Effect of corporate governance on real activity-based earnings management: evidence from Korea/Imoniu valdymo poveikis, pagristas darbo uzmokescio valdymu: Korejos pavyzdys.
Kang, Sun-A ; Kim, Yong-Shik
1. Introduction
Corporate governance is a decision-making structure or process that
monitors and controls firms and their managements in order to achieve
firms' goals. The efficiency of this corporate governance depends
on controlling agency problems that occur between managers and owners,
shareholders, and creditors. However, managers have many incentives to
control their reported earnings such as compensation, debt covenant, or
avoiding loss, even though it may sacrifice shareholder's wealth.
Managers could achieve target earnings by making accounting choices
among Generally Accepted Accounting Principles (GAAP) and/or by making
operating decisions in response to circumstances as they arise. Recent
studies report that managers prefer to use real operating decisions,
such as delaying a new project or reducing expenses, to control
earnings, rather than to use abnormal accruals (Graham et al. 2005).
These preferences seem to have become more prevalent since the SOX Act
came into force because management arbitrary decisions are more
difficult to detect which are protected by 'business judgment
rule' while abnormal accruals are easy to detect.
If corporate governance, in the form of such bodies as the board of
directors or audit committee, is effective, then managers'
discretionary accounting choices and arbitrary operational/investment
decisions could both be reduced. Earlier studies show that earnings
management through accounting accruals is influenced by corporate
governance and identify some of the factors that are significant in
constraining it. However, there is no study that examines the
relationship between corporate governance and the management of earnings
via real operating decisions (henceforth, real activity). The imposition
of constraints on manager's real operating decisions or investment
decisions by corporate governance entails that the control rights that
shareholders and creditors confer on managers are effectively reduced,
with the consequence that firms' future value will not be damaged
by a manager's private interests.
This paper examines the role of corporate governance in the context
of real activity-based earnings management. We focus on board
characteristics and consider three kinds of real activity-based earnings
management: aggressive sales promotions, overproduction, and cutting
discretionary expenses at either the individual or aggregate level. In
order to examine contextual analysis, we examine the relationships
between corporate governance and real activity based earnings management
when the committee operates inside a firm or when firms incur a loss.
For the test of the robustness of our findings, a corporate governance
index and control for endogenous variables are used. Both OLS and 2SLS
regressions were employed to examine the associations between corporate
governance and a firm's real activity-based earnings management.
The empirical results show that overall real activity-based
earnings management is reduced when the board of directors is either
independent or large. Overproducing or cutting discretionary expenses is
reduced as the size of the board increases, and aggressive sales or
overproducing is reduced as the number of outside directors on the board
increases. In the case of firms that have an internal audit committee,
these results are more pronounced at aggregate levels, whereas it seems
that the corporate governance of firms that have made a loss does not
influence manager's real operational decisions incrementally. The
findings are the same when we use a corporate governance index and when
we control for endogenous problems among variables. Finally, we find
that strong corporate governance reduces real activity based earnings
management.
The study reported herein differs from previous studies in that it
examines the relationships between corporate governance and real
activity-based earnings management, whereas previously, corporate
governance has been examined exclusively with regard to accrual-based
earnings management (Klien 2002; Xie et al. 2003). Given the
relationship that identified, managers' private interests, which
hitherto could have been pursued via real operations or investment, can
be controlled effectively by corporate governance in a comprehensive
manner. The results of the study reported herein suggest that certain
board characteristics, such as size or independence, effectively
constrain managers' real activity-based earnings management. This
focus on real activity-based earnings management suggests new avenues
for research on corporate governance. The results offer some insights
for policy makers interested in promoting legislation to ensure strong
corporate governance in their nation. In addition, the results highlight
the importance of strong corporate governance within a firm, because
corporate governance can effectively control real activity-based
earnings management, which in turn affects firm performance.
The remainder of the paper is organized as follows. In Section 2,
we review the relevant literature and develop hypotheses. In Section 3,
we describe the sample and present the empirical models. In Section 4,
we present the empirical results. In Section 5, we provide a summary and
make concluding remarks.
2. Review of the literature and development of hypothesis
2.1. Corporate governance
A board of directors is responsible for monitoring management
effectively in order to maximize shareholder's interests and can
dismiss managers if necessary (Fama, Jensen 1983; Weisbach 1988).
Previous work on corporate governance structures has mainly examined
such matters as earnings management, firm value, and management
compensation (DeJorge, Laborda 2011; Sanchez-Marin et al. 2011). With
respect to earnings management, studies have investigated, as measures
of sound accounting practices, whether certain corporate governance
structures improve the reliability of accounting reports (Ahmed,
Duellman 2007; Kim, Bae 2007); the association of corporate governance
structures with earnings management using abnormal accruals (Xie et al.
2003; Kim 2006); and the relationship of corporate governance structures
with fraud (Beasley 1996; Beasley et al. 2000; Beasley, Salterio 2001;
Uzun et al. 2004). Most of those studies examine the roles of corporate
governance by focusing on factors such as the composition and/or
characteristics of the board of directors or audit committee, and try to
determine which factor effectively controls of conflicts of interest
between owner and manager.
Previous studies that examined the relationship between the
composition of the board and accounting frauds report that fraud is
committed more frequently in financial statements when firms have less
external directors than average; as the close relationship between the
structure of the board and violations of accounting principles were
found (Beasley 1996; Beasley et al. 2000; Beasley, Salterio 2001).
Representative studies that examined relationships between the
characteristics of the board and earnings management include a study by
Klien (2002). This study shows that earnings management decreases when
audit committee is operated 'independently', and that earnings
management increases when the CEO is a member of the board. These
results suggest that as the board becomes more independent from the CEO,
it would be more effective in controlling accounting processes. A study
by Xie et al. (2003) reconfirms the fact that the role of the board is
important in preventing managers from manipulating the accounts.
In particular, the study reports that earnings management decreases
when the proportion of external directors increases, when the size of
the board increases, when the number of directors on the board who have
long experience with the company increases, and when the number of board
meetings increases. It also shows that earnings management decreases
when the independence of the audit committee increases or the number of
meetings of the audit committee increases. Studies that examined
relationships between the characteristics of the audit committee and
earnings management report somewhat different results. A study by Jeon
et al. (2004) reports that there is no significant relationship between
whether or not an audit committee is present and managers' earnings
management. However, a study by Ko et al. (2007) reports that earnings
management is reduced in firms once they have set up an audit committee.
A study by S. C. Lee and K. T. Lee (2003) reports that the greater the
proportion of external directors is in the audit committee, the less is
the extent of earnings management. These previous studies all employ
accrual-based earnings management, discretionary accruals, as a proxy of
earnings management.
To summarize the results of previous studies, it suggests that the
level of firms' accrual-based earnings management decreases if the
proportion of external directors on the board is high and the board is
very active or if there is an independent audit committee that operates
actively.
2.2. Real activity-based earnings management
Most early studies on earnings management focus exclusively on
accrual-based earnings management, which is usually a matter of
accounting choice. 'Real activity-based earnings management'
is defined as actions on the part of a firm's management personnel
that deviate from normal business practices in an attempt to meet target
earnings (Roychowdhury 2006). A few of empirical studies have been
conducted on the actual practice of real activity-based earnings
management, such as sales of fixed assets or investment to avoid
negative earnings growth and violating a debt covenant (Bartov 1993;
Herrmann, Inoue, Thomas 2003; Choi 2004). These practices result from
managers' action who have the opportunity to manage the selling
point of assets because a gain is recognized on the income statement at
the time of sale as the difference between the net book value and the
current market value. This study on real manipulation perspectives has
recently received much more research interest, motivated by Graham et
al. (2005), which provides survey results that CEOs have a preference
for using real activities to manage earnings because accrual-based
earnings management is likely to be detected by regulatory scrutiny and
CEOs can diversify those risks (1) by using both accruals and real
activities. Recent empirical evidences show that firms use multiple real
activities to avoid reporting annual losses, such as giving price
discounts to temporarily boost sales, overproducing to report a lower
cost of goods sold, and reducing discretionary expenditures to improve
earnings(Roychowdhury 2006; Kim et al. 2008). They report that real
activities, such as price discounts, overproducing or reducing
discretionary expenditures are found in firms that are suspected of
trying to avoid losses where earnings are just above zero (called
suspect firm-years). Other studies examine the consequences of real
earnings management (Gunny 2005; Kang, Chun 2010) and the relationship
between accrual- and real activity-based manipulations (Zang 2005; Cohen
et al. 2008). Each study finds that high abnormal real-activity groups
using portfolios show negative subsequent performances and that real
activity- based manipulations are negatively associated with
accrual-based one. Further study indicates that the trade off
associations is linked to the litigation risks (Cohen et al. 2008) and
tried to identify the incentive for real activity-based earnings
management in capital markets, considering such factors as outstanding
number of shares, external audit quality, bonus, and the number of
analysts.
2.3. Hypothesis development
A board must provide active and independent oversight of the
company on behalf of investors, and it should be operated independently
and efficiently in order to mitigate the conflict of interests between
owner and management. A board of directors consigns its decision-making
rights to the managers, but final responsibility for providing financial
reports with credibility and providing effective corporate governance
lies with the directors. The former duty is about managers'
accrual-based earnings management and the latter is about real
activity-based earnings management. In particular, managers'
abnormal real activities are performed in the course of the internal
decision process and most of them have to be discussed or approved by
the board.
Board members monitor management and their effectiveness in
performing this task depends on the independence, professionalism, and
activity of board members (Xie et al. 2003; Ben 2009). Empirical
research has shown that both the size of the board and the number of
board meetings affect accrual-based earnings management. Real
activity-based earnings management could be also influenced by board
characteristics, such as size, activity, and independence, either
directly or indirectly, because a manager's operational or
investment decisions are mostly approved by the board. Evidence for the
effectiveness of board size is inconclusive. Some studies report that
small boards are effective (Yermack 1996; Eisenberg et al. 1998),
whereas others report a significant positive association between board
size and performance (Dalton et al. 1999; Xie et al. 2003). Given the
foregoing, we do not have any expectations as to the direction of the
relationship between board size and real activity-based earnings
management. Boards meet frequently when they have many issues to discuss
and frequent board meetings are correlated positively with improved
financial performance (Vafeas 1999; Xie et al. 2003). We therefore
expect the incidence of real activity-based earnings management to be
inversely related to the number of board meetings. To the extent that
external directors monitor management more effectively than internal
directors (Xie et al. 2003; Lee, S. C., Lee, K. T. 2003; Ko et al.
2007), we hypothesize that a company that has more external than
internal directors on the board will be less likely to engage in real
activity-based earnings management than a company that has more internal
than external directors on the board. If the hypothesis turns out to be
correct, it may be because managers have the authority to make contracts
as to the board's compensation or tenure and internal directors are
likely to have a friendly relationship with managers than external
directors. In sum, we examine empirically whether real activity-based
earnings management is affected by certain characteristics of corporate
governance, such as board size, activity, or independence. We also
examine whether it decreases as external directors participate in board
meetings actively and whether it changes as they are financial experts
or not.
Introducing a system of external directors (2) or organizing an
audit committee within a firm is a way of decentralizing a board's
responsibility and of ensuring that the board discharges its
responsibilities properly. We define the role of audit committee as
being to oversee and monitor a firm's financial reporting and the
managers' day-to-day activity, and view the audit committee as
being first among equals in this monitoring process. The audit committee
reports to a board of directors as a lower branch of the board; its role
is to reduce information asymmetry between management and the board, so
that the likelihood that accounting fraud will be perpetrated might be
reduced (Beasley et al. 2000; Beasley, Salterio 2001). In addition, it
has been found that managers' abnormal real activity is controlled
directly or indirectly by the quality of external audits (Cohen et al.
2008). In order to determine the effectiveness of an audit committee, we
analyze the associations between board structure and real activity based
earnings management when an audit committee operates inside a firm in
the robustness check section. We also examine these associations when
firms incur a loss, because managers are likely to engage in earnings
management, whether it is via abnormal accruals or abnormal real
activity, in order to avoid a loss.
3. Research design
3.1. Model
Firstly, we examine whether board characteristics influence real
activity-based earnings management using multivariate regressions and
confirm the results using 2SLS regressions.
The characteristics of corporate governance that we consider are
board size, activities, independence, external directors'
activities, and external directors' professionalism. We use sales
manipulation, overproduction, and the cutting of discretionary expenses
as a proxy for real activity-based earnings management. In order to
check robustness, we examine those associations when a firm operates an
internal audit committee or incurs a loss. To make sure that our
findings are robust, we consider endogenous relationships among
variables.
We develop [Model 1] to examine these associations between the
characteristics of corporate governance mentioned above and real
activity-based earnings management after controlling for size, leverage,
performance, ownership, and compensation.
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (1)
where, [RM.sub.it]: Individual (aggregated) abnormal real activity
of firm i in year t (Ab.OCF, Ab.PRODcost, Ab.EXP, and RM_Proxy). Ab.OCF:
Abnormal cash flows from operations. Ab.PRODcost: Abnormal production
costs. Ab.EXP: Abnormal discretionary expenses (the sum of employee
welfare, advertising, R&D expense, and education and training
expenses). [B_SCALE.sub.it]: Number of directors on board of firm i in
year t (Log of the number of directors). [B_MEET.sub.it]: Number of
board meetings of firm i in year t (Log of the number of board
meetings). [OUTSIDE.sub.it]: Proportion of external directors of firm i
in year t (External directors/Total board members). [O_ACTIVITY.sub.it]:
Rate of participation in board meetings by external directors of firm i
in year t., [O_EXPERT.sub.it]: Proportion of external directors as
financial experts of firm i in year t. [COMMITTEE.sub.it]: Indicator
variable with a value of 1 if there is an audit committee within firm i
in year t, 0 otherwise. [SIZE.sub.it]: Natural log of total assets of
firm i in year t. [LEVERAGE.sub.it]: Debt ratio of firm i in year t
(Total [Debt.sub.it]/[Total Assets.sub.it-1]). [ROA.sub.it]: Return of
assets of firm i in year t ([Earnings before tax.sub.it]/ [Total
Assets.sub.it-1]). [LOSS.sub.it]: Indicator variable with a value of 1
if the net income of firm i in year t is below zero, 0 otherwise.
[OWNER.sub.it]: Large shareholder's ownership of firm i in year t.
[COMPEN.sub.it]: Managements' compensation of firm i in year t.
[OWNER_DUMMY.sub.it]: Indicator variable with a value of 1 if a manager
of firm i in year t is an owner, 0 otherwise. IND_Dummy: Industry dummy
variables. YEAR_Dummy: Year dummy variables.
We predict that board activities, independence, external
directors' activities, or their professionalism, are each
associated negatively with real activity-based earnings management and
that none of them predicts a larger or smaller board. We anticipate that
the relationships between the board characteristics and real
activity-based earnings management will be more pronounced when a firm
has an internal audit committee or reports a loss.
It is difficult to determine whether corporate governance is
effective overall at controlling earnings management, because internal
control factors interact each other. To overcome the difficulty, we make
a corporate governance index (CORP INDEX) (3) composed of the
significant individual characteristics found in [Model 1] and examine
the associations between corporate governance and real activity-based
earnings management in [Model 2].
[Model 2]
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (2)
where, [RM_proxy.sub.it] The sum of Ab.OCF, Ab.PROD_cost, and
Ab.EXP of firm i in year t (we multiply Ab.OCF and Ab.EXP by -1 so that
the sum of the three variables will be indicative of overall real
earnings management). [CORP_INDEX.sub.it]: The sum of B_SCALE, B_MEET,
and OUTSIDE of firm i in year t (we multiply B_MEET by -1 so that the
sum of the three variables will be indicative of overall corporate
governance).
The number of directors on the board (B_SCALE) represents board
size, the number of meetings (B_MEET) represents directors'
activity, and the proportion of external directors on the board
(OUTSIDE) indicates independence. The rate of participation of external
directors in board meetings (O_ACTIVITY) represents external
directors' activity and the proportion of financial experts among
external directors (O_EXPERT) represents the external directors'
professionalism. We control for firm size (SIZE), capital structure
(LEVERAGE) (4), and performance (ROA). In addition, we use the largest
shareholder's holding (5) (OWNER) and whether or not a firm is
owner-managed (OWNER_DUMMY) to control for ownerships. Finally, we use
dummy variables (IND Dummy and YEAR_ Dummy) to control for internal
director's compensation (COMPEN) and for industrial and yearly
effects that might exist. All these control variables (6) follow
previous studies regarding real activity based earnings management.
For the robustness check, we develop [Model 3] and [Model 4] to
examine the associations between corporate governance and real
activity-based earnings management, especially when the firm has an
internal audit committee or incurs a loss, considering each interaction
term variable as shown below.
[Model 3]
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (3)
[Model 4]
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (4)
where, [COMMITTEE.sub.it]: Indicator variable with a value of 1 if
there is audit committee within firm i in year t, 0 otherwise.
[LOSS.sub.it]: Indicator variable with a value of 1 if firm i in year t
report a loss, 0 otherwise. [CORP_INDEX*COMMITTEE.sub.it]: An
interaction term between CORP_INDEX and COMMITTEE.
[CORP_INDEX*LOSS.sub.it]: An interaction term between CORP_INDEX and
LOSS.
In the literature on corporate governance, there is concern about
endogenous relationships among variables. Managers' abnormal
operational or investment decisions are both influenced by weaker
corporate governance yet also affect the structure of governance. To
address this concern, we employ the following [Model 5] and control for
endogeneity. Following a previous study (Kim 2006), we also consider a
dummy variable for firms' assets (SIZE_DUMMY) in addition to firm
size (SIZE), because certain applications of Korean law depends on
whether firms are large7 or small.
[Model 5: 2SLS regressions]
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (5)
where, [CORP_GOV.sub.it] = Individual characteristic of corporate
governance or aggregate index; B_SCALE, B_MEET, OUTSIDE, O_ACTIVITY,
O_EXPERT and CORP_INDEX. [SIZE_DUMMY.sub.it]: Indicator variable with a
value of 1 the if total assets of firm i in year t are equal to or above
2 trillion Won, 0 otherwise.
3.2. Measurement of variables
We use each abnormal cash flow from operations, abnormal production
costs, abnormal discretionary expenses, and a combined measure as
proxies for overall real activitybased earnings management. We rely on
previous estimation models for normal levels of real activities and
regard abnormal levels of real operations as real activity-based
earnings management. The abnormal level of each measure is computed as
the actual level of a variable minus its normal level. We estimate
normal levels of cash flow from operations, production costs, and
discretionary expenses using a procedure developed by Dechow et al.
(1998), as implemented by Zang (2005) and Roychowdhury (2006), and we
run cross-sectional regressions for every industry and year, as follows.
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (6)
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (7)
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (8)
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (9)
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (10)
where, [OCF.sub.it]: Cash flows from operations of firm i in year
t. [PROD.sub.it] : COGS plus C_INV of firm i in year t. [COGS.sub.it]:
Cost of goods sold of firm i in year t. [C_INV.sub.it]: Change in
inventory of firm i in year t. [DISCEXP.sub.it]: Discretionary expenses
of firm i in year t (Employee welfare + Advertising + R&D expense +
Education and training). [A.sub.it-1]: Total assets of firm i at the
beginning of year t. [S.sub.it]: Sales of firm i in year t.
[C_S.sub.it]: Change in sales of firm i in year t([S.sub.it] -
[S.sub.it-1]). [C_S.sub.it1]: Change in sales of firm i in year
t-1([S.sub.it-1] - [S.sub.it-2]).
We regard normal cash flow from operations as a linear function of
sales and changes in sales in the current period. Production costs are
defined as the sum of the cost of goods sold (COGS) and the change in
inventory (C_INV) during the year. We consider discretionary expenses
including employee welfare, advertising, R&D expense, and education
and training expenses, and also express normal portion of discretionary
expenses as a linear function of lagged sales (8). We use a single
measure (RM_Proxy) (9) combining each individual measure in order to
capture overall effects of abnormal real activities because firms that
manage earnings upwards are likely to use multiple activities (10).
3.3. Sample selection
We took our sample from the firms that were listed on the Korean
Stock Exchange (KSE) from 2005 to 2007. We restricted the sample to
nonfinancial firms, because financial firms operate in highly regulated
industries that have accounting rules different from those in other
industries. We also excluded firms that have negative capital. Financial
data were obtained from the Fn-DataGuidePro database. Data on corporate
governance, such as board size, number of meetings, and the proportion
of external directors, were all hand-collected from annual reports.
Ownership data was obtained from the TS2000. We required that each
firm-year observation has the data necessary to calculate the abnormal
real activities used in this analysis. Further, each firm-year
observation was required to have a fiscal year ending in December, to
ensure homogeneity. We deleted the top and bottom 1% of the distribution
so that the results were not affected by outliers. Our final sample
comprised 1.104 firm-year observations.
Panel A of Table 1 shows the distribution of the sample over time
and Panel B reports industrial composition. Our samples are evenly
distributed by year and the composition by industry is similar to that
of the population.
4. Empirical results
4.1. Descriptive statistics
Table 2 reports the descriptive statistics on proxies for earnings
management and other variables that were examined. Panel A reports the
regression coefficients used to estimate 'normal' level of
proxies. The coefficients are consistent with prior studies
(Roychowedhurry 2006; Kim et al. 2008). Each coefficient of OCF and
production costs on sales is positive and significant, indicating that a
higher sale implies higher OCF and production costs.
The number of directors on the board (B_SCALE) is about 5~6 on
average. The number of meetings (B_MEET), which ranges from 1 to 150, is
about 16 times a year on average (11). External directors account for
32.3% of the total number of directors (OUTSIDE) on average and their
participation rates in board meetings (O_ACTIVITY) are 71.7%.
About 9.4% of the external directors are financial experts
(OEXPERT). 16.3% of the sample firms have an internal audit committee
(COMMITTEE).
Table 3 reports correlations among variables. It shows the Pearson
correlation coefficients based on two tailed tests.
Each board size (B SCALE) or independence (OUTSIDE) is associated
negatively with real activity-based earnings management (RM_Proxy). The
presence of an internal audit committee (COMMITTEE) is correlated
negatively with real activity-based earnings management. Board size (B
SCALE) is associated positively with firm size (SIZE). Each board size
(B_SCALE), independence (OUTSIDE), and whether or not there is an
internal audit committee (COMMITTEE) is associated negatively with the
largest shareholder's holdings (OWNER). Both firm size (SIZE) and
debt ratio {LEVERAGE) are correlated negatively with the largest
shareholder's holdings (OWNER). Real activity-based earnings
management is correlated positively with board compensation (COMPEN).
4.2. Effect on real activity-based earnings management of corporate
governance
Table 4 reports the results of [Model 1], which examines the
associations between the characteristics of corporate governance and
real activity-based earnings management. It shows each result using
sales manipulation, overproduction, and cutting discretionary expenses
as a proxy for real activity-based earnings management at both
individual and aggregate levels.
All adjusted [R.sup.2]s are significant, whether they are examined
using an individual or an aggregated measure of real activity-based
earnings management. From the perspective of an aggregated measure of
real activity-based earnings management, both board size (B_SIZE) and
the proportion of external directors (OUTSIDE) influence earnings
management negatively through abnormal real activities (RM_Proxy). This
result suggests that board scale and independence effectively constrain
managers' abnormal operational/investment decisions. That is, when
boards are large and the proportion of external directors is high, the
board can prevent managers from managing earnings by using abnormal real
activities (12) efficiently. Many studies on the efficiency of board
size report that smaller boards perform better (Yermack 1996 etc.), but
we find a contrary result with respect to earnings management. It is
consistent with the results of previous studies that larger boards may
have more independent directors and a larger board might be better at
preventing earnings management (Dalton et al. 1999; Xie et al. 2003). We
may presume from our findings that larger boards may have a greater
number of experienced directors, who are effective at limiting real
activity-based earnings management. In addition, the estimated value of
earnings management (RM Proxy) is smaller in firms that have an internal
audit committee (COMMITTEE). We may infer from this result that the
existence of an internal audit committee also has the effect of reducing
real activity-based earnings management.
Reviewing the results for real activity-based earnings management
individually, earnings management that uses overproduction (representing
a higher level of Ab.PROD cost) or cutting discretionary expenses
(representing a lower level of Ab.EXP) falls as the board size (B_SCALE)
increases. Earnings management that uses aggressive sales promotions
(representing a lower level of Ab.OCF) or overproduction (Ab.PROD cost)
is effectively controlled as the number of external directors increases
(OUTSIDE). Unexpectedly, aggressive sales promotions (Ab.OCF) are higher
when board meetings (B_MEET) are frequent, while overproduction (Ab.PROD
cost) is higher when the participation of external directors (O
ACTIVITY) is high. From these findings, it may be inferred that holding
board meetings frequently implies that there are many cases to be
discussed or approved by the board and that managers have many
opportunities to make discretionary decisions regarding aggressive sales
promotions or overproduction.
Regarding the results with control variables (13), the debt ratio
is correlated significantly and positive with real activity-based
earnings management, whether it be considered individually or
aggregately. This suggests that firms that have greater debts are likely
to be engaged in real activity-based earnings management. The
association between a firm's performance and real activity-based
earnings management is significant and negative, whether the earnings
management be considered individually or aggregately, which implies that
firms whose performance is poor are more likely to engage in real
activitybased earnings management. Managers are likely to engage in
earnings management, as measured by abnormal real activity (RM_Proxy),
when the board compensation (COMPEN) is higher. Those results on control
variables are consistent with correlations and previous studies.
The characteristics of corporate governance, such as board size,
the number of meetings, independence, or the presence of an internal
audit committee, may work comprehensively rather than individually.
Panel A of Table 5 shows the results for the effect of corporate
governance when we use a corporate governance index. a1 shows a
significantly negative sign (-0.036, P < 0.001) and indicates that
the overall strength of corporate governance constrains real
activity-based earnings management. Results of other variables are all
consistent with the results shown in Table 4.
4.3. Robustness check
Firms that have assets of over 2 trillion Korean Won are required
by law to have an internal audit committee. The presence of an internal
audit committee provides an internal control mechanism for monitoring
management's activity. In light of McMullen's (1996) finding
that earnings management or fraud perpetrated by managers is lower in
firms that have an internal audit committee than in those that do not,
we examine the association between real activity-based earnings
management and corporate governance for those firms that have an
internal audit committee. Firms that incur a loss are likely to be
engaged in earnings management using either accounting choices or
abnormal decisions. The results based on subsamples are shown in Panel A
of Table 5 and the results based on interaction terms in Panel B of
Table 5. We may presume from our findings that the effect of corporate
governance on real activity-based earnings management is negative for
firms that have an internal audit committee of [Model 3] in results of
Panel B is -0.065, P < 0.10). There is a significant negative
relationship between real activitybased earnings management and
ownership (-0.141, P < 0.05, OWNER of loss firm) or the presence of
an owner-manager (-0.071, P < 0.05, OWNER DUMMY of existing an audit
committee), as shown in Panel A. Even though [a.sub.1] of firms that
incur a loss in [Model 2] is significantly negative and its level is
lower than that of firms that make a profit, as shown in Panel A, those
associations are not significant when we test it using interaction terms
in [Model 4] ([a.sub.3] of [Model 4] is -0.025). Other results are
consistent with our main results reported in Table 4.
Hermalin and Weisbach argue that the variables board structure and
performance are endogenous and find that previous studies on boards, the
results of which are inconsistent, often neglect this issue. We test
this argument by employing two-stage least square (2SLS) regressions and
obtain results that are consistent with ours. These are reported in
Table 6. In the first stage, we run real activity-based earnings
management (RM_Proxy) and corporate governance individually
(B_SCALE,B_MEET, OUTSIDE, O_ACTIVITY, and O_EXPERT) or aggregately
(CORP_INDEX), where we control for firm size (SIZE) and size dummy (SIZE
DUMMY) (14). We obtain a result in the second-stage regression that is
consistent with our main results, even though each coefficient of
corporate governance (CORP GOV), [b.sub.1], is larger than the OLS
coefficients in [Model 1] and significant. Regarding the corporate
governance index, we consider all significant individual characteristics
(B_SCALE, B_MEET, and OUTSIDE) and obtain consistent results (15).
5. Conclusions
Previous studies have supported that corporate governance
effectively controls managers' earnings management. However, they
all used abnormal accruals as a proxy for earnings management. Hitherto,
the effect of corporate governance on real activitybased earnings
management has not been investigated. We examined how real activitybased
earnings management is affected by corporate governance, using data for
firms listed on the Korean stock exchange. We found that if corporate
governance influences firms' real operational or investment
decisions and if it is identified which factors are most influential,
then managers' discretionary activities, such as sales
manipulation, overproducing, and cutting expenses, could be controlled
effectively. We considered the following board characteristics. board
size, the number of board meetings, the proportion of external
directors, external directors' activities, and the financial
expertise of the external directors. We examine those characteristics
both individually and aggregately using a corporate governance index. In
particular, we tested the relationships between corporate governance and
real activity-based earnings management when a firm has an internal
audit committee or when it makes a loss. Sales manipulation,
overproduction, and cutting discretionary expenses were used as a proxy
for real activity-based earnings management.
The results show that managers are less likely to be engaged in
real activity-based earnings management when the board of directors is
large enough to control their operational or investment decisions or
when the board of directors consists of more external directors so that
it operates independently to a large extent. Earnings management through
sales manipulation increases as the board's activity increases. We
did not expect this result; we surmise that frequent board meetings
means there are many issues regarding abnormal operational or
investment, such as sales promotions or overproduction, that must be
decided but that cannot be settled for the best. We did not test this
conjecture, leaving it for further work. These associations are more
pronounced when we employ a corporate governance index as an aggregated
measure or when a firm has an internal audit committee. Consistent
results are found when we consider problems regarding endogenous
relationships among variables using 2SLS. Our study also considered
those associations when firms have an internal audit committee inside.
However, it is debatable whether an audit committee is actually
independent of the board of directors or not, because it is a sub
organization of the board. We leave the issue of the independence of
internal audit committees for further study. Our study focus on upwards
earnings management rather than downwards because inflation of earnings
cause damage on shareholder's wealth which is a main concern. Our
analysis and conclusion are based on proxies for earnings management and
these are calculated by estimation models developed by previous studies,
and therefore are subject to any biases inherent in the estimation
models. It is the first empirical evidence and meaningful that a
firm's real operational or investment decisions could be influenced
by well established governance mechanism in practice.
doi: 10.3846/16111699.2011.620164
APPENDIX
Definition of variables
Variables Definition
Earnings management proxies
RM_proxy Sum of Ab.OCF, Ab.PROD cost, and Ab.EXP
(We multiply Ab.OCF and Ab.EXP by
negative one so that the sum of the
three variables will be in-dicative of
overall real earnings management)
Ab.OCF Abnormal cash flows from operations
Ab.PRODcost Abnormal production costs
Ab.EXP Abnormal discretionary expenses (the sum
of employee welfare, advertising, R&D
expense, and education and training
expenses)
Corporate governance variables
BSCALE Number of directors on board, which used
as natural log forms in this study
B_MEET Number of board meetings, which used as
natural log form in this study
OUTSIDE Proportion of outside directors (# of
Outside director + # of total directors
on board)
Outside director Firm's board of directors is not current
employee of the firm, regardless of an
ex-employment in the firm
OACTIVITY Simple participation rate at board
meetings by outside directors
OEXPERT Proportion of outside directors as
financial experts
Financial experts Professor in the field of accounting or
finance, CPAs, or person who had
consulting experience on finance
COMMITTEE Indicator variable with a value of 1 if
there is audit committee within firm, 0
otherwise
CORPINDEX Sum of B SCALE, B_MEET, and OUTSIDE (We
multiply B_MEET by negative one so that
the sum of the three variables will be
indicative of overall corporate
governance)
CORP_[GOV.sub.it] Individual characteristic of
corporate governance or aggregate
index; B SCALE, BMEET, OUTSIDE, O
ACTIVITY, O EXPERT and CORP INDEX
Control variables
SIZE Natural log of total assets
SIZEDUMMY Indicator variable with a value of 1 if
total assets are equal to or above 2
trillion Won, 0 otherwise
LEVERAGE Debt ratio deflated by lagged assets
ROA Return of assets (Earnings before tax/
Total Assets)
LOSS Indicator variable with a value of 1 if
net income is below zero, 0 otherwise
OWNER Large shareholder's ownership which
includes holdings of a majority share-
holder, his or her family, and
affiliated firms which have special
relations with the firm according to
Article 2 of the Securities and Exchange
Act.
COMPEN Board of directors' compensation which
includes salary, bonus, and stock
option.
OWNERDUMMY Indicator variable with a value of 1 if
manager is an owner, 0 otherwise
IND Dummy Industry dummy variables
YEAR Dummy Year dummy variables
Note: This table provides definitions of the variables that were used
in this study. Data were obtained from Fn-DataGuidePro,TS2000, and
annual reports
Received 14 April 2011; accepted 07 July 2011
References
Ahmed, A.; Duellman, S. 2007. Evidence on the role of accounting
conservatism in corporate governance, AAA 2008 Financial Accounting and
Reporting Section (FARS) Paper.
Bartov, E. 1993. The Timing of asset sales and earnings
manipulation, The Accounting Review 68. 840-855.
Beasley, M. S. 1996. An empirical analysis of the relation between
the board of director composition and financial statement fraud, The
Accounting Review 71. 43-465.
Beasley, M. S.; Carcello, J. V.; Hermanson, D. R.; Lapides, P. D.
2000. Fraudulent financial reporting. consideration of industry traits
and corporate governance mechanisms, Accounting Horizons 14. 441-454.
doi.10.2308/acch.2000.14.4.441
Beasley, M. S.; Salterio, S. 2001. The relationship between board
characteristics and voluntary improvements in the capability of audit
committees to monitor, Contemporary Accounting Research 18. 539-570.
doi.10.1506/R[M.sub.1]J-A0YM-3VMV-TAMV
Ben Soltane Bassem. 2009. Governance and performance of
microfinance institutions in Mediterranean countries, Journal of
Business Economics and Management 10(1). 31-43.
doi.10.3846/1611-1699.2009.10.31-43
Black, B. S.; Jang, H. S.; Kim, W. C. 2006. Does corporate
governance predict firms' market value? Evidence from Korea,
Journal of Law, Economics, and Organization 22(2). 366-413.
doi.10.1093/jleo/ewj018
Choi, W. S. 2004. The Trade-off relationship of tax costs and
nontax costs on income from asset sales, Korean Accounting Review 29.
253-291.
Cohen, D.; Dey, A. A.; Lys, T. 2008. Real and accrual-based
earnings management in the Pre- and Post-Sarbanes Oxley Periods,
Accounting Review 83(3). 757-787. doi.10.2308/accr.2008.83.3.757
Cohen, D.; Zarowin, P. 2010. Accrual based and real earnings
management activities around seasoned equity offerings, Journal of
Accounting and Economics 50. 2-19. doi.10.1016/j.jacceco.2010.01.002
Dalton, D. R.; Johnson, J. L.; Ellstrand, A. E. 1999. Number of
directors and financial performance. a meta-analysis, Academy of
Management Journal 42(6). 674-686. doi.10.2307/256988
Dechow, P.; Kothari, S.; Watts, R. 1998. The relation between
earnings and cash flows, Journal of Accounting and Economics 25.
133-168. doi.10.1016/S0165-4101(98)00020-2
DeJorge, J.; Laborda, L. 2011. Corporate growth, age and ownership
structure. empirical evidence in Spanish firms, Journal of Business
Economics and Management 12(1). 164-196.
doi.10.3846/16111699.2011.555449
Eisenberg, T. S.; Sundgren, S.; Well, M. 1998. Larger board size
and decreasing firm value in small firms, Journal of Financial Economics
48. 35-54. doi.10.1016/S0304-405X(98)00003-8
Fama, E. F.; Jensen, M. C. 1983. Separation of ownership and
control, Journal of Law and Economics 26. 301-325. doi.10.1086/467037
Graham, J. R.; Harvey, C. R.; Rajgopal, S. 2005. The economic
implications of corporate financial reporting, Journal of Accounting and
Economics 40. 3-73. doi.10.1016/j.jacceco.2005.01.002
Gunny, K. 2005. What are the Consequences of Real Earnings
Management? Working paper. University of Colorado.
Herrmann, D.; Inoue, T.; Thomas, W. 2003. The sale of assets to
manage earnings in Japan, Journal of Accounting Research 41(1). 89-108.
doi.10.1111/1475-679X.00097
Jeon, K. A.; Choi, J. H.; Park, J. I. 2004. The relationship
between new audit committee establishment and earnings management,
Korean Accounting Review 29(1). 143-177.
Kang, S. A.; Chun, S. B. 2010. Consequences of real activity based
earnings management: evidence of seasoned equity offering firms in
Korea, Korean Management Review 39(3). 595-632.
Kim, B. H. 2006. The influence of the board composition on earnings
management in Korean firms: looking at cases of upwards and downwards
management, Korean Accounting Review 31(1). 1-32.
Kim, C. S. 2006. Outside directors and firm value in Korea, The
Korean Journal of Finance 19(2). 105-153.
Kim, J. H.; Goh, J. M.; Koh, Y. S. 2008. Real earnings management
to avoid loss and smooth income, expectations, Korean Accounting Journal
(17). 31-63.
Kim, J. O.; Bae, G. S. 2007. Corporate governance and accounting
conservatism: evidence from board and audit committee characteristics,
Korean Accounting Review 32(2). 89-115.
Klien, A. 2002. Economic determinants of audit committee, The
Accounting Review 77(2). 30-56.
Ko, D. Y.; Kim, M. T.; Yoon, S. S. 2007. The impacts of
establishment and independence of audit committee on controlling
earnings management, Accounting and Auditing 45. 69-90.
Lee, S. C.; Lee, K. T. 2003. An empirical study on the effect of an
audit committee on earnings management, Korean Accounting Review 28(3).
143-172.
McMullen, D. A. 1996. Audit committee performance. an investigation
of the consequences associated with audit committee, Auditing: a Journal
of Practice and Theory (Spring). 87-103.
Roychowdhury, S. 2006. Earnings management through real activities
manipulation, Journal of Accounting and Economics 42. 335-370.
doi.10.1016/j.jacceco.2006.01.002
Sanchez-Marin, G.; Baixauli-Soler, J. S.; Lucas-Perez, M. E. 2011.
Ownership structure and board effectiveness as determinants of TMT
compensation in Spanish listed firms, Journal of Business Economics and
Management 12(1). 92-109. doi.10.3846/16111699.2011.555371
Uzun, H.; Szewczyk, S. H.; Varma, R. 2004. Board Composition and
corporate fraud, Financial Analysts Journal 60(3). 33-43.
doi.10.2469/faj.v60.n3.2619
Vafeas, N. 1999. Board meeting frequency and firm performance,
Journal of Financial Economics 53. 113-143.
doi.10.1016/S0304-405X(99)00018-5
Weisbach, M. 1988. Outside directors and CEO turnover, Journal of
Financial Economics 20. 413-460. doi.10.1016/0304-405X(88)90053-0
Xie, B.; Davidson III, W. No.; Dadalt, P. J. 2003. Earning
management and corporate governance. the role of the board and the audit
committee, Journal of Corporate Finance 9. 295-316.
doi.10.1016/S0929-1199(02)00006-8
Yermack, D. 1996. Higher market valuation of companies with a small
board of directors, Journal of Financial Economics 40. 185-212.
doi.10.1016/0304-405X(95)00844-5
Zang, A. Y. 2005. Evidence on the Tradeoff Between Real
Manipulation and Accruals Manipulation. Working paper. Duke University.
(1) Using accruals-based earnings management alone is risky,
because real activities can occur during the year and these cannot be
adjusted at the end of the fiscal year. If reported earnings fall below
the target and all attempts at accruals-based earnings management to
meet it fail, managers have no options.
(2) Outside directors are defined as gatekeepers who have a
responsibility to prevent corporate misconduct from its management and
they are not employees of the company.
(3) We make a corporate governance index that includes the
significant factors found in [Model 1-1]. We rank our sample into five
groups based on each board size, activity, or independence, and assign 1
to 5 from lower to higher quintiles. Then we find their mean and get an
equally weighted average score, which is the final corporate governance
index. The results are consistent when we consider all components of the
individual corporate governance variables used in the study.
(4) We use total assets as a denominator because using net assets
may result in negative numbers, which can distort the continuity of a
firm's debt ratio.
(5) Data are available from the TS2000 database of the Korean
listed firms' associations.
(6) One of referee recommends for controlling firm's financing
decision such as seasoned equity offering and we obtain qualitatively
the same results when we include a control variable of SEO, which is
untabulated.
(7) Large firms those that have assets above 2 trillion Won have to
have an internal audit committee.
(8) The reason why lagged sales are used is because modeling
discretionary expense as a function of current sales creates a technical
problem in that unusually low residuals can result if firms manage sales
upwards to increase earnings in a certain year and estimate normal
levels of discretionary expenses.
(9) We multiply abnormal cash flows from operations and abnormal
discretionary expense by -1 to make it easier to interpret the results.
Price discount or channel stuffing has a negative effect on
contemporaneous abnormal OCF. Excessive price discount or overproduction
leads to abnormally high production costs relative to sales. Cutting
discretionary expenses leads to abnormally low discretionary expenses
relative to sales. Each direction implying earnings management is not
the same and we make it to the same direction.
(10) Cohen and Zarowin (2010) combine the 3 individual proxies to
compute 2 kinds of aggregate measures, RM_1 and RM_2, because double
discounting issue may exist. We obtain qualitatively the same
(untabulated) results when we use 2 kinds of combining measures. We
thank the referee for pointing it out.
(11) Simply, this is the number of board meetings and we use log
specification to mitigate heteroskedasticity when we run OLS or 2SLS,
including B SCALE and B MEET.
(12) We have the point of view that earnings management through
real activities is against shareholders interest in this study but
sometimes it may work in favor of shareholders interest. It is still an
issue whether reducing discretionary expenses are good in the
perspective of shareholder.
(13) Multicollinearity is not an issue here and the highest VIF is
1.87.
(14) An asset dummy that indicates large firms is considered in
this study, following Black et al. (2006). Multicollinearity is
insignificant.
(15) Even we make corporate governance index considering all
factors, results are qualitatively the same.
Sun-A Kang [1], Yong-Shik Kim [2]
[1] Business Administration School, Chungnam National University,
Daejeon, Republic of Korea
[2] Business Administration School, Hansung University, Seoul,
Republic of Korea
E-mails: [1] sunakang@hotmail.com (corresponding author); [2]
jacob@hansung.ac.kr
Sun-A KANG got M.A and Ph.D. in Business administration at Sogang
University in Korea. She worked as an AICPA at PWC Samil accounting firm
and as a senior researcher at ETRI (Electronics and Telecommunications
Research Institute). Currently, she is an assistant professor of
Chungnam National University in the field of Management. Her research
interests are financial accounting, especially earnings management,
corporate governance, and regulatory cost accounting.
Yong-Shik KIM got M.A and Ph.D. in Business administration at
Sogang University in Korea. He worked as a KICPA at Deloitte Anjin
accounting firm for 5 years and at finance deportment of Seoul City Hall
for 2 years. Now he is an assistant professor of Hansung University from
2009. He has interested in financial accounting, IFRS, and corporate
governance.
Table 1. Samples distribution
Panel A: Time Distribution
Year Frequency % Cumulative %
2005 377 34.15 34.15
2006 362 32.79 66.94
2007 365 33.06 100
Total 1.104 100 --
Panel B: Industry Distribution
Industry Code Frequency % Industry
Food products I.005 83 7.52% Health
Textile products I.006 49 4.44% Transport
Paper and Paper I.007 50 4.53% Distributions
products
Chemical I.008 184 16.67% Electricity
products and Gas
Medicine I.009 62 5.62% Construction
Nonmetal I.010 51 4.62% Transport and
Minerals Storage
Metals I.011 90 8.15% Service
Machinery I.012 78 7.07% Manufacturing
Electronics I.013 98 8.88% Total
Industry Code Frequency %
Food products I.014 10 0.91%
Textile products I.015 81 7.34%
Paper and Paper I.016 76 6.88%
products
Chemical I.017 22 1.99%
products
Medicine I.018 73 6.61%
Nonmetal I.019 32 2.9%
Minerals
Metals I.026 35 3.17%
Machinery I.027 30 2.72%
Electronics 1.104 100%
Note: Industry classification is by Fn-DataGuidePro database
Table 2. Descriptive statistics
Panel A: Model parameters
Variables [OCF.sub.it]/[A.sub.it-1]
Estimate t-value
Intercept 0.028 9 67 ***
[S.sub.it]1/[A.sub.it-1] -534220 -4.66 ***
0.029 10.11 ***
[S.sub.it]1/[A.sub.it-1]
C_[S.sub.it]/[A.sub.it-1] 0.040 5.24 ***
C_[S.sub.it]/[A.sub.it-1]
Adjusted [R.sup.2] 0.04
[PROD.sub.it]/ [DISCEXP.sub.it]/
Variables [A.sub.it-1] [A.sub.it-1]
Estimate t-value Estimate t-value
Intercept -0.065 -14.19 *** 0.009 9.21 ***
[S.sub.it]1/[A.sub.it-1] -179318 -1.02 -82827 -2.08 **
0.88 189 ***
[S.sub.it]1/[A.sub.it-1] 0.010 11.28
***
C_[S.sub.it]/[A.sub.it-1] -0.005 -0.5
C_[S.sub.it]/[A.sub.it-1] -0.032 -2.81 ***
Adjusted [R.sup.2] 0.90 0.02
Notes: ***: significant at the 1% level, **: significant at the 5%
level, *: significant at the 10%
Panel B: Descriptive statistics
Variables Mean Std.Dev Min 1Q Median 3Q
RM Proxy -0.016 0.165 -0.643 -0.116 -0.016 0.084
Ab.OCF -0.002 0.075 -0.272 -0.051 0.001 0.044
Ab.PRODcost -0.018 0.109 -0.464 -0.074 -0.011 0.044
Ab.EXP 0.000 0.023 -0.121 -0.010 -0.002 0.006
BSCALE 5.941 2.073 3.00 4.00 6.00 7.00
BJMEET 16.27 12.90 1.00 8.00 13.00 20.00
OUTSIDE 0.323 0.102 0.00 0.25 0.286 0.333
OACTIVITY 0.717 0.288 0.00 0.50 0.813 1.00
OEXPERT 0.094 0.243 0.00 0.00 0.00 0.00
COMMITTEE 0.163 0.370 0.00 0.00 0.00 0.00
CORPINDEX 0.599 0.525 -0.40 0.20 0.60 1.00
SIZE 19.33 1.167 16.84 18.44 19.14 20.07
LEVERAGE 0.476 0.206 0.101 0.320 0.467 0.615
ROA 0.056 0.071 -0.221 0.019 0.054 0.094
LOSS 0.151 0.358 0.000 0.000 0.000 0.000
OWNER 0.372 0.175 0.027 0.234 0.361 0.489
COMPEN 0.007 0.009 0.00001 0.002 0.004 0.009
OWNERDUMMY 0.281 0.449 0.000 0.000 0.000 1.000
Variables Max
RM Proxy 0.538
Ab.OCF 0.287
Ab.PRODcost 0.387
Ab.EXP 0.137
BSCALE 14.00
BJMEET 150.00
OUTSIDE 0.750
OACTIVITY 1.00
OEXPERT 1.00
COMMITTEE 1.00
CORPINDEX 1.80
SIZE 22.70
LEVERAGE 1.544
ROA 0.352
LOSS 1.000
OWNER 0.878
COMPEN 0.120
OWNERDUMMY 1.000
Table 3. Correlations
RM_Proxy BSCALE BMEET OUTSIDE OACTIVITY
RM Proxy -0.134 0.095 -0.108 -0.040
<0001 0.002 0.000 0.191
BSCALE 0.024 0.327 -0.053
0.427 <0001 0.079
BMEET -0.078 -0.214
0.010 <0001
OUTSIDE 0.150
<0001
OACTIVITY
OEXPERT
COMMITTEE
SIZE
LEVERAGE
ROA
OWNER
OEXPERT COMMITTEE SIZE LEVERAGE ROA
RM Proxy -0.043 -0.117 -0.133 0.251 -0.453
0.157 0.000 <0001 <0001 <0001
BSCALE -0.050 0.399 0.483 0.109 0.110
0.096 <0001 <0001 0.000 0.000
BMEET -0.071 -0.014 0.185 0.249 -0.028
0.018 0.639 <0001 <0001 0.346
OUTSIDE -0.007 0.519 0.391 0.113 0.026
0.814 <0001 <0001 0.000 0.395
OACTIVITY 0.036 0.120 0.139 -0.093 0.062
0.235 <0001 <0001 0.002 0.042
OEXPERT 0.058 -0.022 -0.063 0.012
0.054 0.465 0.035 0.685
COMMITTEE 0.438 0.127 0.072
<0001 <0001 0.017
SIZE 0.176 0.170
<0001 <0001
LEVERAGE -0.185
<0001
ROA
OWNER
OWNER COMPEN
RM Proxy -0.033 0.150
0.278 <0001
BSCALE -0.053 -0.055
0.082 0.071
BMEET -0.012 -0.034
0.698 0.261
OUTSIDE -0.092 -0.194
0.002 <0001
OACTIVITY 0.002 -0.091
0.942 0.003
OEXPERT 0.110 0.038
0.000 0.212
COMMITTEE -0.056 -0.165
0.064 <0001
SIZE -0.062 -0.429
0.041 <0001
LEVERAGE -0.082 0.056
0.007 0.066
ROA 0.045 -0.152
0.135 <0001
OWNER 0.022
0.467
Note: Pearson correlations based on two-tailed tests
Table 4. Regression of earnings management on corporate governance
(OLS regression)
[Model 1]
RMlt = [a.sub.0] + [a.sub.1] B_[SCALE.sub.it] + [a.sub.2]
B_[MEET.sub.it] + [a.sub.3][OUTSIDE.sub.it] +
[a.sub.4]O_[ACTIVITY.sub.it] + [a.sub.5]O_[EXPERT.sub.it] +
[a.sub.6][COMMITTEE.sub.it] +
[a.sub.7][SIZE.sub.it] + [a.sub.8][LEVERAGE.sub.it] +
[a.sub.9][ROA.sub.it] + [a.sub.10][OWNER.sub.it] +
[a.sub.n][COMPEN.sub.it] + [a.sub.12][OWNER_DUMMY.sub.it] +
[a.sub.13]IND_Dummy + [a.sub.14]YEAR_Dummy + [e.sub.it]
Coefficients Exp. sign Aggregate variable
RM_Proxy Ab.OCF
[a.sub.0] ? 0.020 (0.22) -0.255(-6.15 ***)
[a.sub.1] +/- -0.035(-2.19 **) -0.009(-1.32)
[a.sub.2] - 0.010(1.49) -0.007(-2.36 **)
[a.sub.3] - -0.114(-2.13 **) 0.046(1.96 **)
[a.sub.4] - 0.018(1.12) 0.0001(0.02)
[a.sub.5] - -0.018(-1.03) 0.0009(0.12)
[a.sub.6] - -0.027(-1.68 *) -0.007(-1.11)
[a.sub.7] -0.0003(-0.06) 0.015(6.32 ***)
[a.sub.8] + 0.152(6.56 ***) -0.086(-8.47 ***)
[a.sub.9] - -0.918(-13.7 ***) 0.403(13.7 ***)
[a.sub.10] - -0.002(-0.08) -0.011(-1.00)
[a.sub.11] + 1.121(2.05 **) 0.616(2.57 **)
[a.sub.12] - 0.002(0.21) -0.007(0.10)
[a.sub.13] Inc. Inc.
[a.sub.14] +/- Inc. Inc.
F 29.7 *** 34.9 ***
Adj. [R.sub.2] 24.5 27.7
N 1.104 1.104
Coefficients Individual variable
Ab.PROD cost Ab.EXP
[a.sub.0] -0.210(-3.18 ***) 0.023(1.59)
[a.sub.1] -0.037(-3.33 ***) 0.006(2.78 **)
[a.sub.2] 0.002(0.50) -0.0007(-0.65)
[a.sub.3] -0.059(-1.71 *) -0.001(-0.58)
[a.sub.4] 0.018(1.78 *) -0.000003(-0.00)
[a.sub.5] -0.019(-1.52) -0.001(-0.58)
[a.sub.6] -0.017(-1.70 *) 0.005(2.28 **)
[a.sub.7] 0.013(3.40 ***) -0.001(-2.07 **)
[a.sub.8] 0.062(3.81 ***) -0.003(-1.09)
[a.sub.9] -0.512(-10.8 ***) 0.002(0.23)
[a.sub.10] -0.007(-0.42) 0.005(1.41)
[a.sub.11] 1.321(3.44 ***) -0.416(-4.83 ***)
[a.sub.12] -0.003(-0.51) 0.001(0.97)
[a.sub.13] Inc. Inc.
[a.sub.14] Inc. Inc.
F 18.2 *** 4.64 ***
Adj. [R.sub.2] 16.3 3.95
N 1.104 1.104
Notes: ***: significant at the 1% level, **: significant at the 5%
level, *: significant at the 10% [RM.sub.it]: Individual (aggregated)
abnormal real activity of firm i in year t (Ab.OCF, Ab.PROD cost,
Ab.EXP, and RM Proxy)
Table 5. Results of audit committee inside or loss firm
Panel A: Results of using Corporate Governance Index
[Model 2]
RM_[Proxy.sub.it] = [a.sub.0] + [a.sub.1]CORP_[INDEX.sub.it] +
[a.sub.2][SIZE.sub.lt] + [a.sub.3][LEVERAGE.sub.it] +
[a.sub.4][ROA.sub.lt] + [a.sub.5][OWNER.sub.it] +
[a.sub.6][COMPEN.sub.it] + [a.sub.7][OWNER_[DUMMY.sub.it] + [a.sub.8]
IND_Dummy + [a.sub.9] YEAR_Dummy + [e.sub.it]
Coeffi- Exp. Full sample Audit committee
cients sign
[a.sub.0] ? 0.071(0.78) -0.390(-1.54)
[a.sub.1] - -0.036(-3.91 ***) -0.135(-3.50 **)
[a.sub.2] - -0.004(-1.02) 0.024(1.92 *)
[a.sub.3] - 0.149(6.56 ***) 0.073(1.12)
[a.sub.4] - -0.913(-13.6 ***) -1.074(-6.39 ***)
[a.sub.5] - -0.006(-0.24) 0.035(0.53)
[a.sub.6] - 1.003(1.87 *) 1.715(0.66)
[a.sub.7] - -0.0003(-0.03) -0.071(-2.51 **)
[a.sub.8] +/- Inc. Inc.
[a.sub.9] +/- Inc. Inc.
F 48.8 *** 10.9 ***
Adj. [R.sup.2] 24.0 28.1
N 1.104 180
Coeffi- No audit committee Profit firm
cients
[a.sub.0] 0.119(1.14) 0.078(0.78)
[a.sub.1] -0.020(-2.06 **) -0.030(-3.07 ***)
[a.sub.2] -0.008(-1.50) -0.005(-1.02)
[a.sub.3] 0.157(6.67 ***) 0.151(6.20 ***)
[a.sub.4] -0.869(-12.2 ***) -1.049(-11.4 **)
[a.sub.5] -0.002(-0.08) 0.017(0.62)
[a.sub.6] 0.833(1.56) 0.958(1.63)
[a.sub.7] 0.011(1.17) 0.002(0.25)
[a.sub.8] Inc. Inc.
[a.sub.9] Inc. Inc.
F 40.1 *** 34.9 ***
Adj. [R.sup.2] 23.4 20.6
N 923 937
Coeffi- Loss firm
cients
[a.sub.0] 0.340(1.62)
[a.sub.1] -0.056(-2.64 ***)
[a.sub.2] -0.013(-1.23)
[a.sub.3] 0.101(1.87 *)
[a.sub.4] -0.396(-1.82 *)
[a.sub.5] -0.141(-2.54 **)
[a.sub.6] -0.028(-0.02)
[a.sub.7] -0.017(-0.74)
[a.sub.8] Inc.
[a.sub.9] Inc.
F 4.28 ***
Adj. [R.sup.2] 12.7
N 167
Notes:
RM_proxytt. The sum of Ab.OCF, Ab.PROD cost, and Ab.EXP of firm i in
year t (we multiply Ab.OCF and Ab.EXP by -1 so that the sum of the
three variables will be indicative of overall real earnings
management);
CORP_[INDEX.sub.it]. The sum of B SCALE, B MEET, and OUTSIDE of firm i
in year t (we multiply B_MEET by -1so that the sum of the three
variables will be indicative of overall corporate governance)
Panel B: Results of existing an audit committee or reporting loss
[Model 3]
RM_[Proxy.sub.it] = [a.sub.0] + [a.sub.1]CORP_[INDEX.sub.it] +
[a.sub.2] [COMMITTEE.sub.it] + [a.sub.3] CORPJNDEX *
[COMMITTEE.sub.it] + [a.sub.4][SIZE.sub.it] +
[a.sub.5][LEVERAGE.sub.it] + [a.sub.6][ROA.sub.it] +
[a.sub.7][OWNER.sub.it] + [a.sub.8][COMPEN.sub.it] +
[a.sub.9]OWNER_[DUMMY.sub.it] + [a.sub.10]IND_Dummy +
[a.sub.11]YEAR_Dummy + [e.sub.it]
[Model 4]
RM_[Proxy.sub.it] = [a.sub.0] + [a.sub.1]CORP_[INDEX.sub.it] +
[a.sub.2][LOSS.sub.it] + [a.sub.3]CORP_INDEX * [LOSS.sub.it] +
[a.sub.4][SIZE.sub.it] + [a.sub.5][LEVERAGE.sub.it] +
[a.sub.6][ROA.sub.it] + [a.sub.7][OWNER.sub.it] +
[a.sub.8][COMPEN.sub.it] + [a.sub.9]OWNER_[DUMMY.sub.it] +
[a.sub.10]IND_Dummy + [a.sub.11] YEAR_Dummy + [e.sub.it]
Coefficients Exp. Audit committee [Model 4]
sign Loss firm
[a.sub.0] ? -0.002(-0.02) 0.085(0.95)
[a.sub.1] - -0.023(-2.30 **) -0.029(-3.48 ***)
[a.sub.2] - 0.039(1.09) -0.017(-0.89)
[a.sub.3] - -0.065(-1.89 *) -0.025(-1.13)
[a.sub.4] - -0.001(-0.25) -0.005(-1.15)
[a.sub.5] - 0.150(6.71 ***) 0.146(6.58 ***)
[a.sub.6] - -0.906(13.7 ***) -1.004(-11.8 ***)
[a.sub.7] -0.005(-0.23) 0.0001(0.01)
[a.sub.8] 0.984(1.86 *) 0.938(1.77 *)
[a.sub.9] 0.001(0.10) 0.001(0.18)
[a.sub.10] +/- Inc. Inc.
[a.sub.11] +/- Inc. Inc.
F 39.5 *** 39.7 ***
Adj. [R.sub.2] 24.4 24.5
N 1.104 1.104
Notes:
CORP_[INDEX.sub.it]. The sum of B SCALE, B MEET, and OUTSIDE of firm i
in year t (we multiply B_MEET by-1 so that the sum of the three
variables will be indicative of overall corporate governance);
[COMMITTEE.sub.it]: Indicator variable with a value of 1 if there is
an audit committee within firm i in year t, 0 otherwise;
[LOSS.sub.11]: Indicator variable with a value of 1 if firm i in year
t reports a loss, 0 otherwise; CORP_INDEX * [COMMITTEE.sub.it]. An
interaction term between CORP_INDEX and COMMITTEE; CORP_INDEX *
[LOSS.sub.it]. An interaction term between CORP_INDEX and LOSS
Table 6. Regression of earnings management on corporate governance
(2SLS)
[Model 5]
1st Stage. CORP_[GOV.sub.it] = [a.sub.0] + [a.sub.1]RM_[Proxy.sub.it] +
[a.sub.2S]IZE_[DUMMY.sub.it] + [a.sub.3] [SIZE.sub.it] + [e.sub.it]
2nd Stage. RM_[Proxy.sub.it] = [b.sub.0] + [b.sub.1]CORP_[GOV.sub.it] +
[b.sub.2][SIZE.sub.it] + [b.sub.3][LEVERAGE.sub.it] +
[b.sub.4][ROA.sub.it] + [b.sub.5][OWNER.sub.it] +
[b.sub.6][COMPEN.sub.it] + [b.sub.7]OWNER_[DUMMY.sub.it] +
[b.sub.8]IND_Dummy + [b.sub.9]YEAR_Dummy + [e.sub.it]
Coefficients Exp. B_SCALE B_MEET
sign
[b.sub.0] ? -0.896(-1.91 *) -0.022(0.16)
[b.sub.1] +/- -0.72(-2.37 **) 0.370(2.58 **)
[b.sub.2] - 0.106(2.13 **) -0.051(-3.3 ***)
[b.sub.3] - 0.163(4.29 ***) 0.066 (0.64)
[b.sub.4] - -0.73(-5.45 ***) -0.47(-4.57 ***)
[b.sub.5] - -0.041(-0.90) -0.004(-0.25)
[b.sub.6] - 6.128(2.56 **) 0.515(1.39)
[b.sub.7] +/- -0.011(-0.63) 0.0004(0.03)
[b.sub.8] +/- Inc. Inc.
[b.sub.9] +/- Inc. Inc.
F 17.3*** 16.6***
Adj. 9.63 9.25
[R.sub.2]
N 1.104 1.104
Coefficients OUTSIDE OACTIVITY O_EXPERT
[b.sub.0] 0.009(0.10) 1.046(0.43) 2.747(1.06)
[b.sub.1] -0.44(-3.97 ***) -20.9(-0.30) -13.73(-0.24)
[b.sub.2] 0.004(0.82) 0.762(0.27) -0.079(-0.29)
[b.sub.3] 0.160(7.06 ***) -1.908(-0.16) -0.197(-0.05)
[b.sub.4] -0.93(-13.9 ***) 2.382(1.36) 0.156(0.09)
[b.sub.5] -0.014(-0.55) 0.063(0.14) 0.365(0.04)
[b.sub.6] 0.864(1.61) -7.928(-0.19) 2.162(0.04)
[b.sub.7] 0.002(0.27) -0.029(-0.24) 0.100(0.04)
[b.sub.8] Inc. Inc. Inc.
[b.sub.9] Inc. Inc. Inc.
F 48.4 *** 0.09 0.13
Adj. 23.52 2.49 2.49
[R.sub.2]
N 1.104 1.104 167
Coefficients CORP_INDEX
[b.sub.0] -0.277(-1.78 *)
[b.sub.1] -0.18(-3.57 ***)
[b.sub.2] 0.017(1.81 *)
[b.sub.3] 0.142(6.16 ***)
[b.sub.4] -0.83(-11.75 ***)
[b.sub.5] -0.004(-0.20)
[b.sub.6] 0.907(1.74 *)
[b.sub.7] 0.005(0.59)
[b.sub.8] Inc.
[b.sub.9] Inc.
F 39.3 ***
Adj. 20.1
[R.sub.2]
N 1.104
Notes:
CORP_[GOV.sub.it]. Individual characteristic of corporate governance
or aggregate index; B_SCALE, B_MEET, OUTSIDE, O_ACTIVITY, O_EXPERT
and CORP_INDEX; CORP_INDEXU. The sum of B_SCALE, B_MEET, and OUTSIDE
of firm i in year t (we multiply B_MEET by -1 so that the sum of the
three variables will be indicative of overall corporate governance);
SIZE_[DUMMY.sub.it]: Indicator variable with a value of 1 if the total
assets of firm i in year t are equal to or above 2 trillion Won, 0
otherwise
