Electronic copy available at: http://ssrn.com/abstract=1839449

1

Executive Integrity, Audit Opinion, and Fraud in Chinese Listed Firms

Jiandong Chena, Douglas Cumming

b,*, Wenxuan Hou

c, and Edward Lee

d

a School of Public Finance and Taxation, Southwestern University of Finance and Economics, China

b Schulich School of Business, York University, 4700 Keele Street, Toronto, Ontario M3J 1P3, Canada

c Durham Business School, Durham University, United Kingdom

d Manchester Business School, University of Manchester, United Kingdom

Abstract

The literature on the determinants of fraud mainly focuses on the internal and external

governance mechanisms. Although sound mechanisms still need to be implemented or

coordinated by executives, the role of their integrity is underresearched. Following the

intuition in Jensen et al. (2004), we use earnings management to proxy the lack of integrity of

executives and identify it as the antecedent of fraudulent behavior. The former can develop

into serious malpractice in the absence of necessary intervention. We further show that

auditor opinions play important roles not only in whistleblowing fraud, but also in its

prevention. Nonstandard audit opinions are issued when auditors identify some problems in

financial statements, and their incidence is lower in firms with good governance. More

importantly, nonstandard opinions help deter executives with lower levels of integrity from

committing fraud, presumably because such opinions can alert outside investors, the board,

the regulatory commission, and even the media, putting these executives under severe

scrutiny.

JEL classification: G15; G30; K22; M41

Keywords: Auditor, earnings management, fraud, integrity, ethics, China

We thank Julian Clarke for helpful comments. *Corresponding author. Tel. 416-736-2100 ext. 77942; fax 416-736-5687l; e-mail

dcumming@schulich.yorku.ca.

Electronic copy available at: http://ssrn.com/abstract=1839449

2

1. Introduction

The contributors of fraudulent activities have been widely studied in the literature from the

perspectives of external and internal corporate governance mechanisms. Regarding the

external mechanism, the China Securities Regulatory Commission has been found to provide

favorable regulatory conditions to state-owned listed firms due to their mutual affiliation with

the government (Hou and Moore, 2010). Investors still take heed of its regulatory

enforcements against fraud seriously by revaluing companies downward (Chen et al., 2005).

The regulatory commission perceives that it is the auditors’ responsibility to detect and report

fraud (Dyck et al., 2010), and auditors are sanctioned for failing to do so (Firth et al., 2005);

however, the role of ex ante initiatives of auditors in deterring fraud remains

underresearched.

Regarding the internal mechanism, board independence, the existence of an audit committee,

and the presence of accounting and banking professionals on the committee have been found

to decrease the incidence of fraudulent activities (see Beasley, 1996; Dechow et al., 1996;

Beasley et al., 2000; Uzun et al., 2004; Chen et al., 2006). Denis et al. (2006) show that

option intensity in CEO remuneration encourages risk taking and induces fraud, while

Erickson et al. (2006) show that the exercise of executive options and sales of executive

stocks are not significantly higher for fraudulent firms. Although any theoretically sound

governance mechanisms still need to be implemented or coordinated by the chief executive

officer (CEO) and other executives (Nowak and McCabe, 2003), the effects of their integrity

level on fraud has yet to be investigated in the literature. The importance of integrity became

manifest during the credit crisis. The low levels of integrity of executives have been

identified as a major attributor (Jensen and Walking, 2010), and the recent campaign of

Occupy Wall Street begun in late 2011 is against the insatiable greed of top executives.

3

This paper intends to remedy these shortcomings by arguing that executive integrity matters

for governance quality and therefore the incidence of fraud and auditors play an important

role in disciplining executives with lower levels of integrity. To carry out our analysis, we

include listed firms in the Chinese stock market from 2001 to 2008. Jensen et al. (2004) argue

that managers systematically erode integrity by engaging in the earnings management game,

which amounts to lying to the investors, and this behavior is attributed to the budget-based

remuneration system. We therefore employ earnings management as the proxy for the lack of

integrity, or disintegrity, of executives. Earnings management is measured by the ratio of

non-operating income to sales, following Ding et al. (2007) and Bertrand et al. (2002) and a

dummy variable indicating whether it is above or equal to its industry-median value. This is

believed to be a superior measure to that of Chen and Yuan (2004), who argue that prevailing

related-party transactions in China make it possible for firms to use non-core operating

profits or less as a means of earnings management. Jian and Wong (2004) and Ding et al.

(2007) add that China’s traditionally tax-oriented accounting system makes firms unlikely to

adjust their earnings via non-cash accruals. We find that executive disintegrity measured by

earnings management is positively related to the incidence of fraud in the subsequent

sections. The results suggest that disintegrity is the antecedent of fraud because it induces

unethical decisions and causes fraudulent misconduct to thrive.

We then examine the role audit opinions played in the depravation process in which

executives with poor integrity become fraudsters. There are two categories of audit opinions

in China: standard unqualified opinions and nonstandard ones. Standard unqualified opinions

are issued when a financial statement is true and free from material misstatements, whereas

the nonstandard ones are issued when audit firms identify some problem therein. The

4

occurrence of nonstandard audit opinions is found to be negatively related to the integrity

level of executives and external governance quality in terms of larger analyst following and

mutual fund ownership. Executives with good integrity and those under better external

monitoring tend to give true and fair views in the financial statements that comply with

relevant regulations and requirements.

More importantly, we show that auditors help discipline executives with lower levels of

integrity and prevent them from becoming fraudsters, in that nonstandard audit opinions put

executives under closer scrutiny by altering their integrity problems to outside investors,

other board members, regulatory commissions, and even the media and leave fewer

opportunities for misconduct. This finding suggests that the auditors not only detect ex post

fraud, but also help avoid it ex ante.

This paper contributes to the literature in various ways. First, it proposes a measure for

integrity and provides direct evidence for the argument in Jensen et al. (2004), that without

integrity, corporate governance mechanisms do not work. To our best knowledge, this is the

first paper to investigate whether and how the lack of integrity leads to fraudulent behavior.

Second, it contributes to the literature of audit opinions. We show that executive integrity,

analyst following, and fund ownership help improve the quality of financial reports and

reduce the occurrence of nonstandard audit opinions. Third, and most importantly, we extend

studies of corporate fraud by showing that audit opinions refrain CEOs with poor integrity

from committing fraud. This paper also has policy implications. At the firm level, integrity

needs to be a major criterion in the appointment of CEOs and other executives. At the market

level, the regulatory commission and investors can count on auditors not only in fraud

detection, but also in its prevention.

5

This paper proceeds as follows: Section 2 reviews the literature and develops the hypotheses.

The research design and sample data are introduced in Section 3. Section 4 reports and

discusses the empirical findings and Section 5 concludes the paper.

2. Literature Review and Hypothesis Development

2.1. Executive Integrity and Fraud

Various board characteristics have been identified in the literature as the major determinants

of fraud. For example, board independence, an audit committee, and directors with

accounting and financial backgrounds help enhance the monitoring quality of the board

(Beasley, 1996, 2000; Dechow et al., 1996; Uzun et al., 2004; Chen et al., 2006; Firth et al.,

2011). The supervisory board in China reacts only passively to regulatory enforcement

against fraud by increasing the meeting frequencies rather than deterring the occurrence of

fraud ex ante (Jia et al., 2009; Ding et al., 2010). The executive compensation system also

plays a role. Denis et al. (2006) show that the intensity of executive options encourages

excessive risk taking and induces fraud. In addition, Hou and Moore (2010) show that the

prevailing state ownership in China aggregates agency problems in privatized firms, but

brings about favorable regulatory conditions by playing down the inspection severity in state-

controlled firms.

These studies fail to consider executive integrity, although any good governance mechanisms

still need to be implemented or coordinated by the executives. Management continues to be

seen as a major corporate governance actor (Cohen et al., 2000). In fact, board members

perceive impediments to information flow imposed by CEOs, and Nowak and McCabe

6

(2003) point out that management integrity is central to information flow, and therefore

central to the effectiveness of monitoring and control of the board.

Although integrity is valuable in both material and non-material terms, violation is still

detected when potential gains for violating are estimated to be greater than the costs (see

Bradford, 2007). There are serious concerns about executives’ integrity levels. For example,

the protesters of the recent Occupy Wall Street campaign begun in late 2011 are against the

insatiable greed of top executives. This pervades industrial as well as financial corporations

and goes well beyond financial corporations and the United States. Lack of integrity is also

attributed as being one of the major causes of the current financial crisis (Jensen and

Walking, 2010). As the pioneering advocates of the importance of integrity, Jensen et al.

(2004) highlight how managers erode integrity and destroy firm value by systematically

playing the earnings management game. Since earnings management is not in line with the

maximization of shareholder wealth, it amounts to lying to the investors to whom managers

have a fiduciary responsibility and signals a breakdown in integrity.

Some impediments of integrity have been identified in the literature. Jensen et al. (2004) note

that the integrity problem cannot be handled by executive remuneration and the commonly

used budget-based bonus and promotion systems instead motivate poor integrity. Executives

are rewarded in terms of compensation as well as promotion if they achieve the operating

targets set in their contracts that are mainly used for accounting performance (Tang et al.,

1999; Firth et al., 2006; Cao et al., 2011). In addition, accounting performance is crucial for

firms to raise capital and enhance stock return (see Firth et al., 2005). Therefore, executives

with poor integrity choose to manipulate financial information and engage in earnings

management. Rost (2007) argues that the increasing trend to fill CEO openings through

7

external hires in the external labor market also discourages executives from investing in

integrity and encourages them to invest in networking, since a common past between a

person and a firm downgrades their promotion prospects.

Attributes to promote integrity have also been discussed. The Sarbanes–Oxley Act has been

found to significantly expand the responsibility of auditors and management. Cohen et al.

(2010) find that the requirement for the certification by top management of financial

statements helps improve integrity because it mandates the involvement of the managers.

Kizirian et al. (2005) and Cohen et al. (2011) state that both auditors and audit committees

play an important role in ensuring integrity. Under Section 404 of the Sarbanes–Oxley Act,

auditors are required to assess management integrity and issue an adverse internal control

report noting any material weaknesses in internal control (Kizirian et al. 2005).

Although various factors encourage or discourage integrity, Fuller and Jensen (2002) argue

that human choice is also critical. Executives with low levels of integrity tend to ignore

policies and procedures to pursue self-interests at the costs of other investors, and therefore

unethical decisions and fraudulent behaviors are more likely to flourish among them.

Executives of high integrity are essential to implementing good governance systems, and

therefore integrity is essential for effective monitoring. We hereby hypothesize the following.

H1: Executive disintegrity is positively related to the incidence of fraud.

2.2. Audit Opinion and Corporate Governance

Chinese auditors are not self-regulated but government regulated, by the Chinese Institute of

Certified Public Accountants (CICPA), an agent of the Ministry of Finance. Government-

affiliated CPA firms dominate the audit market, representing 75% of auditors and servicing

8

70% of clients, and around two-thirds of listed firms are controlled by the state (see DeFond

et al., 2000; Yang et al., 2001). Audit independence in China is a concern, in that auditors

play the role of agents of the state audit bureau and bear little economic responsibility for

their improper auditing actions, such as pleasing their clients or pursuing their own interests.

State shareholders have a strong incentive to obtain favorable earnings from executives, but

little demand for independent audits. Xiao (2000) points out that some audited companies

even expect auditors to help them conceal wrongdoings and reappoint auditors if they refuse

to help in the cover-up. In addition, the shortage of qualified accountants and auditors also

holds back the development of professional auditing.

Since audit independence is essential for the auditors’ attesting function (Xiang, 1998),

studies of Chinese audit issues mainly focus on this issue. DeAngelo (1981), Krishnan et al.

(1996), and Yang et al. (2001) regard nonstandard opinions as a proxy for auditor

independence. There are two categories of audit opinions in China: standard unqualified

opinions and nonstandard ones. Nonstandard (standard unqualified) opinions are issued when

audit firms identify (no) problems or potential problems with a financial statement.

Nonstandard opinions include unqualified opinions with explanations, qualified (modified)

audit opinions either with or without explanations, adverse audit opinions, and disclaimers.

A set of reforms has helped improve audit independence in China. DeFond et al. (2000) find

that new auditing standards promulgated in 1995 that prescribe detailed auditing procedures

make auditors less likely to succumb to management pressure in issuing clean opinions when

modified options are appropriate, increasing the frequency of modified opinions from 1% to

9%. Since executives prefer clean audit opinions, they tend to switch from larger audit firms

to smaller ones, which are less independent and will more easily issue clean reports. Such

9

switching is motivated by “shopping” for improved audit opinions (see Krishnan and

Stephens, 1995).

A program launched in 1997 aims to further enhance audit independence operationally by

disaffiliating CPA firms from their sponsoring body and make them financially and

operationally independent (Yang et al., 2001). In affiliated audit firms, auditors are not

exposed to personal risks when performing audits and profits are owned by the sponsoring

government body, whereas in disaffiliated firms, partners are liable for their auditing

practices and profits are retained within the firm for distribution among partners. Following

the reform, the number (percentage) of nonstandard opinions increased from four to 152

(2.20% to 18%), according to Yang et al. (2001). There is no evidence that international firms

and the Big 5 (now Big 4) accounted for the increase in nonstandard opinions. Actually, Kim

et al. (2011) shows that Big 6 auditors have incentives to be more conservative than non-Big

6 auditors in determining reported earnings.

These studies were conducted at the market level. For the determinants of audit opinion at the

firm level in China, DeFond et al. (2000) find that clients receiving qualified (modified) audit

opinions tend to be those with poor operating performance, lower current ratios, and larger

size, and the auditors that issue more qualified (modified) opinions tend to be larger auditors

and joint venture auditors. Firth et al. (2007) argue audit opinion also reflects the

informativeness of earnings and find that private firms and firms with a larger supervisory

board and a greater percentage of independent directors are more likely to receive standard

(clean) audit opinions, implying their earnings are more informative. The threshold for giving

a standard (clean) opinion and auditor reporting conservatism can also be increased by audit

risks. Firth et al. (2011a) show that auditors tend to issue qualified audit opinions for firms

10

that incur financial restatements. Firth et al. (2011b) note that the organizational form of audit

firms can affect their opinion. A partnership firm (limited liability firm) has more (less)

wealth at risk and larger (smaller) risk and liability exposure, and therefore tends to issue

qualified (clean) audit opinions. Joe (2003) shows that auditors overreact to the disclosure of

a negative event in the press that provides no new information, becoming more conservative,

and tend to issue qualified audit opinions.

These studies incorporate the internal governance features and characteristics of the auditor

and auditee, but fail to consider the roles of executive integrity and external monitoring

mechanisms. We argue that executives with higher levels of integrity and those under more

effective external monitoring (as with greater mutual fund ownership and greater analyst

following) are more likely to give a true and fair view in financial statements that complies

with relevant regulations and requirements. We thereby hypothesize the following.

H2: Executives with lower levels of integrity are associated with a higher incidence of

nonstandard audit opinions.

H3: Better external governance is associated with a lower incidence of nonstandard audit

opinions.

2.3. Audit Opinion and Fraud Prevention

Academics, practitioners, and policy makers focus on the role of auditors as fraud

whistleblowers, capable of and responsible for detecting and reporting fraudulent activities.

Dyck et al. (2010) document that auditors account for 10% of detected fraud, which is larger

than the figure from the regulatory commission (7%) and smaller than that from employees

(17%), non-financial market regulators (13%), and the media (13%). On the one hand,

11

auditors who blow the whistle are more likely to lose the accounts of the fraud-committing

firms; on the other hand, if they fail to report fraud, the regulatory commission imposes

administrative sanctions in the form of warnings, fines, and the withdrawal of auditors’

licenses. Yang et al. (2001) show that the number of administrative sanctions against auditors

increased from 100 to about 500 during 1994–1997. During the sample period 1996–2002 of

Firth et al. (2005), 72 auditors were sanctioned by the regulatory commission for failing to

identify material misstatement frauds from listed companies (e.g., revenue-related frauds).

Firth et al. (2005) also show that fewer enforcement actions are taken against auditors that are

more competent and independent in terms of qualified audit opinions (modified audit

opinions, or MAOs) and the logarithm of total clients’ assets.

The role auditors and especially their opinions play in deterring managers from committing

fraud remains underresearched. Chen et al. (2006) find no significant relation between audit

firm size and the incidence of fraud. We argue that nonstandard opinions intervene in the

process of executive disintegrity developing into serious fraudulent activities in that they can

draw the attention of investors, board members, the regulatory commission, and even the

media, putting executives under closer scrutiny and allowing fewer chances for executives

with low levels of integrity to commit fraud. On the contrary, if an executive with low levels

of integrity does not incur a nonstandard opinion as a warning from auditors, disintegrity can

further deteriorate into malpractice. We thereby hypothesize the following.

H4: Nonstandard audit opinions help deter executives with lower levels of integrity from

committing fraud.

12

3. Research Design

3.1. Model Specifications

To empirically test the predictions in H1, we apply the logistic regression model

k

k kk ControltyDisintegriFraud1 110

(1)

where Fraud is a one-year lead dummy variable that is equal to one if the firm is subject to

regulatory enforcement against fraud, and zero otherwise. Following the intuition of Jensen et

al. (2004), Disintegrity is proxied by two earning management measures. The variable EM is

the ratio of non-operating income relative to sales. Chen and Yuan (2004), Jian and Wong

(2004), and Ding et al. (2007) argue that this variable is superior to accruals for measuring

earnings management in China for two reasons. First, related-party transactions are

commonly observed in China, making it easy for firms to use non-core operating income to

manipulate earnings. Second, the traditionally tax-oriented Chinese accounting system makes

it difficult for firms to adjust their earnings via non-cash accruals. Ding et al. (2007) further

show that the correlation between the ratio of non-operating income to sales and discretionary

accruals is nearly 0.5 and significant. To account for industry traits, we propose another

measure, DEM, a dummy variable equal to 1 if EM is equal to or above the industry median

value of yearly observations, and 0 otherwise.

A set of control variables is incorporated to control for the effects of firm characteristics,

performance, and corporate governance, including the natural logarithm of market

capitalization (Size), the price-to-book ratio (PB), a dummy variable of consecutive losses

(STPT is equal to one if a listed firm experiences at least two consecutive year losses, and

zero otherwise), analyst following (Analysts is obtained as the natural logarithm of the sum of

13

the number of analysts following and one), fund ownership (Fund is the number of shares

held by open- and closed-end funds relative to the total number of shares), the number of

restricted shares relative to the total number of shares (RTT), the Herfindahl index of the 10

largest firm blockholders (OwnCon), a dummy variable of duality (Duality is equal to one if

the CEO also holds the position of board chair, and zero otherwise), a dummy variable for

board meeting frequency (Dmeet is equal to one if the number of board meetings is above the

median value of yearly observations, and zero otherwise), a dummy variable of board size

(Dbsize is equal to one if the number of board members is above the median value of yearly

observations, and zero otherwise), a dummy variable of board independence (Drind is equal

to one if the ratio of independent directors is above the median value of yearly observations,

and zero otherwise), and a dummy variable for supervisory board size (Dssize is equal to one

if the number of supervisory board members is above the median value of yearly

observations, and zero otherwise). Regional and industry effects are also controlled for.

Following Firth et al. (2006), firms are grouped into four regions based on levels of economic

development to construct the regional dummy variables: (1) firms located in Shanghai and

Shenzhen, (2) firms located in more developed areas, including the open cities and provinces

along the coast, (3) firms located in the inland provinces, and (4) firms located in the least

developed area, in the northwestern part of the country. The industry dummy variables are

constructed based on the first two digits of the Global Industry Classification Standard

(GICS) codes. If α1 > 0 and is significant, then the lack of integrity is associated with a higher

incidence of fraudulent activities, supporting our prediction in H1.

To test H2 and H3, we apply the following logistic model by regressing the dummy variable

of nonstandard auditor opinions (ANS is equal to one if a nonstandard opinion report is issued

14

by the auditor, and zero otherwise), which indicates (potential) problems with financial

statements, against executive disintegrity and a set of corporate governance measures:

j

j jjk

k

k kk ControlCGMtyDisintegriANS1 11 110

(2)

where the corporate governance measures contain such external features as analyst following

(Analyst), fund ownership (Fund), and internal features such as the ratio of restricted to total

shares (RTT), ownership concentration (OwnCon), CEO duality, (Duality), board meeting

frequency (Dmeet), board size (Dbsize), board independence (Drind), and supervisory board

size (Dssize). Executive integrity is proxied by EM and DEM, as defined earlier. Firm

characteristics and regional and industry effects are controlled for. To support our hypotheses

H2 and H3, we need to observe positive coefficients for executive disintegrity and negative

coefficients for analyst following and fund ownership.

Finally, H4 can be tested by using the regression model

k

k kk ControlAUDtyDisintegriANStyDisintegriFraud3 13210 .

(3)

where the one-year lead dependent dummy variable Fraud is equal to one if the firm is

subject to regulatory enforcement against disclosed fraud, and zero otherwise; Disintegrity is

proxied by EM and DEM; and ANS is equal to one if a nonstandard opinion report is issued

by the auditor, indicating problems with a financial statement, and zero otherwise. The

variable ANS is also interacted with EM and DEM, respectively, and the same set of control

variables is incorporated. To support H4, the coefficient of the interaction term α3 needs to be

significantly negative, showing that when executives with lower levels of integrity are

warned by nonstandard audit opinions, they are less likely to become fraudsters.

15

3.2. Sample Description

The data used in this paper are from two sources. Corporate fraud, industry codes, the

performance measure (STPT), the ratio of restricted to total shares (RTT), and ownership

concentration (OwnCon) data are from the China Center for Economic Research (CCER),

and the rest of the variables are constructed based on data from China Securities Market &

Accounting Research (GTA/CSMAR). Both databases are commonly used in studies of the

Chinese capital market. The sample covers the period from 2001 to 2008, because most of the

corporate governance variables used in this paper are only available since 2001.

Figure 1 shows that the number (ratio) of nonstandard audit opinions in the beginning of the

sample period, from 2001 to 2002, is around 140 (12%). Both the number and ratio are

slightly smaller than the figures in 1998 reported by Yang et al. (2001). One year later, the

number (ratio) drops to around 80 (7%) in 2003, recovers in 2005, and falls back to the level

of 2003 at the end of the sample period, in 2008. Different from the upward trends

documented by Yang et al. (2001), the trends in this paper are downward. Given the fact that

corporate governance, audit independence, and regulatory supervision have all improved in

the past decade, the downward diagram could imply that the quality of financial statements

has improved at the market level, consistent with the finding of Hou et al. (2011), that the

informativeness of Chinese listed firms has improved.

4. Empirical Findings

4.1. Descriptive Statistics

Table 1 presents the summary statistics of the variables used in this study. The ratio of

nonstandard audit opinions is 9.31%, on average, and the ratio of fraud is smaller, at about

4%. The ratio of non-operating profits to sales is around 2%. Table 1 also reports the

16

summary statistics for the split sample, namely, firms with standard opinions and these with

nonstandard audit opinions. Firms with nonstandard opinions have been found to be

associated with a substantially higher incidence of fraud in the subsequent period (21.12%

versus 2.16%); lower levels of executive integrity, as reflected by higher levels of earnings

management (5.86% versus 1.53%); smaller firms; larger price-to-book ratios; a higher

incidence of consecutive losses (40.27% versus 4.70%); smaller numbers of following

analysts; lower fund ownership (0.33% versus 3.00%); lower ownership concentration; a

higher incidence of CEO duality; lower board meeting frequencies; and lower board

independence. These findings support our hypotheses H2 and H3, that the incidence of

nonstandard audit opinions is associated with lower levels of integrity and governance

quality.

Table 2 presents the correlation matrix for the variables used in this study. Fraud and earnings

management are positively related to each other, providing supporting evidence for our

hypothesis H1, that poor integrity tends to induce subsequent fraud. In addition, fraud,

earnings management, and nonstandard audit opinions are negatively related to past operating

performance, corporate quality, and firm size.

4.2. Test of H1

Table 3 presents the results of the test of H1. We examine whether executives with lower

levels of integrity tend to become fraudsters in the subsequent period. Lack of integrity is

measured by the ratio of non-operating profits to sales. Its coefficients are significant and as

high as 2.2355, 0.9538, and 1.0015, respectively in regression I, II, and III. Regression I

controls for firm characteristics only, regression II also incorporates firm performance, and

regression III accounts for industry and regional effects. Supporting our hypothesis H1, the

17

results across the three regressions show that lower levels of integrity serve as an antecedent

of fraud, signal governance malfunction, and lead to a higher incidence of fraud in the

subsequent period.

For an increase of one standard deviation (0.0655) in the ratio of non-operating profits to

sales, the incidence of fraud is increased by 6.56%. This shows the economic significance of

the finding. In addition, firms with consecutive losses and therefore delisting risks (STPT = 1)

are 56.29% more likely to commit fraud. Analyst following, ownership concentration, firm

size, and board meeting frequency are found to help deter fraudulent activities. This is

consistent with the findings of Chung et al. (2002), Hou and Moore (2010), Cumming et al.

(2011), and Hou et al. (2011). As robustness checks, we also use the dummy variable DEM as

an alternative proxy for lack of integrity, assigning it a value of one if the ratio of non-

operating income to sales is above or equal to the industry median value of yearly

observations, and zero otherwise. The results remain consistent, and both the model’s

significance level and explanatory power are increased.

4.3. Tests of H2 and H3

Table 4 presents the results of the determinants of audit opinions and, in particular, the

impact of corporate governance. We regress a dummy variable of audit opinion (ANS is equal

to one if a nonstandard opinion report is issued by the auditor, and zero otherwise) against a

set of proxies for corporate governance. The dependent variable for regressions I and III (II

and IV) is ANS at time t + 1 (time t). The disintegrity measure in regressions I and II (III and

IV) is EM (DEM). The coefficients for DM and DEM across the four regressions are

positively significant, which supports our hypothesis H2, that executives with lower levels of

integrity are more likely to be caught by auditors for material misstatements. With an

18

increase of one standard deviation (0.0655) in the ratio of non-operating profits to sales, the

incidence of nonstandard opinions increases by 10.57%, and executives whose integrity level

is below the industry median level (i.e., DEM = 0) are 31.84% more likely to incur

nonstandard audit opinions.

Various corporate governance variables are also significant. In particular, analyst following

and fund ownership are found to promote the quality of governance and therefore of financial

statements. This supports our hypothesis H3. With an increase of one standard deviation in

analyst following (0.9133) and fund ownership (0.0626), the incidence of nonstandard audit

opinion is decreased by 64.48% and 66.41%, respectively, as suggested by regression I. This

provides direct evidence that external governance features help enhance the quality of

financial statements. The results are robust to the control of other internal governance

attributes, firm characteristics, and regional and industry effects. Consistent results are also

documented by Firth et al. (2007). Ownership concentration, board meeting frequency, and

board independence are found to reduce the incidence of nonstandard opinions. In addition,

large firms and firms free of consecutive losses tend to receive clean opinions from auditors.

4.4. Test of H4

The results for the test of H4 are reported in Table 5. By investigating whether auditors can

intervene in the process of executives with poor integrity becoming fraudsters, we regress the

one-year lead dependent dummy variable Fraud against CEO disintegrity, nonstandard

auditor opinion (ANS), and their interaction term. Panel A measures CEO disintegrity as the

ratio of non-operating profits to revenue (EM), and panel B measures the industry median-

adjusted dummy variable DEM. The coefficient of EM indicates whether executives with

poor integrity tend to commit fraud under the condition that they are not detected by auditors.

19

These coefficients are significantly positive, at 3.2992, 2.8999, and 3.2097, respectively,

from regressions I, II, and III. These findings again reinforce H1. An increase of one standard

deviation (0.0655) in earnings management can lead to a 21.02% increase in the incidence of

fraud, as suggested by regression III. This shows that poor integrity is an antecedent of fraud

and, without necessary intervention, is likely to develop into fraud.

More importantly, the coefficients of the interaction term (EM.ANS) capture the effects of

intervention from auditors. As shown earlier, some executives with low levels of integrity

incur nonstandard opinions from auditors, which serve as a warning to them and as an alert to

outside investors, the board, the regulatory commission, and even the media. Supporting our

hypothesis H4, the coefficients are significantly negative (-4.0126, -3.4746, and -3.9042)

across the three regressions (I, II, and III). Holding the level of earnings management

constant, a nonstandard audit opinion helps reduce the likelihood of fraud by as much as

41.62%. In other words, it is substantially less likely for executives with lower levels of

integrity to become fraudsters if their integrity problem is detected and reported by auditors.

A nonstandard audit report can put such executives under closer scrutiny, leaving them fewer

chances for malpractice. The results are of both statistical and economic significance and

suggest the critical role played by auditors in deterring fraud by disciplining executives with

lower levels of integrity.

The results are robust to controlling firm characteristics (in regression I), firm operating

performance, corporate governance (in regression II), and regional and industry effects (in

regression III). In addition, the coefficients for nonstandard audit opinion are significantly

positive, showing that nonstandard audit opinions can signal poor governance quality and

predict fraud. Larger firms and those with better corporate governance are found to be

20

negatively related to the incidence of fraud. As further robustness checks, we replicate the

tests in panel B of Table 5 by replacing EM with DEM, a dummy variable equal to one if EM

is above or equal to the industry median value of yearly observations, and zero otherwise.

The results remain consistent and, again, support our hypothesis H4.

5. Conclusion

Various governance features are identified as the determinants of fraud in the literature. This

paper, however, focuses on the role played by the integrity of executives. The impact of

integrity cannot be empirically tested unless a proper proxy is available. We therefore employ

earnings management to measure the integrity level of executives by following the intuition

of Jensen et al. (2004), and hypothesize that executives with lower levels of integrity are

inclined to pursue self-interests at the expense of other shareholders. The results confirm our

prediction that lack of integrity is indeed an antecedent of fraud because it induces unethical

decisions, leading to fraudulent behaviors.

We also show the important role auditors play in the deterioration process from poor integrity

to serious malpractice. Auditors are found to be capable of disciplining executives with lower

levels of integrity by using nonstandard audit opinions, which are issued when problems are

detected in financial statements. We show that the occurrence of nonstandard audit opinions

is negatively related to integrity level and indicators of good external governance, such as a

larger analyst following and mutual fund ownership. More importantly, nonstandard opinions

are able to draw attention from investors, the board, the regulatory commission, and the

media to integrity problems, keeping them vigilant and also sending a warning to the

executives. Therefore, when executives of low levels of integrity incur nonstandard audit

opinions, they are less likely to be become fraudsters in the subsequent period.

21

The results have important practical implications. First, they show that executives with good

integrity are essential to implementing governance systems and suggest that integrity needs to

become one of the critical considerations in executive appointments. Executives with lower

levels of integrity tend to divert their abilities and efforts by pursing self-interests and

destroying firm value. Second, in the assessment of executive performance and executive

remuneration decisions, the measurements need to be adjusted for integrity. For example, a

threshold may be imposed on earnings management. Finally, the detection and reporting of

fraud has long been regarded as the ex post responsibility of auditors. However, we provide

evidence that auditors are also able to deter fraud by taking ex ante initiatives. Auditors are

able to spot an integrity problem before it develops into fraudulent activities, and their

opinion can effectively discipline executives, preventing serious misconduct.

22

Appendix. Variable Definitions

Fraud

A dummy variable equal to one if the firm is subject to regulatory enforcement against

disclosed fraud, and zero otherwise.

EM

Earnings management is the ratio of non-operating income relative to revenue. It is used to

proxy for the lack of integrity of executives.

DEM

A dummy of earnings management. It is equal to one if the ratio of non-operating income

relative to revenue is above the median value of yearly observations, and zero otherwise.

ANS

A dummy variable equal to one if a nonstandard opinion report is issued by the auditor,

which indicates (potential) problems with a financial statement, and zero otherwise.

The following control variables are lagged for one year to resolve the causality problem:

Size The natural logarithm of market capitalization.

PB Price-to-book ratio.

STPT

A dummy variable equal to one if a listed firm experiences two or more consecutive years

of loss, and zero otherwise.

Analyst The natural logarithm of one plus the number of analysts following the firm.

Fund The number of shares held by mutual funds relative to the total number of shares.

RTT The number of restricted shares relative to the total number of shares.

OwnCon The Herfindahl index of the top 10 largest firm blockholders.

Duality

A dummy variable equal to one if the CEO holds the position of board chair, and zero

otherwise.

Dmeet

A dummy variable equal to one if the number of board meetings is above the median value

of yearly observations, and zero otherwise.

Dbsize

A dummy variable equal to one if the number of board members is above the median value

of yearly observations, and zero otherwise.

Drind

A dummy variable equal to one if the ratio of independent directors is above the median

value of yearly observations, and zero otherwise.

Dssize

A dummy variable equal to one if the number of supervisory board members is above the

median value of yearly observations, and zero otherwise.

The following industry and regional dummies are also incorporated in our empirical analyses.

The industry dummies are constructed based on the first two digits of the GICS codes.

The regional dummies are constructed according to Firth et al. (2006), who group firms into four

different regions by level of economic development: (1) Shanghai and Shenzhen, (2) the more

developed areas, including the open cities and provinces along the coast, (3) the inland provinces, and

(4) the least developed area, in the northwestern part of the country.

23

References

Beasley, M., J. V. Carcello, D. R. Hermanson, and P. D. Lapides, 2000, Fraudulent financial

reporting: consideration of industry traits and corporate governance mechanisms. Accounting

Horizons, 14: 441–452.

Beasley, M. S., 1996, An empirical analysis of the relation between the board of director

composition and financial statement fraud. Accounting Review, 71, 443–455.

Bertrand, M., P. Mehta, and S. Mullainathan, 2002, Ferreting out tunneling: An application to

Indian business groups. Quarterly Journal of Economics, 117, 121–148.

Bradford, W., 2007, Because that's where the money is: Toward a theory and strategy of

corporate legal compliance. SSRN working paper, http://ssrn.com/abstract=955428.

Cao, J., M. Lemmon, X. Pan, M. Qian, and G. Tian, 2011, Gary Gang, political promotion,

CEO incentives, and the relationship between pay and performance. SSRN working paper,

http://ssrn.com/abstract=1914033.

Chen, G., M. Firth, D. Gao, and O. Rui, 2006, Ownership structure, corporate governance,

and fraud: Evidence from China. Journal of Corporate Finance, 12, 424–448.

Chen, G., M. Firth, D. Gao, and O. M. Rui, 2005, Is China’s securities regulatory agency a

toothless tiger? Evidence form enforcement actions. Journal of Accounting and Public

Policy, 24, 451–488.

Chen, K. C. W., and H. Yuan, 2004, Earnings management and capital resource allocation:

Evidence from China’s accounting-based regulation of rights issue. The Accounting Review,

79, 645–665.

24

Chung, R., M. Firth, and J. Kim, 2002, Institutional monitoring and opportunistic earnings

management. Journal of Corporate Finance 8, 29–48.

Cohen, J., G. Krishnamoorthy, and A. Wright, 2010, Corporate governance in the post-

Sarbanes–Oxley era: Auditors’ experience. Contemporary Accounting Research, 27, 751–

786.

Cohen, J., L. Gaynor, G. Krishnamoorthy, and A. Wright, 2011, The impact of auditor

judgments of CEO influence on audit committee independence and management incentives.

Auditing: A Journal of Practice & Theory, 30, 129–147.

Cumming, D., W. Hou, and E. Lee, 2011, Does the external monitoring effect of financial

analysts deter corporate fraud in China? SSRN working paper,

http://ssrn.com/abstract=1874114.

Dechow, P. M., R. G. Sloan, and A. P. Sweeney, 1996, Causes and consequences of earnings

manipulation: An analysis of firms subject to enforcement actions by the SEC. Contemporary

Accounting Research, 13, 1–21.

DeAngelo, L. E., 1981. Auditor size and audit quality. Journal of Accounting and Economics.

3: 183-199

DeFond, M., T. J. Wong, and S. Li, 2000, The impact of improved auditor independence on

audit market concentration in China. Journal of Accounting and Economics, 28, 269–305.

Denis. D., P. Hanouna, and A. Sarin, 2006, Is there a dark side to incentive compensation?

Journal of Corporate Finance, 12, 467–488.

Ding, S., C. Jia, Y. Li, and Z. Wu: 2010, Reactivity and passivity after enforcement actions:

Better late than never. Journal of Business Ethics, 95, 337–359.

25

Ding, Y., H. Zhang, and J. Zhang, 2007, Private vs state ownership and earnings

management: Evidence from Chinese listed companies. Corporate Governance: An

International Review, 2, 223–238.

Dyck, A., A. Morse, and L. Zingales, 2010, Who blows the whistle on corporate fraud?

Journal of Finance, 65, 2015–2444.

Erickson, M., M. Hanlon, and E. L. Maydew, 2006, Is there a link between executive equity

incentives and accounting fraud? Journal of Accounting Research, 44, 113–143.

Firth, M., P. Fung, and O. Rui, 2006, Corporate performance and CEO compensation in

China. Journal of Corporate Finance, 12, 693–714.

Firth, M., P. Fung, and O. Rui, 2007, Ownership, two-tier board structure, and the

informativeness of earnings – Evidence from China. Journal of Accounting and Public

Policy, 26, 463–496.

Firth, M., P. Mo, and R. Wong, 2005, Financial statement frauds and auditor sanctions: An

analysis of enforcement actions in China. Journal of Business Ethics, 62, 367–381.

Firth, M., P. Mo, and R. Wong, 2011b, Auditors' organizational form, legal liability, and

reporting conservatism: Evidence from China. Contemporary Accounting Research.

forthcoming.

Firth, M., O. Rui, and W. Wu, 2011a, Cooking the books: Recipes and costs of falsified

financial statements in China. Journal of Corporate Finance, 17, 371–390.

Fuller, J., and M. Jensen, 2002, What's a director to do? Harvard NOM Research Paper No.

02-38.

26

Jensen, M., K. Murphy, and E. Wruck, 2004, Remuneration: Where we've been, how we got

to here, what are the problems, and how to fix them. Harvard NOM Working Paper No. 04-

28.

Jensen, M., and R. Walkling, 2010, Pioneers in finance: An interview with Michael C. Jensen

– Part 2. Harvard Business School NOM Unit Working Paper.

Jia, C., S. Ding, Y. Li, and Z. Wu, 2009, Fraud, enforcement action, and the role of corporate

governance: Evidence from China. Journal of Business Ethics, 90, 561–576.

Jian, M., and T. J. Wong, 2004, Earnings management and tunneling through related party

transactions: Evidence from Chinese corporate groups. Working paper, Nanyang

Technological University and Hong Kong University of Science and Technology.

Joe, J., 2003, Why press coverage of a client influences the audit opinion. Journal of

Accounting Research, 41, 109–133.

Hou, W., J. Kuo, and E. Lee, 2011a, State ownership and stock price informativeness:

Evidence of the Split Share Structure Reform in China. EFMA conference paper.

Hou, W., E. Lee, and K. Stathopoulos, 2011b, The role of mutual funds in deterring corporate

fraud in China. SSRN working paper, http://ssrn.com/abstract=1978573.

Hou, W., and G. Moore, 2010, Player and referee roles held jointly: The effect of state

ownership on China’s regulatory enforcement against fraud. Journal of Business Ethics, 95,

317–335.

Kim, J., R. Chung, and M. Firth, 2011, Auditor conservatism, asymmetric monitoring, and

earnings management. Contemporary Accounting Research, 20, 323–359.

27

Kizirian, T. G., B. W. Mayhew, and J. Sneathen, 2005, The impact of management integrity

on audit planning and evidence. Auditing: A Journal of Practice & Theory, 24(2), 49–67.

Krishnan, J., and R. G. Stephens, 1995, Evidence on opinion shopping from audit opinion

conservatism. Journal of Accounting and Public Policy 14, 179–201.

Krishnan, J., J. Krishnan, and R. Stephens, 1996. The simultaneous relation between auditor

and audit opinion: An empirical analysis. Accounting and Business Research, 26, 224-236

Nowak, M. J., and M. McCabe, 2003, Information costs and the role of the independent

corporate director. Corporate Governance: An International Review, 11, 300–307.

Rost, K., 2007, CEO pay and appointments: Putting integrity back into hiring decisions.

SSRN working paper, http://ssrn.com/abstract=1019743.

Skaife, H. A., D. Veenman, and D. Wangerin, 2011, Internal control over financial reporting

and managerial rent extraction: Evidence from the profitability of insider trading. SSRN

working paper, http://ssrn.com/abstract=1928557.

Tang, Q., C. W. Chow, and A. Lau, 1999, Auditing of state-owned enterprises in China:

Practices, problems and further development. International Journal of Accounting 34, 173–

187.

Uzun, H., S. H. Szewczyk, and R. Varma, 2004, Board composition and corporate fraud.

Financial Analyst Journal, 60, 33–43.

Xiang, B., 1998, Institutional factors influencing China’s accounting reforms and standards.

Accounting Horizons, 12, 105–119.

Xiao, J. Z., Y. Zhang, and Z. Xie, 2000, The making of independent auditing standards in

China. Accounting Horizons, 14, 69–89.

28

Yang, L., A. Dunk, A. Kilgore, Q. Tang, and Z. J. Lin, 2003, Auditor independence issues in

China. Managerial Finance, 29, 57–64.

Yang L., Q. Tang, A. Kilgore, and Y. H. Jiang, 2001, Auditor–government associations and

auditor independence in China. British Accounting Review, 33, 175–189.

29

Figure 1

This figure shows the number and ratio of nonstandard opinion of auditors from 2001 to 2008 in the

Chinese stock market. Nonstandard opinions indicate (potential) problems with financial statements.

0%

2%

4%

6%

8%

10%

12%

14%

0

20

40

60

80

100

120

140

160

2001 2002 2003 2004 2005 2006 2007 2008

Nonstandard auditor opinions

No. of Non-standard Opinion Ratio of Non-standard Opinion

30

Table 1 This table presents the descriptive statistics for the total sample and firm subsamples with and without nonstandard auditor opinions. The variable ANS equals

one if a nonstandard opinion report is issued by the auditor, which indicates (potential) problems with a financial statement, and zero otherwise; Fraud is

equal to one if the firm is subject to regulatory enforcement against disclosed fraud, and zero otherwise; EM is the ratio of non-operating income relative to

revenue; and DEM is a dummy variable equal to one if EM is above the median value of yearly observations, and zero otherwise. Other variables are defined

in the Appendix. The sample period covers 2001 to 2008.

All firms Firms with ANS = 0

Firms with ANS = 1 Diff. in Mean Test

Median Mean Std.Dev. Median Mean Std.Dev. Median Mean Std.Dev. Difference t-Value

ANS 0.0000 0.0931 0.2907 N/A N/A Fraud 0.0000 0.0393 0.1942 0.0000 0.0216 0.1453 0.0000 0.2112 0.4084 -0.1896 -30.06***

EM 0.0032 0.0193 0.0655 0.0030 0.0153 0.0501 0.0054 0.0586 0.1415 -0.0433 -19.87***

DEM 0.0000 0.4998 0.5000 0.0000 0.4834 0.4998 1.0000 0.6587 0.4744 -0.1753 -10.40***

Size 20.4613 20.5797 0.9999 20.5170 20.6397 0.9931 19.9633 19.9955 0.8694 0.6443 19.36***

PB 2.6209 3.7035 3.9615 2.6144 3.6124 3.3787 2.7808 4.5902 7.5187 -0.9778 -7.31***

STPT 0.0000 0.0802 0.2716 0.0000 0.0470 0.2117 0.0000 0.4027 0.4907 -0.3557 -41.81***

Analyst 0.0000 0.6386 0.9133 0.0000 0.6908 0.9355 0.0000 0.1310 0.3887 0.5597 18.38***

Fund 0.0009 0.0275 0.0626 0.0016 0.0300 0.0650 0.0000 0.0033 0.0187 0.0267 12.68***

RTT 0.5702 0.5397 0.1651 0.5706 0.5395 0.1661 0.5692 0.5420 0.1550 -0.0025 -0.46

Owncon 0.1708 0.2062 0.1367 0.1773 0.2104 0.1374 0.1251 0.1659 0.1221 0.0445 9.66***

Duality 0.0000 0.0098 0.0985 0.0000 0.0091 0.0949 0.0000 0.0166 0.1280 -0.0076 -2.27**

Dmeet 1.0000 0.5776 0.4940 1.0000 0.5838 0.4930 1.0000 0.5172 0.5000 0.0666 3.98***

Dbsize 0.0000 0.3726 0.4835 0.0000 0.3751 0.4842 0.0000 0.3486 0.4768 0.0265 1.62

Drind 1.0000 0.7754 0.4173 1.0000 0.7830 0.4122 1.0000 0.7014 0.4579 0.0817 5.79***

Dssize 1.0000 0.8991 0.3012 1.0000 0.8997 0.3004 1.0000 0.8928 0.3095 0.0069 0.68

Obs. 10,317 9,356 961

31

Table 2

This table presents the correlation matrix of the variables. The variable Fraud is equal to one if the firm is subject to regulatory enforcement against disclosed

fraud, and zero otherwise; EM is the ratio of non-operating income relative to revenue; DEM is a dummy variable equal to one if EM is above the median

value of yearly observations, and zero otherwise; and ANS equals one if a nonstandard opinion report is issued by the auditor, which indicates (potential)

problems with a financial statement, and zero otherwise. The variable ANS is also interacted with EM and DEM, respectively. The other variables are defined

in the Appendix. The sample period covers 2001 to 2008. The superscript * denote the 1% level of significance.

Fraud EM DEM ANS Size PB STPT Analyst Fund RTT OwnCon Duality Dmeet Dbsize

Fraud 1

EM 0.0687*

DEM 0.0535* 0.2752* 1

ANS 0.2838* 0.1920* 0.1019* 1

Size -0.0990* -0.0603* -0.0821* -0.1873* 1

PB 0.0266* 0.1136* 0.0442* 0.0717* 0.1636* 1

STPT 0.1131* 0.3329* 0.1183* 0.3807* -0.2203* 0.1845* 1

Analyst -0.1048* -0.0840* -0.0850* -0.1781* 0.5078* -0.004 -0.1696* 1

Fund -0.0700* -0.0619* -0.0808* -0.1238* 0.4860* 0.1036* -0.1215* 0.6094* 1

RTT -0.0022 -0.0474* -0.0257* 0.0045 -0.3319* 0.0223 -0.0094 -0.1297* -0.1840* 1

OwnCon -0.0692* -0.0788* -0.0817* -0.0947* 0.0244 -0.0484* -0.1016* 0.0490* -0.0443* 0.5724* 1

Duality 0.0002 0.0336* 0.0109 0.0223 -0.0012 0.0148 0.0105 -0.0208 -0.0026 -0.0316* -0.0351* 1

Dmeet -0.0515* -0.0402* -0.0620* -0.0392* -0.0482* -0.0334* -0.0525* -0.0246 -0.0222 0.0459* 0.0391* -0.0126 1

Dbsize -0.0082 -0.0512* -0.0165 -0.0159 0.0367* -0.0400* -0.0555* 0.0648* 0.0052 0.1020* 0.0239 0.0048 0.0510* 1

Drind -0.0252 0.0143 0.0004 -0.0569* 0.0445* -0.0596* -0.0019 0.2142* 0.1199* -0.1517* -0.0882* 0.004 -0.0445* -0.1272*

Dssize -0.0135 -0.0437* -0.0243 -0.0067 0.0380* -0.0124 -0.0410* -0.0429* -0.0047 -0.0241 0.0351* -0.0223 -0.0194 -0.0067

32

Table 3

The two panels of this table present the empirical results of the regression model

k

k kk ControltyDisintegriFraud1 110

where the one-year lead dependent dummy variable Fraud is equal to one if the firm is subject to

regulatory enforcement against disclosed fraud, and zero otherwise; Disintegrity is proxied by

following two earning management measures; EM is the ratio of non-operating income relative to

revenue; and DEM is a dummy variable equal to one if EM is above the median value of yearly

observations, and zero otherwise. The other control variables are defined in the Appendix. The sample

period covers 2001 to 2008. All t-statistics are reported and adjusted for heteroskedasticity. The

superscripts *,

**, and

*** denote the 10%, 5%, and 1% levels of significance, respectively.

Panel A

Regression I Regression II Regression III

EM 2.2355 4.83 *** 0.9538 1.8 * 1.0015 1.86 *

Size -0.6031 -9.12 *** -0.3341 -3.93 *** -0.3300 -3.8 ***

PB 0.0294 2.91 *** 0.0096 0.97 0.0114 1.14

STPT 0.5380 3.43 *** 0.5184 3.25 ***

Analyst -0.7180 -5.89 *** -0.7370 -6.04 ***

Fund -4.2348 -1.37 -4.1038 -1.34

RTT 0.0360 0.09 -0.0102 -0.02

OwnCon -2.7719 -5.26 *** -2.7673 -4.99 ***

Duality -0.3086 -0.6 -0.1903 -0.37

Dmeet -0.5113 -4.91 *** -0.5407 -5.17 ***

Dbsize 0.0328 0.3 -0.0051 -0.05

Drind -0.1765 -1.47 -0.1588 -1.31

Dssize -0.1907 -1.2 -0.1643 -1.01

Constant 8.8909 6.64 *** 4.7972 2.6 *** 5.3383 2.76 ***

Industry N N Y

Region N N Y

Pseudo-R2 0.0424 0.0914 0.1039

Obs. 10,317 10,317 10,317

33

Panel B

Regression I Regression II Regression III

DEM 0.4566 4.25 *** 0.3050 2.81 *** 0.3012 2.72 ***

Size -0.6027 -9 *** -0.3241 -3.83 *** -0.3188 -3.69 ***

PB 0.0341 3.4 *** 0.0098 0.99

0.0117 1.16 STPT 0.5785 3.97 *** 0.5629 3.8 ***

Analyst -0.7156 -5.89 *** -0.7330 -6.03 ***

Fund -4.1589 -1.35

-4.0782 -1.34 RTT 0.0533 0.13

0.0042 0.01

OwnCon -2.7610 -5.23 *** -2.7405 -4.93 ***

Duality -0.3039 -0.59

-0.1858 -0.36 Dmeet -0.4966 -4.77 *** -0.5241 -5.01 ***

Dbsize 0.0277 0.25

-0.0110 -0.1 Drind -0.1751 -1.46

-0.1556 -1.29

Dssize -0.1969 -1.24

-0.1688 -1.04 Constant 8.6666 6.34 *** 4.4279 2.39 ** 4.9240 2.54 **

Industry

N

N

Y Region

N

N

Y

Pseudo-R2

0.0427

0.0929

0.1051 Obs.

10,317

10,317

10,317

34

Table 4 This table presents the empirical results of the determinants of auditors’ nonstandard opinions, which indicate problems with financial statements.

The dependent dummy variable ANS equals one if a nonstandard opinion report is issued by the auditor, which indicates (potential) problems with

a financial statement, and zero otherwise; EM is the ratio of non-operating income relative to revenue; and DEM is a dummy variable equal to one

if EM is above the median value of yearly observations, and zero otherwise. The other control variables are defined in the Appendix. The sample

period covers 2001 to 2008. All t-statistics are reported and adjusted for heteroskedasticity. The superscripts *,

**, and

*** denote the 10%, 5%, and

1% levels of significance, respectively.

Regression I Regression II Regression III Regression IV

ANSt+1 ANSt ANSt+1 ANSt

EM 1.6234 4.2 *** 1.6135 4.16 ***

DEM 0.3805 4.82 *** 0.3184 4.18 ***

Size -0.3650 -6.11 *** -0.3927 -6.57 *** -0.3514 -5.91 *** -0.3800 -6.38 ***

PB 0.0042 0.5

0.0011 0.13

0.0048 0.57

0.0019 0.22 STPT 1.8762 18.72 *** 1.4023 14 *** 1.9492 20.27 *** 1.4862 15.51 ***

Analyst -0.7060 -7.36 *** -0.7639 -8.24 *** -0.7105 -7.41 *** -0.7662 -8.28 ***

Fund -10.6084 -2.5 ** -7.3983 -2.32 ** -10.5342 -2.48 ** -7.3721 -2.31 **

RTT -0.1496 -0.47

-0.5695 -1.88 * -0.1507 -0.48

-0.5768 -1.89 *

OwnCon -1.8636 -4.75 *** -2.1902 -5.43 *** -1.8382 -4.68 *** -2.1696 -5.35 ***

Duality 0.3241 0.94

-0.3155 -0.84

0.3476 1.01

-0.2822 -0.76 Dmeet -0.1854 -2.43 ** -0.1807 -2.43 ** -0.1655 -2.17 ** -0.1635 -2.2 **

Dbsize 0.0430 0.53

0.1908 2.45 ** 0.0364 0.45

0.1827 2.36 **

Drind -0.3931 -4.54 *** -0.3716 -4.4 *** -0.3889 -4.49 ** -0.3664 -4.34 ***

Dssize 0.1182 0.98

0.0791 0.65

0.1105 0.92

0.0687 0.57 Constant 6.0982 4.59 *** 9.6059 6.99 *** 5.5973 4.23 *** 9.2265 6.73 ***

Industry

Y

Y

Y

Y

Region

Y

Y

Y

Y

Pseudo-R2

0.2165

0.1756

0.2174

0.1755

Obs.

10,317

10,303

10,317

10,303

35

Table 5

The two panels of this table present the empirical results of the regression model

k

k kk ControlAUDtyDisintegriAUDtyDisintegriFraud3 13210 .

where the one-year lead dependent dummy variable Fraud is equal to one if the firm is subject to

regulatory enforcement against disclosed fraud, and zero otherwise; Disintegrity is proxied by following

two earning management measures; EM is the ratio of non-operating income relative to revenue; DEM is

a dummy variable equal to one if EM is above the median value of yearly observations, and zero

otherwise; and ANS equals one if a nonstandard opinion report is issued by the auditor, which indicates

(potential) problems with a financial statement, and zero otherwise. The variable ANS is also interacted

with EM and DEM, respectively. The other control variables are defined in the Appendix. The sample

period covers 2001 to 2008. All t-statistics are reported and adjusted for heteroskedasticity. The

superscripts *,

**, and

*** denote the 10%, 5%, and 1% levels of significance, respectively.

Panel A

Regression I Regression II Regression III

EM 3.2992 4.71 *** 2.8999 3.8 *** 3.2097 4.18 ***

ANS 2.4218 20.1 *** 2.2543 16.83 *** 2.3023 16.78 ***

EM. ANS -4.0126 -4.36 *** -3.4746 -3.63 *** -3.9042 -4.04 ***

Size -0.3137 -4.99 *** -0.2003 -2.4 ** -0.1787 -2.09 **

PB 0.0070 0.79

0.0035 0.38

0.0055 0.59 STPT -0.3112 -1.83 * -0.3353 -1.94 *

Analyst -0.5904 -4.68 *** -0.5918 -4.71 ***

Fund -2.6351 -1

-2.7656 -1.05 RTT 0.0490 0.11

-0.0214 -0.05

OwnCon -2.1577 -4.06 *** -2.1549 -3.86 ***

Duality -0.4818 -0.94

-0.3317 -0.64 Dmeet -0.4441 -4.11 *** -0.4721 -4.35 ***

Dbsize 0.0080 0.07

-0.0272 -0.24 Drind -0.0605 -0.48

-0.0381 -0.3

Dssize -0.2223 -1.34

-0.2257 -1.32 Constant 2.5190 1.96 ** 1.4006 0.77

1.5031 0.77

Industry

N

N

Y Region

N

N

Y

Pseudo-R2

0.1525

0.1776

0.1903 Obs.

10,317

10,303

10,317

36

Panel B

Regression I Regression II Regression III

DEM 0.4595 3.14 *** 0.3636 2.48 ** 0.3766 2.55 **

ANS 2.5852 14.25 *** 2.3664 12.37 *** 2.4188 12.47 ***

DEM*ANS -0.4994 -2.25 ** -0.3869 -1.7 * -0.4162 -1.8 *

Size -0.3121 -4.92 *** -0.1998 -2.4 ** -0.1804 -2.12 **

PB 0.0085 0.95

0.0050 0.55

0.0075 0.81 STPT

-0.2994 -1.8 * -0.3255 -1.92 *

Analyst

-0.5891 -4.71 *** -0.5877 -4.71 ***

Fund

-2.7468 -1.03

-2.8914 -1.09 RTT

0.0320 0.07

-0.0459 -0.1

OwnCon

-2.1310 -4.03 *** -2.1083 -3.8 ***

Duality

-0.4779 -0.93

-0.3329 -0.64 Dmeet

-0.4457 -4.14 *** -0.4707 -4.34 ***

Dbsize

0.0001 0

-0.0377 -0.33 Drind

-0.0663 -0.53

-0.0446 -0.35

Dssize

-0.2262 -1.37

-0.2259 -1.32 Constant 2.3032 1.76

1.2666 0.69

1.3816 0.71

Industry

N

N

Y Region

N

N

Y

Pseudo-R2

0.1509

0.1759

0.1881 Obs.

10,317

10,303

10,317