Comparison of cosmetic earnings management for the developed markets and emerging markets some empirical evidence from the united states and taiwan

Comparison of cosmetic earnings management for the developed
markets and emerging markets: Some empirical evidence from the
United States and Taiwan
Fengyi Lin
a
, Sheng-Fu Wu
b,
⁎
a
Department of Business Management, College of Management, National Taipei University of Technology, Taipei, Taiwan
b
Institute of Industrial and Business Management, College of Management, National Taipei University of Technology, 1, Sec. 3, Chung-Hsiao E. Rd., Taipei, Taiwan
abstractarticle info
Article history:
Accepted 1 October 2013
Available online xxxx
Keywords:
Cosmetic earnings management
Corporate governance
Benford's law
This study examines the effect of the implementation of corporate governance regulations on cosmetic earnings
management in developed and emerging markets respectively. Using Benford's Law, the analysis employs 84,870
positive earnings observations for all publicly listed US and Taiwan companies from 1990 to 2011.
The empirical results show that, regardless of developed markets and emerging markets, the phenomenon of
cosmetic earnings management exists. In contrast to developed markets, corporate managers of emerging
markets have stronger incentives to manipulate earnings. More importantly, it was found that the degree of
earnings management is significantly less after implementing corporate governance regulations both in
developed and emerging markets. This result suggests that the implementation of corporate governance
regulations plays an important role in reducing the earnings manipulative behavior. The findings of the study
add more evidence to the ongoing debate about the effectiveness of corporate governance regulations in
preventing earnings management.

© 2013 Elsevier B.V. All rights reserved.
1. Introduction
Existing information asymmetry problem makes it difficult for
investors to understand the real underlying situation of firms. The
information asymmetry problem is more acute in emerging markets
or developing economies, making financial statements of firms
in developing countries more suspicious than those in developed
countries (Vives, 2006). Several researches suggest that the value
relevance of accounting information is lower inless developed countries
than in more developed countries (Biddle and Hilary, 2006; Biddle et al.,
2009; Hope and Thomas, 2008; McNichols and Stubben, 2008). High-
quality accounting information seems to be desirable in mitigating
information asymmetry for firms (Chen et al., 2011).
Earnings management is the manipulation of accounting numbers
within the scope of the Generally Accepted Accounting Principles
(GAAP) (Jackson and Pitman, 2001). Healy and Wahlen (1999) believed
that managers use subjective judgment in financial reporting or
transaction recognition to manipulate financial reports. Earnings
management is often considered materially misleading and thus a
fraudulent activity to the stakeholders, even though the changes may
follow all of the accounting standards and laws. Obviously the existence
of earnings management will reduce the quality of financial statements.
Excessive earnings management often causes serious corporate
fraud. To reduce the probability of occurrence of corporate fraud,
numerous countries have enacted laws to strengthen the corporate
governance mechanism, such as the United States and Taiwan.
1
World Bank (1999) defines that the complete corporate governance
framework consists of internal and external mechanisms. Good internal
corporate governance mechanisms, including ownership structure, the

board of directors, and timely and accurate disclosure of relevant
information, can reduce the earnings management motive of managers.
These internal mechanisms for corporate governance can be strength-
ened by external laws, rules, and institutions. In developed market
economies, these policies and institutions minimize the divergence
between s ocial and private returns and reduce costly a gency problems,
primarily through greater t ransparency, monit oring by regulatory and
self-regulatory bodies, and compliance mechanisms (World Bank, 1999).
The e xtent o f earnings management is strongly related to the
countries' institutional arrangements (Man and Wong, 2013). Compared
Economic Modelling 36 (2014) 466–473
⁎ Corresponding author.
E-mail addresses: , (S F. Wu).
1
The Sarbanes–Oxley Act of 2002 is one fundamental step toward enhancing the
quality of financial statements in the United States. In 2003, Taiwanese regulatory
authorities began implementing programs to strengthen the corporate governance
mechanism, including the Corporate Governance Best-Practice Principles (CGBPP) for
TWSE/GTSM-Listed Companies, Market Observation Post System, Corporate Governance
Best-Practice Principles for Securities Firms (2003), Corporate Governance Best-Practice
Principles for Future Firms, and Information Disclosure Assessment of Publicly Listed
Companies.
0264-9993/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
/>Contents lists available at ScienceDirect
Economic Modelling
journal homepage: www.elsevier.com/locate/ecmod
with emerging marke ts, deve loped markets have better inve stor
protection and more comprehensive legal systems. Burgstahler et al.
(2006) show that countries with stronger legal systems have lower
earnings management. In addit ion, countries with lower investor

protection usually have a higher extent o f earnings management (Leuz
et al., 2003). Therefore, we e xpect t hat emerg ing markets with weaker
investor protection could give inside managers more incent ive to
manipulate firm performance.
Prior research shows that the critical determinants of earnings
management have been separated into two major categories. In the
first category, zero is adopted as the threshold of earnings management.
Hayn (1995) suggested that firms avoid reported loss, conduct earnings
management, and cross over the zero-earnings thresholds. In the
second category, a key reference point, represented by n × 10
k
,is
used as the threshold of earnings management (Guan et al., 2006;
Herrmann and Thomas, 2005; Lin et al., 2011). For example, if net
income is expected to be $70 million but the actual earnings are only
$69 million, managers may have an incentive to adjust the earnings
data to allow the net income to achieve the expected earnings
threshold. Benford's law has been widely applied to financial data to
investigate instances of digital rounding.
Although substantial studies have been performed on digital
analysis using Benford's law, there is still little research to compare
the difference between developed and emerging markets. This study
aims at comparing the earnings management phenomenon between
developed and emerging markets by utilizing Benford's law. We
observe the existence of earnings adjustments exceeding the key re-
ference point, and analyze whether an earnings management anomaly
exists. Furthermore, has earnings management changed as govern-
ments gradually strengthen corporate governance mechanism? What
is the difference of cosmetic earnings management between a
developed market and an emerging market? Using Benford's law, this

study investigates the difference of cosmetic earnings management
between developed and emerging markets by observing the real
distribution of earnings numbers reported in the United States and
Taiwan. The study could supply more evidence to the ongoing debate
about the effectiveness of corporate governance regulations in
preventing earnings management.
2. Literature review
2.1. Corporate governance mechanism and earnings management
The occurrence of the agency problem results from the separation of
ownership and control. The managers might pursue their self-interest
to maximize their own wealth, perhaps at the expense of other
stakeholders' wealth and interests (Jensen, 1986). Contracts may
request the managers to disclose relevant accounting information in
order to protect the stakeholders' interests. However, due to accounting
information is provided by the managers, who may overstate the
numbers in the financial statements through their accounting estimates
and standards (Watts and Zimmerman, 1986). The existence of
earnings management will reduce the quality of financial statements.
Corporate governance w ould efficiently r educe the agency proble m
between shareholders and managers (Gompers et al., 2003). John
and Senbet (19 98) defined that corporate governance “deals with
mechanisms by which stakeholders of a corporation exercise control
over corporate insiders and manage ment such that their inter ests are
protected.” Man and Wong (2013) consider that an institutional
environment which provides robust legal protection can control
managers' self-interest to a c ertain extent. Prior researches have shown
that firms w ith effective governance mechanisms can successfully
minimize earnings management behavior (Dechow et al., 1995; Liu
and Lu, 2007; M arra et al., 2011; Peasnell et al., 2000; Shen and C hih,
2007; Zéghal et al., 2011).

Dechow et al. (1995) demonstrated that thorough governance
reduces the adverse effects of earnings management. They found that
firms that overstate earnings are more likely to have a board with inside
directors and a CEO serving as the board chair. Peasnell et al. (2000)
showed similar findings in which firms with a high proportion of
outside directors are unlikely to take earnings manipulation when
earnings fall below the threshold. Liu and Lu (2007) investigate the
relation between earnings management and corporate governance in
the Chinese listed companies. They demonstrate that firms with higher
corporate governance levels have lower levels of earnings management.
Chang and Sun (2009) fi
nd earnings management to be negatively
r
elated to the independence of the audit committee and the board of
directors after SOX. Their findings suggest that the SOX provisions
improve the effectiveness of cross-listed foreign firms' corporate-
governance functions in monitoring the quality of ac counting
earnings. Zéghal et al. (2011) show a positive influence of external
audit quality on reducing earnings management. Marra et al.
(2011) found an increase in the influence of audit committees in
strengthening the financial reporting quality after the introduction
of the International Financial Reporting Standards (IFRS). The result
shows that corporate governance mechanis ms are key factors in
earnings quality.
There are differences in corporate governance between emerging
markets and developed markets. Prior studies find that the char-
acteristics of weak investor protection institutions involve severe
earnings management and a low level of earnings information
(DeFond et al., 2007; Leuz et al., 2003). Several studies demonstrate
that it is less likely for managers to manipulate earnings when there is

greater legal protection (Nenova, 2003; Shleifer and Wolfenzon,
2002). Developed markets tend to have more extensive disclosure
requirements, stronger private and public enforcement of security
regulations, and stronger shareholder and creditor rights to reduce the
level of managerial discretion.
Leuz et al. (2003) find that earnings management decreases in
countries with stronger investor protection. Ball et al. (2003) argue
that the institutional arrangement of a country is the most important
feature in controlling managers' self-interest, reducing opportunistic
earnings manipulation, and improving the quality of financial
statements. Shen and Chih (2007) show that earnings management
lowers in countries with stronger investor protection and more
transparent accounting disclosure. Legal systems protect stakeholders'
rights by conferring on their powers to discipline managers as well as
by enforcing contracts designed to limit managers' benefits ( Claessens
et al., 2002; Dyck and Zingales, 2004; La Porta et al., 1998).
Mehmet and Emin (2012) find that whether international big audit
firms provide high quality services or not, the audit environment, which
is affected directly by the legal environment and effectiveness of the
legal system, is more important than audit quality. Firms in a strong
enforcement environment seem to induce a decrease in the level of
discretionary accruals, compared to firms in the weak environment.
The quality of a government depends on other institutional constraints
such as constitution, laws, and the political system. Firms under the
influence of a low quality government tend to have complex orga-
nizational structures, poor transparency and weak corporate gover-
nance (e.g., Fan et al., 2012; Jiang et al., 2010; Leuz and Oberholzer-
Gee, 2006).
Poor disclosure and financial opacity are common characteristics of
emerging market firms. It is well acknowledged that the financial

opacity of emerging market firms cannot be improved by changing
the accounting system alone, because the enforcement of accounting
rules depends on strong institutions which are lacking in these markets
(Ball et al., 2000
). In countries with stronger investor protection laws,
m
anagers and controlling shareholders are less likely to expropriate
the firm's resources and more likely to invest in projects that benefit
shareholders (Bekaert et al, 2010; Shleifer and Wolfenzon, 2002;
Wurgler, 2000).
467F. Lin, S F. Wu / Economic Modelling 36 (2014) 466–473
In the past two decades, many cases of corporate fraud occurred in
the world. Numerous countries have enacted laws to strengthen the
corporate governance mechanism, such as the United States and
Taiwan. Previous studies have demonstrated that good corporate gover-
nance helps to reduce the degree of earnings management. However,
there is less literature to investigate the difference of cosmetic earnings
management between a developed market and an emerging market.
This study attempts to investigate several issues. Do managers reduce
the extent of cosmetic earnings management with the increasingly
stringent legal regulations? What is the difference of cosmetic earnings
management between a developed market and an emerging market?
2.2. Benford's law and earnings management
The earnings management issue has become a concern throughout
the world ( Islam et al., 2011). Earnings management is the managerial
use of procedures to adjust data on financial reports. Earnings manage-
ment may mislead stakeholders of the firm's corporate performance.
This behavior can also explain the contractual behavior of senior
management using accounting data (Healy and Wahlen, 1999).
The major types of earnings management include the selection of

the timing of new accounting principles, the selection of accounting
standards, the control of transaction times, and the adjustment of
discretionary accruals (Dechow et al., 1995; Degeorge et al., 1999;
Jones, 1991).
Each earnings management method applied by managers affects the
presentation of financial statements, which are the main resource for
outside stakeholders (including investors and creditors) to understand
the situation of corporate operating performance. Hence, data
adjustment of financial reports ultimately affects the assessment of a
firm by outside stakeholders. Several studies have shown that managers
have an incentive to manipulate earnings to reach specific thresholds
(Barth et al., 1999; Matsumoto, 2002; Skinner and Sloan, 2001).
Since the mid-1980s, there has been explosive growth in using
accruals to detect earnings management (Sun and Rath, 2010). The
most popular accrual models are the standard Jones and modified
Jones models (Islam et al., 2011). Substantial studies have been
performed in using accrual models to detect earnings management.
However, several studies found that the accrual models are of low
power in detecting earnings management (Beneish, 1997; Islam et al.,
2011; Thomas and Zhang, 2000; Yoon et al., 2006).
Kinnunen and Koskela (2003) define cosmetic earnings management
(CEM) by small upward rounding of reported net income that generates
more than expected zeros and less than expected nines as second digit of
earnings numbers. Thomas (1989) considered two reasons that
managers maybe engage in cosmetic earnings management. One reason
relates to earnings numbers as key cognitive reference points. The use of
lending, bonus and option contracts provides another reason why
managers round earnings numbe rs upward once in a while.
This study detects cosmetic earnings management by observing the
real distribution of earnings numbers. Benford's law has recently

become an accepted tool in the identification of contrived data, both
in academic literature and practice (Carslaw, 1988; Herrmann and
Thomas, 2005; Lin et al., 2011; Reddy and Sebastin, 2012; Thomas,
1989, 2012).
Examining the distribution of digits in earnings numbers to identify
earnings management has a number of appealing features. First, the
researchers don't have to estimate the potentially noisy abnormal
accruals (Healy and Wahlen, 1999). Another appealing feature is that
the researchers can identify a large set of potential earnings
manipulators without invoking specific assumptions about earnings
management motivation or methods (Burgstahler and Dichev, 1997).
Benford (1938) demonstrated that the expected distributions of
naturally occurring numbers are skewed toward one for the
firs
t digits
(because zero cannot be a first digit) and zero for the second digit.
Benford's law provides the basis for the numerical analysis of a
sequence of numbers of a similar nature. The deviation in actual data
from these expected frequencies indicates the presence of manipulation
(Thomas, 2012).
Rodriguez (2004) provided empirical evidence that, in the absence
of earnings management, corporate earnings follow Benford's law.
Durtschi et al. (2004) further examined the use of Benford's law in the
detection of accounting fraud by specifically identifying data sets
expected to follow Benford's law and the types of fraud that can be
detected.
In subsequent studies, researchers have extended the various
analytic methods of Benford's law, such as increasing the number of
digits used in analysis (Diekmann, 2007; Skousen et al., 2004)and
investigating the heaping anomaly (Herrmann and Thomas, 2005; Lin

et al., 2011).
Thomas (2012) examined the extent to which firms manipulate
their financial statement numbers by engaging in cosmetic earnings
management in a post Sarbanes–Oxley Act (SOX) environment. Using
2009 data, Thomas found no evidence of cosmetic earnings manage-
ment, indicating that the SOX has increased financial statement
reliability and reduced earnings management.
The summarized studies have shown clear evidence that Benford's
law can be used to analyze cosmetic earnings management behavior.
Therefore, this study utilizes Benford's law to investigate the extent of
changes on cosmetic earnings management, which uses Taiwan
2
as
the proxy of emerging market and the United States as the proxy of
developed market.
3. Hypotheses and m athematical model
3.1. Hypotheses
Man and Wong (2013) consider that an institutional environment
which provides robust legal protection can control managers' self-
interest. Prior studies show that it is less likely for managers to
manipulate earnings when there is greater legal protection (Nenova,
2003; Shleifer and Wolfenzon, 2002). Developed markets tend to have
more extensive disclosure requirements, more complete regulatory
mechanisms and laws, and stronger shareholder and creditor rights so
as to reduce the level of managerial discretion. Therefore, this study
hypothesized that the degree of earnings management of developed
markets is weaker than that of emerging markets.
Formally, our first hypothesis is stated as follows: (in the null form)
H1. The degree of earnings management of publicly listed companies in
developed markets will not be significantly weaker than companies in

emerging markets.
As discussed, corporate governance mechanisms can successfully
minimize earnings management behavior (Dechow et al., 1995; Jo and
Kim, 2007; Marra et al., 2011; Park and Shin, 2004; Peasnell et al.,
2000). Effective corporate governance mechanisms can substantially
reduce incentives for management to manipulate earnings. Since
2002, the United States and Taiwan have enacted laws to strengthen
the corporate governance mechanism. Cohen et al. (2010) demonstrate
2
Taiwan is classified as the emerging market by many research institutions, including
the Economist, Standard and Poor's (S&P), FTSE (Financial Times Stock Exchange) Group,
Columbia University EMGP (Emerging Market Global Players) List, and BBVA (Banco
Bilbao Vizcaya Argentaria) Research. The FTSE Group (2010) distinguishes between
advanced and secondary emerging markets on the basis of their national income and
the development of their market infrastructure. The advanced emerging markets are
classified as such because they are upper or lower middle income GNI countries with
advanced market infrastructures or high income GNI countries with lesser developed
market infrastructures, such as Taiwan and Turkey. The secondary emerging markets
include some low income, lower middle income, upper middle income and high income
GNI countries with reasonable market infrastructures and significant size and some upper
middle income GNI countries with lesser developed market infrastructures, such as China
and India.
468 F. Lin, S F. Wu / Economic Modelling 36 (2014) 466–473
that the corporate governance environment has significantly improved
in th e Post-Sarbanes–Oxley Era. We expect that the earnings
management level will continue to decline because of the gradual
implementation of corporate governance mechanisms since 2003. To
investigate the differences before and after the strengthening of the
corporate governance mechanism, the sample was divided into two
periods in this study, using 2003 as the division point. Formally, the

second hypothesis is stated as follows:
H2a. The degree of cosmetic earnings management of publicly listed
companies in developed markets after 2003 will not be significantly weaker
than before 2003.
H2b. The degree of cosmetic earnings management of publicly listed
companies in emerging markets after 2003 will not be significantly weaker
than before 2003.
3.2. Mathematical model
3.2.1. Benford's law
To test our hypotheses, we identify the expected proportions of each
of the 10 digits (zero to nine) in each place of the earnings numbers
under the null hypothesis. The true distribution of the digits without
managerial manipulation of the reported earnings is not publicly
observable (Thomas, 1989). Therefore, we approximate this distribu-
tion with Benford's law (Carslaw, 1988).
Benford (1938) demonstrated that, contrary to basic intuition, the
expected distributions of naturally occurring numbers are skewed
toward one for the firstdigits(becausezerocannotbeafirst digit)
and zero for the second digit. If earnings management is conducted by
achieving the key reference point represented by n × 10
k
,anabnormal
distribution of the digits in the place to the right of the reference point
is expected. For example, if the key reference point is the second digit
of positive earnings and management tends to distort earnings to
achieve this key point, more zeros and fewer nines are expected in the
third place of the earnings numbers.
Benford postulated that the expected proportions or occurrences of
anumberasthefirst di git in a number series ca n be approximated by
the following relation:

proportion a is the first digitðÞ¼log
10
a þ 1ðÞ− log
10
aðÞ: ð1Þ
Furthermore, the expected proportion of the given number a as the
first digit and the number b as the second digit can be found in the
following relation:
log
10
a þ
b þ 1
10

− log
10
a þ
b
10

: ð2Þ
Using the established equations and summing all possible a values
for any b value produce an overall expected proportion for b as the
second digit. This equation is presented as follows:
proportion b is the second digitðÞ¼
X
log
10
a þ
b þ 1

10

− log
10
a þ
b
10

:
ð3Þ
The expected proportion of numbers in the third, fourth, fifth digits,
and so on, can be similarly derived.
3.2.2. Chi-square test
The chi-square test has often been used to test for conformity to
Benford's law (Nigrini, 2012). The chi-square test is an extension of
the z test, which tests only one digit at a time. The chi-square test
combines the results of testing each digit's expected frequency with
each digit's actual frequency into one test statistic. If the chi-square
test rejects the hypothesis that the probability of all digits conforms to
the Benford law, then the entire account warrants further examination.
The chi-square test is generally less discriminatory than the individual
z-test results, but results in fewer false positives (Durtschi et al., 2004).
The chi-square test is presented as follows:
χ
2
¼
X
9
i¼1
nP

0
−nP
e
½
2
nP
0
for the first digit ð4Þ
χ
2
¼
X
9
i¼0
nP
0
−nP
e
½
2
nP
0
for the other digits ð5Þ
where P
e
and P
0
are the observed and expected proportions,
respectively. The sample size is represented by n.
3.2.3. Z statistic

If the chi-square value is significant, the number that has deviated
from Benford's law and the degree of the deviation can be identified
by examining the Z statistic on numbers zero to nine. To test our null
hypothesis of no managerial effort to round earnings, we compared
the observed frequency of each number x in various places of the
earnings numbers to the expected occurrences of the number as
predicted by Benford's law [Eqs. (1) through (3)]. To perform a sig-
nificance test of the observed deviations from the expected proportions,
we used a normally distributed Z statistic:
Z ¼
P
0
−P
e
jj
−
1
2n
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
P
0
1−P
0
ðÞ
n
r
: ð6Þ
The second term in the numerator is a correction term; it should be
applied only when it is smaller than |P
0

− P
e
|(Thomas, 1989). These Z
statistics reject the null hypothesis at the 10%, 5%, and 1% levels if
their values exceed 1.64, 1.96, and 2.57, respectively.
3.2.4. Cramer's V
Cramer's V was used as a post test to determine the strengths of
association after the chi-square t est determined significance. The chi-
square test demonstrated that a significant relationship existed between
variables, but cannot identify the extent to which the significance occurs.
Cramer's V is based on adjusting the chi-square significance to factor out
sample size. Cramer's V varies between zero and one. A value close to
zero shows li ttle association between the vari ables. Values close to o ne
indicate a strong associ ation between the variables.
We used Cramer's V to compare the level of deviation from Benford's
law by different groups, and the related equation is as follows:
Cramer
0
sV¼
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
χ
2

nk−1ðÞ
s
ð7Þ
where n and k are the sample size and number of variables, respectively.
4. Empirical results
4.1. Data
The data used in this study were obtained from the Compustat

Research Insight (Compustat) database and the Taiwan Economic
Journal (TEJ) database. The analysis includes the annual net incomes
of firms listed on the stock exchanges of the United States and Taiwan
from 1990 to 2011. After deleting incomplete data and 1% extreme
values, t he final sample consisted of 84,870 positive earnings
observations, including 65,977 observations in the United States and
18,893 observations in Taiwan.
469F. Lin, S F. Wu / Economic Modelling 36 (2014) 466–473
4.2. Test of hypothesis 1
Tables 1 and 2 list the distributions of each number (zero to nine)
appearing in the first through third places of positive earnings of U.S.
companies and Taiwanese companies, respectively.
3
The first number
in each cell of the table represents the observed proportion of the
sample data (in terms of a percentage of the sample). The second
number is the proportion predicted by Benford's law. The third number
represents the difference between the actual and expected proportions
of the sample (in terms of a percentage of the sample). The last number
represents the Z statistic result.
In Table 1, the results of a chi-square test of the first through third
places were 16.58, 36.57, and 53.86, respectively. The results indicate
that managers in the U.S. have strong incentives to manipulate earnings
by exaggerating the earnings numbers.
The distribution of the first digits reveals that number seven was
observed more frequently than expected, suggesting that firms are
more likely to round to numbers when seven is set as the first di git. In
addition, number nine was observed less frequently than expected.
The lack of number nine as the first d igit suggests that firms are more
likely to round when the first digit is any of these numbers than they

are to round to a figure that has any of these numbers as the first digit.
Moreover, consistent with our expectations, significantly more zeros
(or ones) and fewer nines occurred in the second and third places,
suggesting that firms may use the first and second digits as reference
points.
4
The proportion of zeros as the second digit, expected to be
11.97% of the sample, was actually 3.84% higher, and the Z statistic
was 3.67. The number nine exhibited a rate of deviation of −5.53%
and a Z statistic of 4.33. This indicates that firms are likely to use the
number zero as the key reference point for the second digit, causing
the anomaly of more zeros than nines. Similarly, the proportion
of ones, twos, threes and fives as the third digit was higher than
expected. The proportion of eights and nines was lower than expected
simultaneously.
In addition, we found the similar earnings management situation in
Taiwan. In Table 2, the results of a chi-square test of the first through
third places were 26.96, 54.43, and 19.89, respectively, suggesting that
managers in the Taiwan also have strong incentives to manipulate
earnings by exaggerating the earnings numbers. There are significantly
more zeros and fewer nines occurring in the second and third places.
The proportion of zeros as the second and third digits was 9.11% and
8.25% higher than expected, respectively (The Z statistics were 4.6 and
3.82). The proportion of nines as the second and third digits was
12.47% and 4.27% lower than expected, respectively (The Z statistics
were 5.23 and 1.93). This result concurs with prior studies (Carslaw,
1988; Herrmann and Thomas, 2005; Lin et al., 2011; Thomas, 1989)
and suggests that window dressing is a pervasive phenomenon.
A chi-square test is affected by the size of a sample. To eliminate the
problem of sample size, we used Cramer's V to compare the level of

deviation from Benford's law of different groups. After controlling for
the size effect, the results show that the Cramer's V of the first through
Table 1
Distribution of first through third digits in positive annual net income of companies in the United States from 1990 to 2011.
Number 0 1 2 3 4 5 6 7 8 9 Chi-square Cramer's V
First digit
(n = 65,977)
Observed proportion – 30.18 17.65 12.29 9.65 8.06 6.61 5.97 5.22 4.36 16.58** 0.0056
Expected proportion – 30.1 17.61 12.49 9.69 7.92 6.7 5.8 5.12 4.58
Deviation rate – 0.27 0.23 −1.60 −0.41 1.77 −1.34 2.93 1.95 −4.80
Z-statistics – 0.45 0.30 1.59 0.36 1.35 0.84 1.86* 1.25 2.68***
Second digit
(n = 65,875)
Observed proportion 12.43 11.44 10.8 10.6 9.84 9.77 9.45 8.91 8.73 8.03 36.57*** 0.0079
Expected proportion 11.97 11.39 10.88 10.43 10.03 9.67 9.34 9.04 8.76 8.5
Deviation rate 3.84 0.44 −0.74 1.63 −1.89 1.03 1.18 −1.44 −0.34 −5.53
Z-statistics 3.67*** 0.40 0.69 1.41 1.63 0.85 0.97 1.11 0.21 4.33***
Third digit
(n = 64,868)
Observed proportion 10.1 10.46 10.37 10.39 9.88 10.35 9.8 9.72 9.35 9.57 53.86*** 0.0096
Expected proportion 10.18 10.14 10.1 10.06 10.02 9.98 9.94 9.9 9.86 9.83
Deviation rate −0.79 3.1 6 2.67 3.28 −1.40 3.71 −1.41 −1.82 −5.17 −2.64
Z-statistics 0.66 2.70*** 2.30** 2.79*** 1.15 3.17*** 1.16 1.52 4.35*** 2.19**
Notes: Theobserved, expected proportion and deviation rate are measured asthe percentage of the sample. The deviation rate=(observed proportion − expected proportion)∕ (expected
proportion). *, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively (two-tailed test).
Table 2
Distribution of first through third digits in positive annual net income of companies in Taiwan from 1990 to 2011.
Number 0 12 3456 789 Chi-squareCramer'sV
First digit
(n = 18,893)

Observed proportion – 29.74 16.64 12.47 9.67 8.18 7.29 6.08 5.32 4.6 26.96*** 0.0134
Expected proportion – 30.1 17.61 12.49 9.69 7.92 6.7 5.8 5.12 4.58
Deviation rate – −1.20 −5.51 −0.16 −0.21 3.28 8.81 4.83 3.91 0.44
Z-statistics – 1.08 3.50*** 0.09 0.08 1.34 3.28*** 1.62 1.29 0.17
Second digit
(n = 18,893)
Observed proportion 13.06 11.27 11.06 10.62 10.25 10.02 8.92 8.75 8.61 7.44 54.43*** 0.0179
Expected proportion 11.97 11.39 10.88 10.43 10.03 9.67 9.34 9.04 8.76 8.5
Deviation rate 9.11 −1.05 1.65 1.82 2.19 3.62 −4.50 −3.21 −1.71 −12.47
Z-statistics 4.60*** 0.49 0.78 0.82 1.00 1.65* 1.96** 1.36 0.72 5.23***
Third digit
(n = 18,893)
Observed proportion 11.02 10.19 10.28 10.04 9.84 10.03 9.69 9.8 9.69 9.41 19.89** 0.0108
Expected proportion 10.18 10.14 10.1 10.06 10.02 9.98 9.94 9.9 9.86 9.83
Deviation rate 8.25 0.49 1.78 −0.20 −1.80 0.50 −2.52 −1.01 −1.72 −4.27
Z-statistics 3.82*** 0.24 0.84 0.06 0.78 0.22 1.15 0.45 0.78 1.93*
Notes: The observed, expected proportion and deviation rate are measured asthe percentage of the sample. The deviation rate=(observed proportion − expected proportion)∕ (expected
proportion). *, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively (two-tailed test).
3
Based on the following two reasons, this study focuses on the first through third
places. The first reason is that previous studies show that the fourth place is less
significant. The second reason is that investors have relatively less attention to the fourth
digit data.
4
The phenomenon of earnings manipulation still exists in companies with negative
earnings. There are significantly more nines and fewer zeros occurring in the second or
third places. This result concurs with prior studies (Aono and Guan, 2008; Thomas, 1989).
470 F. Lin, S F. Wu / Economic Modelling 36 (2014) 466–473
third digits of the samples in Taiwan was larger than that of the samples
in the United States. For example, the Cramer's V of the second digit in

Taiwan was 0.0179, which is larger than the Cramer's V of the second
digit in the United States (0.0079). Overall, the degree of deviation in
Taiwan is larger than the degree of deviation in the United States. The
results reject the hypothesis H1, indicating that corporate managers of
emerging markets have stronger incentives to manipulate earnings by
exaggerating the earnings numbers.
4.3. Test of hypothesis 2
Since 2002, the United States and Taiwan have enacted laws to
strengthen the corporate governance mechanism. We expect that the
cosmetic earnings management level will continue to decline because
of the gradual implementation of corporate governance mechanisms
since 2003. Therefore, the sample period was divided into two groups,
1990 to 2002 and 2003 to 2011, to investigate whether differences in
cosmetic earnings management behavior occurred between these two
periods. The results are presented in Tables 3 and 4.
For developed markets, such as the United States, the first through
third digits in positive earnings before 2003 and the third digit after
2003 revealed that the chi-square value is s ignificant. W e found that
the distribution of t he first and second digits in the earnings number s
shows s ignificant differe nces between t hese two p eriods. The firs t digit
wasmorelikelytobemanipulated before 2003 (with a chi-square
value of 16.78) than after 2003 (with a chi-square value of 12.57).
Similarly, the second digit was more likely to be manipulated before
2003 (with a chi-square value of 38.22) than after 2003 (with a chi-
square value of 12.47). In contrast, the first t hrough third digits w ere
less frequently manipulated after 2003 than before 2003. The results
reveal that manipulation behavior is dec re asi ng in develo ped markets.
Moreover, to control for the effect of inconsistency of the number of
samples of different groups, we also calculated the Cramer's V. We
found that the first through third digits of the samples before 2003

obtain higher Cramer's V (0.0082; 0.0116; and 0.0122, respectively)
than the samples after 2003 (0.0068; 0.0063; 0.009). Therefore, our
results reject the hypothesis H2a, indicating that improvements to the
corporate governance mechanisms in the United States led to less
cosmetic earnings management in publicly listed firms.
For emerging markets, such as Taiwan, the first through second
digits in positive earnings before 2003 and the second through third
digits after 2003 revealed that the chi-square value is significant. We
first look at the distribution of digits in the second place of the earnings
numbers. Our results show that the proportion of nines as the second
digit, expected to be 8.5% of the sample, was actually significantly
lower in both periods. The Z statistics of nine as the second digit in the
first and second periods were 3.23 and 4.14, respectively. Moreover,
Table 4 also reveals a greater number of zeros in the second place of
earnings (the Z statistics were 1.92 and 4.54, respectively). This implies
that the second-place digit was manipulated in both periods.
In addition, we found that the distribution of the first and third digits
in the earnings numbers shows significant differences between the two
periods. The first digit was more like ly to be ma nipulated bef ore 2003
(with a chi-square value of 26.33) than after 2003 (with a chi-square
value of 7.03). And the third digit was more likely to be manipulated
after 2003 (with a chi-square value of 19.45) than before 2003 (with a
chi-square value of 14.08). These results show that managers used the
first and second digits as reference points to round earnings before
2003. In contrast, the second through third digits were more frequently
manipulated after 2003 than before 2003.
The Cramer's V of the second and third digits of the samples after
2003 was larger than the samples before 2003 in Taiwan. The result
shows that t he effect of r educing cosmetic earnings m anagement is not
full

y revealed although Taiwan is also committed to the improvement
Table 3
Distribution of first through third digits in positive annual net income of companies in the United States at different periods.
PeriodNumber0123456789Chi-squareCramer'sV
First digit
(n = 65,977)
1990–2002
(n = 31,536)
Deviation rate – 0.52 −0.49 −4.70 0.49 2.88 − 1.01 4.82 2.04 −1.67 16.78** 0.0082
Z-statistics – 0.60 0.39 3.14*** 0.27 1.49 0.47 2.11** 0.83 0.64
2003–2011
(n = 34,441)
Deviation rate – 0.04 0.94 1.16 −1.28 0.81 −1.42 1.19 2.18 −7.62 12.57 0.0068
Z-statistics – 0.04 0.80 0.81 0.77 0.43 0.70 0.54 0.93 3.08***
Second digit
(n = 65,875)
1990–2002
(n = 31,473)
Deviation rate 7.12 −0.24 −1.75 −0.29 −3.17 2.67 1.48 −0.93 −0.40 −6.62 38.22*** 0.0116
Z-statistics 4.65*** 0.14 1.08 0.17 1.87* 1.54 0.83 0.51 0.21 3.57***
2003–2011
(n = 34,402)
Deviation rate 0.92 1.07 0.12 3.37 −0.75 −0.48 0.93 −1.78 −0.15 −4.55 12.47 0.0063
Z-statistics 0.62 0.70 0.07 2.12** 0.46 0.28 0.54 1.03 0.08 2.56**
Third digit
(n = 64,868)
1990–2002
(n = 30,912)
Deviation rate −0.
39 3.67 6.50 2.13 −2.09 3.90 −3.28 −2.15 −5.42 −3.15 41 .59*** 0.0122

Z-statistics 0.22 2.16** 3.82*** 1.24 1.22 2.27** 1.9 0* 1.25 3.14*** 1.82*
2003–2011
(n = 33,956)
Deviation rate −1.13 2.69 −0.74 4.34 −0.70 3.62 0.35 −1.50 −4.94 −2.12 24.6*** 0.009
Z-statistics 0.69 1.65* 0.45 2.66*** 0.42 2.21** 0.20 0 .91 3.00*** 1.28
Notes: The deviation rate is measured as the percentage of the sample. The deviation rate = (observed proportion − expected proportion)/(expected proportion). *, **, and *** denote
significance at the 10%, 5%, and 1% levels, respectively (two-tailed test).
Table 4
Distribution of first through third digits in positive annual net income of companies in Taiwan at different periods.
Period Number 0 1 2 3 4 5 6 7 8 9 Chi-square Cramer's V
First digit
(n = 18,893)
1990–2002
(n = 9222)
Deviation rate – −2.42 −7.14 −0.36 1.49 4.90 11.59 6.59 8.75 −2.37 26.33*** 0.0189
Z-statistics – 1.51 3.15*** 0.12 0.45 1.36 2.96*** 1.55 1.93* 0.47
2003–2011
(n = 9671)
Deviation rate – −0.04 −3.99 −0.02 −1.84 1.86 6.41 3.06 −0.34 3.49 7.03 0.0095
Z-statistics – 0.02 1.80* 0.01 0.57 0.52 1.67* 0.73 0.05 0.73
Second digit
(n = 18,893)
1990–2002
(n = 9222)
Deviation rate 5.46 −2.98 2.24 3.52 2.26 7.34 −4.07 −2.91 −3.17 −11.08 23.76*** 0.0169
Z-statistics 1.92* 1.01 0.73 1.14 0.71 2.29** 1.24 0.86 0.92 3.23***
2003–2011
(n = 9671)
Deviation rate 12.58 0.87 1.10 0.10 2.15 0.21 −4.87 −3.41 − 0.34 −13.87 37.96*** 0.0209
Z-statistics 4.54*** 0.29 0.36 0.02 0.69 0.05 1.52 1.04 0.09 4.14***

Third digit
(n = 18,893)
1990–2002
(n = 9222)
Deviation rate 8.24 −0.
74 −3.88 3.94 − 0.42 − 0.57 0.47 0.42 −0.51 −7.20 14.08 0.013
Z-statistics 2.65*** 0.22 1.23 1.25 0.12 0.17 0.13 0.12 0.15 2.26**
2003–2011
(n = 9671)
Deviation rate 8.30 1.79 7.32 − 4.07 − 2.98 1.55 −5.44 −2.36 − 2.93 −1.51 19.45** 0.0149
Z-statistics 2.73*** 0.57 2.40** 1.32 0.96 0.49 1.76* 0.75 0.94 0.47
Notes: The deviation rates are measured as the percentage of the sample. The deviation rate = (observed proportion − expected proportion)∕ (expected proportion). *, **, and *** denote
significance at the 10%, 5%, and 1% levels, respectively (two-tailed test).
471F. Lin, S F. Wu / Economic Modelling 36 (2014) 466–473
of corporate governance mechanisms. Therefore, the results do not reject
the hypothesis H2b. Nevertheless, even if this is the case, the effect of the
improvement of c orporate governance cannot b e ignored. The Cramer's
Vofthefirst d igit from years 1990 t o 2002 was 0.0189, which is l arger
than the Cramer's V of the first digit from 2003 to 2011 (0.0095).
Therefore, the d egree of deviation of the first digits from y ears 1990 to
2002 is larger than the degree of deviation from 2003 to 2011. This is
because improvement of the corporate governance mechanisms has
caused earnings management to become more difficult than in the
years preceding 2003. Consequently, the adjusted digits have moved
backwards. In contrast, the developed markets have g reater improve-
ment than the emerging markets.
Comparing different periods for Taiwan and the United States by
using Cramer's V, regardless of any period, this study found that the
degree of deviation in Taiwanese companies is larger than the degree
of deviation in the U.S. companies (Table 5). For example, the first

through third digits of earning in Taiwanese companies after 2003
obtain the higher Cramer's V (0.0095; 0.0209; 0.0149, respectively)
than U.S. companies after 2003 (0.0068; 0.0063; 0.0090). Our results
showed that cosmetic earnings management of emerging markets
may be more serious than developed markets. This result also supports
with hypothesis H1 again.
5. Conclusion
This study examines the earnings of companies publicly listed in
Taiwan and America from 1990 to 2011. This study investigated whether
corporate managers adopt accounting adjustments to engage in
cosmetic earnings management behavior by using Benford's law. T he
results show that in both developed markets and emerging markets,
publicly listed companies engage in the practice of cosmetic e arnings
management. It is generally recognized that the developed markets
have more complete regulatory mechanisms and laws, and earnings
are less likely to be manipulated by corporate managers. This study
provides empirical evidence that corporate managers of emerging
markets have stronger incentives to manipulate earnings by exag-
gerating the e arnings numbers.
Over the last two decades, corp orate governance has attracted a
considerable amount of public interest because of increased instances of
corporate fraud both domestically and abroad. Numerous countries
have enacted laws to strengthen their corporate governance mechanisms.
This study documents pervasive evidence that improvements to the cor-
porate governance mechanisms led to less cosmetic e arnings manage-
ment in publicly listed firms, regardless of developed markets and
emerging markets. This result suggests that the implementation of
corporate governance regulations plays an important role in reducing
the earnings manipulative behavior. In contrast, the developed markets
have greater improvement than the emerging markets. Overall, our

results show that cosmetic earnings management of financial statements
has become more difficult for managers because of improvements to the
corporate governance mechanisms. The findings of the study add more
evidence to the ongoing debate about the effectiveness of corporate
governance regulations in preventing earnings management. These
findings have implications for supervisory authority and investors. In
recent years, the International Accounting Standards (AIS) in the world
is being widely recognized and promoted, the accounting items of
financial s tatements are be coming more flexible, and t he importance o f
vigorous corporate governance mechanisms is increasing. We recom-
mend supervisory authority to further enhance corporate governance
regulations and mechanisms to minimize earnings manipulation. For
international investors, this study suggests that investors should pay
more attention t o the accuracy of financial statements when investing
in emerging mar kets.
References
Aono, J.Y., Guan, L., 2008. The impact of Sarbanes–Oxley act on cosmetic earnings
management. Res. Account. Regul. 20, 205–215.
Ball, R., Kothari, S.P., Robin, A., 2000. The effect of institutional factors on properties of
accounting earnings: international evidence. J. Account. Econ. 29, 1–52.
Ball, R., Robin, A., Wu, J., 2003. Incentives versus standards: properties of accounting
income in four East Asian countries. J. Account. Econ. 361 (3), 235–270.
Barth, M., Elliot, J., Finn, M., 1999. Market rewards associated with patterns of increasing
earnings. J. Account. Res. 37, 387–413.
Bekaert, G., Harvey, C., Lundblad, C., 2010. Financial openness and productivity. World
Dev. 39, 1–19.
Benford, F., 1938. The law of anomalous numbers. Proc. Am. Philos. Soc. 784, 551–572.
Biddle, G., Hilary, G., 2006. Accounting quality and firm-level capital investment. Account.
Rev. 81 (5), 963–982.
Biddle, G., Hilary, G., Verdi, R.S., 2009. How does financial reporting quality relate to

investments efficiency? J. Account. Econ. 48 (2–3), 112–131.
Burgstahler, D.C., Hail, L., Leuz, C., 2006. The importance of reporting incentives: earnings
management in European private and public firms. Account. Rev. 815, 983–1016.
Carslaw, C., 1988. Anomalies in income numbers: evidence of goal oriented behavior.
Account. Rev. 632, 321–327.
Chang, J.C., Sun, H.L., 2009. Crossed-listed foreign firms' earnings informativeness,
earnings management and disclosures of corporate governance information under
SOX. Int. J. Account. 44 (1), 1–32.
Chen, F., Hope, O.K., Li, Q., Wang, X., 2011. Financial reporting quality and investment
efficiency of private firms in emerging markets. Account. Rev. 86 (4), 1255–1288.
Claessens, S., Djankov, S., Fan, J., Lang, L., 2002. Disentangling the incentive and
entrenchment effects of large shareholdings. J. Finance 57, 2741–2772.
Cohen, J., Krishnamoorthy, G., Wright, A., 2010. Corporate governance in the post-
Sarbanes–Oxley era: auditors' experiences. Contemp. Account. Res. 27 (3), 751–786.
Dechow, P.M., Sloan, R.G., Sweeney, A.P., 1995. Detecting earnings management. Account.
Rev. 70, 193–225.
DeFond, M., Hung, M.Y., Trezevant, R., 2007. Investor protection and the information
content of annual earnings announcements: international evidence. J. Account.
Econ. 431, 37–67.
Degeorge, F., Patel, J., Zeckhauser, R., 1999.
Earnings management to exceed thresholds.
J.
Bus. 72, 1–33.
Diekmann, A., 2007. Not the first digit! Using Benford's Law to detect fraudulent scientific
data. J. Appl. Stat. 34, 321–329.
Durtschi, C., Hillison, W., Pacini, C., 2004. The effective use of Benford's Law to assist in
detecting fraud in accounting data. J. Forensic Account. 51, 17–34.
Dyck, A., Zingales, L., 2004. Private benefits of control: an international comparison.
J. Finance 59 (2), 537–600.
Fan, J.P.H., Wong, T.J., Zhang, T., 2012. Institutions and organizational structure: the case of

state-owned corporate pyramids. J. Law Econ. Org.
Gompers, P., Ishii, J., Metrick, A., 2003. Corporate governance and equity prices. Q. J. Econ.
1181, 107–155.
Guan, L., He, D., Yang, D., 2006. Auditing, integral approach to quarterly reporting, and
cosmetic earnings management. Manag. Audit. 216, 569–581.
Hayn, C., 1995. The information content of losses. J. Account. Econ. 202, 125–153.
Healy, P.M., Wahlen, J.M., 1999. A review of the earnings management literature and its
implications for standard setting. Account. Horiz. 13, 365–383.
Herrmann, D., Thomas, W., 2005. Rounding of analyst forecasts. Account. Rev. 803, 805–823.
Hope, O.K., Thomas, W.B., 2008. Managerial empire building and firm disclosure.
J. Account. Res. 46 (3), 591–626.
Islam, M.A., Ali, R., Ahmad, Z., 2011. Is modified Jones model effective in detecting
earnings management? Evidence from a developing economy. Int. J. Econ. Finance
3(2),116–125.
Jackson, S., Pitman, M., 2001. Auditors and earnings management. CPA J. 71 (7), 39–44.
Jensen, M.C., 1986. Agency costs of free cash flow, corporate finance, and takeovers. Am.
Econ. Rev. 762, 323–329.
Jiang, G., Lee, C.M.C., Yue, H., 2010. Tunneling through intercorporate loans: the Chinese
experience. J. Financ. Econ. 98, 1–20.
Jo, H., Kim, Y., 2007. Disclosure frequency and earnings management. J. Financ. Econ. 84,
561–590.
John, K., Senbet, L., 1998. Corporate governance and board effectiveness. J. Bank. Finance
22 (4), 371–403.
Jones, J., 1991. Earnings management during import relief investigations. J. Account. Res.
292, 193–228.
La Porta, R., Silanes, F., Shleifer, A., Vishny, R., 1998. Law and
fina
nce. J. Polit. Econ. 106,
1113–1155.
Leuz, C., Oberholzer-Gee, F., 2006. Political relationships, global financing, and corporate

transparency. J. Financ. Econ. 81, 411–439.
Leuz, C., Nanda, D., Wysocki, P.D., 2003. Earnings management and investor protection:
an international comparison. J. Financ. Econ. 693, 505–527.
Table 5
The Cramer's V of first through third digits in positive annual net income of companies at
different periods.
Period Country First digit Second digit Third digit
1990–2002 United States 0.0082 0.0116 0.0122
Taiwan 0.0189 0.0169 0.0130
2003–2011 United States 0.0068 0.0063 0.0090
Taiwan 0.0095 0.0209 0.0149
472 F. Lin, S F. Wu / Economic Modelling 36 (2014) 466–473
Lin, F., Guan, L., Fang, W., 2011. Heaping in reported earnings: evidence from
monthly financial r eport s of Taiwanese firms. Emerg. Mark. Finance Trade 472,
62–73.
Liu, Q., Lu, Z., 2007. Corporate governance and earnings management in the Chinese listed
companies: a tunneling perspective. J. Corp. Finance 13 (5), 881–906.
Man, C.K., Wong, B., 2013. Corporate governance and earnings management: a survey of
literature. J. Appl. Bus. Res. 29 (2), 391–418.
Marra, A., Mazzolab, P., Prencipea, A., 2011. Board monitoring and earnings management
pre- and post-IFR. Int. J. Account. 46 (2), 205–230.
Matsumoto, D.A., 2002. Management's incentives to avoid negative earnings surprises.
Account. Rev. 77, 483–514.
McNichols, M.F., Stubben, S.R., 2008. Does earnings management affect firms' investment
decisions? Account. Rev. 83 (6), 1571–1603.
Mehmet, U.M., Emin, H.C., 2012. Earnings management, audit quality and legal
environment: an international comparison. Int. J. Econ. Finance Issues 2 (4), 460–469.
Nenova, T., 2003. The value of corporate votes and control benefits: a cross-country
analysis. J. Financ. Econ. 683, 325–351.
Nigrini, M.J., 2012. Benford's Law: Applications for Forensic Accounting, Auditing, and

Fraud Detection. John Wiley and Sons, Inc., Hoboken, NJ (350 pages).
Park, Y.W., Shin, H.H., 2004. Board composition and earnings management in Canada.
J. Corp. Finance 103, 431–457.
Peasnell, K.V., Pope, P.F., Young, S., 2000. Board monitoring and earnings management: do
outside directors influence abnormal accruals? J. Bus. Finance Account. 32 (7, 8),
1311–1346.
Reddy, Y.V., Sebastin, A., 2012. Entropic analysis in financial forensics. IUP J. Account. Res.
Audit Pract. 6 (3), 42–57.
Rodriguez, R.J., 2004. Reducing false alarms in the detection of human influence on data.
J. Account. Audit. Finance 192, 141–158.
Shen, C.H., Chih, H.L., 2007. Earnings management and corporate governance in Asia's
emerging markets. J. Compilation 155, 999–1021.
Shleifer, A., Wolfenzon, D., 2002. Investor protection and equity markets. J. Financ. Econ.
661, 663-27.
Skinner, D.J., Sloan, R.G., 2001. Earnings surprises, growth expectations and stock return
or don't let an earnings torpedo stock sink your portfolio. Rev. Acc. Stud. 7 (2, 3),
289–312.
Skousen, C.J., Guan, L., Wetzel, T.S., 2004. Anomalies and unusual patterns in reported
earnings: Japanese managers round earnings. J. Int. Financ. Manag. Account. 153,
212–234.
Sun, L., Rath, S., 2010. Earnings management research: a review of contemporary research
methods. Glob. Rev. Account. Finance 11, 121
–13
5.
Thomas, J.K., 1989. Unusual patterns in reported earnings. Account. Rev. 644, 773–787.
Thomas, E.W., 2012. Further evidence on the extent of cosmetic earnings management by
U.S. firms. Acad. Account. Financ. Stud. J. 16 (3), 57–64.
Vives, X., 2006. Banking and regulation in emerging markets: the role of external
discipline. World Bank Res. Obs. 21, 179–206.
Watts, R.L., Zimmerman, J.L., 1986. Positive Accounting Theory. Prentice-Hall.

World Bank, 1999. Corporate governance: framework for implementation, overview.
www.worldbank.org.
Wurgler, J., 2000. Financial markets and the allocation of capital. J. Financ. Econ. 58,
187–214.
Zéghal, D., Chtourou, S., Sellami, Y.M., 2011. An analysis of the effect of mandatory
adoption of IAS/IFRS on earnings management. J. Int. Account. Audit. Tax. 20, 61–72.
473F. Lin, S F. Wu / Economic Modelling 36 (2014) 466–473