings with management,
engaging in vigorous debate, and asking probing questions. Thus, the amount of realized
earnings management, M(a2, e), is a function of both the CEO's manipulation activity, a2,
and the board's monitoring effort, e, with dM/da2 > 0 and dM/de < 0. For a given level
of earnings management, the report in stage 3 is given by R = vlal + mM + r\ with m
> 0 and the long term-firm value is V = vxax - v2M + t], with v2 > 0.
Earnings management has a negative effect on firm value for various reasons. Examples
include costs for the firm due to reputation loss, legal liability, earnings restatements, or
increases in audit fees and D&O insurance premiums. By analyzing and discussing the
firm's financial and pro forma statements, the audit committee is able to detect and prevent
manipulation attempts early, which positively affects firm value. In addition, firm employees
may receive bonus payments that depend on the earnings report, the stock price, or both.
In this case, a higher level of oversight reduces the firm's wage cost. Moreover, manipu
lation may lead to distortions in the firm's operating and investment decisions. Graham et
al. (2005) find strong evidence that managers engage in real earnings management, which

2 A review of the 2006 Spencer Stuart Board Index shows that it is not uncommon for committee members to
receive retainers, and that the levels of these retainers differ depending on the type of the committee. (See
http://content.spencerstuart.com/sswebsite/pdf/lib/SSBI-2006.pdf)

The Accounting Review May 2009

American Accounting Association

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
 874 Laux and Laux

includes reducing discretionary spending on R&D, advertising, and maintenance, and post
poning positive net present value investments at the expense of long-term firm value. Again,
an attentive board should be better able to restrict management from undertaking or con
tinuing real activities that are aimed at manipulating the report.
We normalize the monitoring effort to take values between 0 and 1, e E [0,1], and
assume that M = (1 - e)a2. To guarantee a concave optimization problem and to focus on
interior solutions with a strictly positive pay-performance sensitivity, we assume that earn
ings management has a relatively low impact on the report relative to the productive act,
1. e., m is relatively small compared to vv
The activities al9 a2, and e are not observable. Directors' personal cost of performing
the minimum level of oversight, e = 0, and the personal cost of setting CEO compensation
are normalized to 0. Choosing an oversight effort above the minimum level involves a pri
vate cost, c(e) = Q.5kee2, with ke sufficiently high to induce interior solutions. The CEO's
private cost of undertaking the activity a{ is c(at) = 0.5&f<2?, for / = 1, 2. c(ax) is the
standard effort cost (or, equivalently, reduced private benefits). The cost of manipulation,
c(a2), may stem from litigation, reputation, or psychic costs.

Stage 4
The firm realizes its long-term cash flow, V. Shareholders' net payoff is the cash flow
V net of expected compensation for the CEO and the board. The terminal net firm value is
therefore given by V16* = V - wc - bS2 - wB - wA, where S2 is the stock price in stage
2, bS2 is firm's cost of buying the shares that are granted to the CEO in stage 2, wB is the
board members' base wage, and wA is the retainer for members of the audit committee.

Shareholders' Preferences and Interim Stock Prices

We assume that shareholders and potential investors are risk-neutral and that the risk
free rate is 0. The number of shares is normalized to 1. In equilibrium, the share price in
stage 2 equals the expected net terminal value, i.e., S2 = E[V,ef] = E[S3(R)].
In stage 3, shareholders observe the earnings report R. The stage 3 stock price S3 equals
the expected net firm value, given the report R, and the investors' conjectures about
the level of earnings management, denoted M. We obtain 53 = E[Vnet\R, M], with
E[Vnet\R, M] = (R - mM - v2M - wc - bS2 - wB - wA). We assume that investors know
the parameters of the model and hold rational expectations. Investors correctly anticipate
the CEO's incentive to bias the earnings report and the audit committee's incentive to
engage in monitoring. In equilibrium, those expectations are met, i.e., M = M*, where M*
is the equilibrium level of earnings management.

CEO Preferences
We assume the CEO can sell his stock in stage 3 at the price 53, which leads to a
compensation of wc 4- bS3.3 The CEO is risk-averse with CARA utility function u(y)
= -exp[-ry], where y is his compensation minus personal cost of effort, and r is his risk
aversion coefficient. The goal of the CEO is to maximize his expected utility, which is
equivalent to maximizing the certainty equivalent:

3 Of course, this is a simplifying assumption. In practice, legal or contractual constraints may prohibit the CEO
from unloading all his shares at the same time. In this case, the CEO's ability to benefit from earnings manage
ment is lower.

The Accounting Review May 2009

American Accounting Association

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
Board Committees, CEO Compensation, and Earnings Management 875

CE = wc + bE[S3(R)] - - 0.5k2a22 - 0.5rb2^. (1)

The CEO's reservation certainty equivalent is zero.

Board Preferences and Structure

To model the preferences of the risk-neutral board, we follow the approach in Hermalin
and Weisbach (1998), Hermalin (2005), and Adams and Ferreira (2007). The preferences
of individual committee members can be aggregated so that a committee acts as if it were
a single player with a utility function that puts a positive weight, 0, on the expected terminal
net value of the firm, E[Vne% and a negative weight on monitoring effort. The audit com
mittee's utility function can be stated as:

UAC = fflV""} + wB + wA- 0.5kee2. (2)

As in Hermalin (2005), 0 is exogenously given and depends on directors' long-term
stakes in the firm, the concern for public opinion, and reputation. The assumption that the
audit committee cares only about the long-term firm value is stronger than we need. Instead,
we could assume that the audit committee is also interested in the intermediate share price,
for example, because it wants to sell some of its shares in the short-run. As long as the
audit committee's interest in the intermediate share price is not too large relative to its
interest in the firm's long-run perspective, the level of board oversight will be positive and
our qualitative results continue to hold.
The preferences of the compensation committee are more intricate. First, compensation
committee members who also sit on the audit committee have the same preferences as
above, pEiy^'] + wB + wA - 0.5kee2. Second, directors who only sit on the compensation
committee do not incur a personal cost of monitoring and do not receive the retainer wA.
Their preferences are given by fiElV '] + wB. We introduce 7 E [0,1] (respectively (1
- 7)) to measure the weight that the first (respectively the second) group gains in the
compensation committee's decision process and assume that decisions in the compensation
committee are made to maximize the following (mixed) preferences of its members:

Uccil) = l(mVnet] + wB + wA- 0.5kee2) + (1 - 7)0?[V""] + wB)

= $E[V"e'] + wB + y(wA - 0.5kee2). (3)
If 7 = 1, then tasks are not separated: the two committees comprise exactly the same
directors, and the preferences of the compensation committee coincide with the preferences
of the audit committee. This case also captures the board's decision making when there are
no committees. In contrast, for 7 = 0, task separation is at its maximum: the two committees
are composed of different directors, and the compensation committee fully ignores the cost
of monitoring when choosing the CEO's incentive contract.
In general, we expect 7 to be positively related to the fraction of directors on the
compensation committee who also sit on the audit committee. Moreover, 7 also depends
on the relative importance of the individual board members who serve on both committees.
For example, if the chair of the compensation committee also sits on the audit committee,
then we would expect a greater impact on 7 than if a "regular" director is serving on both
committees. For convenience, we refer to 7 as director overlap. A lower director overlap
implies that the compensation committee is less involved in the oversight function of the
audit committee, which reflects situations where the level of task separation is greater.

The Accounting Review May 2009

American Accounting Association

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
876 Laux and Laux

Board members are willing to serve on the audit committee if they get compensated
for the expected cost of monitoring; that is, w*A = 0.5fe*2, where e* is the equilibrium
effort level. The base salary wB paid to all directors depends on the directors' expected
utility if they do not serve on the board. If, for example, p?[Vner] represents private benefits
of control that are only available to board members, then directors are willing to "pay" for
the privilege to sit on the board. In this case, the expected private benefit can be extracted
by choosing wB - -fiE[Vnet]. Alternatively, we can assume that board members are long
term investors who are holding a share of p in the firm independent of whether they sit on
the board; if they do not sit on the board, then another long-term investor with the same
share in the firm will. In this case, the participation constraint is binding for wB = 0. Since
our qualitative results do not depend on the level of board compensation, we assume for
simplicity that wB = 0.

IV. BENCHMARK: COMMITMENT TO MONITORING

In this section, we examine a benchmark setting in which the audit committee's mon
itoring activity is observable and contractible. Shareholders are therefore able to write a
contract with the audit committee in stage 1 that determines not only the salary wA, but
also the level of monitoring. We assume that the level of monitoring cannot be renegotiated
at a later stage, i.e., full commitment is possible.
We solve the model backward and first analyze the CEO's optimal action choices in
stage 3, given the incentive pay parameters wc and b and the committed level of monitor
ing, e.

Definition: An equilibrium consists of action choices for the CEO, (af, a$), and a
pricing function for the market, S3(R), that satisfy:

(i) Given S3(R), the CEO's optimal action choices (af, a%) solve:

max wc + bE[S3(R)] - OMrf - 0.5k2a2 - 0.5rZ?2cx2; (4)

(ii) Investors have rational expectations about the CEO's action choices:

S3(R) = R - m(l - e)a% - v2(l - e)a^ - wc - bS2 - wA. (5)

The action choices for the CEO and the pricing function for the market are mutually
consistent. Investors correctly anticipate the CEO's equilibrium level of manipulation and
the CEO understands the effect of the earnings report on the stock price.

Lemma 1: The CEO's equilibrium action choices are given by:

af=fb, (6)
m(l ? e) ,
a* = b. (7)

The CEO engages in earnin

ception about future cash f

May 2009
The Accounting Review
American Accounting Associ

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
Board Committees, CEO Compensation, and Earnings Management 877

CEO's incentive to manipulate and perfectly back out the bias in the report. Hence, similar
to Stein (1989), the CEO is "trapped" into manipulating the report in an attempt to distort
investors' belief even though nobody is fooled.
The CEO's incentive to deliver productive effort and his incentive to manipulate the
report both increase with the pay-performance sensitivity, b. Keeping the level of b fixed,
the CEO's incentive to manipulate decreases if the board commits to a higher level of
oversight.
In stage 2, the compensation committee sets the pay parameters wc and b, taking into
account the effects of b on effort and manipulation. The compensation committee solves:

max $E[Vnet] + y(wA - 0.5kee2l (8)

b,wc

subject to (6), (7), and the CEO's participation constraint:

wc + bE[S3(R)] - 0.5^1 - 0.5k2a22 - O.Srb2^ = 0. (9)

Observe that there is no conflict of interest between the compensation committee and
shareholders, since e and wA are fixed parameters in stage 2. Hence, the goal of the com
pensation committee is to maximize E[Vnet].

Lemma 2: The compensation committee's optimal choice of b is determined by:

v2 m(l - e? v2 m\\ - e)2 7

- - v2 ?--y b-k-b ~ rb^ = ?* (10)
The optimal pay-performance sensitivity trades off the benefit of inducing productive
effort against the cost of inducing earnings manipulation and the cost of imposing risk on
the risk-averse agent. If the marginal productivity of effort, vv increases or if the marginal
cost of earnings management, v2, decreases, then the compensation committee will find it
optimal to increase b in order to induce greater effort. In addition, when the audit committee
is committed to a higher level of oversight, earnings management becomes less of a con
cern, which, again, makes it optimal to increase the pay-performance sensitivity.
In stage 1, shareholders choose the level of monitoring, e, taking into account its effects
on subsequent decisions and activities. The optimal level of monitoring maximizes the
expected net terminal firm value, E[Vnet], subject to the constraints (6), (7), (9), (10), and
the audit committee's participation constraint, wA = 0.5kee2. Solving the shareholders' op
timization problem leads to the next lemma.4

Lemma 3: The optimal choice of e in stage 1 is determined by:

2v2(l - e) -K2
b* + K2
?^-- b*2 = kee, (11)

where satisfies (10).

4 Equivalently, shareholders can delegate the choice of monitor

the cost of monitoring since monitoring is observable and ver

The Accounting Review May 2009

American Accounting Association

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
 878 Laux and Laux

The left-hand side of (11) is the marginal benefit of monitoring, which equals the
marginal cost of monitoring, kee, for the optimal monitoring level. Monitoring has both a
direct and an indirect effect on the level of earnings management, M(e,a2). The direct effect
follows because board oversight helps to prevent earnings management; the indirect ef
fect arises because commitment to better oversight weakens the CEO's incentive to take
manipulative activities in the first place. A lower level of a2 is beneficial not only because
it reduces M, but also because the CEO must be compensated for his personal cost of
manipulation in order to ensure his participation. This latter effect implies that the cost
of CEO compensation decreases if the induced level of a2 declines. It may come as a
surprise that the CEO is compensated for his manipulation activity. However, the CEO's
participation constraint is binding. Since the CEO is "trapped" into manipulating the report
to satisfy the earnings expectations of the market, he will not participate unless this cost is
taken into account when the board sets his pay.
Note that we assume here that the actions ax and a2 are independent from one another.
Instead, we could assume a dependency between the two actions in the sense that earnings
manipulation distracts the CEO from working on his productive task (Holmstrom and
Milgrom 1991). This can be modeled by assuming that the two actions are substitutes in
the CEO's cost function, e.g., c(ax,a2) = 0.5kxa2 + 0.5k2al + k3axa2. This cost function
implies that spending time on the manipulation task increases the CEO's marginal cost of
working on the productive task. In such a situation, an increase in monitoring is not only
beneficial because it reduces the magnitude of earnings manipulation, but also because it
has a positive spillover effect on the productive action. Intuitively, increasing the level of
oversight makes manipulation less attractive and hence steers the CEO's attention to his
main job. However, since the introduction of effort substitution does not change our qual
itative results, we retain our original assumption that actions are independent.

V. MAIN RESULTS
Equilibrium Outcome
We now examine our original problem where the audit committee's choice of moni
toring is unobservable. In this case, it is not possible to commit to a specific level of
monitoring. The analysis proceeds as follows. In this subsection we analyze how the lack
of commitment changes the equilibrium results. In the next subsection we discuss the im
pact of a change in director overlap on the equilibrium outcome of the game and derive
the optimal level of director overlap from the shareholders' perspective.
We begin by analyzing the choice of activities in stage 3, given the compensation
parameters wc and b. Since ax, a2, and e are unobservable, we assume without loss of
generality that they are chosen simultaneously.

Definition: An equilibrium consists of action choices for the CEO, (af, a$), an action
choice for the audit committee, e*, and a pricing function for the market,
S3(R), that satisfy:

(i) Given S3(R) and e*, the CEO's optimal action choices (af, a$) solve:

max
a\,a2
wc + bE[S3(R(ax, a2, e*))] - 0.5kxa2x - 0.5k2a2 - 0.5rb2v2,

(12)

The Accounting Review May 2009

American Accounting Association

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
Board Committees, CEO Compensation, and Earnings Management 879

(ii) Given S3(R) and (af, a$), the audit committee's optimal action choice
e* solves:

max
e
$[vxaf - v2(l - e)a% - wc - bS2 - wA] + [wA - Q.5kee2],

(13)
(iii) S3(R) = R - m(\ - e*)af - v2(l ~ e*)a% - wc - bS2 - wA.

In equilibrium, each party correctly anticipates the other party's actions.

That is, af, af, and e* are best responses.

Lemma 4: The equilibrium strategies af, af, and e* are determined by the following
first-order conditions:

"T = fb> (14>

4 =*2 T0^?)b9 (15)
0i;2a:
(16)

The CEO's first-order conditions (14) and (15) are akin to those with commitment in
the benchmark case. However, with unobservable monitoring, the CEO bases his manipu
lation choice on the conjectured equilibrium level of monitoring rather than the actual level
of monitoring. Hence, the audit committee is unable to directly affect the CEO's manipu
lation incentive through its actual choice of monitoring. Oversight of financial reporting is
nevertheless valuable because, for a given level of CEO incentives, more diligent monitoring
helps to prevent earnings management. Since the audit committee puts a weight of 0 on
the net terminal firm value, the audit committee's marginal benefit of monitoring is
pi;2a|.
It is instructive to briefly compare the optimal levels of monitoring with and without
commitment. In the commitment case, monitoring is beneficial because it deters and pre
vents earnings manipulation. In the absence of commitment, monitoring no longer plays a
deterrence function. In addition, since monitoring is not observable in the current setting,
there is a standard moral hazard problem with respect to the monitoring choice, which
results in an underprovision of monitoring (for p < 1). For these reasons, for any given
levels of b and a2, the audit committee's monitoring level is lower in the current setting
than in the commitment setting.
Before we analyze the optimal size of the stock grant, b, we first discuss the equilibrium
effects of a change in b on the CEO's and the board's action choices and the magnitude
of earnings management, M. The equilibrium level of monitoring is determined by substi
tuting (15) into (16), which yields:

- -
kek2 + $v2mb'
^mb (17)

The Accounting Review May 2009

American Accounting Association

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
 880 Laux and Laux

Equation (17) shows that the audit committee's optimal choice of monitoring increases
with the pay-performance sensitivity, b. The audit committee anticipates that an increase
in stock-based compensation increases the CEO's direct incentive to engage in manipulative
activities. Since the marginal benefit of monitoring, $v2a$, increases in the CEO's equilib
rium choice of manipulation, an increase in the pay-performance sensitivity induces the
audit committee to be a more vigilant overseer.

Proposition 1: The audit committee's choice of monitoring, e*9 increases with the pay
performance sensitivity, b.

The total effect of an increase in the pay-performance sensitivity on the amount of

earnings management, M(a2,e), is ambiguous. On the one hand, holding the level of mon
itoring fixed, an increase in stock-based compensation increases the CEO's incentive to
engage in manipulation, a2, which increases M. This effect is weaker if the equilibrium
level of oversight is larger (for example, if monitoring is perfect, e = 1, an increase in a2
has no impact on M). On the other hand, since the audit committee anticipates the CEO's
manipulative behavior, an increase in b also increases the audit committee's incentive to
carefully oversee the reporting process. An increase in the equilibrium level of oversight
reduces M not only directly, but also indirectly. The indirect effect occurs because the CEO
correctly anticipates the equilibrium level of monitoring, and a higher conjectured level of
e reduces the CEO's ex ante incentive to manipulate.
In Appendix B we show that there exists a threshold, bt = kek2/fiv2m, so that the
magnitude of earnings management, M, increases with b for b < bp and decreases with b
for b > bt (bt is the pay-performance sensitivity that induces e* = 0.5). Intuitively, for
b > bn the CEO pay scheme induces a high equilibrium level of oversight (e* > 0.5),
which implies that the increase in a2 has a weaker impact on M than the increase in
resulting in a decline in M. Thus, for b > bp an attempt to limit earnings management by
reducing the use of stock-based CEO compensation actually leads to an increase in the
equilibrium level of M. In the remainder of the paper, we assume that the parameters of
our model are such that bt G (0,1).

Proposition 2: If the pay-performance sensitivity, b, increases, then the productive ac

tion, af, and the manipulative attempt, a$, increase, and the magnitude
of earnings management, M*, increases for b < bt and decreases for
b > br

In stage 2, the compensation committee sets the parameters wc and b of the CEO's
incentive scheme, taking into account the effects of b on the equilibrium choices of ax, a2,
and e. The compensation committee solves:

max $E[Vnet] + y(wA - 0.5kee2\ (18)

wc,b

subject to the action constraints (14), (15), and (16), and the CEO's participation constraint
(9). The solution to this problem is given in Proposition 3.

The Accounting Review May 2009

American Accounting Association

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
Board Committees, CEO Compensation, and Earnings Management 881

Proposition 3: The compensation committee's optimal choice of b satisfies:

v2 m(l - e*)2 v\
-v7-:-? 1 m\\ --e*)2
b-:-b rbvi119
2

+ ((2-,w.-^) = ? + =^??)f-a
where e* is determined by (17).

Without commitment to monitoring, the pay-performance sensitivity of th

scheme plays an additional incentive role that is absent in the commitment
stock-based CEO compensation induces the audit committee to become a mo
overseer. This positive spillover effect of CEO compensation is reflected in the
term:

which is missing in (10) in the benchmark case.

When choosing the pay-performance sensitivity, the board uses this spillover effect to
indirectly commit to monitoring. That is, by increasing the link between CEO pay and
performance, the board credibly signals to the CEO that it will take its oversight function
seriously. This indirect commitment to monitoring is beneficial, since it curbs the CEO's
incentive to engage in manipulation in the first place. In other words, a higher pay
performance sensitivity allows for monitoring to function as a deterrent in a situation in
which a direct commitment to monitoring is not possible.
There is another reason why the compensation committee will use b as a tool to induce
greater monitoring. Both the audit and the compensation committees weight the benefits of
monitoring with 0, but the compensation committee considers only a fraction 7 of the cost
of monitoring, while the audit committee considers the full cost. For this reason, the com
pensation committee finds it beneficial to increase b to induce the audit committee to work
harder on the oversight task. However, this effect is only present if tasks are separated to
some extent (7 < 1).
The feature that the pay-performance sensitivity serves as a tool to indirectly commit
to higher monitoring has the following implication. If the cost of earnings management, v2,
increases, then the board does not necessarily respond to this change by reducing the pay
performance sensitivity. Instead, the board may actually increase stock-based CEO com
pensation in order to commit to a higher level of board oversight, which deters and prevents
earnings management.

Proposition 4: There exist parameters for which the equilibrium pay-performance sen
sitivity increases if the firm's marginal cost of earnings management,
v2, increases.

The Role of the Committee Structure

The noncontractability of monitoring creates a role for task separation on boards. The
degree of director overlap, 7, determines the extent to which the compensation committee
is willing to use the pay-performance sensitivity, b, to indirectly commit to higher levels

The Accounting Review May 2009

American Accounting Association

This content downloaded from

1f:ffff:ffff:ffff:ffff:ffff on Thu, 01 Jan 1976 12:34:56 UTC
All use subject to https://about.jstor.org/terms
 882 Laux and Laux

of monitoring. Intuitively, if director overlap, 7, decreases, the compensation committee is

less involved in the oversight task and hence less concerned about the effort cost of mon
itoring. As a consequence, the compensation committee is willing to increase stock-based
CEO compensation in order to implement a higher equilibrium level of oversight (see
Equation (19)).

Proposition 5: As director overlap, 7, decreases, the compensation committee chooses

a higher pay-performance sensitivity, ft.

The next corollary directly follows from Propositions 1, 2, and 5.

Corollary 1: As director overlap,^, decreases, the monitoring effort, e*, the productive
action, af, and the manipulative attempt, af, increase, and the magnitude
of earnings management, M*, increases for ft* < bt and decreases for
ft* > ftr

Corollary 1 implies that for the parameters for which it is optimal to select a relatively
low (high) pay-performance sensitivity, the link between director overlap and earnings man
agement is negative (positive). Clearly, the equilibrium pay-performance sensitivity is higher
in firms in which the CEO's productivity of effort, vv is higher and/or where the environ
ment is less uncertain (u2 is small). We expect that for these firms, a decrease in director
overlap, which triggers an increase in ft*, reduces the level of earnings management.
Shareholders choose the structure of the board, 7, in stage 1 to maximize the expected
net terminal value of the firm, E[Vne% At the same time, the audit committee members
obtain a fixed salary wA, which compensates them for the expected equilibrium cost of
monitoring. The shareholders' problem in stage 1 is the following:

max E[Vne%

subject to (14), (15), (16), (19), and the audit committee's participation constraint, wA
= 0.5kee2.

Proposition 6: The optimal degree of director overlap for shareholders is given by

7 = P.

When the compensation committee chooses the pay-performance sensitivity, it weights

the positive effect of monitoring with p and the personal cost of monitoring with 7. As a
result, for 7 > p, the compensation committee cares "too much" about the cost of moni
toring because it is to a large extent directly involved in the oversight task. In anticipation
of burdensome monitoring, the compensation committee under-exploits the positive spill
over effect of CEO incentive pay on board oversight and chooses a ft that is too small from
shareholders' perspective. For 7 < P, the reverse problem occurs: the compensation com
mittee cares "too little" about the cost of monitoring, because tasks are separated to a great
extent. From the perspective of shareholders in stage 1, monitoring is costly because the
audit committee must be compensated for the expected personal cost of oversight. As a
result, for 7 < P, the compensation committee chooses a ft that is larger than the level
desired by shareholders. For 7 = p, the two effects are optimally balanced such that the

The Accounting Review May 2009

American Accounting Association

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
Board Committees, CEO Compensation, and Earnings Management 883

compensation committee will find it optimal to choose the level of b that is optimal for
shareholders in stage 1.
In practice, the director overlap in boards is likely also affected by other factors that
we do not examine here. Several considerations could cause shareholders to choose a level
of director overlap higher than the one identified in Proposition 6. For example, shareholders
may decide to limit the size of the board to save compensation costs, reduce free-rider
problems, or avoid untalented board members. Doing so will likely increase the level of
director overlap, since regulatory standards establish requirements for the minimum size
of committees as well as the level of independence and financial literacy. For example, the
NYSE and NASD require listed companies to maintain audit committees with at least three
directors, all of whom must be independent from management and financially literate.5
Moreover, our model ignores the potential benefits of communication between board mem
bers who perform different functions. The benefits of information sharing between com
mittees may provide a counterweight to the effects we outline in this study.

VI. EMPIRICAL IMPLICATIONS

The academic literature has identified and analyzed governance factors that are expected
to influence the quality of the internal control process in corporations. Popular governance
indicators are, for instance, the size of boards or committees, directors' degree of indepen
dence from the CEO, their stock ownership, and whether the CEO is also the chair of the
board (Weisbach 1988; Jensen 1993; Yermack 1996; Core et al. 1999).
Our paper contributes to this literature by carving out another channel through which
board structure has an impact. We focus on task separation on boards, which is arguably
an important consequence of forming committees. The model provides predictions relating
the board structure to board decision-making and organizational performance. What matters
for our predictions is the degree of task separation on boards, 7. A straightforward way to
measure the level of task separation is to look at the extent to which boards have adopted
committees. Generally, boards with committees are expected to employ a greater degree of
task separation than boards without committees. For those firms that are required to have
board committees by regulation, possible proxies for the level of task separation are, e.g.,
whether the chairman of the compensation committee also sits on the audit committee or
the fraction of compensation committee members who also serve on the audit committee.
For the purpose of illustration, it is interesting to take a look at the degree of task
separation in the 30 corporations that comprise the Dow Jones Industrial Average. A review
of these corporations' 2005 proxy statements shows that the size of the compensation
committee ranges from three to seven members, and that on average, the members of the
compensation committee sit on 1.47 other committees. However, for 12 corporations no
member of the compensation committee also sits on the audit committee, and for ten
corporations only one member sits on both committees. For only seven corporations does
the chairman of the compensation committee also serve on the audit committee. The 2006
proxy statements for the S&P 100 firms show similar numbers: for nearly 50 percent (23
percent) of the companies there is no (one) member of the compensation committee sitting
on the audit committee. In about 20 percent of the cases, the chairman of the compensation
committee also serves on the audit committee. Thus, for most firms, the degree of director
overlap between the compensation committee and the audit committee is relatively small.
This observation suggests that the introduction of committees does indeed result in a sep
aration of the two functions of setting CEO compensation and oversight.

5 See NYSE Listed Company Manual ?303.01, ?303A.06, and ?303A.07.

The Accounting Review May 2009

American Accounting Association

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
884 Laux and Laux

Board Structure and CEO Compensation

Our analysis indicates a link between the structure of the board and CEO incentive
pay. In particular, the model predicts that a reduction in the degree of director overlap leads
to a greater pay-performance sensitivity of CEO compensation. We are not aware of any
existing empirical research that examines this relation.

CEO Compensation and Earnings Management

Policymakers and regulators have devoted special attention to determining the cause of
recent accounting scandals. Some observers have suggested that the increased use of stock
based compensation has motivated executives to manipulate accounting information. How
ever, the extant empirical evidence on the link between CEO equity incentives and ac
counting irregularities is at best mixed (see Armstrong et al. [2008] for a detailed discussion
of the literature). While there are studies that find some evidence of a positive association
between the level of stock-based and, in particular, option-based compensation and earnings
manipulation (e.g., Cheng and Warfield 2005; Bergstresser and Philippon 2006; Burns and
Kedia 2006; Peng and Roell 2008), Erickson et al. (2006) do not find consistent evidence
of such a link, and Armstrong et al. (2008) provide some evidence suggesting a negative
effect of CEO equity incentives on the frequency of accounting irregularities.
The lack of a clear relation between CEO incentive pay and accounting manipulation
is consistent with the results in our paper. In our setting, an increase in the pay-performance
sensitivity not only has a positive effect on the CEO's direct incentive to engage in manip
ulation, but also on the audit committee's incentive to diligently oversee the accounting
process. Due to these two countervailing forces, the amount of earnings management can
increase or decrease with the level of CEO incentive pay. However, we do not want to
overstress this point, since there are alternative explanations for the lack of correlation
between CEO incentives and accounting fraud. For example, boards may offer executives
high-powered incentive schemes only in firms in which, for exogenous reasons, it is more
difficult or costly to manipulate the earnings report.

Board Structure and Oversight

Our model predicts that a reduction in the degree of director overlap leads to better
oversight of the financial reporting process. However, the link between board structure and
monitoring is indirect. A lower director overlap results in a greater pay-performance sen
sitivity of CEO compensation, which in turn enhances the audit committee's incentive to
carefully oversee financial reporting.
There are no studies that we know of that directly test our prediction, since the level
of board oversight is hard to measure. However, there is a large body of empirical research
that examines the relation between board structure and earnings management. Our model
suggests that it is important to control for the pay-performance sensitivity when analyzing
this relation. Without such a control variable, it is difficult to interpret the empirical findings,
because a lower magnitude of earnings management can be the result of better monitoring
or the use of compensation systems that are less sensitive to performance.
To elaborate on this issue, consider a hypothetical study that finds no association be
tween the presence of audit committees and the extent of earnings management. One might
be tempted to conclude that audit committees do not play an important role in the corporate
governance process. However, as our model shows, this conclusion might be wrong. If one
accepts the assumption that the formation of committees leads to an increase in task sep
aration, then our model predicts that the presence of audit committees positively affects the
pay-performance sensitivity of CEO compensation. This increase in CEO incentive pay

The Accounting Review May 2009

American Accounting Association

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
Board Committees, CEO Compensation, and Earnings Management 885

changes both the CEO's direct incentive to manipulate and the level of board oversight.
The two effects work in opposite directions, implying that, on average, the amount of
earnings management may not change. However, this does not indicate that the adoption
of audit committees has no effect on corporate governance; the move toward more
performance-sensitive pay systems, which is triggered by greater task separation, also en
hances the CEO's incentive to work hard on productive tasks, which has a positive effect
on firm value.
The empirical studies that analyze whether the existence of audit committees influences
the quality of the financial reporting process produces mixed evidence. Dechow et al. (1996)
and Beasley et al. (2000) find a positive association between audit committee presence and
financial reporting quality, but Beasley (1996) and Peasnell et al. (2005) find no such
relation.

VII. CONCLUSION
Arguably, two of the most important board functions are setting CEO pay and over
seeing the financial reporting process. Different committees are responsible for performing
these functions, yet both functions are very closely related. Stock-based compensation
schemes encourage the CEO to manipulate earnings, which in turn makes it necessary for
the board to act as a vigilant overseer. Compensation committee members who are not
involved in the oversight process will favor a higher pay-performance sensitivity of CEO
compensation than will board members who are responsible for overseeing financial re
porting. As a consequence, the separation of board functions is associated with greater
stock-based CEO compensation. However, the increase in CEO equity incentives does not
necessarily lead to a higher level of earnings management because the audit committee will
adjust its oversight effort in response to a change in CEO incentives. Our analysis provides
testable predictions concerning the relations between board committee structure, CEO com
pensation, board oversight, and earnings management.
Important incentives for board members to act as overseers of the financial reporting
process also stem from their fiduciary duty and potential liability claims (Gutierrez 2003).
In our model we do not consider claims against directors. However, our qualitative results
continue to hold when directors face legal penalties. The Sarbanes-Oxley Act of 2002
assigns directors serving on the audit committee with the special responsibility for over
seeing the firm's financial accounting process. Hence, when task separation on boards is
greater, the compensation committee will be less concerned about potential legal sanctions.
This again implies that the separation of board functions will lead to a stronger pay
performance sensitivity of CEO compensation, which, in turn, will increase the audit com
mittee's incentive to carefully oversee the reporting process.

APPENDIX A
NOTATION DESCRIPTIONS

V = long-term gross firm value (stage 4);

ynet _ iong_term net firm value (V net of compensation and direct cost of
manipulation);
St = interim share price in stage t\
wc = CEO's fixed pay;
b = pay-performance sensitivity of CEO compensation;
R = interim earnings report (stage 3);
P = weight that board puts on long-term net firm value;

The Accounting Review May 2009

American Accounting Association

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
 886 Laux and Laux

wB = non-contingent board members' base salary;

wA = non-contingent audit committee's compensation;
ax = CEO's productive activity;
a2 = CEO's manipulative activity;
e E [0,1] = audit committee's monitoring effort;
c(at) = 0.5kiOf = CEO's cost of undertaking activity at, at E{al9a2};
c(e) = 0.5kee2 = audit committee's cost of monitoring;
vx > 0 = marginal productivity of CEO effort;
v2 > 0 = marginal direct cost of earnings management;
M = (1 - e)a2 = level of earnings management;
7 E [0,1] = director overlap or degree of task separation on the board; and
r = CEO's risk-aversion coefficient

APPENDIX B
Proof of Lemma 1
The CEO takes his actions in stage 3. At that time, the expected share price is given
by:

E[S3(R(ax,a2))] = vxax + m(l - e)a2 - m(l - e)af

? v2(l - e)a% ? wA ? wc - bS2, (21)

where S2 is a constant. Substituting (21) into (4) yields:

max
a\,a2
wc + b[(vxax + m(l - e)a2 - m(l - e)a%

- v2(l - e)a% - wA - wc - bS2)] ? 0.5kxa2 - 0.5^2 ~ 0.5rb2cr2.

Calculating the first-order conditions with respect to ax and a2 yields (6) and (7).

Proof of Lemma 2
Substituting E[Vnet] - vxax - v2(l - e)a2 - wc - bE[S3(R)] - wA, (9), (6), and (7)
into (8) gives:

max (3
b Kx K2 /C| K2 v2

+ 7(wA - 0.5kee2).

Calculating the first-order condition with respect to b yields (10).

Proof of Lemma 3
Substituting (6), (7), (9), and wA = 0.5kee2 into the shareholders' objective function
E[Vnet] = vxax - v2(l - e)a2 - wc - bE[S3(R)] - wA9 shareholders' problem can be stated
as:

max ?p-V2 TtLzir

e ?vjb*K>2
- o.5
KxVA
K2b*2 _ 0.5m2(1 - 6)2 b*
- 0.5rfc*2a2 - 0.5kee2,

The Accounting Review May 2009

American Accounting Association

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
Board Committees, CEO Compensation, and Earnings Management 887

where b* satisfies (10). The first-order condition for an optimal level of e is:

Jft*
-r - v2 -^-t?- ~ b*-?-- b* - rb*v%
/C| AC2 ?v2 fife

+ 2u2(l - e) ft* + V 7, ; ft*2 - kee = 0.

Since the expression in brackets is zero in equilibrium (see (10)), the above condition
simplifies to (11).

Proof of Lemma 4
Substituting:

E[S3(R(ax,a2,e*))] = vxax -f m(l ? e*)a2 ? m(\ ? e*)a%

- v2(l ? e*)a% ? wc ? bS2 ? wA

into (12) yields:

max wc

+ b[vxax + m(l ? e*)a2 ? m(l - e*)a* ? u2(l ? e*)a* ? wc ? bS2 ? wA]

- 0.5kxa2x - 0.5k2a2 - 0.5rft2cx2.

Calculating the first-order conditions with respect to ax and a2 gives Equations (14) and
(15).
Note that at the time the audit committee chooses e, the stage 2 stock price, S2, is a
constant. Using (13), calculating the first-order condition with respect to e yields (16).

Proof of Proposition 2
We first show that the equilibrium level of manipulation, af, increases with ft. The first
derivative of (15) with respect to ft is:

da$ m(l - i*) m ^ de*^

db k2 k2 db'
de* By ink k de*
From (17) it follows that ? = /T , * 2 Substituting ? into (22) and rear
db (kek2 + 0u2mft)2 B db v '
ranging terms yields:

da* m
? db
= Tk2
(1 - e*)2.

We next show that ^?db

> 0 for ft <db
ft, and < 0 for ft > ft.. Given M = (1 - e)a2,
we have:

The Accounting Review May 2009

American Accounting Association

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
888 Laux and Laux

dM ^ de* _ , v m
2J

de* $v2mkek2 _ $v2m

Substituting ?db= (kek2
/u * +2 $v2mb)2
a: = nr , (kek2
2o^ ^ +(1$v2mb)
- e*) and (15) into the equa
tion above and rearranging yields:

? = -(1 - e*)2 ? [2e* - 1].

do k2

Hence, if follows that ^ > 0 for b < btdb

= -^L $v2m
and ^ < 0 for
db & > $u2m
bt = -^2-.

Proof of Proposition 3
Substituting E[Vnet] = vxat - v2(l - e)a2 - wc - bE[S3(R)] - wM
into (18) yields:

max&
p hlb _ k2
V2 ??LZJ*1 kx
b-WA-0.5vAb2-
k2 Q.5 ;^ y -/
o.5^)
+ y(wA - 05kee*2\

where satisfies (17). Recog

an optimal choice of b is

Pn (v\ mil
\kx k2 kx k2 - rbv2
77 ~ y2-r-t- 6-Z-b ? e*)2

Rewriting (17) as e*ke = $Vl b (l - ?*) and substituting this into the equation above
yields (19) in Proposition 3.

Proof of Proposition 4
Let

_ v2 m(\ - e*)2 v2 m2(l - e*)2 _ _ 2

T ~ Vl-jfc-Tb-1-b ~ rb<
/Vj IK2 /Vj IK2

+ |a-TWi-^) = ? + =^*>)f-a
Recall that in the optimum it holds that F = 0 (see (19)). The first derivative of b* with
. . , db* dF /dF , a/7. .
F 26 J dv2 dv2/ db db 5
respect to u2 is given by ?? =-/ ?. Note that ? is negative since we assumed that

The Accounting Review May 2009

American Accounting Association

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
Board Committees, CEO Compensation, and Earnings Management 889

m is small enough to ensure a concave optimization problem. Hence, in order to determine

db* dF
the
dv2 dv2 ??
sign of we need to find ?. I

dF mil ? e*)2 ? mil ? e*) de* ? m2

? =-?7-dv2+k2 2v2
k2 dv2?- k2
? +dv22 ?-b
i /, m , x de* m i ,r* x de* m2 10\ de*
+ (1 - e*) - b(2 - 7) " v2 ? - b(2 - 7) " ? -r- b2 ?
k2 dv2 k2 dv2 k2 / db
/i ^ m urn x i ra2Q - e*) \ d2e*
v2(l - e*) - b(2 - 7) + ^ b>) ?. (23)

Using (17), wedu2

have ^) =+ ^$v2mb)
(A^A:2 (1 - ,*), = (1 + $v2mb)
db (kek2
d2e*(b,v2) (1 - e*)Bra
- e*), and- = ?-? [1 - 2e*]. Substituting this into (23) and rear
dbdv2 (kek2 + pv2mb)
ranging gives:

dF m(l ? - =-(1
e*)2 -m(l
2e*)- +
e*)2
(2 - 7)-e*[2 - 3e*]
vu2 K,2 /C2

+ 3 ^ _ e*)2$mb m2 ^ ^ ^
(kek2 + $v2mb) k2
db*
Equation (24) shows that a sufficient condition for ?? > 0 is that e* > 0.5 and e*
dv2
2
< -. Let b*(vx) denote the equilibrium level of b* as a function of vx. Let uf be the

of vx that satisfies fe*(yf) = bn and let vx be the level of vx that satisfies ft*(i;f) = b

Z?r satisfies bT = 2 e 2 . Note that < vx. Since ft*(ut) is increasing in vx, and ft, a
yjU2m
are independent of vx, we have ft*^) > bt for i;1 > uf (which means that e* > 0.5) and
dft*
ft*(i;,) < bT for < vx (which means that e* < 2/3). It follows that ?? > 0 for all yf
ax) 2

Proof of Proposition 6
The shareholder's choice of 7 determines the compensation committee's choice of b in
stage 2. Shareholders' optimal strategy in stage 1 is therefore to choose the 7 that induces
the compensation committee to choose the ft that maximizes stage 1 firm value. Share
holder's optimal level of b in stage 1, denoted bs, can be found by solving:

max E[Vnet]9
b

subject to (14), (15), (16), (9), and wA = 0.5kee2. Using E[Vnet] = uxax -
? wc - bE[S3(R)] ? wA9 (9), and wA = 0.5kee2, the objective function can be writ

max vxax ? v2(l - e)a2 - 0.5kxa2x - 0.5k2a\ ? 0.5rb2cr2 - 0.5kee2.

b

The Accounting Review May 2009

American Accounting Association

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
 890 Laux and Laux

Recall that the compensation committee chooses the b in stage 2 that solves:

max
b
$E[Vnet] + y(wA - 0.5kee2\ (25)

subject to (14), (15), (16), and (9). Using (9), for 7 = p, the compensation com
objective function can be written as:

max
b
$[vxax - v2(l - e)a2 - 0.5kxa} - 0.5^ - 0.5rb2v2 - 0.5kee

This shows that the compensation committee will find it optimal to choose b* =
only if 7 = p.

REFERENCES
Adams, R. B. 2003. What do boards do? Evidence from board committee and director compensation
data. Working paper, Stockholm School of Economics.
-, and D. Ferreira. 2007. A theory of friendly boards. The Journal of Finance 62 (1): 217-250.
Almazan, A., and J. Suarez. 2003. Entrenchment and severance pay in optimal governance structures.
The Journal of Finance 58 (2): 519-547.
Armstrong, C. S., A. D. Jagolinzer, and D. F. Larcker. 2008. Chief executive officer equity incentives
and accounting irregularities. Working paper, University of Pennsylvania and Stanford
University.
Baker, G. 1992. Incentive contracts and performance measurement. The Journal of Political Economy
100 (3): 598-614.
Beasley, M. S. 1996. An empirical analysis of the relation between the board of director composition
and financial statement fraud. The Accounting Review 71 (4): 443-465.
-, J. V. Carcello, D. R. Hermanson, and R D. Lapides. 2000. Fraudulent financial reporting:
Consideration of industry traits and corporate governance mechanisms. Accounting Horizons 14
(4): 441-454.
Bergstresser, D., and T. Philippon. 2006. CEO incentives and earnings management. Journal of
Financial Economics 80 (3): 511-529.
Bums, N., and S. Kedia. 2006. The impact of performance-based compensation on misreporting.
Journal of Financial Economics 79 (1): 35-67.
Cheng, Q., and T. D. Warfield. 2005. Equity incentives and earnings management. The Accounting
Review 80 (2): 441-476.
Core, J. E., R. W. Holthausen, and D. F. Larcker. 1999. Corporate governance, chief executive officer
compensation, and firm performance. Journal of Financial Economics 51 (3): 371-406.
-, W. R. Guay, and R. E. Verrecchia. 2003. Price versus non-price performance measures in
optimal CEO compensation contracts. The Accounting Review 78 (4): 957-981.
Dechow, P. M., R. G. Sloan, and A. P Sweeney. 1996. Causes and consequences of earnings manip
ulation: An analysis of firms subject to enforcement actions by the SEC. Contemporary
Accounting Research 13 (1): 1-36.
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 (1): 113-143.
Ewert, R., and A. Wagenhofer. 2005. Economic effects of tightening accounting standards to restrict
earnings management. The Accounting Review 80 (4): 1101-1124.
Farber, D. B. 2005. Restoring trust after fraud: Does corporate governance matter? The Accounting
Review 80 (2): 539-561.
Feltham, G. A., and J. Xie. 1994. Performance measure congruity and diversity in multi-task principal/
agent relations. The Accounting Review 69 (3): 429-453.

The Accounting Review May 2009

American Accounting Association

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms
 Board Committees, CEO Compensation, and Earnings Management 891

Fischer, P. E., and R. E. Verrecchia. 2000. Reporting bias. The Accounting Review 75 (2): 229-245.
Fudenberg, D., and J. Tirole. 1986. A "signal-jamming" theory of predation. The RAND Journal of
Economics 17 (3): 366-376.
Goldman, E., and S. L. Slezak. 2006. An Equilibrium model of incentive contracts in the presence
of information manipulation. Journal of Financial Economics 80 (3): 603-626.
Graham, J. R., C. R. Harvey, and S. Rajgopal. 2005. The economic implications of corporate financial
reporting. Journal of Accounting and Economics 40 (1): 3-73.
Gutierrez, M. 2003. An economic analysis of corporate directors' fiduciary duties. The RAND Journal
of Economics 34 (3): 516-535.
Hermalin, B. E., and M. S. Weisbach. 1998. Endogenously chosen boards of directors and their
monitoring of the CEO. The American Economic Review 88 (1): 96-118.
-s and-. 2003. Boards of directors as an endogenously determined institution: A survey
of the economics literature. Economic Policy Review 9(1): 7-26.
-. 2005. Trends in corporate governance. The Journal of Finance 60 (5): 2351-2384.
Holmstrom, B., and P. Milgrom. 1991. Multi-task principal agent analysis: Incentive contracts, asset
ownership and job design. Journal of Law Economics and Organization 7 (Special Issue): 24
52.
-. 1999. Managerial incentive problems: A dynamic perspective. The Review of Economic
Studies 66 (1): 169-182.
Jensen, M. 1993. The modem industrial revolution, exit, and the failure of internal control systems.
The Journal of Finance 48 (3): 831-880.
Klein, A. 2002. Audit committee, board of director characteristics, and earnings management. Journal
of Accounting and Economics 33 (3): 375-400.
Larcker, D. F., S. A. Richardson, and I. Tuna. 2007. Corporate governance, accounting outcomes, and
organizational performance. The Accounting Review 82 (4): 963-1008.
Lorsch, J., and E. Maclver. 1989. Pawns or potentates. Boston, MA: Harvard Business School Press.
Peasnell, K. V., P. F. Pope, and S. Young. 2005. Board monitoring and earnings management: Do
outside directors influence abnormal accruals? Journal of Business Finance & Accounting 32
(7-8): 1311-1346.
Peng, L., and A. Roell. 2008. Executive pay and shareholder litigation. Review of Finance 12 (1):
141-184.
Stein, J. C. 1989. Efficient capital markets, inefficient firms: A model of myopic corporate behavior.
The Quarterly Journal of Economics 104 (4): 655-669.
Weisbach, M. S. 1988. Outside directors and CEO turnover. Journal of Financial Economics 20 (1
2): 431-460.
Yermack, D. 1996. Higher market valuation for firms with a small board of directors. Journal of
Financial Economics 40 (2): 185-211.

The Accounting Review May 2009

American Accounting Association

This content downloaded from

117.214.24.212 on Tue, 15 Jun 2021 01:44:52 UTC
All use subject to https://about.jstor.org/terms