Accrual-Based and Real Earnings Management:

An International Comparison for Investor Protection

Masahiro Enomoto

Kobe University, Kobe, Japan

Fumihiko Kimura

Tohoku University, Sendai, Japan

Tomoyasu Yamaguchi

Tohoku Gakuin University, Sendai, Japan

March 2013

Corresponding author.
E-mail: menomoto@rieb.kobe-u.ac.jp
Tel: +81-78-803-7031; fax: +81-78-803-7031.
Research Institute for Economics & Business Administration, Kobe University, 2-1 Rokkodaicho, Nada-ku, Kobe
657-8501, JAPAN.

E-mail: fkimura@econ.tohoku.ac.jp
Tel: +81-22-217-6282; fax: +81-22-217-6282.
Graduate School of Economics and Management, Tohoku University, 2-2-1 Katahira, Aoba-ku, Sendai 980-8576,
JAPAN.

E-mail: yamaguchi@tscc.tohoku-gakuin.ac.jp
Tel: +81-22-721-3471; fax: +81-22-721-3471.
Faculty of Business Administration, Tohoku Gakuin University, 1-3-1 Tsuchitoi, Aoba-ku, Sendai 980-8511, JAPAN.

Accrual-Based and Real Earnings Management:

An International Comparison for Investor Protection

Abstract
The purpose of this paper is to examine the differences between accrual-based and
real earnings management across 38 countries. Following the international comparative
research on earnings management and the research on the choice of the method of
earnings management, we predict that real earnings management is preferred over accrualbased earnings management in countries with stronger investor protection. We regard legal
enforcement, outside investor rights, disclosure regulations, and analyst following as
constituting investor protection. Our examination is based on the data of 222,513 firm-year
observations from 1991 to 2010. We show that accrual-based earnings management is
inhibited, while real earnings management is implemented in the countries with stronger
investor protection except for analyst following. With regard to analyst following, we find
evidence that real earnings management is limited in the presence of analysts. Moreover,
this result is not affected by the control of audit quality and the calculation of earnings
management measures according to country and year.

JEL Classification: G34 K22 M41 G38

Key Words: Corporate Governance, Corporation and Securities Law, Accounting,
Government Policy and Regulation

1. Introduction
The purpose of this paper is to examine the differences in accrual-based and real
earnings management across 38 countries. Following Scott (2011), we define earnings
management as the choice by a manager of accounting policies, or real actions that affect
earnings so as to achieve a specific reported earnings objective. The methods of earnings
management are categorized as either the change in the accrual process or the deviation
from normal business activity. The former is called Accrual-based Earnings Management
(AEM) and the latter, Real Earnings Management (REM).
Previous international comparative studies on earnings management have
concentrated on AEM. Leuz et al. (2003) applied investor protection as an institutional factor
that influences managerial behavior, investigated whether or not investor protection in each
country is related to AEM, and examined the relationship between outside investor protection
and earnings management in 31 countries from 1990 to 1999. Investor protection is defined
as the power to prevent manager from expropriating minority shareholders and creditors
within the constraints imposed by law (La Porta et al., 2002; Leuz et al., 2003). Researchers
have found that earnings management decreases in countries with stronger investor
protection. Boonlert-U-Thai (2006) also investigated the relationship between investor
protection and earnings management in 31 countries from 1996 to 2002, and suggested that
earnings are smoothed in countries where investor protection has progressed. Shen and
Chih (2005) focused on banks in 48 countries and showed that earnings management
declined in countries with stronger investor protection and more transparent accounting
disclosure.
Prior research mainly employs accrual-based measures as a proxy of earnings
management. However, aside from the accrual process, managers can also use real activities to
manage earnings. REM also serves as an important topic in recent researches on

earnings management (e.g., Roychowdhury, 2006). While REM could make a negative
impact on the future value of the firms (e.g., Bhojraj et al., 2009; Cohen and Zarowin, 2010),
neither regulators nor auditors could restrain it. 1 Moreover, the scrutiny of REM by regulators
and auditors could not exceed that of AEM. Based on such arguments, we predict that REM,
rather than AEM, is implemented in countries with stronger investor protection.
We measure AEM in three proxies, following Leuz et al. (2003): (1) the ratio of the
standard deviation of operating income to that of operating cash flow, which is calculated by
time-series data from each firm, (2) the correlation between changes in accruals and
changes in operating cash flow computed from the pooled data in each country, and (3) the
ratio of the absolute value of accruals to that of operating cash flow calculated in each firmyear. On the other hand, we measure REM in two proxies: (1) the correlation between
changes in sales and production costs and (2) the correlation between changes in sales and
discretionary expenses.2
As variables of investor protection, we use the strength of legal enforcement and the
extent of outside investor rights that are related to corporate law and security law (La Porta
et al., 1998; La Porta et al., 2006). Furthermore, we emphasize the disclosure regulations
and the role of the analyst as important factors for investor protection, including the
existence of analyst following and the disclosure index.
Our examination is based on the data of 222,513 firm-year observations from 38
countries between 1991 and 2010. Our findings suggest show that AEM is less prevalent
than REM in the countries that have stronger investor protection. Moreover, this result is not
affected by the control of audit quality and the calculation of earnings management
measures according to country and year.
1

Contrary
and Zarowin (2010), Guuny (2010) suggests that REM makes a positive impact on the
future
valuetoof Cohen
the firms.

2Roychowdury
The correlation
between
the changes
in sales
and cash
flows 4.from
(2006),
is not included.
The reason
is shown
in footnote
2

operations, which is employed in

Most international comparative researches on earnings management (e.g., Leuz et

al., 2003; Shen and Chih, 2005; Boonlert-U-Thai, 2006), has only analyzed AEM. We,
however, also investigate REM based on the theoretical background of REM and AEM.
Moreover, compared to previous studies, our results are more reliable because of the larger
sample size and longer term.
The remainder of this paper is organized as follows. Section 2 provides a discussion
on AEM and REM and institutional background. Section 3 shows sample selection
procedures and the measure of earnings management and investor protection. Section 4
contains the results for the relationship between earnings management and investor
protection. Section 5 discusses additional analyses, and Section 6 concludes.

2. Accrual-based and real earnings management, and institutional background

Managers can opportunistically manage earnings by changing the accrual process
because various estimations and judgments go into the process of preparing financial
statements. Thus, managers can implement AEM after the end of the fiscal year. However,
AEM is more visible than REM due to scrutiny from auditors and regulators, among others,
and has the potential for reversal in future periods.
In addition to the accrual process, managers can also manage earnings by changing the
timing or structure of operating, investing, or financial decisions. These activities are referred to
as real earnings management (REM). REM could be more costly with respect to future firm cash
flow, as it often cuts funding for research and development (R&D) and marketing, increases price
discounts, and reduces capital investments (Graham et al., 2005, Chen et al., 2008). However, it
is hard for stakeholders to detect REM in comparison to AEM, as REM is easier to camouflage
as normal activities compared to AEM (Kothari et al. 2012).

Thus, managers may choose AEM and/or REM under the various economic contexts. Our
work focuses on both AEM and REM.
Regulatory authorities impose various restrictions on managers when the latter
carry out AEM because earnings management has the potential to reduce the outside
investors profits. The degree of restrictions is believed to vary with the institutional factors of
each country. Many studies attach importance to investor protection, including corporate law,
accounting standards, and security markets as institutional factors. La Porta et al. (1997;
1998) show that each countrys investor protection is based on its legal origin, and countries
with stronger investor protection have larger and open capital markets. Leuz et al. (2003)
suggest that AEM decreases in countries with stronger investor protection. In a similar vein,
disclosure systems and analyst following serve as components of investor protection,
thereby reducing AEM (for example, Yu, 2008 and Degeorge et al., 2013). However, even
though AEM is restrained, the motives for earnings management are not necessarily
controlled. Moreover, in countries with developed capital markets, investors should be
cautious about earnings information. Therefore, it is believed that in countries with stronger
investor protection, managers tend to choose REM over AEM.
Previous research supports our prediction. In a survey of chief financial officers, Graham
et al. (2005) present evidence suggesting that managers opt for REM because they fear the risk
wrought by overzealous regulators. Through the analysis of mathematical models, Ewert and
Wagenhofer (2005) also suggest that as an effect on tighter accounting standards, manager
prefer REM. Cohen et al. (2008) imply that managers switched from AEM to REM after the
passage of Sarbanes Oxley Act (SOX) in United States, which was seen as strengthening
investor protection. In summary, in the presence of stricter regulations, managers tend to avoid
AEM but implement REM to achieve income targets. In other words, the tightening of regulations
leads to restrain AEM but tends to induce REM. Given the
foregoing discussion, we propose two hypotheses. 4

H1: Ceteris paribus, accrual-based earnings management is more constrained in

countries with stronger investor protection.
H2: Ceteris paribus, real earnings management is more often implemented in countries
with stronger investor protection.

3. Research design
This study tests hypotheses pertaining to the relationship between both types of
earnings management and investor protection. The following subsections present a detailed
explanation of two earnings management and four investor protection variables.

3.1. The variables of accrual-based and real earnings management

Earnings management is classified as either AEM or REM. The purpose of this paper
is to explore the relationship between two types of earnings management on the one hand
and investor protection, on the other. We calculate three measures for AEM (based on
accruals) and two measures for REM (correlated to sales), then aggregate the results.
Though our country-level earnings management measures may be parsimonious due
to the limited amount of balance sheet and income statement information for each firm in
Standard & Poors Capital IQ database, we can make a comparison between our results and
those of Leuz et al. (2003) by using these measures. Thus, we compute aggregate the
measures for each type in order to evaluate the countrys level of earnings management,
avoiding the concerns from calculation methods as much as possible. The measures are
ranked from the lowest to highest level of earnings management (e.g. the country with the
least amount of earnings management gets rank one). The aggregate measure of accrualbased (or real) earnings management is the average rank of the three (two)
measures. For aggregation, all the measures should have similar tendencies to some extent.
5

3.1.1. The variables of accrual-based earnings management

Three measures of AEM are employed in this studythree of the four Leuz et al.
(2003) employedand indicate the extent of earnings management by using accruals. The
explanation of the three measures was drawn from Leuz et al. (2003). We concentrate on
the comparison between AEM and REM, and small loss avoidance in Leuz et al. (2003) is
not adopted.3
To obtain these measures, accruals are calculated as follows:
Accruals (Acc) = (CAit Cashit) (CLit STDit TPit) Depit (1), CAit = change in
total current assets, Cashit = change in cash/cash equivalents, CLit = change in total
current liabilities, STDit = change in short-term debt included in current liabilities, TPit
= change in income taxes payable, and Depit = depreciation and amortization expense. The
subscripts i and t refer to firm i and year t. It is the same as in Dechow et al. (1995) and
used generally. We can obtain operating cash flow by subtracting accruals from earnings:
CFO = Earnings Accruals. We do not directly compute accruals from the cash flow
statement because the data is not available in the cash flows of all the countries in the
sample. For reliability, our statistical analysis eliminates outliers, trimming the firm-years
within the top and bottom 1% of accrual and operating cash flow data. All of the variables to
compute the measures of AEM and REM are divided by the total assets at the beginning of
the year.
The first measures of AEM are the standard deviation of operating earnings divided
by the standard deviation of operating cash flow (AEM1). This measure assumes that a
manager might use accruals to mitigate the variation of operating cash flow. If so, this
3

In this study, we focus on the managers choice of AEM and/or REM. However, small loss avoidance in Leuz

et al. (2003) does not mean the use of AEM only. Firms might avoid loss by using REM (for example,
Roychowdhury 2006).

measure becomes smaller. AEM1 is calculated by time-series data of each firm, and the
median of AEM1 in each country is its representative value.
(E)
AEM1 = (CFO)

(2)

(E) the standard deviation of the operating income, and (CFO) the standard deviation of
the operating cash flows.
The second measure is the correlation between the changes in accruals and changes in
operating cash flow (AEM2). As is well known, the correlation is negative. However, we employ it
as an AEM measure because the large magnitudes of the correlation can be interpreted as
income-smoothing behavior. AEM2 is computed from the pooled data in each

country.
AEM2 = (Acc, CFO)

(3)

Acc change in accruals and CFO change in operating cash flows.

The third measure is the ratio of the absolute value of accruals to the absolute value of
operating cash flows (AEM3). Accruals are reflected by various kinds of manager behaviors in
achieving certain earnings targets. Therefore, we use the absolute value of accruals as an
earnings discretion measure. The reason for dividing by the operating cash flow is to control firm
size and performance. AEM3 is calculated in each firm-year, and we use each countrys

median data.
|Acc|
AEM3 = |CFO |

(4)

|Acc| absolute value of accruals and |CFO| absolute value of operating cash flows.
AEM1 and AEM2 are employed as smoothing measures, and AEM3 is the measure
of discretion used by Leuz et al. (2003). We regard these three measures as the degree of
earnings management through accruals.

3.1.2. The variables of real earnings management

As previously mentioned, earnings can be managed not only through AEM but also
through REM. For example, previous research (Roychowdhury, 2006; Cohen et al., 2008;
Cohen and Zarowin, 2010) examine sales manipulation with price discounts and lenient
credit terms, the reduction of discretionary expenditure such as R&D, and overproduction,
which lowers the unit cost of goods sold. Based on these researches, this paper considers
three types of REM: (1) sales manipulation, (2) manipulation of discretionary expenditures,
and (3) production manipulation.
According to Roychowdhury (2006), sales manipulation with price discounts and
overproduction leads to abnormally high production costs relative to sales. 4 Thus, sales and
production manipulation induce an imbalance in production costs and sales, and result in a
lower correlation between the change in production costs and the change in sales. For this
reason, we use the contemporaneous correlation between the change in production costs
and the change in sales (indicated by REM1) as measures of sales and production
manipulation. REM1 is computed from the pooled sample in each country. A lower REM1
means that more sales and production manipulation has been carried out.
REM1 = (Prod, Sales) (5),
Prod = change in production costs, production costs = cost of goods sold + change in
inventory, and Sales = change in sales.
The manipulation of discretionary expenditures leads to departures from the normal
level of discretionary expenses. Discretionary expenses are expressed as a linear function of
sales in previous research (Dechow et al., 1998; Roychowdhury, 2006). Based on these
4

Roychowdhury (2006) also said that sales manipulation and overproduction lead to an abnormally low cash flow

from operation, while the reduction of discretionary expenses leads to an abnormally high cash flow from
operation. Therefore, the net effect on cash flow from operation by real earnings management is ambiguous, as
stated in Roychowdhury (2006). Consequently, consistent with Gunny (2010) and Zang (2012), we do not use the
proxy with cash flow from operation.

researches, we employ the contemporaneous correlation between the change in

discretionary expenses and the change in sales (REM2) as a proxy for the manipulation of
discretionary expenditures. If discretionary expenditures are manipulated to manage
earnings, REM2 should show a lower value. We define discretionary expenses as selling,
general, and administrative expenses (SG&A), which frequently include discretionary
expenses such as R&D, advertising, employment costs, etc. REM2 is also computed from
the pooled sample in each country.
REM2 = (DE, Sales) (6)
DE = change in selling, general, and administrative expenses.

3.2 The variables of investor protection

3.2.1. Outside investor rights
Following Leuz et al. (2003), we adopt the anti-director right index created by La
Porta et al. (1998) as variables pertaining to outside (minority) investor protection. The index
is variable if outside investors can reflect their views easily in shareholders meetings; it is
calculated by adding 1 when: (1) the country allows shareholders to mail their proxy vote to
the firm, (2) shareholders are not required to deposit their shares prior to the general
shareholders meeting, (3) cumulative voting or proportional representation of minorities in
the board of directors is allowed, (4) an oppressed minorities mechanism is in place, (5) the
minimum percentage share of capital that entitles a shareholder to call for an extraordinary
shareholders meeting is less than or equal to 10% (the sample median), or (6) shareholders
have preemptive rights that can be waived only by a shareholders vote. In our study, the
index ranges from 0 to 5. Leuz et al. (2003) show that strong and well-enforced outsider
rights limit the acquisition of private control benefits by insiders, and consequently, reduce
the incentives of insiders to engage in accrual-based earnings management. Based on the
9

argument of Leuz et al. (2003) and the discussion in section 2, we predict that the rights of
outside investors negatively affect AEM and positively affect REM.
3.2.2. Legal enforcement
The research in investor protection addresses legal enforcement. As La Porta et al.
(1998) point out, a strong system of legal enforcement could substitute for weak rules.
Employing the framework of La Porta et al (1998)., Leuz et al. (2003) suggested legal
enforcement as the institutional factors influencing AEM. Following Leuz et al. (2003), we
measure Legal Enforcement as the mean score across three legal variables used by La
Porta et al. (1998): (1) the efficiency of the judicial system, (2) an assessment of rule of law,
and (3) the corruption index. All three range from 0 to 10. Similar to the rights of outside
investors, legal enforcement is negatively related to AEM and positively related to REM.
3.2.3. Disclosure regulations
Disclosure regulations play an important role in investor protection because less
information is available to minority investors than to insiders, and the former have to depend
on the information that firms are required to disclose. We predict that since it is difficult to
implement AEM in countries that have stricter disclosure regulations, which might lower
earnings quality, managers tend to apply REM instead.
Previous studies have often used two indexes relating to the disclosure regulations of
each country (e.g., Jirasakuldech et al., 2011). One is the index that comes from the Center
for International Financial Analysis and Research (CIFAR) database and the other, from the
Doing Business project of the World Bank (the World Bank disclosure index). 5 We use the
latter as the variable for disclosure regulation due to the limited data in the CIFAR index. The
World Bank disclosure index measures the extent to which investors are protected through

This data can be obtained at the World Bank website, http://data.worldbank.org/ (last accessed 15

2013).

10

th

March

the disclosure of ownership and financial information. The index ranges from 0 to 10, with
higher values indicating more disclosure.
3.2.4. Analyst following
Financial analysts also play a key role in investor protection, as they are among the
important information intermediaries in securities markets (Healy and Palepu, 2001). The
extent of analyst following could affects earnings management in a country. Some previous
studies indicate that monitoring and pressure from analysts affects AEM. For example, Yu
(2008) and Degeorge et al. (2013) provide evidence that analyst following negatively affects
AEM. If analysts are sophisticated monitors of AEM, it is difficult for managers to engage in
AEM, and they therefore carry out REM in the presence of analysts. Thus, we predict that
analyst following will negatively affect AEM and positively affect REM.
In this regard, however, Cohen and Zarowin (2010) find a negative association
between analyst following and total EM, which is a metric that includes both AEM and REM.6
Their results indicate that AEM and REM are restrained because analysts scrutinize and
monitor firm activities.7 Thus, note that analysts may pay attention to both accounting and
real activity. In this case, both AEM and REM could be reduced.

3.3. Regression models

To test the hypotheses on the relationship between the two types of earnings
management and investor protection, we constructed the following regression models, which
are based on Leuz et al. (2003). Equations (7) to (10) are single regression models, and
6

However, they do not compare the relationships of REM and AEM with analyst following.

7 According

to Cohen and Zarowin (2010), there could be two different views regarding analyst following: (1)
AEM and REM are restrained because analysts scrutinize and monitor firm activities and (2) AEM and REM are
accelerated because analyst forecasts serve as incentives for managers, who strive to meet or beat them.

11

equation (11) is a multiple regression model that includes the four independent variables
used in single regression models. Following Leuz et al. (2003), we estimate these models by
rank regression.8
AEMi (or REMi) = 0 + 1

Outside Investor Rightsi + i

(7)

AEMi (or REMi) = 0

+ 1 Legal Enforcementi + i

(8)

AEMi (or REMi) = 0

+ 1 Disclosure Indexi + i

(9)

AEMi (or REMi) = 0

+ 1 Analyst Followingi + i

(10)

AEMi (or REMi) = 0

+ 1 Outside Investor Rightsi + 2 Legal Enforcementi

+ 3 Disclosure Indexi + 4 Analyst Followingi + i

AEM = the aggregate

accrual-based earnings management

(11)
score, REM = the

aggregate real earnings management score, Outside Investor Rights = the anti-director
rights index from La Porta et al. (1998), Legal Enforcement = the average score across three
legal variables used by La Porta et al. (1998), Disclosure Index = the disclosure index from
the World Bank, which ranges from 1 to 10, Analyst Following = the average number of
analyst following for all firms in each country, i = country, and = error term. Under AEM
(REM), the predicted signs of the coefficients of Outside Investor Rights, Legal Enforcement,
Disclosure Index, and Analyst Following are negative (positive).

3.4. Sample selection procedure

Our sample is based on the 49 countries studied by La Porta et al. (1998) and
constructed by the listed firms in nonfinancial sectors, from 1991 to 2010, which were selected
from Capital IQ. The former criterion is for comparison to the many studies that refer to La Porta
et al. (1998). To be included in our analysis, the firms had to have available balance sheets and
income statements for at least three consecutive years and the countries,
8

OLS regression with unranked variables produces similar results.

12

at least 300 firm-years, to allow us to obtain sufficient data for calculating the representative
data from each country. Finally, we excluded the countries that experienced hyperinflation
over the sample period, as hyperinflation could seriously affect our earnings management
measures. Argentina, Brazil, Peru, Turkey, Venezuela, and Zimbabwe were thus cut from our
sample, leaving a final sample of 222,513 firm-year observations from 38 countries.
Table 1 provides the descriptive statistics for each country. Panel A shows the
number of firm-years in each country and year. Our total sample (222,513) is much larger
than that of Leuz et al. (2003) due to the extended sample periods and the use of Capital IQ.
The rightmost column shows the number of firm-years in each country; the United States
has the largest at 53,072, followed by Japan, India, and Canada. Nigeria has the smallest at
300. The bottom row provides the number of firm-years in each year, the smallest of which
was in 1991 and the biggest, in 2010. In general, the observations we can obtain increase by
the year. Canada and India, for example, had scanty data in the early 1990s, but after the
second half of the decade, a large amount of data was included in Capital IQ.
Panel B shows the descriptive statistics of fundamental variables in each country.
Considerable variations of other measures are also observed in Panel B due to the
economic environment and the difference in the development of the capital market from
country to country. Hence, all of the financial statement variables we use are divided by total
assets to mitigate the effect of the size.
Panel C provides the descriptive statistics of institutional variables in each country.9
The institutional variables mainly rely on La Porta et al. (1997; 1998) and Leuz et al. (2003).
Legal Origin and Legal Tradition are shown in columns 1 and 2. There are two steps to the
calculation of the Importance of Equity Market. First, in La Porta et al. (1997), the three variables
that follow are ranked: (1) the ratio of the aggregate stock market capitalization held by minorities
to the gross national product (GNP), (2) the number of listed domestic firms
9

The institutional variables for the regression analysis are given in Table 3.

13

relative to the population, and (3) the number of IPOs relative to the population. Second, we
compute the average rank of the three variables. Ownership concentration is the median
ownership stake of the three largest shareholders among the 10 largest publicly traded
companies (La Porta et al., 1998).
[Insert Table 1 here]
4. Empirical results
4.1. Descriptive statistics
Table 2 shows the scores for AEM and REM. Panel A provides three AEM measures
and their aggregated scorethe average rank of three measures (rightmost column). Lower
AEM1 and AEM2 scores imply that accrual management is used to reduce the variation of
earnings and conceal economic shocks to the firms operating cash flow. For AEM1, Australia
has the highest score and Portugal, the lowest, followed by Spain, Italy, Austria, and South
Korea. This ratio is maintained in AEM2. Both measures are higher in Anglo-American countries
than in Asia and Continental Europe, indicating that Anglo-American countries employ the lower
extent of AEM. AEM3 (accruals divided by CFO) is reflected by various manager behaviors in
their attempt to achieve specific earnings targets. All three variables,

AEM1, AEM2, and AEM3 share similar tendencies.10 Therefore, we may use the aggregated
accrual-based earnings management score (AEM), which is the average rank of the three
measures.
[Insert Table 2 here]
REM1 is the extent to which production manipulation for earnings management is
pervasive in a country. The mean is 0.787, with Japan posting the highest value of 0.928.
10

In our research, the calculation of AEM1, AEM2, and AEM3 is related to that of EM1, EM2, and EM3, as in

Leuz et al. (2003). The country scores seem to be similar to those of Leuz et al. (2003) in total. The calculated
means of AEM1 and AEM2 are close to those of EM1 and EM2. However, AEM3 and EM3 have a comparatively
large difference of 0.25, possibly due to the extended sample periods.

14

South Korea, Taiwan, Austria and the Netherlands show a lower level of earnings
management.
REM2 is the correlation between the change in sales and the change in discretionary
expense. If discretionary expense is rarely managed, REM2 is expected to rise, as
discretionary expenses typically increase in proportion to sales. The balance between
discretionary expenses and sales is negatively affected when the former is managed. The
results of REM2 are similar to those of REM1. REM2 provides the evidence that Oceanic
countries tend to employ REM more than AEM.
The rightmost column of Panel B shows the aggregated score of REM measures.
The result of REM2 is also similar to that of REM1. Hence, we may calculate the aggregated
real earnings management score (REM) in the same manner as we calculated AEM.
Table 3 contains the institutional variables for the regression analysis of the 38
countries. The Outside Investor Rights and Legal Enforcement are similar to those of Leuz et
al. (2003). A characteristic of the Disclosure Index is that all the developed countries do not
have high scores (e.g., Germanys index is five). Overall, the Analyst Following of European
countries is relatively high. The two variables are added to the Leuz et al. (2003) model and
tested for association with AEM and REM.
[Insert Table 3 here]
Table 4 provides the correlation matrix for the AEM, REM, and investor protection
variables. AEM is negatively correlated to REM, and the p-value is under 5% (two-tailed).
The results imply the substitutability between AEM and REM. The correlation between AEM
(REM) and Outside Investors Rights is significantly negative (positive). These results are
consistent with our predictive analysis.
[Insert Table 4 here]

15

4.2. Regression results

Panels A and B of Table 5 show the regression results. In Section 2, we hypothesized
that AEM (REM) is constrained (more prevalent) in countries with better investor protection.
Higher values for Outside Investor Rights, Legal Enforcement, and Disclosure Index mean
better investor protection. Thus, the predicted signs for the coefficient on these variables are
negative (positive) when the dependent variable is AEM (REM). However, we do not predict
the signs for the coefficients of Analyst Following because the effect of analysts on earnings
management is still obscure.
Since the results of single and multiple regressions are similar in Panels A and B, we
describe only the results of multiple regressions. Panel A contains the results for AEM. As
predicted, and consistent with Leuz et al. (2003), the coefficients of Outside Investor Rights
and Legal Enforcement are significantly negative (-0.395, t = -2.549 and -0.474, t = -2.756).
These results support Hypothesis 1 and indicate that AEM is less prevalent in countries with
better investor protection. However, the coefficients of the Disclosure Index and Analyst
Following are insignificant (-0.058, t = -0.391 and 0.081, t = 0.460).
Panel B carries the results for REM. As expected, they are contrary to the results for
AEM. The coefficient of Outside Investor Rights is significantly positive (0.373, t = 2.298),
which is consistent with Hypothesis 2, and indicates that REM is more prevalent in countries
with better investor protection. The coefficient on Legal Enforcement is not significant (0.160,
t = 0.887). As with the results for AEM, the coefficient of the Disclosure Index is insignificant
(0.065, t = 0.421). In contrast to the results for AEM, the coefficient of Analyst Following is
significantly negative (-0.416, t = -2.262). This result contradicts our hypothesis but is
consistent with Cohen and Zarowin (2010), who find a negative association between analyst
following and earnings management. This finding is consistent with the notion that
monitoring by analysts influences the behavior of managers and limits REM.
16

Overall, the results in this section indicate that AEM is constrained in countries with
better investor protection, thus making REM more prevalent. Managers in countries with
better investor protection tend to engage in REM instead of AEM. In addition, we find
evidence that REM is constrained by analyst following, but no evidence of a relationship
between AEM (REM) and the level of the disclosure index.
[Insert Table 5 here]

5. Robustness check
Previous research considers the per capita gross domestic product (Leuz et al., 2003)
and Big 4 audits (Francis and Wang, 2010) as other factors that could affect earnings
management. In this paper, we omit the regression analysis of per capita GDP because of a
11

serious multicollinearity.

Consequently, we reestimate equation (11), including the ratio of firms

audited by Big 4 in each country as an additional independent variable (indicated by the

Big4 Ratio).12 The rank regression results, including the Big4 Ratio, are presented in Table
6. Although the coefficients of the Big4 Ratio are not significant (-0.051, t = -0.294 and
0.046, t = 0.254), the coefficients of the other variables are essentially similar to those in the
previous section. Hence, the results in the previous section are robust when we add the
regression models to the Big4 Ratio.
[Insert Table 6 here]
In Section 4.2, we follow Leuz et al. (2003) and compute the legal enforcement score
as the average across three variables used in La Porta et al. (1998): (1) the efficiency of the

11

TheGDP
correlation
rank
of per capita GDP and analyst followings is 0.71. In addition, the VIF for per
capita
in rank between
regression
is 5.63.

12

We exclude the Big4 Ratio from the main regression models to ensure the sample size as much as

possible in the main analysis.

17

judicial system, (2) an assessment of rule of law, and (3) the corruption index. 13 In this
section, we create a newer legal enforcement variable using available data in the sample
period and reestimate equation (5). In particular, to recompute the value of Legal
Enforcement, we replace an assessment of the rule of law and the corruption index with their
counterparts from the year 2000, based on Kaufmann et al. (2004). Table 7 shows the rank
regression results using the new proxy for legal enforcement. The results are the same as
before and are thus robust when substituting them with the variable of Legal Enforcement.14
[Insert Table 7 here]
Next, due to the aggregation of EM by country in Table 5, the analysis of the
relationship between investor protection and earnings management in our sample focuses
on an in-depth examination of 222,513 firm-years, which is a larger sample base when
compared to the standard data from 38 countries. To address this problem of decreased
sample size, we use measures of EM calculated by country and year and re-estimate
equation (11). 15
The recalculation increases the sample size from 38 to 549. To run the regression
equation (11), two types of EM measures are ranked in the pooled data and converted from 1 to
38. In addition, we add year indicator variables to equation (11) to control for year effects. The
results in Table 8 are similar to those in Table 5 for AEM, whereas for REM, the coefficient of
Legal Enforcement is significantly positive at a marginal level and the coefficient of the
Disclosure Index is significantly positive. In summary, the results of this additional analysis
support Hypothesis 2, which corroborates the results of previous analyses.
13

The first variable ranges from 0 to 10 and the second and third variables, from -2.5 to +2.5.

14

Furthermore,
when the
disclosure
index
is replaced
by Disclosure Requirement in La Porta et al. (2006), the
results
are also similar
to those
shown
in Table
5.

15

Because each country-year requires at least 30 observations to obtain two EM measures, the amount

of data for each country is unbalanced (for example, Japan has data for 20 years, whereas Nigeria has data
only for five years). Hence, we do not adopt country-year based analysis as the primary analysis.

18

[Insert Table 8 here]

6. Conclusion
This paper examined the relationships between earnings management and investor
protection across 38 countries. Our tests were based on the data from 1991 to 2010 over
280,000 firm-years observations from the countries. Leuz et al. (2003) show evidence that
strong investor protection limits earnings management. While their measures of earnings
management are mainly accrual-based, we focused on both REM and AEM. REM is based
on operational activities, and following Roychowdhury (2006, 338), accrual manipulation is
more likely to draw the scrutiny by auditor and regulator than real decisions about pricing
and production. Therefore, we hypothesized that investor protection has an adverse effect
on REM as compared to AEM.
Consistent with our hypothesis, outside investor rights is negatively correlated to
REM and positively correlated to AEM. However, we provide evidence that the more analysts
investigate firms, the less REM is carried out. The results do not change under the control of
audit quality, the alternative definitions of investor protection variables, and the use of AEM
and REM measures that are calculated by country and year. Our results for REM conflict
with those of Leuz et al. (2003). In summary, differences in investor protection among
countries have an effect on their earnings management methods.
REM is a departure from normal operational practices (Roychowdhury, 2006, 337),
and will likely impact future performance negatively. For example, Bhojraj et al. (2009)
suggest that the effort to meet or beat analyst forecasts by using REM and AEM has a
negative effect on a firms future ROA and stock price.

19

Our results show that strong investor protection restricts AEM, but induces a shift to
REM. Moreover, our results also imply that strong investor protection heightens the risk of
firm value reduction by REM and the existence of analysts is effective in monitoring REM.
As with prior research, we have a parsimonious model in calculating the proxies for
AEM and REM due to the limited data from Capital IQ. Future research is required to
improve the model to estimate AEM and REM using detailed international accounting data.

Acknowledgments
The authors gratefully acknowledge the financial support from The Japan Securities
Scholarship Foundation.

References
Bhojraj, S., P. Hribar, M. Picconi, J. Mcinnis, 2009. Making sense of cents: An examination of
firms that marginally miss or beat analyst forecasts. The Journal of Finance. 64, 23612388.
Boonlert-U-Thai, K., G. K. Meek, S. Nabar, 2006. Earnings attributes and investor-protection:
International evidence. The International Journal of Accounting. 41, 327-357.
Chen, J. Z., L. L. Rees, S. Sivaramakrishnan, 2010. On the use of accounting vs. real
earnings management to meet earnings expectations-A market analysis. Working paper.
Cohen, D., A. Dey, T. Lys, 2008. Real and accrual based earnings management in the pre
and post sarbanes oxley periods. The Accounting Review 83. 757-787.
Cohen, D. A., P. Zarowin, 2010. Accrual-based and real earnings management activities
around seasoned equity offerings. Journal of Accounting and Economics. 50, 2-19.
Dechow, P. M., R. G. Sloan, A. P. Sweeney. 1995. Detecting Earnings Management. The
Accounting Review 70, 193-225.

20

Dechow, P. M., S. P. Kothari, R. L. Watts. 1998. The relation between earnings and cash
flows. Journal of Accounting and Economics 25, 133-168.
Degeorge F., Ding Y, Jeanjean T., Stolowy H. 2013. Analyst coverage, earnings management
and financial development: An international study. Journal of Accounting and Public
Policy 32, 1-25.
Ewert, R., A. Wagenhofer, 2005. Economic effect of tightening accounting standards to
restrict earnings management. The Accounting Review 80, 1101-1124.
Graham, J. R., C. R. Harvey, S. Rajgopal, 2005. The economic implications of corporate
financial reporting. Journal of Accounting and Economics, 40, 3-73.
Healy, P., K. Palepu, 2001. A review of the empirical disclosure literature. Journal of
Accounting and Economics, 31, 405440.
Jirasakuldech, B., T. Zorn, J. Geppert, D. Dudney, 2011, Financial disclosure, investor
protection, and stock market behavior: an international comparison, Review of
Quantitative Finance and Accounting, 37, 181-205.
Kothari, S.P., N. Mizik, S. Roychowdhury. 2012. Managing for the Moment: The Role of Real
Activity versus Accruals Earnings Management in SEO Valuation. Working paper.
La Porta, R., F. Lopez-de-Silanes, A. Shleifer, R. Vishny, 1997. Legal determinants of
external finance. Journal of Finance, 52, 1131-1150.
La Porta, R., F. Lopez-de-Silanes, A. Shleifer, R. Vishny, 1998. Law and finance. Journal of
Political Economy, 106, 1113-1155.
La Porta, R., F. Lopez-de-Silanes, A. Shleifer, R. Vishny, 2002, Investor protection and
corporate Valuation. Journal of Finance, 75, 1147-1170.
La Porta, R., F. Lopez-de-Silanes, A. Shleifer, 2006, What works in securities laws? Journal
of Finance 61, 1-32,
Leuz, C., D. Nanda, P. Wysocki, 2003. Earnings management and investor protection: an
international comparison. Journal of Financial Economics, 69, 505527. 21

Roychowdhury, S., 2006. Earnings management through real activities manipulation. Journal
of Accounting and Economics, 42, 335-370.
Scott, W., 2011. Financial accounting theory 6th edition, Prentice Hall.
Shen, C. H., H. L. Chih, 2005. Investor protection, prospect theory, and earnings
management: An international comparison of the banking industry, Journal of Banking
and Finance 29, 26752697.
Yu, F., 2008. Analyst coverage and earnings management. Journal of Financial Economics
88, 245-271.
Zang, A. Y. 2012. Evidence on the trade-off between real activities manipulation and accrualbased earnings management, The Accounting Review, 87, 675703.

22

Table 1 Descriptive statistics

Panel A: Firm-year observations in each country
Country
Australia
Austria
Belgium
Canada
Chile
Denmark
Finland
France
Germany
Greece
Hong Kong
India
Indonesia
Ireland
Israel
Italy
Japan
Jordan
Malaysia
Mexico
Netherlands
New Zealand
Nigeria
Norway
Pakistan
Philippines
Portugal
Singapore
South Africa
South Korea
Spain
Sri Lanka
Sweden
Switzerland
Taiwan
Thailand
UnitedKingdom

United States

Total

1991
45
10
16
2
13
28
25
95
113
0
30
1
0
18
0
16
856
0
70
0
35
5
0
18
0
1
1
38
18
0
7
0
24
55
0
2
272
246
2,060

1992
49
14
16
2
14
28
25
99
115
0
32
1
2
18
0
16
910
0
73
0
37
5
0
17
0
1
1
38
18
0
8
0
25
56
0
2
279
311
2,212

1993
50
14
16
3
24
28
27
100
115
0
48
4
25
18
1
15
954
0
76
6
37
5
0
19
0
5
2
38
18
0
28
0
25
57
0
36
281
749
2,824

1994
52
15
17
6
26
26
26
100
111
0
49
4
28
18
2
18
979
0
76
23
37
5
0
19
0
5
2
40
18
0
41
0
27
57
1
39
280
1,018
3,165

1995
53
15
15
10
29
26
28
100
111
0
57
5
31
19
4
20
1,017
0
80
15
35
7
0
23
0
2
3
49
18
104
44
0
26
59
1
43
281
1,415
3,745

1996
52
13
22
27
37
28
27
122
100
4
86
7
52
18
4
33
1,038
0
114
15
39
9
0
25
11
3
9
96
22
113
47
0
30
63
1
68
260
1,932
4,527

1997
65
14
17
201
35
33
30
122
110
6
113
5
63
21
5
30
1,005
0
164
26
46
11
1
21
40
10
7
99
24
136
46
0
38
60
1
59
241
2,498
5,403

1998
107
25
27
463
40
44
43
176
184
26
150
25
85
23
33
56
1,091
0
200
41
63
15
1
32
76
22
12
115
40
149
57
1
77
88
2
72
346
2,747
6,754

1999
244
31
36
555
70
54
50
223
260
37
246
37
117
27
44
67
1,156
1
261
50
71
22
1
36
97
37
12
153
63
161
64
2
101
102
16
112
420
2,888
7,924

2000
278
42
61
656
84
70
74
303
343
51
344
650
125
28
65
91
1,564
3
351
56
89
40
0
51
86
44
16
206
80
197
74
5
138
128
180
176
497
2,974
10,220

2001
368
49
66
716
89
80
88
341
414
58
439
770
132
35
77
121
1,728
3
429
63
94
47
0
62
97
60
23
258
109
298
78
5
157
135
240
209
542
3,172
11,652

2002
491
49
66
736
101
83
91
360
419
66
579
875
122
33
78
132
1,980
7
458
68
91
53
4
69
74
63
27
292
127
355
83
5
174
141
296
227
580
3,106
12,561

2003
626
54
66
840
110
86
97
402
462
79
678
1,098
126
40
102
144
2,128
45
500
72
99
64
13
81
70
89
35
334
146
442
88
6
205
152
367
247
673
3,572
14,438

2004
686
57
69
864
114
86
98
430
470
83
716
1,585
125
40
116
153
2,216
53
542
77
101
78
32
88
92
93
37
360
149
534
92
15
225
157
871
271
697
3,613
16,085

2005
706
57
78
914
115
88
100
460
499
93
760
1,770
128
43
123
163
2,315
58
594
79
103
83
42
96
110
96
38
394
162
599
99
18
238
167
985
312
739
3,730
17,154

2006
748
57
91
952
119
92
101
495
541
123
805
1,889
143
44
142
176
2,435
67
656
79
104
83
56
109
127
104
38
414
176
699
99
24
262
171
1,033
338
796
3,875
18,263

2007
816
58
94
1,043
123
95
102
509
555
138
819
1,978
157
49
148
191
2,663
77
707
81
111
87
54
123
151
106
39
425
187
1,323
100
30
301
176
1,192
353
865
3,979
20,005

2008
811
61
88
1,012
127
95
104
508
548
164
836
2,050
182
53
143
198
3,141
37
727
84
113
85
38
133
160
103
40
454
198
1,317
103
71
319
178
1,323
365
889
3,611
20,469

2009
915
60
89
1,083
129
95
105
513
537
176
850
2,125
196
54
165
205
3,233
34
759
87
115
91
31
144
175
108
40
471
214
1,307
105
140
344
180
1,399
377
924
3,698
21,273

2010
1,003
59
88
1,146
130
92
107
518
524
183
881
2,306
217
54
212
207
3,241
5
788
88
116
93
27
150
189
112
41
491
222
936
103
158
366
183
1,461
390
954
3,938
21,779

Total
8,165
754
1,038
11,231
1,529
1,257
1,348
5,976
6,531
1,287
8,518
17,185
2,056
653
1,464
2,052
35,650
390
7,625
1,010
1,536
888
300
1,316
1,555
1,064
423
4,765
2,009
8,670
1,366
480
3,102
2,365
9,369
3,698
10,816
53,072
222,513

23

Panel B: Descriptive statistics of fundamental variables and investor protection variables in each country
Fundamental variables
Median of
Percentage
Variance of
Median of
Per Capita
Country
Capital
of Mfg.
Inflation
GDP
Firm Size
GDP
Intensity
Firms
Growth
Australia
14.538
0.585
0.270
28652
2.61
1.23
Austria
247.907
0.508
0.647
31794
2.13
0.87
Belgium
193.295
0.466
0.577
30196
2.07
0.83
Canada
13.580
0.615
0.319
28164
2.01
0.61
Chile
234.318
0.721
0.407
6080
5.82
1.57
Denmark
135.082
0.451
0.575
39361
2.09
0.79
Finland
234.122
0.464
0.628
30720
1.70
1.61
France
147.184
0.375
0.460
29783
1.72
0.79
Germany
115.243
0.425
0.526
30354
1.89
0.95
Greece
220.073
0.493
0.458
16997
6.35
1.02
Hong Kong
99.948
0.447
0.475
24848
2.97
0.47
India
25.963
0.505
0.741
605
7.69
0.95
Indonesia
127.126
0.545
0.603
1279
11.55
3.81
Ireland
636.282
0.526
0.547
33015
2.52
1.10
Israel
77.257
0.373
0.479
19775
5.90
0.62
Italy
406.999
0.465
0.570
25416
2.95
0.92
Japan
328.640
0.442
0.546
35147
0.29
0.81
Jordan
26.205
0.544
0.581
2160
3.91
0.49
Malaysia
56.700
0.512
0.598
4819
2.88
1.51
Mexico
1007.584
0.679
0.444
6701
11.69
1.96
Netherlands
391.558
0.474
0.467
32078
2.23
0.75
New Zealand
64.701
0.591
0.342
19668
2.26
1.79
Nigeria
66.197
0.417
0.730
593
21.75
8.87
Norway
182.353
0.569
0.415
49284
2.18
0.88
Pakistan
41.850
0.546
0.856
662
9.08
0.56
Philippines
97.271
0.681
0.344
1249
6.68
1.16
Portugal
545.510
0.653
0.359
14715
3.67
0.92
Singapore
60.004
0.429
0.450
25702
1.68
1.02

Legal Origin
English
German
French
English
French
Scandinavian
Scandinavian
French
German
French
English
English
French
English
English
French
German
French
English
French
French
English
English
Scandinavian
English
French
French
English

Investor protection variables

Importance
Legal
Ownership
of Equity
Tradition
Concentration
Markets
CM
29.00
0.28
CD
8.67
0.51
CD
13.33
0.62
CM
28.33
0.24
CD
23.00
0.38
CD
23.17
0.40
CD
16.17
0.34
CD
11.00
0.24
CD
6.67
0.50
CD
13.50
0.68
CM
34.50
0.54
CM
17.67
0.43
CD
5.33
0.62
CM
21.00
0.36
CM
26.33
0.55
CD
8.17
0.60
CD
20.17
0.13
CD
26.00
N.A
CM
30.67
0.52
CD
6.50
0.67
CD
23.50
0.31
CM
25.17
0.51
CM
10.25
0.45
CD
23.83
0.31
CM
10.25
0.41
CD
7.33
0.51
CD
13.83
0.59
CM
34.50
0.53

24

South Africa
South Korea
Spain
Sri Lanka
Sweden
Switzerland
Taiwan
Thailand
United Kingdom
United States
Mean
Median
Std

151.670
122.685
466.207
20.964
39.488
423.616
92.637
58.380
88.635
91.902
193.518
118.964
207.338

0.475
0.498
0.521
0.660
0.454
0.463
0.429
0.559
0.465
0.472
0.513
0.495
0.086

0.342
0.750
0.541
0.523
0.514
0.634
0.851
0.584
0.378
0.489
0.527
0.525
0.141

4090
13265
20484
1118
34608
45284
14014
2723
28627
35682
20256
22666
14247

7.50
4.15
3.35
10.21
1.92
1.41
1.91
3.58
2.39
2.60
4.45
2.75
4.09

2.22
2.11
1.12
0.48
1.41
0.66
0.49
1.49
0.76
0.05
1.31
0.93
1.42

English
German
French
English
Scandinavian
German
German
English
English
English

CM
CD
CD
CM
CD
CD
CD
CM
CM
CM

19.67
14.33
8.50
8.67
20.67
29.75
16.33
17.67
30.33
28.33
18.74
18.67
8.66

0.52
0.20
0.50
0.61
0.28
0.48
0.14
0.48
0.15
0.12
0.42
0.48
0.16

Firm Size is represented by US$ total assets (in million). Capital Intensity is the percentage of long-term assets to total assets. Percentage of Manufacturing Firms is computed from
the firm-year percentage, whose SIC code is from 2000 to 3999. Per Capita GDP is the mean value from 1991 to 2010. Inflation is the mean value of the change in consumer prices
from 1991 to 2010. Volatility of GDP Growth is the standard deviation of the GDP growth rate from 1991 to 2010.
Legal Origin and Legal Tradition in La Porta et al. (1998) are applied. The calculation of Importance of Equity Market has two steps in La Porta et al. (1997). We first rank the following three
variables: (1) the ratio of the aggregate stock market capitalization held by minorities to the GNP, (2) the number of listed domestic firms relative to the population, and (3) the number of IPOs
relative to the population. Each variable is ranked in such a way that higher scores indicate a greater importance of the stock market. Then we compute the average rank of the three variables.
Ownership Concentration is the median ownership stake of the three largest shareholders among the 10 largest publicly traded companies (La Porta et al., 1998).

25

Table 2 Score for earnings management

Panel A: Score for accrual-based earnings management
AEM measures
Country
AEM1
AEM2
AEM3
Australia
0.766
-0.532
0.421
South Africa
0.759
-0.747
0.495
Sweden
0.755
-0.659
0.579
United States
0.679
-0.614
0.606
Canada
0.704
-0.564
0.635
United Kingdom
0.583
-0.69
0.561
Hong Kong
0.599
-0.714
0.677
Finland
0.679
-0.816
0.662
New Zealand
0.591
-0.813
0.576
Ireland
0.492
-0.749
0.512
Philippines
0.667
-0.783
0.727
Norway
0.568
-0.738
0.695
Denmark
0.572
-0.777
0.689
Switzerland
0.554
-0.785
0.621
Israel
0.62
-0.733
0.799
Netherlands
0.51
-0.781
0.616
Mexico
0.539
-0.826
0.565
Thailand
0.592
-0.832
0.663
Chile
0.577
-0.839
0.607
Pakistan
0.61
-0.886
0.73
Jordan
0.499
-0.801
0.759
India
0.824
-0.86
0.909
Malaysia
0.61
-0.837
0.817
Singapore
0.528
-0.803
0.787
Sri Lanka
0.469
-0.803
0.728
Germany
0.513
-0.797
0.81
Belgium
0.471
-0.835
0.726
Taiwan
0.574
-0.854
0.78
Austria
0.452
-0.847
0.735
Nigeria
0.508
-0.916
0.751
Spain
0.391
-0.899
0.696
France
0.432
-0.869
0.779
Indonesia
0.461
-0.846
0.821

Score for AEM

1.3
4.7
5
6
6.3
8
11
13
13
13.3
13.7
14.7
15
15
15.3
15.3
15.7
16.3
16.7
21.7
22.3
22.7
22.7
22.7
22.7
23
23.7
24
27.7
28.3
29.7
30.7
30.7

Panel B: Score for real earnings management

REM measures
Country
REM1
REM2
Japan
0.753
0.928
Austria
0.677
0.865
Mexico
0.732
0.853
Netherlands
0.651
0.859
Greece
0.645
0.861
Switzerland
0.714
0.827
South Korea
0.597
0.895
Chile
0.641
0.828
Ireland
0.645
0.824
Belgium
0.600
0.846
Denmark
0.646
0.817
Taiwan
0.549
0.877
Thailand
0.575
0.852
Finland
0.609
0.818
Germany
0.570
0.824
Indonesia
0.585
0.797
Spain
0.640
0.782
Italy
0.556
0.796
Nigeria
0.479
0.825
France
0.560
0.781
United States
0.535
0.791
Singapore
0.456
0.796
Israel
0.517
0.774
United Kingdom
0.524
0.765
Norway
0.451
0.788
Portugal
0.402
0.804
India
0.490
0.747
South Africa
0.404
0.792
Sweden
0.483
0.733
Hong Kong
0.422
0.776
Jordan
0.467
0.715
Pakistan
0.444
0.726
Malaysia
0.345
0.744

Score for REM

1
4
4.5
5.5
6.5
7
7.5
9.5
10
10.5
11
11
11.5
13
15
16
17
18.5
18.5
21
21
23.5
24.5
24.5
25.5
25.5
26
27
27.5
28
29.5
30.5
33.5

26

Japan
South Korea
Italy
Portugal
Greece
Mean
Median
Std.

0.475
0.452
0.406
0.267
0.458
0.558
0.561
0.117

-0.875
-0.869
-0.907
-0.932
-0.924
-0.799
-0.815
0.094

0.819
0.877
0.816
0.798
0.906
0.704
0.727
0.116

31.3
33.3
34.3
35
35.3

Philippines
New Zealand
Canada
Sri Lanka
Australia
Mean
Median
Std.

0.412
0.418
0.372
0.371
0.336
0.533
0.542
0.113

0.644
0.556
0.695
0.701
0.622
0.787
0.796
0.076

34
34.5
35
35
37.5

In Panel A, countries are sorted by accrual-based earnings management scores. Each variable is computed using 222,513 firm-year observations from 1991 to 2010. AEM1 is based on the ratio
of the standard deviation of the operating income to that of operating cash flow, which is calculated by time-series data from each firm. AEM1 is the median of the ratio in each country. AEM2 is
the correlation between changes in accruals and changes in operating cash flows, and is computed from the pooled data in each country. AEM3 is the ratio of the absolute value of accruals to that
of operating cash flows, and is calculated in each firm-year. AEM3 represents each countrys median data. The score for AEM is the average rank of

AEM1, AEM2, and AEM3.

In Panel B, countries are sorted by real management scores. Each variable is computed using 222,513 firm-year observations from 1991 to 2010. REM1 and REM2 are the
correlation between the change in sales and the change in production costs, and the change in discretionary expenses. REM1 and REM2 are computed from the pooled data in each
country. The score for REM is the average rank of REM1 and REM2.

27

Table 3 Institutional variables in each country

Outside Investor
Legal Enforcement Disclosure Index
Analyst Following
Rights
Australia
4
9.51
8
2.496
Austria
2
9.36
2
4.647
Belgium
0
9.44
8
5.650
Canada
5
9.75
8
2.651
Chile
5
6.52
8
1.112
Denmark
2
10.00
7
4.755
Finland
3
10.00
6
7.590
France
3
8.68
10
5.665
Germany
1
9.02
5
5.509
Greece
2
6.82
1
2.212
Hong Kong
5
8.91
10
2.230
India
5
5.58
7
1.967
Indonesia
2
2.88
8
2.970
Ireland
4
8.36
10
9.362
Israel
3
7.72
7
1.482
Italy
1
7.07
7
5.204
Japan
4
8.24
7
2.244
Jordan
1
6.16
5
0.365
Malaysia
4
7.72
10
2.011
Mexico
1
5.37
7
4.227
Netherlands
2
10.00
4
9.692
New Zealand
4
10.00
10
2.525
Nigeria
3
4.34
5
0.872
Norway
4
10.00
7
5.642
Pakistan
5
3.67
6
1.036
Philippines
3
3.47
1
1.285
Portugal
3
7.19
6
5.403
Singapore
4
8.93
10
1.811
South Africa
5
6.45
8
3.211
South Korea
2
5.55
6
2.384
Spain
4
7.14
5
9.349
Sri Lanka
3
4.66
4
0.155
Sweden
3
10.00
2
4.270
Switzerland
2
10.00
2
7.990
Taiwan
3
7.37
6
2.200
Thailand
2
4.89
10
2.730
United Kingdom
5
9.22
10
6.046
United States
5
9.54
7
5.209
Mean
3.13
7.62
6.58
3.846
Median
3
7.98
7
2.850
Std
1.40
2.14
2.64
2.571
Outside Investor Rights mean anti-director rights, which is an index aggregating shareholder rights. The antidirector rights index is formed by adding 1 when: (1) the country allows shareholders to mail their proxy vote to
the firm, (2) shareholders are not required to deposit their shares prior to the general shareholders meeting; (3)
cumulative voting or proportional representation of minorities in the board of directors is allowed; (4) an
oppressed minorities mechanism is in place; (5) the minimum percentage of share capital that entitles a
shareholder to call for an extraordinary shareholders meeting is less than or equal to 10 percent (the sample
median) or (6) shareholders have preemptive rights that can be waived only by a shareholders vote (La Porta et
al. 1998). In our study, the index ranges from 0 to 5. Following Leuz et al. (2003), we compute Legal
Enforcement as the average score across three variables used in La Porta et al. (1998): (1) the efficiency of the
judicial system, (2) an assessment of rule of law, and (3) the corruption index. All three variables range from 0 to
10. The Disclosure Index is obtained from World Bank 2005. Analyst Following is the number of analyst following
per firm-year in each country.
Country

28

Table 4 Correlation between AEM, REM, and institutional variables

Score for
AEM

Score for
REM

-0.381
(0.019)
Outside Investor
-0.423
0.465
Rights
(0.008)
(0.003)
Legal
-0.392
-0.076
Enforcement
(0.015)
(0.649)
Disclosure
-0.080
-0.411
Index
(0.633)
(0.010)
Analyst
-0.159
0.174
Following
(0.341)
(0.296)
The variables are defined in Tables 2 and
3. n=38.

Outside
Investor
Rights

Legal
Enforcemen
t

Disclosure
Index

Score for REM

0.073
(0.665)
-0.164
(0.325)
0.321
(0.050)

0.525
(0.001)
0.081
(0.629)

-0.048
(0.774)

29

Table 5 Regression results of earnings management and investor protection

Panel A: Regression results for accrual-based earnings management (Dependent variable = AEM)
Dependent variable = AEM
Constant

28.545***
(8.462)

28.414***
(8.549)

23.570***
(6.386)

22.007***
(5.956)

36.002***
(7.611)

Outside Investor Rights

-0.464***
(-3.061)

-0.395**
(-2.549)

Legal Enforcement

-0.457***
(-3.075)

-0.474***
(-2.756)

Disclosure Index

-0.209
(-1.260)

-0.058
(-0.391)

Analyst Following

-0.129
(-0.778)

0.081
(0.460)

R (or Adjusted R )
Number of Observations

0.207

0.208

0.042

0.017

0.319

38

38

38

38

38

Panel B: Regression results for real earnings management (Dependent variable = REM)
Dependent variable = REM
Constant

10.074***
(3.019)

20.561***
(5.511)

16.160***
(4.345)

27.002***
(7.849)

15.949***
(3.220)

0.483***
(3.224)

0.373**
(2.298)

Legal Enforcement

-0.054
(-0.326)

0.160
(0.887)

Disclosure Index

0.171
(1.026)

0.065
(0.421)

Analyst Following

-0.385**
(-2.502)

-0.416**
(-2.262)

0.224

0.003

0.028

0.148

0.254

Outside Investor Rights

R (or Adjusted R )

Number of Observations
38
38
38
38
38
The t-statistics are in parentheses. ***, **, and * indicate significance at 1%, 5%, and 10% levels (two-tailed).
2
2
For equation (11), R shows the adjusted R . All variables are defined in Tables 2 and 3.The equations are:
AEMi (or REMi) = 0 + 1 Outside Investor Rightsi + i
(7),
AEMi (or REMi) = 0 + 1 Legal Enforcementi + i
(8),
AEMi (or REMi) = 0 + 1 Disclosure Indexi + i
(9),
AEMi (or REMi) = 0 + 1 Analyst Following + i
(10), and
AEMi (or REMi) = 0 + 1 Outside Investor Rightsi + 2 Legal Enforcementi
+ 3 Disclosure Indexi + 4 Analyst Followingi + i
(11).

30

Table 6 Regression results of earnings management and investor protection: Controlling for Big 4 audits ratio
Constant
Outside Investor Rights
Legal Enforcement
Disclosure Index
Analyst Following
Big4 Ratio

AEM

REM

30.623***

14.771***

(6.988)

(3.203)

-0.451**

0.371*

(-2.473)

(1.934)

-0.428**

0.132

(-2.403)

(0.703)

-0.009

0.048

(-0.059)

(0.285)

0.138

-0.466**

(0.813)

(-2.603)

-0.051

0.046

(-0.294)

(0.254)

Adjusted R
0.3485
0.2790
Number of Observations
33
33
The t-statistics are in parentheses. ***, **, and * indicate significance at 1%, 5%, and 10% levels (two-tailed).
Big4 Ratio is the ratio of firms audited by the Big 4 in each country. The other variables are defined in Tables
2 and 3.The equations are as follows:
AEMi (or REMi) = 0 + 1 Outside Investor Rightsi + 2 Legal Enforcementi
+ 3 Disclosure Indexi + 4 Analyst Followingi + 5 Big4 Ratioi + i.

31

Table 7 Regression results of earnings management and investor protection: Another proxy for Legal
Enforcement
AEM

REM

Constant

36.409***
(7.583)

16.291***
(3.235)

Outside Investor Rights

-0.391**
(-2.493)

0.391**
(2.375)

Legal Enforcement

-0.437**
(-2.638)

0.065
(0.376)

Disclosure Index

-0.075
(-0.502)

0.062
(0.396)

Analyst Following

0.036
(0.215)

-0.354*
(-1.992)

0.308

0.240

Adjusted R

Number of Observations
38
38
The t-statistics are in parentheses. ***, **, and * indicate significance at 1%, 5%, and 10% levels (two-tailed).
Legal Enforcement is the average score across three legal variables: (1) the efficiency of the judicial system
used in La Porta et al. (1998), (2) an assessment of rule of law based on Kaufmann et al. (2004), and (3) the
corruption index based on Kaufmann et al. (2004). First variables range from 0 to 10. Second and third
variables range from -2.5 to +2.5. The other variables are defined in Table 2 and Table 3.The equations are:
AEMi (or REMi) = 0 + 1 Outside Investor Rightsi + 2 Legal
Enforcementi + 3 Disclosure Indexi + 4 Analyst Followingi
+ i

32

Table 8 Regression results of earnings management and investor protection based on country-year earnings
management measures
AEM
42.806***
(13.812)

REM
11.409***
(3.195)

Outside Investor Rights

-0.324***
(-8.468)

0.150***
(3.409)

Legal Enforcement

-0.393***
(-9.211)

0.085*
(1.73)

Disclosure Index

-0.054
(-1.495)

0.106**
(2.54)

Analyst Following

-0.051
(-1.147)

-0.160***
(-3.145)

Year_Dummy

included

included

0.362

0.153

Constant

Adjusted R

Number of Observations
548
548
The t-statistics are in parentheses. ***, **, and * indicate significance at 1%, 5%, and 10% levels (two-tailed).
All variables are defined in Tables 2 and 3.The equations are:
AEMit (or REMit) = 0 + 1 Outside Investor Rightsit + 2 Legal Enforcementit
+ 3 Disclosure Indexit + 4 Analyst Followingit + Year_Dummy +it

33