Industry Product Market Competition and Earnings Management

advertisement
Industry Product Market Competition and Earnings Management
Christo Karuna†
K R Subramanyam#
Feng Tian*
May, 2012
Abstract
This study examines the relation between industry product market competition and earnings
management in firms. Using several determinants of competition, namely, product substitutability,
market size, and entry costs, and a range of variables to proxy for accruals and real activities
manipulation, and accounting restatements that constitute material violations of accounting practices,
we find a robust positive relation between competition and earnings management. This study shows
that industry factors play a vital role in influencing the variation in extent of earnings management
across firms. In contrast to much research on competition, it also provides evidence on the perverse
effects of competition.
JEL classification: D4; G34; J33; L1; M40; M41; M46
Keywords: Managerial incentives; Competition; Corporate Governance; Earnings Quality, Earnings
Management
We thank Jeremy Michels (discussant at the 2011 American Accounting Association Annual Conference), Laura Li
(discussant at the 2012 American Accounting Association Financial Accounting and Reporting Section mid-year
conference), Gil Sadka, and workshop participants at the University of Hong Kong and University of Texas at San Antonio
for their comments and suggestions. All errors are our own.
†
Christo Karuna, C.T. Bauer College of Business, 390 Melcher Hall, University of Houston, Houston, TX 77004; Phone:
(713) 743-8953; Email: ckaruna@uh.edu
# K R Subramanyam, Marshall School of Business, University of Southern California, CA 90089; Phone: (213) 740-5017;
Email: krs@marshall.usc.edu
*Feng Tian, Faculty of Business and Economics, The University of Hong Kong, Hong Kong; Phone: (852) 2857-8348;
Email: ftian@hku.hk
1. Introduction
From Smith (1776), who remarks that “monopoly… is a great enemy to good management”, to
Caves (1980), who states that economists have a “vague suspicion that competition is the enemy of
sloth”, the economic literature has widely held the belief that industry product market competition
(hereafter competition) is a vital determinant of a nation’s economic well-being. Additional support to
this notion is provided by much of Porter’s work by showing that greater competition leads to higher
firm profitability (e.g., Porter, 1990).
The reasoning underlying this stream of research is that
competition promotes firm efficiency as firms in more competitive industries are encouraged, with
great employee effort, to differentiate their products from rivals’ or reduce costs to maintain a
competitive edge.
However, more recent studies suggest that the benefits of competition to a nation’s economy
may not extend beyond economic efficiency. For example, Milgrom and Roberts (1992) state that
competitive pressure aggravated the moral hazard problems in the savings and loan industry in the U.S.
by forcing executives in the industry to gamble on risky investments in order to survive. In another
study, Shleifer (2004) argues that competitive pressure results in unethical behavior in firms by
encouraging activities such as the employment of child labor and bribery/corruption, and Cummins and
Nyman (2005) show that competitive pressure leads firms to make inefficient decisions using their
private information to cater to their customers because the customers may take their business to a rival
firm that shares the same opinion as the customers.
In his study, Shleifer also claims that competitive pressure increases the likelihood of earnings
manipulation in firms.
This is because when competition is intense, managers are pressured to
manipulate earnings to influence prospective and incumbent investors by influencing the stock price
1
and to show good earnings numbers right before they make acquisitions. 1 Further, the greater
discretion in decision-making managers receive with greater competition gives them greater latitude to
engage in such opportunistic actions as their actions are less observable or understandable due to their
complex nature (Christie, Joye, and Watts, 2003; Kole and Lehn, 1997, 1999). In a more recent study,
Bagnoli and Watts (2010) more formally examine the relation between competition and earnings
management by showing how competition in an industry affects the costs versus benefits of earnings
management for firms that operate in the industry. Specifically, they show that a firm’s biased report
leaves its rivals rationally believing that the reporting firm’s costs are lower than they actually are,
which leads the rival to believe that the reporting firm will produce more than it actually does. As a
result, in a symmetric equilibrium, total output is less than the full-information level and so both firms
benefit.
Our study provides a direct test of the relation between competition and earnings management
in firms.
We hypothesize that earnings management in firms is more prevalent with greater
competition in the industry. To support our theoretical reasoning, we draw from prior studies that
show that competitive pressure leads to increased myopia in firms like delaying or cutting research and
development expenditure and other actions to increase short-term profitability. According to these
studies, competition leads to higher risk (Raith, 2003; Gaspar and Massa, 2006). In this case, riskaverse managers become more conservative by avoiding or delaying investment projects that
potentially yield cash flows in the future.
Greater competition also leads to a higher threat of
liquidation (Schmidt, 1997), which leads managers to increase equity and reduce debt. We posit that
this myopic orientation due to competitive pressure encourages the managers to engage in earnings
manipulation to influence the market’s perception of their short-term performance to attract financing,
firm survival and other reasons.
1
Anecdotal evidence for a relation between competition and earnings manipulation is provided by recent accounting
scandals like Enron and WorldCom – competitive pressure led to a range of earnings management activities in these firms.
2
To proxy for competition, we use several determinants of competition, namely, product
substitutability, market size, and entry costs, which have been used in recent studies to denote
competition (Raith, 2003; Karuna, 2007). Greater product substitutability, larger market size, and
lower entry costs denote more intense competition. We use a range of variables that have been
documented in prior research to denote earnings management: accruals quality measures based on
Dechow and Dichev (2002), discretionary accruals measures derived from the modified Jones model
(Dechow, Sloan, and Sweeney, 1995), and real activities manipulation measures that are used by
Roychowdhury (2006). Larger values of accruals quality and larger absolute values of discretionary
accruals and abnormal real activities are generally regarded in the literature as denoting greater
nondirectional earnings manipulation. However, it is arguable that these variables capture normal
operational managerial activities and not willful misrepresentation by the firm. To test whether our
findings are subject to such concerns, we also use two other variables, financial restatements triggered
by accounting irregularities (Hennes, Leone, and Miller, 2008) and Accounting and Auditing
Enforcement Releases (AAERs) due to accounting enforcement actions brought about by the Securities
and Exchange Commission (SEC) when a firm commits a violation, including intentional falsification
of financial statements (Dechow, Sloan, and Sweeney, 1995). Although firms restate earnings due to
“tolerable” reasons like changes in accounting policies, restatements that are irregularities constitute
clear and material violation of appropriate accounting practices as they represent a direct admission by
managers of misrepresentation (Agrawal and Chadha, 2005). Although many such restatements are
not criminally fraudulent in nature, such a measure is used as a proxy for fraud and a failure in ethics
(Staubus, 2005; Connor, Priem, Coombs, and Gilley, 2006).
Therefore, it is a good proxy for
intentional financial misrepresentation by managers. We further collect the AAERs issued by the SEC
to identify fraudulent financial reporting since the AAERs help avoid any coding biases from
3
researchers in determining earnings management and covers a longer sample period than commonly
used restatement data (Dechow et al. 2010).
We find robust evidence of a strong positive relation between competition and earnings
manipulation across our different competition and earnings management proxies. Specifically, we find
that firms in more competitive industries have lower earnings quality, larger absolute values of
discretionary accruals and abnormal real activities, and a higher likelihood of irregular accounting
restatements and SEC accounting enforcement actions, indicating greater earnings management and
willful financial misrepresentation. Intriguingly, we find a positive relation between industry size and
earnings management but a negative relation between firm size and earnings management. Our results
are also robust to controlling for industry fixed and random effects, which control for unobserved
industry factors that are constant for a given industry over time but vary across industries, and to using
competition variables constructed using economic census data collected from the U.S. Census Bureau.2
To more carefully address causality/endogeneity issues, we examine whether an exogenous increase in
competition via deregulation in the telecommunications industry in 1996 led to an increase in the
prevalence of earnings management. We find confirmatory evidence supporting a positive causal
relation between competition and earnings management.
Our study contributes to the literature in at least three ways. First, it sheds some light on the
debate as to whether excessive competition is beneficial or harmful to a nation’s economy. Much of
the earnings management literature (e.g., Dechow, Sloan, and Sweeney, 1996; Ronen, Tzur, and Yaari,
2006) assumes that earnings management constitutes financial misrepresentation and is therefore
harmful to a firm’s shareholders and other stakeholders. According to these studies, such corporate
2
Our primary analysis uses competition measures constructed using data collected from the Compustat database. However,
Compustat leaves out many firms in the industry, especially private firms (Karuna, 2007; Ali, Klasa, and Yeung, 2009), and
hence, our competition measures could inaccurately reflect the true competition level in the industry. In contrast, although
the Economic Census does not provide data on overhead and other capacity costs, leading to our measure of product
substitutability possibly being overestimated, and although it enables us to construct competition measures for only the
manufacturing sector, the Census data includes many more firms within an industry.
4
misconduct distorts the quality of reported earnings and is therefore misleading to the firm’s
stakeholders who rely on such information to make decisions. If earnings management is indeed
harmful to a firm’s stakeholders, then our findings provide evidence of the perverse effects of
excessive competition. In this sense, our study is one of a few studies (e.g., Cummins and Nyman,
2005) to demonstrate the “dark side” of competition. The belief that greater competition is beneficial
to a nation’s economy has been influencing policy making worldwide; some examples include trade
liberalization and deregulation in certain industries. Our finding that greater competition is associated
with a higher likelihood of earnings manipulation has policy implications for anti-trust authorities and
other governmental bodies by showing that a more careful examination of the consequences of
excessive competition has to be considered first before promoting greater competition in an economy.
This in turn suggests that a particular competition level may be optimal in a given industry (i.e., the
benefits of competition are context-specific) and that this level could vary across industries. In this
regard, our study complements recent studies that show that excessive competition may lead to less
innovation and efficiency. For example, Aghion, Bloom, Blundell, and Howitt (2005) find that while
competition increases innovation in firms, excessive competition decreases innovation. Similarly,
Green and Mayes (1991) and Schmidt (1997) show that excessive competition leads to lower firm
efficiency.
A contrasting premise is that earnings management in firms is an optimal response to the firms’
operational and environmental needs and is not harmful to a firm’s stakeholders. According to this
view, if earnings management were indeed harmful to firm stakeholders, and if it was detectable, then
these stakeholders would take necessary measures to prevent earnings management and we would not
observe such behavior by managers. If competition indeed promotes firm efficiency, then our findings
show that greater earnings management in firms is an optimal response to greater competition and is
tolerated by firms’ stakeholders.
5
Although it is debatable whether earnings management is harmful or beneficial to firms, as
discussed above, there is overwhelming evidence in the literature to support the notion that earnings
management is detrimental to a nation’s economy because both industry peer firms and firms engaging
in earnings management are forced to make sub-optimal investment and operational decisions that lead
to distorted economic resource allocation in the economy (e.g., Sadka, 2006; Kedia and Philippon,
2009; Durnev and Mangen, 2009). Kedia and Philippon (2009) document that firms that manipulate
earnings hire and invest too much during their manipulation period. Durnev and Mangen (2009)
further show that rival firms in an industry tend to make sub-optimal investment decisions due to
distorted signals about investment profitability they obtain from the misreported statements of other
firms in the industry that manage earnings. Overall, earnings management has significant negative
social welfare effects on the economy because low productivity firms produce too much relative to
high productivity firms (e.g., Sadka, 2006).
Second, our study contributes to the earnings management literature by showing that industry
attributes like competition play a vital role in influencing the likelihood of earnings management
across firms. Key (1997) provides evidence of income-decreasing earnings management in the Cable
television industry. Our study provides comprehensive evidence on earnings management across
industries by showing that varying competition levels across industries provide an explanation for the
observed heterogeneity in earnings quality, discretionary accruals, abnormal real activities, irregular
accounting restatements, and SEC accounting enforcement actions across firms. Much prior research
has focused on firm-level determinants of earnings management, while merely controlling for industry
factors by including industry indicator variables in regressions. Our study shows that researchers need
to consider higher order determinants of earnings management beyond firm-level attributes.
Finally, our robust evidence of a positive competition-earnings management relation across the
plethora of earnings management proxies we use provides additional validation that these measures are
6
suitable proxies of earnings management.
By showing that competition is related to earnings
management, our study also shows that earnings management is an important consideration in a
nation’s welfare, especially in light of incidences of widespread earnings management in several
companies in recent times.
The rest of the paper is organized as follows. The next section discusses the theoretical
background for our study. Section 3 provides the sample selection procedure and measures for our
study while Section 4 outlines our methodology. We discuss our results in Section 5 and robustness
checks in Section 6. Section 7 concludes this study.
2. Theoretical development
The theoretical literature is inconclusive on whether competition is beneficial or harmful to
firms. Some studies show that competition disciplines managerial behavior by providing additional
information for performance evaluation (e.g., Hart, 1983; Nalebuff and Stiglitz, 1983). Specifically, if
industry shocks affecting each firm’s costs are correlated, then an increase in competition generates
additional information which the firms’ owners can use to evaluate their managers and mitigate moral
hazard problems. Other studies show that greater competition leads to a higher takeover threat (e.g.,
Kole and Lehn, 1997; 1999) and a higher threat of a firm’s liquidation if the firm faces managerial
slack and high costs (e.g., Schmidt, 1997). Here, managers are motivated (disciplined) to engage in
beneficial actions to avoid liquidation, takeover, loss of profitability or market share, and to protect
their jobs.
In contrast, another stream of research shows that competition increases managerial slack
(Scharfstein, 1988).
For example, some studies predict that greater competition discourages
innovation/productivity growth by reducing the monopoly rents that reward new innovation and hence
managerial effort pertaining to such activities (e.g., Dixit and Stiglitz, 1977; Grossman and Helpman,
1991; Aghion and Howitt, 1992). This ambiguity is also seen in the empirical literature with some
7
studies finding that competition increases innovation (Nickell, 1996; Blundell, Griffith, and Reenen,
1999) whereas others show that excessive competition could hinder innovation (e.g., Scherer, 1967;
Aghion et al., 2005).
The above discussion is based on how competition affects managerial behavior pertaining to
economic productivity.
The impact of competition on managerial behavior beyond economic
productivity has received much less attention in the literature. Recent studies allude to the perverse
effects of competition (e.g., Milgrom and Roberts, 1992; Shleifer, 2004). For example, Milgrom and
Roberts argue that the design of the deposit insurance program and lax regulation led to a moral hazard
problem in the management of savings and loans in the Savings and Loan industry. 3 According to
Milgrom and Roberts, the risk taking and fraudulent behavior that ensued was exacerbated by intense
competition in this industry in the 1980s. To survive the excessive competition in this industry, many
executives, including conservative ones, had to gamble on risky investments to mimic their less risk
averse rivals in the industry and exploit the deposit insurance system. In their enthusiasm to attract
substantial new deposits, some S&Ls offered higher interest rates than their competitors. Risky
investments were necessary for some firms to meet the promised higher interest rate payment to
depositors and make a profit, given operating costs. In the end, competitive pressure forced out of the
industry many conservative S&Ls who were not prepared to make the risky investments and the
biggest losers were the FSLIC and the taxpayer.
More recently, Shleifer (2004) argues that more intense competition leads to unethical/criminal
behavior in firms.4 Under the assumption that the proprietor of the firm values ethical behavior, but
that such behavior is a normal good, Shleifer posits that the proprietor’s demand for ethical behavior
3
A savings and loan association (S&L) is a for-profit financial institution that borrows money from the public in the form
of deposits and then invests it by lending it out again, like banks. A U.S. federal government agency (until 1990, the
Federal Savings and Loan Insurance Corporation - FSLIC) insures the deposits of individual depositors by repaying the
deposits in the event that the savings and loan association couldn’t.
4
The economics of crime, based on insights in Becker (1968), stipulates that an individual can rationally indulge in crime
depending on the benefits and costs involved. A link between competitive pressure and a detrimental impact on corporate
conduct is also documented in the sociology literature (e.g., Coleman, 1987).
8
decreases with more intense competition due to a reduction in his/her income. For example, Shleifer
illustrates that the demand for child labor increases with competition due to the lower costs associated
with such labor. If a given firm hires children, its rivals in the industry is also compelled to hire
children to stay competitive or be forced out of business. In the event that a firm chooses not to hire
children even though competition increases, its willingness to pay for not hiring children declines
when competition becomes intense and profits are drastically reduced. In another illustration of the
effect of competition on unethical behavior, Shleifer reasons that if a firm is able to reduce taxes or
some other necessary payment via bribery or corruption, its rival in the industry is also compelled to do
the same or face the prospect of going out of business. Even here, the proprietor’s willingness to pay
for ethical conduct declines as his/her profits do with more intense competition, leading him/her to
engage in bribery/corruption.
In another recent study, Cummins and Nyman (2005) analytically show that competitive
pressure can cause firms to “cater” to their customers when the firms are better informed than their
customers, leading the firms to not utilize their private information efficiently. For example, when
faced with selecting between two types of firms (e.g., financial versus retail firms) to invest in for her
client, an investment manager may invest in one type of firm, say retail, because her client is of the
opinion that the retail firm is a better investment. Although, based on her private information, the
financial firm is a better investment to maximize her client’s expected return, the investment manager
is pressured to make this inefficient investment to please her client who may take her business to a
rival in the industry who shares the same opinion as the client. The likelihood of losing the client is
greater with greater competition, say if there are more rival firms in the industry for the client to
choose from.
The preceding discussion on the effects of competition on undesirable behavior can be
extended to earnings manipulation in firms. Although it is arguable that earnings management is a
9
necessary part of running a business and therefore beneficial to shareholders in some firms,5 there is
growing evidence that earnings are being manipulated to consciously manipulate shareholder opinion.
For example, Bergstresser and Philippon (2006) find that executives manipulate earnings to show good
earnings numbers just before the executives exercise their stock options. Shleifer (2004) argues that
competitive pressure leads to aggressive accounting practices, including earnings management, in
firms. With greater competition, managers are pressured to manipulate earnings to drive up share
prices.6 Higher share value reduces the cost of capital, enabling managers to make acquisitions for
stock, and also to attract better executives and workers with stock options, or even to issue new shares,
all being important for firms to sustain a competitive edge or survive. Moreover, Linck, Netter, and
Shu (2010) suggest that firms can ease financial constraints and gain access to external funds by
managing earnings, thereby reducing costs of investments and increasing firm value. If competition is
high in an industry, a firm with lower costs of capital can better price products and drive its
competitors out of the industry.
Gerety and Lehn (1997) provide evidence supporting the notion that external market forces
shape corporate activities like earnings management more than internal firm structures. In their study,
they find that accounting fraud appears to fool the market in the short term. However, they also find
that announcement of SEC charges of accounting fraud result in cumulative abnormal returns of 3.05% during a 3 day window surrounding these announcements.
Another finding in Gerety and Lehn’s study is that greater uncertainty/complexity across
industries increases the probability of accounting fraud.
observability
and
understandability
uncertainty/complexity.
of
managerial
They attribute this to the reduced
actions
in
settings
with
greater
Prior studies show that greater competition is associated with greater
complexity of managerial actions (e.g., Kole and Lehn, 1997; 1999).
5
Managers receive greater
Demski (1998) shows that earnings management can be optimal in some settings.
Worldcom’s senior executives encouraged and sustained a culture of opportunistic behavior, including earnings
management, due to competitive pressure, among other reasons.
6
10
discretion to engage in complex actions to attain/sustain a competitive advantage in the industry
(Christie et al., 2003). The preceding discussion suggests that managers in more competitive industries
are more likely to engage in earnings management due to the greater latitude in decision making they
have and because they perceive that their actions are not easily observable or understandable, thus
making earnings management less detectable.
A recent theoretical study by Bagnoli and Watts (2010) more directly examines the relation
between competition and earnings management by showing how competition in an industry affects the
costs versus benefits of earnings management for firms that operate in the industry. Specifically,
Bagnoli and Watts (2010) show that a firm’s biased report leaves its rivals rationally believing that the
reporting firm’s costs are lower than they actually are, which leads the rival to believe that the
reporting firm will produce more than it actually does. As a result, in a symmetric equilibrium, total
output is less than the full-information level and so both firms benefit.7 A possible result is that a firm
is encouraged to manage earnings if it believes that its rivals in the industry are also managing earnings.
This behavior is exacerbated when competition is greater in the industry, thus leading to firms in the
industry managing earnings on average.
Our study tests the hypothesis that greater competition leads to greater earnings manipulation.8
We posit that competitive pressure encourages the managers to be myopic and thus manipulate
earnings to boost short-term performance. Increased short-term performance is beneficial for several
reasons, namely, financing by banks and other lenders, firm survival, and other short-term career
prospects. According to Stein (1988; 1989) and Von Thadden (1995), since the true value of a firm’s
7
Bagnoli and Watt’s study is motivated by the then CEO of AT&T, Michael Armstrong’s, assertion that accounting fraud
at Worldcom led to AT&T’s strategic failures, inability to compete with Worldcom, and the decision to break up the
company, in addition to layoffs, cost-cutting, and other ill-fated corporate decisions. William Estry, Sprint’s CEO at that
time, also echoed the same sentiment as AT&T’s CEO.
8
Although, as discussed above, it is conceivable that greater competition could discipline managers into working harder
due to providing additional information for managerial performance evaluation, a higher threat of takeover and/or
liquidation, we argue that these forces exacerbate than prevent earnings manipulation. Further, as shown in Bagnoli and
Watts (2000), if competitive forces in the industry encourage a firm in the industry to manage earnings, its rivals would also
manage earnings even though there is enhanced comparability of information for performance evaluation purposes. This is
because the firm’s rivals perceive that the firm is also managing earnings and feel the pressure to do the same.
11
assets cannot be known to the market and since a firm’s current earnings are positively correlated with
the true value of its assets, the firm will try to boost current earnings in order to increase its stockmarket value or to avoid being cut off from further financing. According to Shleifer and Vishny
(1990), firms have short-term horizons because long-term assets are more likely to be mispriced
because the arbitrage cost for long-term assets is higher than for short-term assets. Risk-averse
managers would therefore choose short-term projects in order to avoid the underpricing of their equity.
Several studies show that competitive pressure encourages managers to engage in short-term
actions to improve short-term performance. This myopic behavior is due to several reasons. One is
based on competition leading to higher risk (Raith, 2003; Gaspar and Massa, 2006). In this case, riskaverse managers become more conservative by delaying investment projects that potentially yield cash
flows in the future. Greater competition also leads to reduced market power and declining levels of
profitability/cash flow (Tirole, 2006) and a higher threat of liquidation for firms (Schmidt, 1997).
Mackay and Phillips (2005) and Xu (2012) find evidence that greater competition leads to less debt
usage in firms because the tax benefit of debt is reduce due to lower expected cash flows. For
example, Xu finds that competitive pressure leads to firms issuing equity and selling assets to reduce
debt. In a more recent study, Fresard and Valta (2012) find evidence that competitive pressure leads to
firms reducing capital and R&D investment, increasing cash reserves and equity, and decreasing debt.
Then, managers are likely to focus on short-term actions aimed at enhancing their firms’ survival
prospects in the industry and/or influencing financing to increase liquidity (Feinberg, 1995) and/or to
react quickly to enhance or sustain competitive advantage (Christie et al., 2003)
Earnings management provides an important avenue for managers to influence short-term firm
performance. Several studies have shown that firms manage earnings to survive in the short-run. For
example, Teoh, Welch, and Wong (1998a; 1998b) find that firms offer higher prices in initial public
offerings and seasoned equity offerings, respectively, by manipulating pre-offer earnings.
12
Other
studies show that firms manage their reported earnings around other corporate events like management
buyouts (Perry and Williams, 1994) and stock-for-stock mergers (Erickson and Wang, 1999; Louis,
2004). A theoretical study by Chen (1994) shows that firms face a dilemma when deciding whether to
reveal information related to investment opportunities, technological expertise, or simply business
plans. While, from a capital market perspective, it is beneficial for firms to reveal such information,
from a product market perspective, it could be harmful to reveal such information to rivals. Thus, the
future profitability of a firm’s assets is no longer exogenous, but depends on the strategic actions of its
competitors, which in turn depend on what information the firm reveals. When the firm improves the
long-term quality of its assets, it has to face the dilemma of either bearing diminished returns on the
product market if it reveals the information or being undervalued on the stock market if it does not.
This motivates the firm to switch resources from future prospects to short-term profits.
Much research has shown that managers manage accruals to manipulate reported earnings (e.g.,
Paul M. Healy, 1985; Dechow et al., 1995). Managers have more discretion over short-term than longterm accruals (Guenther, 1994). For example, current accrual adjustments involve short-term assets
and liabilities which support the day-to-day operations of the firm. Managers can increase current
accruals by, among other ways, advancing recognition of revenues with credit sales (before cash is
received), by delaying recognition of expenses through assumption of a low provision for bad debts, or
by deferring recognition of expenses when cash is advanced to suppliers. Based on the discussion
above, we hypothesize that firms in more competitive industries will use accruals to manage earnings
more intensively than firms in less competitive industries. This leads to our study’s first hypothesis
(stated in alternate form):
H1: Firms in more competitive industries engage in greater accruals manipulation to influence
reported earnings than firms in less competitive industries, ceteris paribus.
13
Some studies show that firms manipulate real activities to influence reported earnings. For
example, Roychowdhury (2006) shows that managers manipulate earnings to avoid losses through real
activities like offering price discounts to temporarily increase sales, engaging in overproduction to
lower cost of goods sold, and reducing discretionary expenditures aggressively to improve short-term
earnings. He also finds that industry factors (e.g. manufacturing versus nonmanufacturing) influence
the extent of such earnings manipulation. Furthermore, Graham, Harvey and Rajgopal’s (2005) survey
finds that executives are willing to manipulate real activities to meet earnings targets, even though the
manipulation reduces firm value.
Real activities manipulation is less detectable than accruals
manipulation because it is difficult for external auditors or outsiders to challenge the validity of
managerial operational decisions. We posit that competitive pressure in an industry may also induce
these executives to engage in real activities manipulation. This leads to the second hypothesis for our
study:
H2: Firms in more competitive industries engage in greater real activities manipulation to influence
earnings than firms in less competitive industries, ceteris paribus.
By nature, earnings management consisting of accruals and real activities manipulation is
difficult to detect and measure, and correlates with operational activities in a firm. Therefore, using
these forms of earnings management, it is empirically challenging to tease out the portion of deliberate
earnings management that arises due to competitive pressure. To more precisely determine whether
greater competition leads to deliberate financial misconduct via earnings manipulation, we examine the
relation between competition and financial restatements that are triggered by accounting irregularities
(Hennes et al., 2008). The reason is that if firms manage earnings aggressively, they have a greater
14
chance to restate earnings later than firms that do not manipulate earnings. Although firms frequently
restate earnings due to reasons like changes in accounting policies, restatements that are irregularities
constitute clear and material violation of appropriate accounting practices as they represent a direct
admission by managers of misrepresentation (Agrawal and Chadha, 2005). We postulate that greater
competition leads to a higher proportion of financial restatements that are labeled as accounting
irregularities, leading to our third hypothesis:
H3: Firms in more competitive industries are more likely to have financial restatements that constitute
accounting irregularities than firms in less competitive industries, ceteris paribus.
Similarly, if greater competitive pressure gives managers greater incentives or latitude to
commit fraud, then managers are more likely to engage in outright accounting fraud in more
competitive industries.
SEC enforcement actions against firms for material violations of GAAP
constitute instances of accounting fraud. We collect the AAERs issued by the SEC to identify these
accounting fraud cases. Therefore, we hypothesize that firms in more competitive industries are more
likely to have accounting fraud that is identified in the SEC enforcement actions as follows:
H4: Firms in more competitive industries are more likely to have accounting fraud identified in the
SEC enforcement actions against them than firms in less competitive industries, ceteris paribus.
To summarize, this section provides a discussion on the relation between competition and the
likelihood of earnings manipulation. We hypothesize that firms in more competitive industries are
more likely to manipulate earnings through accruals and real activities, and are more likely to have
financial restatements that comprise accounting irregularities and commit outright accounting fraud
15
than firms in less competitive industries.
In the next section, we discuss our sample selection
procedure and construction of measures we use to test our hypotheses.
3. Sample selection and empirical measures
3.1. Sample
For our analyses, we primarily use data obtained from Standard and Poor’s Compustat database
between 1992 and 2003. To be included in the sample, data must be complete across the Segments and
Annual Industrial databases in Compustat. Further, to be included in the final sample, firms must have
identical four-digit Standard Industrial Classification (SIC) codes across all the databases for
observations. For accounting irregularities data, we use the General Accounting Office (GAO) 2006
version, as in Hennes et al. (2008). Finally, we collect accounting enforcements data from the SEC’s
Accounting and Auditing Enforcement Releases (AAERs) to identify accounting fraud. Following
prior studies (e.g., Subramanyam, 1996; McNichols, 2002; Roychowdhury, 2006), we exclude
financial industries in our analysis. For our analyses, we compute average values for all our variables
for each firm over our sample period.9,10 Our final sample comprises 8,930 firms (observations) for the
data we obtain from Compustat and 7,743 firms (observations) for the data we obtain from GAO.
Table 1 presents the descriptive statistics for our sample. Overall, our sample descriptive statistics are
consistent with those reported by prior studies (e.g., Dechow and Dichev, 2002; Karuna, 2007). Table
2 reports correlations among the independent variables.11 The correlations are generally consistent
9
The reason that we compute averages for our variables across the sample period is that, consistent with prior studies in the
Industrial Organizations literature (e.g., Sutton, 1991), we assume that product market competition has attained its long run
equilibrium during our sample period and should have a persistent effect on earnings management. Therefore, if we average
all earnings-management estimates across the sample period for each firm, we can eliminate noise associated with earningsmanagement measures in any single year and increase the likelihood that we will find out the true underlying relation
between product market competition and earnings management. Doing this also minimizes concerns of artificially inflated t
statistics from having repeated observations for a firm over time.
10
For several earnings management measures discussed later, we use time-series models to estimate them over the sample
period.
11
Consistent with Karuna (2007), we do not provide correlations between our dependent variables and the independent
variables because it is necessary that we examine the effect of each of our competition variables on our earnings
16
with prior studies (e.g., Karuna, 2007; Hribar and Nichols, 2007).
Cash flow volatility
(AVG_CFO_VOLATILITY) and revenue volatility (AVG_REVENUE_VOLATILITY) are highly
correlated, which is expected because both measures are designed to measure operational uncertainties.
In untabulated analyses, we include these two variables separately in our regressions and find that our
results remain the same. With the exception of these two variables, the other variables are not highly
correlated with each other.
3.2 Measures of earnings management
Based on prior research (e.g., Dechow et al., 1995; Dechow and Dichev, 2002; Roychowdhury,
2006; Hennes et al., 2008), we use four comprehensive sets of earnings management measures in our
analyses to capture accruals and real activities manipulations, as well as financial restatements that
constitute irregularities, to capture different dimensions of earnings management. Based on Dechow
and Dichev’s (2002) model, which measures how current accruals map into past, present, and future
cash flows, we construct our first set of earnings management measures to capture accruals quality.
We refer to these as AQ measures in our study. Based on the modified Jones model, where total
accruals are a function of property, plant and equipment, and change in sales with adjustment of
change in accounts receivable (e.g., Jones, 1991; Dechow et al., 1995), we estimate our second set of
earnings management proxies for nondirectional earnings management.
measures.
We label these as DA
The AQ (accruals quality) and DA (discretionary accruals) measures are proxies for
accruals management.
Our third set of measures capture real activities management as in
Roychowdhury (2006). We refer to these as RM measures. Our final set of earnings management
measures comprises financial restatements that are accounting irregularities.
These accounting
irregularities represent a clear and willful misrepresentation of financial statements (Hennes et al.,
management measures controlling for the effect of the other competition variables, and especially industry concentration
(Sutton, 1991). Due to the effect of the industry-level variables on one another, univariate associations between our
earnings management variables and the competition variables could be misleading.
17
2008). A detailed discussion of the earnings management measures we use in our study is provided
below.
3.2.1. Accrual quality (AQ) measures:
We follow Dechow and Dichev’s (2002) study to construct our AQ measures. Dechow and
Dichev argue that the quality of accruals, and therefore earnings, is decreasing in the magnitude of
estimation errors (intentional or unintentional) in accruals and derive empirical AQ measures as the
residuals from firm-specific regressions of current accruals on past, present, and future cash flows.
Thus, larger values of our AQ measures capture lower accruals quality. Following Dechow and
Dichev’s (2002) model, we estimate AQ_TS01, our first AQ measure, as follows:
Similar to Dechow and Dichev (2002), we first run regressions for each firm j that has been in
existence for at least eight years:
TCAi,t  0  1CFOi,t 1  2 CFOi,t  3CFOi,t 1  i,t
(1)
where:
TCA = total current accruals and CFO = cash flows from operations.
We scale all variables by
beginning-of-year total assets.
Second, we estimate the standard deviation of the residual (νjt) for firm j and obtain a value of
AQ_TS01 for firm j over the sample period. In addition, we estimate the model’s Root Mean Squared
Error (RMSE), which is an unbiased estimator of the standard deviation for residuals (νjt), to obtain a
different AQ measure: AQ_TS01_RMSE (Kutner, Nachtsheim, Neter, and Li, 2005).
Dechow and Dichev’s (2002) model assumes that cash flows realizations related to current
accruals occur within one year. We further relax this assumption to construct our third AQ measure
and take into consideration that cash flows realizations associated with current accruals may need more
than one year to be realized. We first expand Dechow and Dichev’s original model as follows:
18
TCAi,t  0  1CFOi,t 2  2 CFOi,t 1  3CFOi,t  4 CFOi,t 1  5CFOi,t 2  i,t
(2)
With a procedure similar to that used for AQ_TS01, we estimate our third and fourth AQ
measures, AQ_TS02, which is the standard deviation of the residuals from Equation (2), and
AQ_TS02_RMSE, which is the model’s RMSE from estimations of Equation (2).
Studies in accounting further modify Dechow and Dichev’s time-series model into a crosssectional model, because it then can reserve more data observations in samples for analyses without
requiring at least eight years of observations for each firm; it also can increase the model’s explanatory
power (e.g., McNichols, 2002; Francis, LaFond, Olsson, and Schipper, 2005). To estimate crosssectional AQ measures, we follow prior studies (e.g., Francis et al., 2005; Ashbaugh-Skaife, Collins,
Kinney, and LaFond, 2008) to run Equations (1) and (2) cross-sectionally each year using all firm-year
observations in the same industry group and then calculate the standard deviations of the residuals. 12
These standard deviations are estimated accruals quality (AQjt) for firm j at year t using firm j’s
estimation of residuals from industry-year group regressions of year t-4 to year t. Next, we average
AQjt for each firm across the whole sample period to obtain our cross-sectional AQ measures 13 :
AQ_CS01, which is based on Equation (1), and AQ_CS02, which is based on Equation (2).
The cross-sectional model may give us a larger sample, but we run the regression crosssectionally each year using all firm-year observations in the same industry group to infer whether a
firm’s reported earnings deviate from the normal earnings that are unmanaged, and thus it is possible
that unobserved industry characteristics other than competition may influence the relation between
product-market competition and earnings management. Therefore, we use AQ measures estimated
12
Specifically, we first group firm-year observations by the three-digit SIC code industry group each year provided the
group has at least 20 observations. If the industry group has less than 20 observations, we keep all the firm-year
observations in this group and combine them with other three-digit SIC industry groups that do not have at least 20
observations under the same two-digit SIC industry grouping in the same year. We then determine whether the combined
two-digit SIC industry group has at least 20 observations. If it does not, we repeat the above procedure and move up to the
one-digit SIC group. In this way, we maximize our sample size. In untabulated results, we also use the Fama-French 48industry classification or two-digit SIC industry classification and obtain similar results through this paper.
13
Since the estimation of cross-sectional AQ measures does not require information from Segments data, we also use
Compustat Annual data from 1988 to 1992 to estimate the value of AQ in 1992.
19
from both time-series and cross-sectional Dechow-Dichev models to ensure that our results are not
sensitive to such a concern. In addition, as discussed later, we conduct industry random effects
regressions to rule out such a possibility.
3.2.2. Discretionary accruals (DA) measures
We construct our DA measures based on variations of the modified Jones’ model, which
assumes that the normal level of accruals are a function of property, plant and equipment, and change
in sales with adjustment of credit revenue changes (e.g., Jones, 1991; Dechow et al., 1995; Kothari,
Leone, and Wasley, 2005). Below we provide details of the estimations.
Following prior studies (e.g., Jones, 1991; Dechow et al., 1995), we estimate a time-series
modified Jones model for each firm, as follows:
TAccit   0  1 (1/ Assetsi,t 1 )   2 (Sales it  ARit )   3 PPEit  it
(3)
where:
TAccit = EBEIit - (CFOit – EIDOit),
TAccit = firm i’s total accruals in year t;
EBEIit = firm i’s income before extraordinary items in year t;
CFOit = firm i’s cash flows from operations in year t;
EIDOit = firm i’s extraordinary items and discontinued operations included in CFOit in year t.
Assetsi,t-1 = firm i’s total assets in year t-1;
∆Salesit = change in firm i’s sales from year t-1 to year t;
∆ARit = change in firm i’s accounts receivable from operating activities from year t-1 to year t;
PPEit = firm i’s gross property, plant, and equipment in year t.
20
Consistent with Kothari et al. (2005), we include ∆ARit in the estimation because we have no
prior knowledge to distinguish between nonearnings management firms and earnings management
firms.
In addition, we exclude firms with less than 11 observations, following Jones’s (1991)
procedure. We scale all variables by beginning-of-year total assets. We then estimate the RMSE from
each individual regression as the nondirectional earnings management measure DA_RMSE.14
Kothari et al. (2005) suggest that the modified Jones model can be more reliable to infer
earnings management if firms’ performance is taken into consideration. Therefore, we further include
return on assets (ROA) in the model to control for performance and generate our second DA measure,
DA_RMSE_ROA.
Since the cross-sectional modified Jones model is widely used in the literature, to determine the
robustness of our results to different specifications of the modified Jones model, we also estimate the
cross-sectional modified Jones Model (Equation (3)) each year using all firm year observations by each
SIC industry group and calculate the absolute value of discretionary accruals as DA_CMJ. 15, 16 As
discussed previously, controlling firm performance can ensure the reliability of DA measures to infer
earnings management. Therefore, we further include ROA in the cross-sectional model and have a
different DA measure, DA_CMJ_ROA. Moreover, following prior studies (e.g., Kothari et al., 2005;
Li, Pincus, and Rego, 2008), we conduct a performance-matching procedure and have our
performance-matched DA measure: DA_PMMJ, which is the absolute value of performance-matched
DAs for firm i in year t. In our study, larger values of our DA measures capture greater earnings
management.
3.2.3. Real activities management (RM) measures
14
We also estimate the standard deviations of the model residuals as DA measures, and our results are similar.
Similar to the procedures for the estimation of AQ measures, we estimate equation (3) by three-, two-, or one digit SIC
codes, conditional on having at least 20 firms in each SIC group to obtain our cross-sectional DA measures.
16
Since we examine nondirectional earnings management, we follow previous studies (e.g., Klein, 2002) and take the
absolute value of these DA measures from the cross-sectional modified Jones models.
15
21
The set of earnings management measures in this section are designed to capture three types of
real-activities management: over- or under-production, discretionary expenditure (e.g., R&D)
management, and sales manipulation. Based on Roychowdhury (2006), we conduct cross-sectional
regressions to compute real activities management measures. Similar to how we estimate time-series
AQ and DA measures, we modify Roychowdhury’s model and use the time-series observations from
each individual firm to estimate real activities management measures. Specifically, we compute
abnormal levels of production, discretionary expenditure, and cash flows from operations to proxy for
our real activities management measures.17 The estimation methods for these measures are discussed
below:
Abnormal production costs:
For the time-series measure of abnormal production
(PROD_ABNORMAL_RMSE), we first estimate the following regression for each firm:
PRODit / Asseti ,t 1   0  1 (1 / Asseti ,t 1 )  1 (Sales i ,t / Asseti ,t 1 )
  2 (Sales it / Asseti ,t 1 )   3 (Sales i ,t 1 / Asseti ,t 1 )   i ,t
(4)
where:
PRODit = firm i’s production costs in year t, which equals cost of goods sold (COGS) +
change
in inventory;
Asseti,t-1 = firm i’s total assets in year t-1;
Salesit = firm i’s sales in year t; and
Δ Salesit = firm i’s change in sales from year t-1 to year t.
Then, we
obtain the RMSE of
regression
for
each firm
observation, which
PROD_ABNORMAL_RMSE.
17
As discussed by Roychowdhury (2006), when managers attempt to temporarily manipulate sales during the year by
offering price discounts or changing credit terms, the abnormal levels of cash flows relative to sales and production costs
will be higher.
22
is
To compute the cross-sectional measure of abnormal production (AVG_PROD_ABS), we first
estimate the regression above (i.e., Equation (4)) cross-sectionally each year by each SIC industry.
The three-, two-, or one-digit SIC codes are used to identify an industry, conditional on having at least
20 firms in each SIC group. We obtain the residual for firm i in year t, which is the value of the
abnormal production cost for firm i in year t.
We then average each year’s absolute value of the abnormal production cost across the whole
sample period to obtain a value of AVG_PROD_ABS for a given firm i.
Abnormal discretionary expenses: For the time-series measure of abnormal discretionary
expenditure (DISEXP_RMSE), we first estimate the following regression for each firm:
DISEXPt / Assett 1   0  1 (1 / Assett 1 )  1 (Sales t 1 / Assett 1 )   t
(5)
Where:
DISEXP = discretionary expenses (R&D+ advertising
+ selling, general, & administrative
expenses (SG&A); as long as SG&A is available, advertising and R&D are set to zero if they are
missing;
Asset = total assets; and
Sales = sales.
We scale all variables by beginning-of-year total assets.
Then, we obtain the RMSE of the regression for each firm observation, which is DISEXP_RMSE.
To
compute
the
cross-sectional
measure
of
abnormal
discretionary
expenditure
(AVG_DISEXP_ABS), we first estimate the regression above (i.e., Equation (5)) cross-sectionally each
year by each SIC industry. The three-, two-, or one-digit SIC codes are used to identify an industry,
23
conditional on having at least 20 firms in each SIC group. We obtain the residual for firm i in year t,
which is the value of the discretionary expense for firm i in year t.
We then average each year’s absolute value of discretionary expenditure across the whole
sample period to obtain a value of AVG_DISEXP_ABS for a given firm i.
Abnormal cash flows from operations: For the time-series measure of abnormal cash flows
from operations (CFO_ABNORMAL_RMSE), we first estimate the following regression for each firm:
CFOit / Asseti ,t 1   0  1 (1 / Asseti ,t 1 )  1 (Sales i ,t / Asseti ,t 1 )
  2 (Sales it / Assetit 1 )   it
(6)
Where:
CFOit = firm i’ cash flow from operations in year t;
Asseti,t-1 =firm i’s total assets in year t-1;
Salesit = firm i’s sales in year t;
Δ Salesit = firm i’s change in sales from year t-1 to year t.
Then, we obtain the RMSE of the regression for each firm observation to compute
CFO_ABNORMAL_RMSE.
To compute the cross-sectional measure of abnormal cash flows from operations
(AVG_CFO_ABS), we first estimate the regression above cross-sectionally each year by each SIC
industry (i.e., Equation (6). The three, two, or one-digit SIC codes are used to identify an industry,
conditional on having at least 20 firms in each SIC group (please refer to footnote 9 for details). We
obtain the residual for firm i in year t, which is the value of the abnormal cash flows for firm i in year t.
Then, we average each year’s absolute value of abnormal cash flows across the whole sample period to
obtain a value of AVG_CFO_ABS for a given firm i.
24
3.2.4. Restatements and AAER fraud
For our final set of earnings management measures, we use earnings restatements in financial
statements that comprise accounting irregularities. When a company restates earnings, it merely
suggests that the company has unintentionally or intentionally misstated earnings in the past. However,
an earnings restatement that is an accounting irregularity represents a more serious, and, arguably,
intentional misrepresentation of the company’s financial statements (Agrawal and Chadha, 2005).
Such misstatements constitute a clear and material violation of appropriate accounting practices as they
represent a direct admission by managers of misrepresentation. Although many such restatements are
not criminally fraudulent in nature, such a measure is used as a proxy for fraud and a failure in ethics
(Staubus, 2005; O’ Connor et al., 2006). Therefore, it is an ideal proxy for intentional financial
misrepresentation by managers. Our sample period starts from 1992 for most earnings management
measures while the GAO restatements start from 1997. To ensure the robustness of our results, we
further collect the AAER sample from Lexis-Nexis and the SEC website to identify accounting frauds.
We identify the starting year of each AAER fraud that is issued by the SEC since the AAER releases
are considerably lagged compared to the actual earnings management period. The AAER fraud sample
also represents the most egregious accounting manipulations that are identified by the SEC, thereby
reducing any coding biases in determining earnings management (e.g., Dechow et al. 2010)
To measure earnings restatements that are accounting irregularities, we create an indicator
variable (IRREGULARITY) that equals one if a firm makes such a restatement during our sample
period, and equal to zero otherwise. To measure accounting fraud that is identified from the AAERs,
we crease an indicator variable (AAER_FRAUD) that equals one if a firm commits accounting fraud in
our sample period that is identified by the SEC. Next, we discuss the measures of product market
competition we use in our study.
25
3.3. Measures of competition
Following Karuna (2007) and Sutton (1991), we measure product-market price competition
along three dimensions: product substitutability, market size, and entry costs. Product substitutability
refers to the extent to which there are close substitutes to a product in an industry. The greater the
availability of close substitutes, the greater the intensity of price competition (Nevo, 2001). Market
size reflects market demand in an industry and reflects the density of consumers in an industry. When
market demand for a product increases at a given price, sales of that product also increase. Attracted
by the prospects of greater profitability, firms enter the market (industry), and thus price competition
increases (Sutton, 1991). Entry costs refer to the barriers to entry in an industry (Sutton, 1991). The
higher these barriers, the lower the intensity of price competition. These determinants of price
competition reflect product market fundamentals and more directly capture the nature of competition
in an industry. In contrast, much prior research uses industry concentration measures as competition
proxies (e.g., Aggarwal and Samwick, 1999). However, several studies in the industrial organizations
literature (e.g., Demsetz, 1973; Sutton, 1991; Raith, 2003) show that industry concentration is
endogenous and misleading in terms of how it reflects competition, especially in cross-industry studies,
as it is not clear whether low or high concentration reflects high competition. Moreover, industry
concentration fails to capture the threat from potential competition in the industry (Karuna, 2007).
Like Karuna, we obtain industry-level data at the primary four-digit SIC code level to construct
our competition measures. Adopting such a narrowly defined industrial classification allows us to
more precisely determine how product market competition in an industry affects earnings management
in that industry. Our measure for product substitutability is AVG_GMARGIN, which is estimated as
26
the industry’s gross margin.18 This measure is similar to the Lerner index (Lerner, 1934) and is used
extensively in the industrial organizations literature as a measure of product substitutability (e.g., Nevo,
2001). Larger values of AVG_GMARGIN denote lower product substitutability in the industry and
thus less intense competition. We measure market size by using the variable, AVG_MKTSIZE, which
is the natural logarithm of industry sales (Sutton, 1991). Larger values of AVG_MKTSIZE denote
greater market size and hence greater competition. Finally, similar to Sutton (1991), we proxy for
entry costs using exogenous set up costs in an industry. We use the variable, AVG_ENTCOST to
measure entry costs in an industry. Due to the absence of industry-level data to construct this measure,
we estimate it as the natural logarithm of the weighted average gross value of the cost of property,
plant, and equipment for firms weighted by each firm’s market share in the industry. Larger values of
AVG_ENTCOST reflect higher entry costs and less intense competition. We average all competition
measures across the sample period. Using our different proxies for competition, our study alludes to
the multi-dimensional nature of competition and enables us to more robustly determine the precise
nature of the relation between competition and earnings management while controlling for the level of
industry concentration and other determinants of earnings management.19
4. Empirical regression model
To test our hypotheses, we conduct the following regression for all our earnings management
variables as dependent variables:
Earnings Management measure = β0+ β1AVG_HINDEX +β2AVG_GMARGIN+
+β3AVG_MKTSIZE+β4AVG_ENCOST+β5AVG_ln(SALES)+β6AVG_ln(MB)
18
Karuna (2007) uses the price-cost margin, an industry profitability measure, computed as sales divided by operating
costs, as the measure of product substitutability. Using this measure, we obtain several outliers that confound our analyses
that arise due to the sales figure being much larger than the operating cost figure for several industries. Therefore, we
modify this measure by using the industry gross margin, another industry profitability measure. Our overall findings remain
when we use the price-cost margin as our measure of product substitutability as in Karuna (2007).
19
Bagnoli and Watts (2010) specifically predict a positive relation between product market size and earnings management
in an industry.
27
+β7AVG_OPERATING_CYCLE+β8AVG_CFO_VOLATILITY
+β9AVG_REVENUE_VOLATILITY+β10AVG_FOREIGN
+β11AVG_ln(SEGMENT)+ε
(7)
Our dependent variables for the regression in equation (7) are AQ, DA, RM measures, and
IRREGULARITY. These denote accruals quality, discretionary accruals, real activities management
and irregular earnings restatements. Where AQ, DA, and RM measures are used as the dependent
variables, we conduct robust regressions to examine the competition-earnings-management relation.20
The robust regression weights the observations in proportion to their proximity to the mean value of
the dependent variable and therefore minimizes the impact of outliers on the results. Such a procedure
also enables us to compute robust standard errors. Where IRREGULARITY is the dependent variable,
we conduct a logistic regression. All variables are defined in the appendix.
Following prior research, we include various control variables that might be correlated with our
earnings management measures and competition (e.g., Karuna, 2007).
We control for firm size
(AVG_ln(SALES)) because large firms may have better control systems that deter earnings
management and because studies document that larger firms have better accruals quality (e.g., Dechow
and Dichev, 2002). We also control for growth opportunities (AVG_ln(MB)) because it is possible that
growth firms’ earnings may trigger stronger capital-market reactions that motivate managers to
manipulate earnings (Collins and Kothari, 1989; Skinner and Sloan, 2002; Lee, Li, and Yue, 2006).
Furthermore, we control for AVG_CFO_VOLATILITY and AVG_REVENUE_VOLATILTIY in the
regression because Hribar and Nichols (2007) suggest that unsigned earnings management measures
are correlated with volatility of cash flows and revenues which might be associated with our
competition metrics as well. Finally, we control for firm operating cycle (AVG_OPERATING_CYCLE)
20
Using OLS regressions instead of robust regressions doesn’t alter our overall findings.
28
and complexity of businesses (AVG_ln(SEGMENT), AVG_FOREIGN) because managers have greater
leeway to manage earnings in firms with long operating cycles and complex operations without being
detected. In the next section, we discuss the results of the test we conduct in this section.
5. Regression results
Table 3 provides results of regressions based on the Dechow and Dichev (2002) model to
examine the relation between competition (AVG_GMARGIN, AVG_MKTSIZE, and AVG_ENTCOST)
and earnings quality (our AQ measures). Column (1) provides regression results for AQ_TS01, which
is the accrual quality measure from the time-series Dechow-Dichev model, as the dependent variable,
under the assumption that current accruals are realized within a year. The coefficient estimate on
AVG_GMARGIN is -0.0306, which is significant at the 1% level. This indicates that firms that operate
in industries with higher product substitutability have lower accruals quality (i.e., higher values of our
accruals quality measures). Economically, the results from column (1) suggest that for an interquartile
increase in product substitutability, a firm’s accruals quality decreases by 8.7% relative to the median
value of accruals quality.21 The coefficient estimate on AVG_MKTSIZE is 0.0057, which is positive
and significant (1% level). This result indicates that firms in industries with greater market size have
lower accruals quality; for a 1% increase in industry market size, a firm’s accruals quality decreases by
0.16% relative to the median value of accruals quality. 22 Finally, the coefficient for AVG_ENTCOST
is -0.0047, which is negative and significant (1% level), and indicates that firms in industries with
higher barriers to entry have higher accruals quality. For a 1% increase in industry entry costs, a firm’s
21
We estimate the percentage change of 8.7% as follows: -0.0306(the coefficient on AVG_GMARGIN)×[0.054(the first
quartile value of AVG_GMARGIN)-0.153(the third quartile value of AVG_GMARGIN)]/0.035(the median value of
AQ_TS01).
22
We estimate the percentage change of 0.16% as follows: 1%×0.0057(the coefficient on AVG_MKTSIZE))/0.035(the
median value of AQ_TS01).
29
accruals quality increases by 0.13% relative to the median value of accruals quality.23 Collectively,
these results indicate that firms in more competitive industries have lower accruals quality, providing
evidence supporting our first hypothesis. The economic significance from results in Column (1) is
nontrivial. Columns (2) and (3) provide similar results using the alternative time-series AQ measure
(AQ_TS01_RMSE) and the cross-sectional AQ measure (AQ_CS01) as the dependent variables,
respectively, which also indicate a negative relation between competition and accruals quality.
In Table 4, we expand Dechow and Dichev’s model and estimate AQ measures by assuming
that it takes longer than a year, specifically, two years, for current accruals to be realized as discussed
in Section 3.2. The results in Table 4 continue to support and corroborate our findings in Table 3.
Similar to its counterparts in Table 3, regression results using AQ_TS02, AQ_TS02_RMSE, and
AQ_CS02 also indicate that firms in industries with higher product substitutability, market size, and
lower barriers to entry have lower accruals quality, providing additional evidence supporting our first
hypothesis.
Overall, our results using the AQ measures in Tables 3 and 4 provide evidence supporting our
first hypothesis that firms in more competitive industries are more likely to manipulate accruals
compared to firms in less competitive industries.
AQ measures are often used as proxies for earnings quality, but poor earnings quality may not
be the consequence of intentional earnings management as Dechow and Dichev (2002) suggest. To
further verify that earnings management is positively related to competition, we use the absolute values
of discretionary accruals (i.e., DA measures) as the dependent variable and report the results in Table 5;
DA measures reflect greater managerial discretion than AQ measures which are primarily associated
with the properties of earnings (e.g., Dechow et al., 1995; Bergstresser and Philippon, 2006). Columns
(1) and (2) report the results using DA_RMSE and DA_RMSE_ROA as the dependent variables,
23
The calculation of the percentage change (0.13%)
AVG_ENTCOST)/0.035(the median value of AQ_TS01).
30
is
as
follows:
1%×-0.0047(the
coefficient
on
respectively.
These two variables are estimated from the time-series modified Jones model.
Coefficient estimates on the three competition measures in Columns (1) and (2) generally suggest a
positive relation between competition and accruals management. DA_CMJ, the absolute value of the
discretionary accruals from the cross-sectional modified Jones model is the dependent variable used in
Column (3).
The results show that the coefficients on AVG_GMARGIN, AVG_MKTSIZE, and
AVG_ENTCOST are significantly negative, positive, and negative, respectively, also indicating that
competition intensity is positively related to accruals management. Finally, Columns (4)-(5) report
similar results for regressions using the DA measures but where controls for firm performance are
included. The results in Table 5 provide additional confirmatory evidence to Table 4 supporting our
first hypothesis that firms in more competitive industries are more likely to manipulate accruals
compared to firms in less competitive industries.
In Table 6, we explore how competition intensity affects the extent of real activities
management, including over- or under-production, discretionary expenditure management, and sales
manipulation. To test the relation between competition and the extent of real activities management in
a time series sense, Columns (1)-(3) use root mean square errors from the firm-specific regressions
(Equations (4)-(6)) to estimate the extent of real activities for each firm and regress these estimated
real activities management measures on AVG_GMARGIN, AVG_MKTSIZE, and AVG_ENTCOST and
other control variables. The coefficients on AVG_MKTSIZE and AVG_ENTCOST are significantly
positive and negative, respectively, indicating that larger industry market size and lower entry costs are
associated with more real activities management. The coefficient estimate on AVG_GMARGIN is
negative and significant in Column (3), but is insignificant in Columns (1) and (2). Based on the crosssectional real activities management model as in Roychowdhury (2006), Columns (4)-(6) use the
estimated absolute values of abnormal production costs, discretionary expenses, or abnormal cash
flows (AVG_PROD_ABS, AVG_DISEXP_ABS and AVG_CFO_ABS) as the dependent variables and
31
examine the relation between competition and real activities manipulation. The results in Columns
(4)-(6) are similar to those in Columns (1)-(3). Overall, the results in Table 6 provide evidence
suggesting that the extent of real activities management is greater in firms in more competitive
industries, supporting our second hypothesis.
To test our third hypothesis that firms in more competitive industries are more likely to restate
earnings due to accounting irregularities than firms in less competitive industries, we conduct a logistic
regression in Table 7 where we regress IRREGULARITY on our competition variables and control
variables.24 The results generally support our third hypothesis.
To test our fourth hypothesis that firms in more competitive industries are more likely to
commit accounting fraud that is identified by the AAERs than firms in less competitive industries,
Table 8 uses AAER_FRAUD as the dependent variable and reports the logistic regression results. Our
results suggest that there is a positive relation between competition and accounting fraud, as identified
by the most egregious cases identified by the SEC.
The coefficients for the competition variables in all the tables are economically and statistically
significant. Intriguingly, we also find evidence that greater industry size is associated with greater
earnings management whereas greater firm size is associated with less earnings management. The
coefficient for AVG_MKTSIZE is positive in all our regressions whereas the coefficient for
AVG_ln(SALES) is generally negative. This provides one example of how an economic attribute can
have a different impact on earnings management at the firm and industry levels.
To summarize, we provide comprehensive evidence of a positive competition-earnings
management relation across several measures of competition and earnings management for both crosssectional and time-series earnings management models. Accruals quality is lower, accruals and real
activities management is greater, and restatements due to accounting irregularities are more likely in
24
Since the restatement data starts from 1997, we adjust the averaging period to estimate our independent variables in this
regression. In Section 6, we also make corresponding adjustments to ensure that the independent and dependent variables
are estimated in the consistent sample period.
32
firms that operate in more competitive industries than firms that operate in less competitive industries.
In the next section, we conduct additional checks to test the robustness of our results.
6. Additional Tests
6.1 AAER fraud and deregulation
Our tests above document consistent evidence for a positive relation between competition and
earnings management. One may have a concern about the causal relation since these regressions are
cross-sectional in nature. To address this issue, we investigate a natural experiment setting on how an
external shock affects competition and earnings management. Specifically, we examine whether the
deregulation occurred to Telecommunication (TELECOM) industry in 1996 affects earnings
management in this industry compared to other industries. The deregulation is relatively exogenous to
the competition-earnings management relation, and the 1996 deregulation event happens to be the midpoint of our sample period. To have sufficient power for this investigation, we want to reduce our
coding biases for earnings management measures. Ideally, we may use irregularity restatement data,
but such data only start from 1997. Fortunately, our accounting fraud data from the AAERs cover all
our sample period, and can help avoid any coding biases from the accounting scholars in determining
earnings management. In Table 9, we first examine the competition change from the pre-deregulation
period to the post deregulation period. The results show that all three competition measures
consistently show TELECOM industry’s competition increases after the deregulation. Table 10 uses a
difference in difference test to examine the effect of de-regulation on earnings management in
TELECOM industry. The coefficient on DEREGULATION×TELECOM is significantly positive. This
suggests that deregulation is related with high earnings management measured as accounting fraud.
Table 9 and 10 together suggest that deregulation increases earnings management via increasing
competition.
33
6.2 Robustness checks
In our first set of robustness tests, we rerun the above regressions including industry fixed and
random effects. Including these effects controls for the possibility that unobserved industry factors that
are constant for a given industry over time but vary across industries could have confounded our
analyses above.25 It also allows us to group firms by industry as firms in the same industry are likely
to be connected to each other by having common industry characteristics. Our overall findings are
unchanged.
Our analysis above is primarily based on the Compustat database. A criticism with Compustat
is that it does not comprise the universe of firms in an industry, e.g., private firms. Consequently, our
competition measures may distort the true level of competition in an industry. We also conduct the
regressions above using an alternative dataset to the one we use above based on Compustat.
Specifically, we collect data to construct our competition measures from the U.S. Economic Census
Bureau. This data source comprises many more firms in an industry including both public and private
firms. A limitation with this alternative dataset, however, is that we are only able to obtain data for the
manufacturing sector as many of the data items necessary to construct our competition measures are
not available for the other sectors. We find that our overall findings are unaltered using this dataset.
7. Conclusion
In this paper, we examine how product market price competition in an industry affects the
extent of earnings management in firms that operate in the industry. Using different determinants of
price competition that have been used in the industrial organizations literature, namely, product
substitutability, market size, and entry costs, on a range of earnings management variables that capture
accruals and real activities management as well as financial restatements that comprise accounting
25
Our results are robust to running OLS regressions and clustering by industry.
34
irregularities, we find a robust positive relation between competition and the extent of earnings
manipulation.
We find that our results hold for both cross-sectional and time-series earnings
management models, attesting to the robustness of our results.
Further to the evidence documented in studies supporting the notion that earnings management
is considered harmful to firms and a nation’s economy (e.g., Sadka, 2006; Kedia and Philippon, 2009;
Durnev and Mangen, 2009), our findings suggest that governmental organizations and other regulatory
bodies should consider both the beneficial as well as the perverse effects of competition in policy
choices and other welfare considerations in general. While competition may enhance efficiency and
productivity in firms, it could come at a cost. Under this perspective, a more careful consideration of
the benefits and costs of competition is necessary in policy-making. An alternative perspective is that
earnings management is optimal to shareholders in some settings, e.g., in more competitive industries.
Future researchers could consider examining different aspects of competition that affect financial
misrepresentation in firms, or how competition affects specific instances of egregious earnings
management. More work could also be conducted on how competition affects directional earnings
management in firms. Conducting detailed field studies in different industries is a worthwhile step in
that direction.
35
Appendix: Variable Definitions:
Variable Name
Definitions
Estimation Method
AVG_GMARGIN
Gross Margin in Industry (at the
four-digit SIC code level)
is equal to average operating profit/sales in the
industry
AVG_HINDEX
Herfindahl-Hirschman Index (at the
four-digit SIC code level)
is the sum of the squared market shares of the firms
in the industry
AVG_MKTSIZE
Level of market size in industry (at
the four-digit SIC code level)
is equal to the natural log of industry sales
AVG_ENTCOST
Level of entry costs in industry (at
the four-digit SIC code level)
AVG_ln(SALES)
Firm size
AVG_ln(MB)
Market-to-book ratio
is equal to the natural log of the weighted average
gross value of cost of property, plant, and equipment
for firms in the industry, weighted by each firm's
market share in industry
is equal to the natural log of firm sales over the
sample period
is equal to the natural log of market value/book
value of shareholder equity
is equal to the natural log of the average of
[(sales/360)/(average accounts receivable) + (cost
of goods sold/360) / average inventory)], calculated
during the sample period.
AVG_OPERATING_CYCLE
Operating cycle
AVG_CFO_VOLATILITY
CFO volatility
The standard deviation of cash from operations
scaled by beginning assets during the sample period
AVG_REVENUE_VOLATILITY
Sales volatility
The standard deviation of sales scaled by beginning
assets during the sample period
AVG_FOREIGN
Frequency of foreign transactions
AVG_ln(SEGMENT)
Number of segments
36
The frequency with which the firm has a nonzero
foreign currency transaction during the sample
period
is equal to the natural log of 1 plus the number of
business segments
AQ_TS01
Time-series accruals quality measure
based on Equation (1) in Section 3.
AQ_CS01
Cross-sectional accruals quality
measure based on Equation (1) in
Section 3.
AQ_TS02
Time series accruals quality measure
based on Equation (2) in Section 3.
AQ_CS02
Cross-sectional accruals quality
measure based on Equation (2) in
Section 3.
AQ_TS01_RMSE
AQ_TS02_RMSE
DA_RMSE
DA_RMSE_ROA
The standard deviation of the residuals from
Dechow and Dichev’s (2002) accruals-quality
measure during the sample period
The standard deviation of the residuals from
Dechow and Dichev’s (2002) accruals-quality
measure, as adjusted by McNichols (2002) and
Francis et al. (2005), measured during the sample
period
The standard deviation of the residuals from
Dechow and Dichev’s (2002) accruals-quality
measure during the sample period, considering two
leading and lagged periods of cash flows in the
model
The standard deviation of the residuals from
Dechow and Dichev’s (2002) accruals-quality
measure, considering two leading and lagged
periods of cash flows in the model, as adjusted by
McNichols (2002) and Francis et al. (2005)
Time series accruals quality
measure, which is the root mean
square error of each individual firm
regression based on Equation (1) in
Section 3.
The root mean square deviation from Dechow and
Dichev’s (2002) accruals-quality model
Time series accruals quality
measure, which is the root mean
square error of each individual firm
regression based Equation (2) in
Section 3.
The root mean square deviation from Dechow and
Dichev’s (2002) accruals-quality model, considering
two leading and lagged periods of cash flows in the
model
Time-series accrual earnings
management measure using the
modified Jones' model
Time-series accrual earnings
37
The root mean square deviation from the time-series
Modified Jones' Model
The root mean square deviation from the time-series
Modified Jones' Model controlling for ROA
management measure using the
modified Jones' model with control
of ROA
DA_CMJ
Absolute value of discretionary
accruals
The average absolute value of discretionary accruals
from the cross-sectional Modified Jones' model
DA_CMJ_ROA
Absolute value of discretionary
accruals with control of ROA
The average absolute value of discretionary accruals
from the cross-sectional Modified Jones' model
controlling for ROA
Absolute value of performancematched discretionary accruals
The average absolute value of performance-matched
accruals from the cross-sectional Modified Jones'
model
PROD_ABNORMAL_RMSE
Time-series abnormal production
costs measure
The root mean square deviation from the time-series
abnormal production costs model
DISEXP_RMSE
Time-series abnormal expenses
measure
The root mean square deviation from the time-series
abnormal expense model
CFO_ABNORMAL_RMSE
Time-series abnormal cash flows
from operations
The root mean square deviation from the time-series
abnormal cash flows model
AVG_PROD_ABS
Absolute value of abnormal
production costs
The average absolute of abnormal production costs
from Roychowdhury’s (2006) model
AVG_DISEXP_ABS
Absolute value of discretionary
expenses
The average absolute of abnormal discretionary
expenses from Roychowdhury’s (2006) model
Absolute value of abnormal cash
flows from operations
The average absolute of abnormal cash flows from
operations according to the estimation developed by
Roychowdhury (2006)
an indicator variable that captures
whether firms have restatements due
to accounting irregularities during
sample period
an indicator variable that equals 1 if the firm has
made a restatement due to accounting irregularities
during the sample period, and equals to 0 otherwise
DA_PMMJ
AVG_CFO_ABS
IRREGULARITY
AAER_FRAUD
an indicator variable that captures
38
an indicator variable that equals 1 if the SEC has
taken an enforcement action against a firm due to
DEREGULATION
whether a firm commits accounting
fraud in the period that are
investigated by the SEC.
An indicator variable that captures
whether firms are in the deregulation
period.
39
accounting fraud caused by intentional falsification
of financial statements in the period, and equals to 0
otherwise
An indicator variable that equals 1 if the sample
period is after 1996.
References
Aghion, P., Bloom, N., Blundell, R., Griffith, R., Howitt, P., 2005. Competition and Innovation:
An Inverted-U Relationship. Quarterly Journal of Economics, 120(2), 701-28.
Aghion, P., Howitt, P., 1992. A Model of Growth through Creative Destruction. Econometrica,
60(2), 323-51.
Aggarwal, R. K. , Samwick, A.A, 1999. Executive Compensation, Strategic Competition, and
Relative Performance Evaluation: Theory and Evidence. Journal of Finance, 54(6), 19992043.
Agrawal, A., Chadha, S., 2005. Corporate Governance and Accounting Scandals. Journal of Law
& Economics, 48(2), 371-406.
Ashbaugh-Skaife, H., Collins, D.W, Kinney, W.R., LaFond, R., 2008. The Effect of Sox Internal
Control Deficiencies and Their Remediation on Accrual Quality. The Accounting Review,
83(1), 217-50.
Ali, A., Klasa, S., Yeung, E., 2009. The Limitations of Industry Concentration Measures
Constructed with Compustat Data: Implications for Finance Research. Review of
Financial Studies, 22(10), 3839-71.
Bagnoli, M., Watts, S.G., 2000. The effect of relative performance evaluation on earnings
management: A game theoretic approach. Journal of Accounting and Public Policy 19,
377-397.
_________, _______, 2010. Oligopoly, Disclosure, and Earnings Management. The Accounting
Review, 85(4), 1191-214.
Becker, G.S, 1968. Crime and Punishment - Economic Approach. Journal of Political Economy,
76(2), 169-217.
Belsley, D.A., Kuh, E., Welsch, R.E.,1980. Regression Diagnostics. New York: John Wiley &
Sons.
Bergstresser, D., Philippon, T., 2006. CEO Incentives and Earnings Management. Journal of
Financial Economics, 80(3), 511-29.
Blundell, R., Griffith, R., Reenen, J.V., 1999. Market Share, Market Value and Innovation in a
Panel of British Manufacturing Firms. Review of Economic Studies, 66(3), 529-54.
Caves, R. E, 1980.Industrial Organization, Corporate Strategy and Structure. Journal of
Economic Literature, 18(1), 64-92.
Chen, Y., 1994. Conflicting interests in information disclosure and short-term orientation of
firms. International Journal of Industrial Organization, 12, 211-225.
Christie, A. A., Joye, M.P., Watts, R.L., 2003. Decentralization of the Firm: Theory and
Evidence. Journal of Corporate Finance, 9(1), 3-36.
Collins, D.W., Kothari, S. P., 1989. An Analysis of Intertemporal and Cross-Sectional
Determinants of Earnings Response Coefficients. Journal of Accounting & Economics,
11(2-3), 143-81.
Coleman, J. W., 1987. Toward an Integrated Theory of White-Collar Crime. American Journal
of Sociology, 93(2), 406-39.
Cummins, J.G., Nyman, I., 2005. The Dark Side of Competitive Pressure. Rand Journal of
Economics, 36(2), 361-77.
Dechow, P. M., Sloan,R.G., Sweeney,A.P., 1995. Detecting Earnings Management. The
Accounting Review, 70(2), 193-225.
40
_________________,_______________ ,_________________, 1996. Causes and Consequences
of Earnings Manipulation: An Analysis of Firms Subject to Enforcement Actions by the
SEC. Contemporary Accounting Research, 13(1), 1-36.
____________ , Dichev, I.D., 2002. The Quality of Accruals and Earnings: The Role of Accrual
Estimation Errors. The Accounting Review, 77, 35-59.
____________, Ge, W., Schrand, C.,2010. Understanding earnings quality: A review of the
proxies, their determinants and their consequences. Journal of Accounting and
Economics 50 (2-3), 344-401.
Demsetz, H., 1973. Industry structure, market rivalry, and public policy. Journal of Law and
Economics 16, 1-10.
Demski, J. S., 1998. Performance Measure Manipulation. Contemporary Accounting Research,
15(3), 261-85.
Dixit, A. K., Stiglitz, J.E., 1977. Monopolistic Competition and Optimum Product Diversity.
American Economic Review, 67(3), 297-308.
Durnev, A., Mangen, C., 2009. Corporate Investments: Learning from Restatements. Journal of
Accounting Research, 47(3), 679-720.
Erickson, M., Wang, S., 1999. Earnings management by acquiring firms in stock for stock
mergers. Journal of Accounting and Economics, 27, 149–176.
Feinberg, R., 1995. In defense of corporate myopia. Managerial and Decision Economics, 16(3),
205-210.
Francis, .J., LaFond, R., Olsson, P., Schipper, K., 2005. The Market Pricing of Accruals Quality.
Journal of Accounting & Economics, 39(2), 295-327.
Fresard, L., Valta, P., 2012. Competitive pressure and corporate policies. Working paper.
Gaspar, J., Massa, M., 2006. Idiosyncratic volatility and product market competititon. Journal of
Business, 79, 3125-3152.
Gerety, M., Lehn, K., 1997. The Causes and Consequences of Accounting Fraud. Managerial
and Decision Economics, 18, 587-99.
Graham, J.R, Harvey, C.R., Rajgopal, S., 2005. The Economic Implications of Corporate
Financial Reporting. Journal of Accounting & Economics, 40(1-3), 3-73.
Grossman, G., Helpman, E., 1991. Innovation and Growth in the Global Economy. Cambridge,
MA: MIT Press.
Green, A., Mayes, D., 1991. Technical Inefficiency in Manufacturing Industries. Economic
Journal, 101(406), 523-38.
Guenther, D., 1994. Earnings management in response to corporate tax rate changes: Evidence
from the 1986 tax reform act, Accounting Review 69, 230-243.
Hart, O. D., 1983. The Market Mechanism as an Incentive Scheme. Bell Journal of Economics,
14(2), 366-82.
Healy, P. M., 1985. The Effect of Bonus Schemes on Accounting Decisions. Journal of
Accounting & Economics, 7(1-3), 85-107.
Hennes, K.M., Leone, A.J., Miller, B.P., 2008. The Importance of Distinguishing Errors from
Irregularities in Restatement Research: The Case of Restatements and CEO/CFO. The
Accounting Review, 83(6), 1487-519.
Hribar, P., Nichols, C., 2007. The Use of Unsigned Earnings Quality Measures in Tests of
Earnings Management. Journal of Accounting Research, 45(5), 1017-53.
Jones, J.J., 1991. Earnings Management During Import Relief Investigations. Journal of
Accounting Research, 29(2), 193-228.
41
Karuna, C., 2007. Industry Product Market Competition and Managerial Incentives. Journal of
Accounting & Economics, 43(2-3), 275-97.
Kedia, S., Philippon, T., 2009. The Economics of Fraudulent Accounting. Review of Financial
Studies, 22(6), 2169-99.
Key, K.G., 1997. Political Cost Incentives for Earnings Management in the Cable Television
Industry. Journal of Accounting & Economics, 23(3), 309-37.
Klein, A., 2002. Audit Committee, Board of Director Characteristics, and Earnings Management.
Journal of Accounting & Economics, 33(3), 375-400.
Kole, S., Lehn,K., 1997. Deregulation, the Evolution of Corporate Governance Structure, and
Survival. American Economic Review, 87(2), 421-25.
__________________, 1999. Deregulation and the Adaptation of Governance Structure: The
Case of the Us Airline Industry. Journal of Financial Economics, 52(1), 79-117.
Kothari, S. P., Leone, A.J. , Wasley, C.E., 2005. Performance Matched Discretionary Accrual
Measures. Journal of Accounting & Economics, 39(1), 163-97.
Kutner, M.H., Nachtsheim, C. J., Neter, J., Li, W., 2005. Applied Linear Statistical Models. New
York: McGraw-Hill Irwin.
Lee, C.W.J., Li, L.Y., Yue, H., 2006. Performance, Growth and Earnings Management. Review
of Accounting Studies, 11(2-3), 305-34.
Lerner, A. P., 1934. The concept of monopoly and the measurement of monopoly power. The
Review of Economic Studies 1(3), 157-175.
Li, H.D., Pincus, M., Rego, S.O., 2008. Market Reaction to Events Surrounding the SarbanesOxley Act of 2002 and Earnings Management. Journal of Law & Economics, 51(1), 11134.
Linck, J. S., Netter, J.M., Shu, T., 2010. Can Earnings Management Ease Financial Constraints
Evidence from Earnings Management Prior to Investment. SSRN eLibrary.
Louis, H., 2004. Earnings management and the market performance of acquiring firms. Journal
of Financial Economics, 74, 121–148.
MacKay, P., Phillips, G., 2005. How does industry affect firm financial structure? Review of
Financial Studies, 18, 1433-1466.
McNichols, M.F., 2002. Discussion of the Quality of Accruals and Earnings: The Role of
Accrual Estimation Errors. The Accounting Review, 77, 61-69.
Milgrom, P., Roberts, J., 1992. Economics, Organization and Management. Englewood Cliffs:
Prentice Hall.
Nalebuff, B. J., Stiglitz, J.E., 1983. Prizes and Incentives - Towards a General Theory of
Compensation and Competition. Bell Journal of Economics, 14(1), 21-43.
Narayanan, M., 1985. Managerial incentives for short-term results. Journal of Finance, 40, 14691484.
Nevo, A., 2001. Measuring Market Power in the Ready-to-Eat Cereal Industry. Econometrica,
69(2), 307-42.
Nickell, S. J., 1996. "Competition and Corporate Performance." Journal of Political Economy,
104(4), 724-46.
O'Connor, J. P., Priem, R.L., Coombs, J.E., Gilley, M.K., 2006. Do CEO Stock Options Prevent
or Promote Fraudulent Financial Reporting? Academy of Management Journal, 49(3),
483-500.
42
Perry, S., Williams, T., 1994. Earnings management preceding management buyout offers.
Journal of Accounting and Economics, 18, 157–179.Porter, M. E., 1990. The
Competitive Advantage of Nations. Macmillan Press, London.
Raith, M., 2003. Competition, Risk, and Managerial Incentives. American Economic Review,
93(4), 1425-36.
Reynolds, J. K., Francis, J.R., 2000. Does Size Matter? The Influence of Large Clients on OfficeLevel Auditor Reporting Decisions. Journal of Accounting & Economics, 30(3), 375-400.
Ronen, J., Tzur, J., Yaari, V., 2006. The Effect of Directors' Equity Incentives on Earnings
Management. Journal of Accounting and Public Policy, 25(4), 359-89.
Roychowdhury, S., 2006. Earnings Management through Real Activities Manipulation. Journal
of Accounting & Economics, 42(3), 335-70.
Sadka, G., 2006. The Economic Consequences of Accounting Fraud in Product Markets: Theory
and a Case from the U.S. Telecommunications Industry (Worldcom). American Law and
Economics Review, 8(3), 439-75.
Scharfstein, D., 1988. Product Market Competition and Managerial Slack. Rand Journal of
Economics, 19(1), 147-55.
Scherer, F., 1967. Market Structure and the Employment of Scientists and Engineers. American
Economic Review, LVII, 524-31.
Schmidt, K.M. 1997. Managerial Incentives and Product Market Competition. Review of
Economic Studies, 64(2), 191-213.
Skinner, D.J., Sloan, R.G., 2002. Earnings Surprises, Growth Expectations, and Stock Returns or
Don't Let an Earnings Torpedo Sink Your Portfolio. Review of Accounting Studies, 7(2),
289-312.
Stein, J., 1988. Takeover threats and managerial myopia. Journal of Political Economy, 96, 6180.
_______, 1989. Efficient capital markets, inefficient firms: A model of myopic corporate
behavior. Quarterly Journal of Economics, 104, 655-669.
Subramanyam, K. R, 1996. The pricing of discretionary accruals. Journal of Accounting &
Economics 22 (1-3):249-281.
Shleifer, A., 2004. Does Competition Destroy Ethical Behavior? American Economic Review,
94(2), 414-18.
_________, Vishny, R., 1990. Equilibrium short horizons of investors and firms. American
Economic Review, 80, 148-153.
Smith, A., 1776. An Inquiry into the Nature and Causes of the Wealth of Nations. London:
Electric Book Co., c2001.
Staubus, G.J., 2005. Ethics Failures in Corporate Financial Reporting. Journal of Business Ethics,
57(1), 5-15.
Sutton, J., 1991. Sunk costs and market structure. The MIT Press, Cambridge, Massachusetts.
Teoh, S., Welch, I, Wong, T., 1998a. Earnings management and the long-run market
performance of initial public equity offerings. Journal of Finance, 53, 1935–1974.
_______,_______, ________, 1998b. Earnings management and the underperformance of
seasoned equity offerings. Journal of Financial Economics, 50, 63–99.
Von Thadden, E., 1995. Long-term contracts, short-term investment, and monitoring. Review of
Financial Studies, 62, 557-575.
Xu, J., 2011. Profitability and capital structure: Evidence from import penetration. Journal of
Financial Economics, Forthcoming.
43
Table 1: Descriptive Statistics
Variables
AVG_GMARGIN
AVG_HINDEX
AVG_MKTSIZE
AVG_ENTCOST
AVG_Ln(SALES)
AVG_ln(MB)
AVG_OPERATING_CYCLE
AVG_CFO_VOLATILITY
AVG_REVENUE_VOLATILITY
AVG_FOREIGN
AVG_ln(SEGMENT)
AQ_CS01
AQ_TS01
AQ_TS01_RMSE
AQ_CS02
AQ_TS02
AQ_TS02_RMSE
DA_CMJ
DA_CMJ_ROA
DA_PMMJ
DA_RMSE
DA_RMSE_ROA
AVG_PROD_ABS
AVG_DISEXP_ABS
AVG_CFO_ABS
PROD_ABNORMAL_RMSE
DISEXP_RMSE
CFO_ABNORMAL_RMSE
IRREGULARITY
AAER_FRAUD
Mean
0.084
0.205
10.196
7.381
4.216
1.140
210.251
1.651
6.289
0.203
0.383
0.073
0.046
0.058
0.055
0.030
0.046
0.177
0.139
0.173
0.073
0.057
0.161
0.239
0.139
0.057
0.079
0.069
0.021
0.029
Median
0.091
0.162
10.266
7.267
4.281
0.989
108.757
0.094
0.274
0.000
0.000
0.055
0.035
0.043
0.044
0.022
0.032
0.108
0.091
0.105
0.055
0.045
0.119
0.154
0.093
0.045
0.047
0.054
0.000
0.000
Std
0.557
0.169
1.873
1.888
2.648
0.542
1,062.434
71.910
341.730
0.353
0.499
0.063
0.039
0.049
0.040
0.027
0.041
0.195
0.159
0.192
0.065
0.043
0.162
0.378
0.218
0.046
0.112
0.056
0.142
0.168
P25
0.054
0.081
9.010
6.050
2.598
0.794
67.122
0.049
0.139
0.000
0.000
0.032
0.018
0.022
0.027
0.012
0.018
0.057
0.049
0.057
0.033
0.027
0.069
0.078
0.053
0.027
0.024
0.034
0.000
0.000
P75
0.153
0.259
11.500
8.920
5.955
1.320
174.754
0.200
0.553
0.273
0.693
0.094
0.061
0.076
0.074
0.039
0.059
0.218
0.170
0.214
0.093
0.075
0.205
0.285
0.157
0.071
0.089
0.086
0.000
0.000
44
Table 2: Correlation Table
AVG_GMARGIN
A
A
B
C
D
E
F
G
H
I
J
AVG_HINDEX
B
AVG_MKTSIZE
C
AVG_ENTCOST
D
AVG_Ln(SALES)
E
AVG_ln(MB)
F
AVG_OPERATING_CYCLE
G
AVG_CFO_VOLATILITY
H
AVG_REVENUE_VOLATILITY
I
AVG_FOREIGN
J
AVG_ln(SEGMENT)
K
-0.161
(<.0001)
0.225
(<.0001)
0.165
(<.0001)
0.067
(<.0001)
-0.053
(<.0001)
-0.009
(0.3722)
-0.001
(0.9353)
0.000
(0.9735)
0.022
(0.0355)
0.001
(0.9234)
-0.553
(<.0001)
-0.249
(<.0001)
0.010
(0.337)
-0.072
(<.0001)
0.017
(0.116)
0.006
(0.608)
0.002
(0.873)
-0.023
(0.03)
0.062
(<.0001)
0.795
(<.0001)
0.132
(<.0001)
0.085
(<.0001)
-0.002
(0.861)
-0.014
(0.202)
-0.009
(0.42)
0.078
(<.0001)
0.092
(<.0001)
0.170
(<.0001)
-0.020
0.069
0.010
(0.35)
-0.011
(0.341)
-0.006
(0.618)
0.047
(<.0001)
0.115
(<.0001)
-0.439
(<.0001)
-0.167
(<.0001)
-0.021
(0.067)
-0.006
(0.608)
0.207
(<.0001)
0.209
(<.0001)
0.086
(<.0001)
0.025
(0.029)
0.005
(0.678)
-0.020
(0.069)
-0.002
(0.841)
0.002
(0.882)
-0.001
(0.944)
-0.004
(0.684)
-0.003
(0.79)
0.980
(<.0001)
-0.010
(0.365)
0.015
(0.174)
-0.007
(0.519)
0.016
(0.173)
0.117
(<.0001)
45
Table 3: Competition and Accruals Quality (AQ) assuming accruals realization within a year
Intercept
AQ_TS01
0.0219 ***
(0.0069)
AVG_HINDEX
0.0124
(0.0086)
**
(0.0060)
AVG_GMARGIN
-0.0306
0.0057
***
-0.0047
***
-0.0053
***
0.0147
***
0.0000
***
-0.0149
***
0.0091
***
AVG_ln(SEGMENT)
R2
# of observations
0.0056
***
-0.0067
0.0186
0.0000
-0.0187
***
0.0118
***
0.0072
0.0096
***
-0.0091
***
-0.0074
***
(0.0003)
***
0.0178
***
(0.0020)
***
0.0000
(0.0000)
***
0.0119
***
(0.0013)
***
(0.0016)
**
***
(0.0008)
(0.0042)
(0.0012)
AVG_FOREIGN
-0.0059
-0.0383
(0.0009)
(0.0000)
(0.0033)
AVG_REVENUE_VOLATILITY
***
(0.0027)
(0.0000)
AVG_CFO_VOLATILITY
0.0072
***
(0.0098)
(0.0004)
(0.0022)
AVG_OPERATING_CYCLE
***
(0.0010)
(0.0004)
AVG_ln(MB)
-0.0388
0.0342
(0.0058)
(0.0012)
(0.0008)
AVG_ln(SALES)
**
(0.0141)
(0.0010)
AVG_ENTCOST
0.0158
AQ_CS01
0.0387 ***
(0.0062)
(0.0075)
(0.0113)
AVG_MKTSIZE
AQ_TS01_RMSE
0.0275 ***
-0.0013
***
(0.0003)
***
0.0098
(0.0022)
(0.0027)
(0.0021)
0.0004
0.0000
0.0008
(0.0016)
(0.0020)
(0.0015)
24.2%
24.6%
22.7%
967
967
2,264
***
*,**, and *** correspond to 10%, 5% and 1% significance levels based on two-tailed hypotheses tests, respectively; standard errors are reported
in parentheses.
46
Table 4: Competition and Accruals Quality (AQ) assuming accruals takes longer than a year to be realized
Intercept
AVG_HINDEX
AVG_GMARGIN
AQ_TS02
0.0252 ***
(0.0091)
(0.0062)
0.0033
0.0041
0.0190
(0.0050)
(0.0078)
(0.011)
-0.0295
***
0.0025
-0.0027
AVG_ln(SALES)
***
AVG_OPERATING_CYCLE
AVG_CFO_VOLATILITY
AVG_REVENUE_VOLATILITY
AVG_FOREIGN
AVG_ln(SEGMENT)
R2
# of observations
0.0057
0.0028
-0.0029
-0.0040
0.0094
0.0078
***
-0.0082
-0.0064
***
0.0134
(0.0033)
(0.0022)
0.0000
0.0000
(0.0000)
(0.0000)
(0.0000)
0.0400
(0.0062)
(0.0096)
0.0026
0.0198
(0.0017)
(0.0026)
0.0048
***
0.0068
***
***
(0.0008)
0.0000
***
***
(0.0009)
***
(0.0021)
0.0263
**
(0.0056)
(0.0005)
***
-0.0204
***
(0.011)
**
(0.0011)
***
(0.0003)
AVG_ln(MB)
***
(0.0013)
(0.0007)
-0.0026
-0.0425
(0.0163)
***
(0.0009)
AVG_ENTCOST
AQ_CS02
0.0438 ***
(0.0058)
(0.0105)
AVG_MKTSIZE
AQ_TS02_RMSE
0.0353 ***
***
0.0087
***
**
***
(0.0028)
***
-0.0008
**
(0.0004)
**
0.0095
(0.0018)
(0.0028)
(0.0021)
0.0000
-0.0001
0.0017
(0.0013)
(0.002)
(0.0014)
18.1%
21.0%
26.2%
592
592
1,427
***
*,**, and *** correspond to 10%, 5% and 1% significance levels based on two-tailed hypotheses tests, respectively; standard errors are reported
in parentheses.
47
Table 5: Competition and Accruals Management
Intercept
DA_RMSE
0.0221 ***
(0.0069)
AVG_HINDEX
AVG_GMARGIN
AVG_MKTSIZE
0.0192
(0.0052)
***
(0.0044)
-0.0089
-0.0172
(0.0086)
(0.0064)
0.0054
***
-0.0054
-0.0022
***
-0.0062
***
0.0000
***
0.5248
***
AVG_FOREIGN
AVG_ln(SEGMENT)
R2
# of observations
0.0053
-0.0016
-0.0047
0.0000
***
0.4856
(0.0053)
***
***
-0.0094
**
(0.0039)
***
0.0240
***
(0.0011)
***
-0.0180
***
-0.0163
***
0.0462
***
***
***
0.0648
(0.0069)
0.0010
-0.0117
(0.0029)
(0.0036)
0.0210
***
-0.0147
-0.0137
0.0494
0.0226
***
-0.0166
***
***
***
-0.0157
***
(0.0004)
***
0.0518
***
(0.0021)
0.0000
0.0000
0.0000
(0.0000)
(0.0000)
(0.0000)
-0.0001
-0.0001
-0.0001
(0.0001)
(0.0001)
(0.0001)
0.0000
0.0000
0.0000
0.0000
(0.0018)
(0.0014)
(0.0000)
(0.0000)
(0.0000)
0.0022
-0.0004
0.0170
(0.0021)
(0.0016)
(0.0027)
0.0140
***
(0.0009)
(0.0017)
***
***
(0.0010)
(0.0003)
(0.0023)
***
***
(0.0056)
(0.0007)
(0.0004)
***
0.0533
DA_PMMJ
0.0090
(0.0075)
(0.0008)
(0.0009)
(0.0000)
(0.0071)
AVG_REVENUE_VOLATILITY
***
(0.0018)
(0.0000)
AVG_CFO_VOLATILITY
-0.0046
0.0614
DA_CMJ_ROA
-0.0076
(0.0060)
(0.0075)
(0.0003)
(0.0024)
AVG_OPERATING_CYCLE
***
(0.0006)
(0.0004)
AVG_ln(MB)
0.0044
DA_CMJ
0.0183 **
(0.0081)
(0.0007)
(0.0008)
AVG_ln(SALES)
0.0178
(0.0059)
(0.0010)
AVG_ENTCOST
DA_RMSE_ROA
0.0190 ***
***
(0.0020)
0.0152
*
**
***
(0.0025)
-0.0005
0.0000
0.0027
0.0004
0.0048
(0.0016)
(0.0012)
(0.0019)
(0.0014)
(0.0018)
35.9%
42.9%
16.7%
22.9%
18.3%
1,385
1,385
6,871
6,879
6,872
***
*,**, and *** correspond to 10%, 5% and 1% significance levels based on two-tailed hypotheses tests, respectively; standard errors are reported in parentheses.
48
Table 6: Competition and Real-Activities Management
Intercept
PROD_ABNOR
MAL_RMSE
0.0351 ***
(0.0067)
AVG_HINDEX
AVG_GMARGIN
0.0177
(0.0090)
***
-0.0081
-0.0138
-0.0384
0.0039
(0.0112)
***
-0.0037
-0.0018
-0.0157
***
AVG_OPERATING_CYCLE
0.1427
***
AVG_FOREIGN
AVG_ln(SEGMENT)
2
R
# of observations
0.0341
(0.0020)
-0.0050
0.0321
0.0000
0.1678
***
**
***
(0.0118)
***
-0.0032
-0.0015
-0.0039
-0.0135
0.0246
**
-0.0247
***
-0.0027
(0.0006)
*
0.0347
(0.0061)
0.0092
-0.0181
0.0346
-0.0360
-0.0129
0.0711
0.0108
***
-0.0070
-0.0107
***
0.0644
(0.0042)
(0.002)
0.0000
0.0000
0.0000
(0.0000)
(0.000)
(0.0000)
(0.0000)
**
-0.0008
***
(0.0001)
(0.0002)
(0.0001)
0.0000
(0.0000)
0.0000
(0.0000)
0.0000
0.0043
(0.0046)
(0.0022)
0.0001
0.0006
0.0241
(0.0027)
***
-0.0008
(0.0016)
0.0000
(0.0000)
-0.0032
0.0094
***
0.0005
-0.0096
(0.0020)
(0.0028)
(0.0017)
(0.0033)
0.0003
0.0044
0.0009
-0.0070
(0.0014)
(0.0020)
(0.0012)
(0.0024)
(0.0034)
(0.0016)
21.8%
19.8%
44.2%
7.7%
10.9%
16.4%
1,212
1,212
1,212
5,570
5,570
5,570
**
***
***
***
***
0.0000
(0.0071)
***
***
(0.0004)
(0.003)
-0.0003
***
(0.0008)
***
0.0000
***
***
(0.0009)
(0.0021)
0.6391
***
(0.0036)
***
(0.0008)
***
0.0273
(0.0131)
(0.0017)
***
AVG_CFO_ABS
0.0191 ***
(0.0066)
***
(0.002)
(0.0012)
***
0.1201
(0.0076)
***
(0.0014)
***
AVG_DISEXP_
ABS
0.0413 ***
(0.0141)
***
(0.0055)
***
(0.0003)
(0.0000)
***
***
(0.0007)
***
(0.0034)
***
(0.0087)
AVG_REVENUE
_VOLATILITY
***
0.0819
(0.0094)
(0.0008)
(0.0005)
(0.0000)
AVG_CFO_VOLATILITY
-0.0100
0.0028
AVG_PROD_
ABS
0.0274 ***
(0.0101)
***
(0.0067)
***
(0.0011)
***
(0.0025)
0.0000
0.0113
(0.0014)
(0.0004)
AVG_ln(MB)
0.0121
(0.0046)
(0.0008)
AVG_ln(SALES)
(0.0054)
***
(0.0077)
(0.001)
AVG_ENTCOST
0.0381
CFO_ABNOR
MAL_RMSE
0.0210 ***
(0.0057)
(0.0083)
AVG_MKTSIZE
DISEXP_RMS
E
-0.0245 ***
**
*,**, and *** correspond to 10%, 5% and 1% significance levels based on two-tailed hypotheses tests, respectively; standard errors are reported in parentheses.
49
Table 7: Competition and Restatements due to Accounting Irregularities
IRREGULARITY=1/0
Intercept
-6.4462
***
(0.8018)
AVG_HINDEX
0.7721
(0.6608)
AVG_GMARGIN
1.5593
(1.4356)
AVG_MKTSIZE
0.3513
***
(0.0963)
AVG_ENTCOST
-0.3516
***
(0.0802)
AVG_ln(SALES)
0.1601
***
(0.0419)
AVG_ln(MB)
0.3465
*
(0.1977)
AVG_OPERATING_CYCLE
-0.0002
(0.0005)
AVG_CFO_VOLATILITY
-0.1401
(0.1842)
AVG_REVENUE_VOLATILITY
0.0000
(0.0051)
AVG_FOREIGN
0.0449
(0.2157)
AVG_ln(SEGMENT)
0.3600
**
(0.1553)
Pseudo R2
# of observations
4.15%
6,144
*,**, and *** correspond to 10%, 5% and 1% significance levels based on two-tailed hypotheses tests, respectively; standard errors are reported
in parentheses.
50
Table 8: Competition and AAER Fraud
AAER FRAUD=1/0
Intercept
-6.6489
***
(0.7375)
AVG_HINDEX
0.8919
(0.5031)
AVG_GMARGIN
0.7593
(1.5299)
AVG_MKTSIZE
0.3290
***
(0.0937)
AVG_ENTCOST
-0.2705
***
(0.0915)
AVG_ln(SALES)
0.2054
***
(0.0391)
AVG_ln(MB)
0.3567
***
(0.0894)
AVG_OPERATING_CYCLE
0.0008
***
(0.0002)
AVG_CFO_VOLATILITY
-0.1370
(0.1348)
AVG_REVENUE_VOLATILITY
0.0788
(0.0872)
AVG_FOREIGN
0.1272
(0.1722)
AVG_ln(SEGMENT)
0.2691
*
(0.1475)
Pseudo R
2
# of observations
4.88%
6,276
*,**, and *** correspond to 10%, 5% and 1% significance levels based on two-tailed hypotheses tests, respectively; standard errors are reported
in parentheses.
51
Table 9: Deregulation effects on competition
Panel A: Variable=GMARGIN
OTHER
TELECOM
TELECOM-OTHER
pre-Deregulation
0.14
0.14
0.00
post-Deregulation
0.12
0.10
-0.02***
-0.02***
-0.04 ***
OTHER
TELECOM
TELECOM-OTHER
pre-Deregulation
9.45
10.86
1.41***
post-Deregulation
10.71
12.13
1.42***
1.26***
1.27***
OTHER
TELECOM
TELECOM-OTHER
pre-Deregulation
6.43
9.05
2.62***
post-Deregulation
7.09
9.46
2.37***
0.66***
0.41***
OTHER
TELECOM
TELECOM-OTHER
pre-Deregulation
0.19
0.15
-0.04***
post-Deregulation
0.20
0.13
-0.07***
0.01***
-0.02***
Post-Pre Diff
Panel B: Variable=MKTSIZE
Post-Pre Diff
Panel C: Variable=ENTCOST
Post-Pre Diff
Panel D: Variable=HINDEX
Post-Pre Diff
*,**, and *** correspond to 10%, 5% and 1% significance levels based on two-tailed hypotheses tests, respectively
52
Table 10: Deregulation effects on earnings management
AAER FRAUD=1/0
Intercept
-5.2816 ***
(0.221)
DEREGULATION
0.6805 ***
(0.1318)
TELECOM
-13.9001 ***
(0.5212)
DEREGULATION×TELECOM
13.5151 ***
(0.5723)
AVG_ln(SALES)
0.1619 ***
(0.0327)
AVG_ln(MB)
0.3233 ***
(0.0548)
AVG_OPERATING_CYCLE
0.0000 ***
(0.0000)
AVG_CFO_VOLATILITY
-0.0120
(0.0146)
AVG_REVENUE_VOLATILITY
-0.0297
(0.0245)
AVG_FOREIGN
0.3045 **
(0.1429)
AVG_ln(SEGMENT)
-0.2362 **
(0.1193)
Pseudo R2
4.04%
# of observations
13,060
*,**, and *** correspond to 10%, 5% and 1% significance levels based on two-tailed hypotheses tests, respectively; standard errors are reported
in parentheses.
53
Download