a r t i c l e
i n f o
a b s t r a c t
We use the unique economic, legal, and political landscape of China to examine the impact of auditors on the incidence of accounting fraud. In particular, we examine whether large audit rms reduce the incidence of nancial
statement fraud in China, an emerging market in which auditors face strong government sanctions but low litigation risk associated with audit failures. We nd that companies audited by large audit rms are less likely to
commit nancial statement fraud. This effect is stronger for regulated industries and for revenue-related frauds.
Our results are robust to considering the severity of fraud, excluding rms cross-listing in other jurisdictions,
using alternative measures of fraud, accounting for the self-selection of auditors, and controlling for other corporate governance mechanisms. Our results highlight the role of government sanctions in assuring audit quality and
have important practical implications to help international audit rms and businesses more generally successfully compete in China.
2014 Elsevier Inc. All rights reserved.
1. Introduction
China is now the world's second largest economy and growing at a
faster rate than most western countries. The size and growth of
China's economy present many business opportunities. However,
understanding China's unique economic, legal, and political landscape
is imperative for any rm to successfully compete in China. One of the
prominent differences from the west is the direct role the government
plays in detecting and dealing with accounting fraud. We focus on the
relation between auditor size and the incidence and consequences of
nancial statement fraud in China, an environment where auditors
face negligible litigation risk but severe government sanctions for
audit failures.
The goal of an independent audit is to provide reasonable assurance
that nancial statements are free from fraud or material error. Auditing
is thus essential in ensuring the proper functioning of the nancial
reporting system. Some studies nd that large audit rms provide
We thank three anonymous reviewers, Robert Grosse (the Guest Editor), Zhaoyang Gu,
Tom Omer, Roger Simnett, TJ Wong, Arch Woodside (The Editor in Chief), Xing Xiao, and participants at the George Mason University Competing in China Conference for their helpful
comments and suggestions. Song and Wang gratefully acknowledge research funding from
the National Natural Science Foundation of China (Project 70902004, Project 71372048,
and Project 71072116).
Corresponding author.
E-mail addresses: llisic@gmu.edu (L.L. Lisic), ssilveri@binghamton.edu (S.(D.) Silveri),
songyh@zju.edu.cn (Y. Song), wangk@sem.tsinghua.edu.cn (K. Wang).
http://dx.doi.org/10.1016/j.jbusres.2014.11.013
0148-2963/ 2014 Elsevier Inc. All rights reserved.
higher quality audits using data from countries where auditors face signicant litigation risk (Francis & Krishnan, 1999; Khurana & Raman,
2004; Lennox & Pittman, 2010). In contrast, other studies nd no
difference in audit quality between large and small audit rms in
countries with a relatively low level of legal protection for claimholders
(Hope & Langli, 2010; Jeong & Rho, 2004).
Using China as a setting, where auditors face negligible litigation risk
but harsh sanctions from government agencies for providing low
quality audits, we examine whether large audit rms reduce the
incidence of nancial statement fraud. On the one hand, litigation
against auditors in China is very rare. The court only accepts fraud
case allegations after the government has already sanctioned the
auditor for fraud (see Several Regulations about Fake Statement Litigation in Security Market (2003) enacted by the Supreme People's
Court). Although few cases have been brought to court, there has
never been a successful case of shareholder litigation against an auditor
due to low audit quality. Given this legal environment, it is not clear
whether auditors in general and large auditors in particular have
sufcient incentives to provide high quality audit services.
On the other hand, government agencies such as the Chinese
Security Regulatory Committee (CSRC) and the Ministry of Finance
sanction audit rms that fail to detect and report fraud in clients'
nancial statements (Firth, Mo, & Wong, 2005). The Chinese government is motivated to improve nancial statement quality in order to
attract foreign investment. Depending on the severity of the fraud, the
penalties for audit rms range from nes, to reprimands, to suspension
of audit work, to revoking licenses.
We seek to shed light on whether government sanctions can substitute for claimholder litigation in ensuring audit quality. Using all government enforcement actions against companies in China whose nancial
statements are challenged by government agencies for accounting
malfeasance, we identify 270 nancial statement frauds committed
during 19992005. We nd that the incidence of fraud declines over
our sample period. We also nd that, despite a very different legal
landscape, government sanctions in China appear to have a similar effect
on audit quality as litigation does in the U.S. Importantly, we show that
clients audited by large accounting rms are less likely to commit nancial statement fraud.
Certain industries are more heavily regulated and monitored by the
CSRC due to strategic considerations (Tian & Estrin, 2008; Wei, Xie, &
Zhang, 2005). These industries include energy, iron and steel, oil renery and petrochemicals, communications, and heavy machinery. Since
these industries potentially pose a higher risk of government sanctions
for nancial fraud, we expect and nd a more negative relation between
audit rm size and nancial statement fraud in these industries.
In China, revenue-related fraud is more consequential than assetrelated fraud because income performance is an important criterion
for initial public offerings, rights offerings and maintaining exchangetrading status (Aharony, Lee, & Wong, 2000; Chen & Yuan, 2004). In
addition, investors rely on accounting earnings to evaluate a company's
performance. Firth et al. (2005) nd that auditors are more likely to be
sanctioned when they fail to detect and report revenue-related fraud
compared with asset-related fraud. Accordingly, we partition the
sample into revenue-related and asset-related fraud. We nd that the
negative association between fraud and auditor size is stronger for
revenue-related fraud than for asset-related fraud.
Some rms in our sample are cross-listed in other countries and
regions such as the U.S. or Hong Kong, or have B-shares. As a result,
such rms are potentially subject to both government sanctions and
litigation risk. In order to provide a cleaner sample of rms that are
subject to government sanctions only, we drop these cross-listed rms
and nd that our results remain robust.
Our results are also robust to considering the severity of fraud as
measured by the punishment imposed after the fraud is discovered.
Our inference regarding the relation between auditor size and the
incidence of fraud remains robust to accounting for the self-selection
of auditors and controlling for corporate governance mechanisms that
potentially reduce the incidence of fraud. Finally, we examine alternative measures of fraud by investigating high abnormal accruals and
nd similar results.
Our study contributes to the academic literature on three levels and
has important implications for rms looking to compete in China. First,
our study furthers our understanding of China's unique economic environment. As a developing country with a very different legal landscape
to the U.S., China presents a low litigation risk environment for businesses including audit rms. However, the Chinese government is
motivated to improve nancial statement quality in order to attract
foreign investment. For example, the former Chinese Premier Zhu
Rongji set No Fictitious Records as a motto for the Shanghai National
Accounting Institute at its Inauguration Ceremony in 2001. Thus, the
government uses an alternative mechanism to litigation, namely government sanctions, to ensure high quality audits. Understanding this
feature is important for international audit rms who intend to successfully compete in China. For example, if an audit rm draws from its
experience in the U.S. where the threat of litigation is a key driver
of audit quality, the rm might erroneously conclude that it can get
away with lower quality audits in China absent such litigation risk.
However, our ndings suggest that government sanctions in China
substitute for litigation risk to ensure audit quality.
Second, our paper also has implications for client companies
successfully competing in China's capital markets. We document that
rms using larger audit rms are less likely to be sanctioned by the
Chinese government. The accounting literature generally suggests that
1187
larger auditors provide higher quality audits and this brings economic
benets to client rms in the capital markets. For example, clients of
larger auditors exhibit higher earnings quality (Teoh & Wong, 1993)
and lower underpricing when they undertake an initial public offering
(Hogan, 1997). In addition, rms switching from small auditors to
large auditors tend to experience a positive stock market reaction
(Knechel, Naiker, & Pacheco, 2007). Thus, a key implication of our
results is that hiring a large audit rm in China ensures higher audit
quality (i.e., lower fraud rate), which translates into economic benets
for client rms in the capital markets. Hiring a large audit rm in
China helps client rms successfully compete in China.
Third, we add to the audit literature by suggesting that government
discipline potentially serves as an alternative mechanism to litigation to
ensure high quality audits. As Allen, Qian, and Zhang (2011) argue, the
alternative mechanisms found in many fast-growing economies can be
superior to using the law as the basis for nance and commerce. Our
ndings thus have potential policy implications for other countries
with a less litigious environment. Public enforcement (government
sanctions) may be able to substitute for private enforcement (class
action law suits) to ensure high quality audits in these countries.
Our paper proceeds as follows. In Section 2, we discuss the institutional background of the audit market in China and review the relevant
literature. In Section 3, we develop our main hypothesis and specify
the research design. Section 4 presents the sample selection process
and the descriptive statistics. Section 5 reports the empirical results
and additional analyses and Section 6 concludes.
2. Institutional background and literature review
Since the 1978 economic reforms, interest in business practices in
China has grown substantially. China's fast growing economy presents
numerous opportunities for rms in China as well as for foreign rms
considering entering China's capital markets. The subsequent rapid
development of China's capital market has created a demand for high
quality external audits. Chinese regulators have also taken steps to
improve audit quality in order to attract foreign investment and to
restructure state-owned enterprises. For example, the Chinese Institute
of Certied Public Accountants was established in 1988 to regulate
Certied Public Accountants (CPAs). The CPA Act, which mandates
that auditors be indicted for ctitious audit reports, was enacted in
1993. In 1995, the Ministry of Finance adopted a new set of auditing
standards that are closely modeled after the International Auditing
Standards. The China Securities Regulatory Commission (CSRC) and
the Ministry of Finance introduced the Audit Firm Disafliation program
in 1997 to sever ties between CPA rms and government agencies. The
CSRC has also issued several regulations since 2000 to encourage the
mergers of small- and medium-sized audit rms into larger rms.
Despite signicant regulatory reforms, however, the role of the auditor in assuring accounting information quality in China's capital markets
is still unclear. The major concern is derived from the legal landscape in
China. Litigation risk is a major factor ensuring high quality audits in
developed countries such as the U.S. In contrast, suing auditors in
China is rare and to date there has not been a successful case of
shareholder litigation against auditors.
Although litigation risk is negligible, auditors who provide low
quality audit services in China do bear other risks, namely government
penalties. Both the CSRC and the Ministry of Finance have authority to
monitor and sanction audit rms that fail to detect and report nancial
statement fraud such as misreporting income, misreporting assets and
liabilities and facilitating ctitious transactions. Chapter 10 of The Security Law delegates the CSRC as the main regulator of security markets in
China, which has authority to investigate and sanction listed companies
and market intermediates (including auditors) involved in securities
fraud and malpractice. The CSRC regularly reviews and randomly
inspects listed companies, especially when they receive complaints
from investors or the media. Once a nancial statement fraud is
1188
nancial statement fraud raises a larger concern to regulatory authorities due to its substantial negative impact on the development of
China's capital market. Regulators in China believe that auditors are
responsible for detecting and reporting fraud and therefore should be
heavily sanctioned for their failures (Firth et al., 2005). Auditors'
concerns about regulatory sanctions can dominate their economic
incentives (Chen, Sun, & Wu, 2010). Large auditors in particular are
more prominent and therefore more likely to be monitored by the
CSRC and the Ministry of Finance. They also have more to lose once
their licenses are revoked due to their larger client base (DeAngelo,
1981). In addition, the Ministry of Finance requires peer reviews as
part of its annual inspection of auditors. Large auditors draw substantially more competition from their peer rms and therefore, ceteris paribus,
are more likely to receive an unfavorable peer review. Consistent with
these arguments, Zheng and Xu (2011) nd that CSRC sanctions have
a larger negative impact on larger accounting rms ex-post. In particular, larger accounting rms suffer a larger loss of market share and a
greater reduction in audit fees after they are sanctioned by the CSRC.
Therefore, ex-ante, we expect larger accounting rms in China to have
stronger incentives to provide high quality audits relative to smaller
accounting rms. Our main hypothesis appears in its alternative form.
H1. Ceteris paribus, companies audited by larger accounting rms are
less likely to engage in nancial statement fraud.
3.2. Model specications
We use the following probit model to examine H1:
Fraudi;t 0 1 Ln AuditorSizei;t 2 Auditor Tenurei;t
3 ROE Dummyi;t 4 StateOwnedi;t
5 Ln CompanyAgei;t 6 Ln CompanySizei;t
7 Leveragei;t 8 ROAi;t 9 Book toMarket i;t
10 Lossi;t1 i;t
2008). As such, we include a dummy variable ROE Dummyi,t that takes the
value one if rm i's ROE in year t is between 0 and 1%, 6 and 7% or 10 and
11%. Ding, Zhang, and Zhang (2007) and Chen, Chen, Lobo, and Wang
(2011) suggest that state-owned companies manage earnings less so we
include an indicator variable, State-Ownedi,t, which equals 1 for stateowned companies and 0 otherwise. We note that it is also possible that
the Chinese government might be less likely to sanction state-owned
companies, again predicting a negative sign on State-Ownedi,t. We follow
Lennox and Pittman (2010) to control for company age and size. Company
Agei,t is the number of years that the company is listed on a stock
exchange. We use total assets to measure Company Sizei,t. As debt holders
have incentive to monitor management and obtain high quality accounting information, it is more difcult for rms to manage earnings if they
have high leverage (Frankel, Johnson, & Nelson, 2002). On the other
hand, highly levered rms face larger expected costs of nancial distress.
As a result they may be more likely to misreport (Burns & Kedia, 2006)
or even commit fraud (Erickson, Hanlon, & Maydew, 2006). We thus
include Leveragei,t, dened as total debt divided by total assets for rm i
in year t, as a control variable but do not predict the sign.
To control for rm performance, we include Return on Assetsi,t (ROAi,t)
which is dened as net income divided by total assets for rm i in year t.
Some studies nd that rms with rapid growth are more likely to commit
fraud (e.g., Loebbecke, Eining, & Willingham, 1989) although other papers
fail to nd a signicant association between growth and misreporting/
fraud (Burns & Kedia, 2006; Erickson et al., 2006). We use the Book-toMarketi,t ratio, dened as the book value of equity divided by the market
value of equity for rm i in year t, to proxy for growth opportunities.
Since Chinese rms are marked as Special Treatment by stock exchanges
if they incur losses in two consecutive years, rms have strong incentives
to avoid a loss in the current year if they also incur a loss in the prior year.
As such, we include Lossi,t 1, a dummy variable that equals 1 if rm i
incurs a loss during year t 1 and 0 otherwise.
4. Sample and descriptive statistics
To identify rms that have committed accounting fraud, we search
the China Stock Market and Accounting Research Database (CSMAR)
for enforcement reports issued by government agencies including the
CSRC, the Ministry of Finance, and the Shanghai and Shenzhen stock
exchanges. On average, there is a four-year gap between the fraud commitment date and the government report date. Thus, we identify frauds
committed during the period 1999 to 2005 as our sample is based on
enforcement reports up to 2009.
CSMAR identies four types of fraudulent activities: nancial statement frauds, delayed or fake announcements, shareholders' violation
of fund provisions or other expropriation activities and illegal share
purchases or stock price manipulations. We focus on the rst type,
which is often perceived as the auditor's responsibility to detect and
report. We exclude nancial rms following Khurana and Raman
(2004) and Ettredge, Sun, Lee, and Anandarajan (2008). A total of 270
fraud-year observations during 1999 to 2005 are identied. As an
example of our classication scheme, Shenxin Taifeng Group Co.
received an enforcement report from the CSRC on October 9, 2007.
The report stated that the company had fabricated nancial numbers
in its 2003 annual report. Specically, the company inated inventory
by 20 million Yuan and inated prepaid accounts by 6.5 million Yuan.
Thus, we identify 2003 for this company as a fraud-year observation.
It is further dened as an asset-related fraud since the misstatement
affects only assets but not revenue (or income). If the accounting
fraud was committed in more than one year, we identify each year as
a fraud-year observation. Our results, however, are robust to examining
only the rst year of consecutive frauds.
Table 1 summarizes the sample distribution. Panel A breaks down
the number of frauds by year. The frequency of accounting frauds in
China is higher than the U.S. (see Lennox and Pittman (2010) for U.S.
data). For example, in 1999 the frequency of accounting frauds is 3.9%
1189
Table 1
Financial statement frauds by year and industry.
Panel A: Frauds by year
Years
Entire sample
Total
1999
2000
2001
2002
2003
2004
2005
7843
862
1000
1076
1146
1211
1273
1275
270 (3.44)
34 (3.94)
36 (3.60)
38 (3.53)
42 (3.66)
45 (3.72)
47 (3.69)
28 (2.20)
Agriculture
Exploring
Manufacturing
Utilities
Construction
Transportation
Technology
Commerce
Properties
Services
Media
Conglomerate
Entire sample
Frauds
Proportion (%)
180
96
4530
331
132
321
435
625
312
267
77
537
22
2
171
3
1
6
15
8
12
3
8
19
12.22
2.08
3.77
0.91
0.76
1.87
3.45
1.28
3.85
1.12
10.39
3.54
in China and 1.1% in the U.S. There is a decreasing trend in the frequency
of frauds over the years, from 3.9% in 1999 to 2.2% in 2005. This pattern
is partially due to more stringent government scrutiny on listed companies and auditors in recent years. Panel B of Table 1 reports the distribution of frauds across industries. The industry classication follows the
CSRC guidelines. The distribution of fraud is in general similar across
industries, except for the agriculture industry and the media industry.
These industries are more difcult to audit due to the nature of the
products (Ding, 2013; Si, 2012). This potentially explains why these
industries have the highest incidence of accounting fraud (12.22% and
10.39%, respectively).
Table 2 presents the ranking and the market share of the Top 10
auditors where market share is dened as the combined client total
assets of the audit rm divided by the combined total assets of all listed
companies. The total number of auditors on the list is more than ten
because the composition of the Top 10 auditors changes over time. In
terms of market share, large auditors dominate China's market. For
example, as of 2005 the Top 10 auditors as a group account for more
than 50% of the market. We dene a company as being audited by a
Top 10 auditor only if the auditor is among the largest 10 auditors in
the corresponding scal year.
Table 3 compares the number and the frequency of nancial statement fraud among companies audited by the Top 10 auditors versus
non-Top 10 auditors. For our sample period, the average frequency of
accounting fraud is 1.8% in companies audited by the Top 10 auditors
compared to 3.7% for non-Top 10 auditors. The difference is statistically
signicant at the 1% level (t = 3.43). When we break down the
sample by year, the difference is negative in each year and is signicant
in four out of seven years. This preliminary evidence supports our main
hypothesis that companies audited by Top 10 auditors are less likely to
commit accounting fraud.
Table 4 presents the descriptive statistics for the fraud versus the
non-fraud sub-samples. The number of observations is slightly reduced
due to missing nancial information for the ROA computation. Altogether, 8% of the fraud rm-years are audited by the Top 10 auditors while
15% of the non-fraud rm-years are audited by Top 10 auditors. Auditor
Size is also greater in the non-fraud sample. The differences in the means
and medians of Top 10 and Auditor Size between the fraud and nonfraud samples are both statistically signicant at the 1% level. These
1190
Table 2
Ranking and market share of top 10 auditors.
Anderson HuaQiang
BeiJing Jingdu
Deloitte Hu Jiang/HuaYong
E&Y DaHua
E&Y HuaMing
Grant Thornton ZhongHua
HeBei Huaan
KPMG HuaZhen
PWC DaHua/ZhongTian
ShangHai
ShangHai Lixin
ShenZhen ZhongTian
Shinewing
SiChuan JunHe
TianQin
ZhongTianQin
ZheJiang TianJian
1999
2000
2001
2002
2003
2004
2005
Ranking
(Share %)
Ranking
(Share %)
Ranking
(Share %)
Ranking
(Share %)
Ranking
(Share %)
Ranking
(Share %)
Ranking
(Share %)
12 (2.57)
8 (2.98)
34 (0.98)
1 (6.19)
15 (2.06)
3 (4.08)
6 (3.37)
2 (4.22)
19 (1.74)
4 (3.95)
9 (2.81)
5 (3.70)
22 (1.63)
10 (2.77)
7 (3.31)
8 (3.04)
5 (3.73)
46 (0.83)
2 (5.64)
3 (3.77)
7 (3.36)
12 (2.34)
9 (2.88)
32 (1.22)
6 (3.42)
4 (3.75)
2 (4.95)
6 (3.23)
32 (1.08)
3 (4.60)
5 (3.35)
8 (2.26)
13 (1.97)
1 (17.05)
9 (2.25)
12 (2.13)
4 (3.97)
6 (2.92)
14 (1.77)
3 (4.21)
5 (3.09)
8 (2.37)
10 (1.99)
1 (14.80)
2 (11.54)
13 (1.78)
4 (3.56)
7 (2.57)
11 (2.11)
4 (3.75)
5 (2.95)
8 (2.39)
10 (2.13)
1 (14.39)
2 (9.96)
14 (1.86)
3 (3.94)
7 (2.68)
10 (2.29)
5 (2.78)
4 (4.11)
11 (2.11)
9 (2.33)
1 (14.57)
2 (9.19)
15 (1.66)
3 (4.98)
7 (2.53)
5 (3.78)
8 (2.44)
3 (6.06)
9 (2.38)
11 (2.28)
1 (15.74)
2 (8.77)
19 (1.38)
4 (4.27)
19 (1.49)
13 (2.26)
10 (2.18)
17 (1.56)
7 (2.60)
19 (1.40)
6 (2.61)
21 (1.34)
6 (2.75)
32 (1.05)
6 (2.73)
35 (0.88)
1 (6.89)
10 (2.75)
7 (2.30)
9 (2.30)
9 (2.27)
8 (2.36)
10 (2.32)
11 (2.71)
The market share of each audit rm is dened as the total assets of the audit rm's client base divided by the combined total assets of all listed companies on the stock market.
results are consistent with our hypothesis that frauds are less likely to
be committed by companies audited by larger auditors.
Variables associated with nancial statement fraud at the 5% signicance level include Auditor Tenure, State-Owned, Company Size, Leverage,
ROA and the Book-to-Market ratio. For example, Auditor Tenure is on average 3.9 years in the fraud sub-sample, and 4.7 years in the non-fraud subsample, which supports the view that longer auditor tenure is associated
with a lower likelihood of nancial statement fraud. Firms that commit
fraud are generally smaller than non-fraud rms. The mean of Company
Size is 1.5 billion Yuan (median 1.0 billion Yuan) for the fraud sample
compared with 2.7 billion Yuan (median 1.3 billion Yuan) for the nonfraud sample. We also nd that fraud-sample rms have lower ROA,
which suggests that rms are more likely to manipulate accounting
numbers when their performance is relatively poor. In addition, the
Book-to-Market ratio is lower for fraud rms when compared to nonfraud rms. Our ndings are, in general, consistent with the results of
Lennox and Pittman (2010) who use U.S. data. Thus, despite a very different legal landscape, it appears that government sanctions in China have a
similar effect on audit quality as litigation does in the U.S.
Table 3
Comparison of nancial statement frauds in companies audited by top 10 versus non-top 10 auditors.
Years
Total sample
1999
2000
2001
2002
2003
2004
2005
Top 10 auditors
Non-top 10 auditors
Obs.
Frauds
Obs.
Frauds
1169
46
112
152
198
205
235
221
21
1
2
1
4
5
6
2
1.80
2.17
1.79
0.66
2.02
2.44
2.55
0.90
6674
816
888
924
948
1006
1038
1054
249
33
34
37
38
40
41
26
3.73
4.04
3.83
4.00
4.01
3.98
3.95
2.47
***, ** and * denote signicance at the 1%, 5% and 10% level, respectively.
t-Statistic
1.93
1.87
2.04
3.34
1.99
1.54
1.40
1.57
3.43***
4.64***
1.77*
1.73*
1.18
0.34
0.83
2.24**
1191
Table 4
Descriptive statistics for the fraud versus non-fraud samples.
Variable
Fraud sample
Top 10
Auditor Size (billion Yuan)
Auditor Tenure (years)
ROE Dummy
State-Owned
Company Age (years)
Company Size (billion Yuan)
Leverage
ROA
Book-to-Market
Losst 1
Non-fraud sample
Obs.
Mean
Median
Obs.
Mean
Median
270
270
270
270
270
270
270
270
270
270
270
0.08
44.04
3.85
0.84
0.68
6.14
1.53
0.51
0.02
0.34
0.20
0
32.90
3
1
1
6
1.02
0.49
0.02
0.29
0
7573
7573
7573
7573
7545
7573
7570
7569
7569
7546
7569
0.15
57.39
4.66
0.82
0.83
6.19
2.68
0.48
0.02
0.41
0.17
0
39.47
4
1
1
6
1.30
0.47
0.03
0.35
0
t-Statistic
z-Statistic
4.38***
4.01***
5.38***
1.06
5.35***
0.31
6.88***
2.29**
5.72***
3.94***
1.44
3.35***
4.29***
4.00***
1.00
6.60***
0.13
4.17***
2.11**
6.95***
3.76***
1.55*
Note. Variables used in this table are dened as follows: Top 10 is a dummy variable that equals 1 if the auditing rm is among the 10 largest in terms of the total assets of the auditing rm's
client base in a given year and 0 otherwise. Auditor Size is the combined total assets of the auditor's client base measured in billion Yuan. Auditor Tenure is the number of years that the
company has been audited by the same auditing rm. ROE Dummy takes the value one if the listed company's ROE is between 0 and 1%, 6 and 7% or 10 and 11% and zero otherwise.
State-Owned equals 1 for state-owned companies and 0 otherwise. Company Age is the number of years that the company is listed on a stock exchange. Company Size is the year-end
total assets of the company. Leverage is a company's total debt divided by total assets. ROA (Return On Assets) is a company's net income divided by total assets. Book-to-Market is the
ratio of a company's book value of equity to the market value of equity. Losst 1 equals 1 if the company had a loss in year t 1 and 0 otherwise. The t-statistic tests for whether
there is a difference in means across the two samples. The z-statistic tests for whether there is a difference in medians across the two samples. ***, ** and * denote signicance at the
1%, 5% and 10% level, respectively.
with the Top 10 variable as a proxy for auditor size. We thus report in
Table 6 the results using Auditor Size. For regulated industries, the coefcient on Auditor Size is signicantly negative (0.2701, t = 2.70),
suggesting that clients in regulated industries audited by a larger
auditor are less likely to commit nancial statement fraud. For unregulated industries, although the coefcient on Auditor Size is signicantly
negative (0.1196, t = 2.45), its economic signicance is less than
half of that for regulated industries. These results are consistent with
the effect of auditor size on nancial statement fraud being larger in
regulated industries due to more severe government monitoring.
5.3. Revenue- versus asset-related fraud
We classify frauds into two types; revenue-related frauds that
include revenue (or expenses) misstatement, and asset-related frauds
that only affect assets (or liabilities). In China, revenue-related fraud is
more consequential than asset-related fraud because income performance is an important criterion for initial public offerings, rights
offerings and maintaining exchange-trading status (Aharony et al.,
Table 6
Probit regression of the incidence of fraud on auditor size for regulated versus unregulated
industries.
Table 5
Probit regression of the incidence of fraud on auditor size.
Fraud
Top 10
Ln(Auditor Size)
Auditor Tenure
ROE Dummy
State-Owned
Ln(Company Age)
Ln(Company Size)
Leverage
ROA
Book-to-Market
Losst 1
Intercept
Year xed effects
Industry xed effects
N
Pseudo R2
2000; Chen & Yuan, 2004). For example, during the 1990s the CSRC
required a minimum ROE of no less than 10% for a company to apply
for a rights offering (at present, this ROE requirement is no less than
6% on average for three years). Consequently, Chinese listed companies
have strong incentives to manipulate income numbers in order to raise
capital and to maintain listing status (Aharony et al., 2000; Chen & Yuan,
2004; Jian & Wong, 2010). In addition, investors rely on accounting
earnings to evaluate a company's performance. Due to the protability
regulations and the prominent nature of accounting earnings in the
capital markets, regulators perceive revenue-related fraud as more
devastating than those solely related to assets. Consistent with this,
Firth et al. (2005) nd that auditors are more likely to be sanctioned
when they fail to detect and report revenue-related fraud compared
with asset-related fraud. As a result, we expect the negative association
between auditor size and the incidence of nancial fraud to be stronger
for revenue-related fraud than for asset-related fraud.
To investigate this, we split the fraud sample into revenue- and
asset-related fraud. Untabulated statistics show that revenue-related
Fraud
Coefcient
t-Statistic
0.2598**
(1.96)
0.0537***
0.0899
0.3888***
0.1418
0.0089
0.2202
2.2199***
0.1012
0.0468
0.7612
Included
Included
7809
0.0879
(2.91)
(1.22)
(3.09)
(1.57)
(0.15)
(0.85)
(5.14)
(0.43)
(0.56)
(0.63)
Coefcient
t-Statistic
0.1078**
0.0473***
0.0876
0.3938***
0.1408
0.0173
0.2185
2.2250***
0.1186
0.0487
1.5423
Included
Included
7809
0.0854
(2.36)
(2.63)
(1.19)
(3.15)
(1.57)
(0.29)
(0.84)
(5.19)
(0.51)
(0.58)
(1.02)
Noe. This table reports the results of probit regressions investigating the association between
nancial statement fraud and auditor size. The dependent variable is Fraud, which equals 1 if
the rm committed nancial statement fraud in year t and 0 otherwise. See Table 4 for all
other variable descriptions. t-statistics clustered by rm are presented in parentheses. ***,
** and * denote signicance at the 1%, 5% and 10% level, respectively.
Ln(Auditor Size)
Auditor Tenure
ROE Dummy
State-Owned
Ln(Company Age)
Ln(Company Size)
Leverage
ROA
Book-to-Market
Losst 1
Intercept
Year xed effects
Industry xed effects
N
Pseudo R2
Regulated industries
Unregulated industries
Coefcient
t-Statistic
Coefcient
t-Statistic
0.2701***
0.2203***
0.3292
0.3292
0.3213
0.5439***
0.9951
3.1604**
1.7830
0.1638
15.9239***
Included
Included
885
0.25
(2.70)
(2.73)
(1.52)
(0.08)
(1.15)
(3.00)
(1.47)
(1.97)
(1.43)
(0.88)
(4.12)
0.1196**
0.0453**
0.1319*
0.3900***
0.0959
0.0552
0.2536
2.2990***
0.2017
0.0774
0.4585
Included
Included
6924
0.06
(2.45)
(2.49)
(1.76)
(3.14)
(1.02)
(0.86)
(0.93)
(5.24)
(0.87)
(0.91)
(0.28)
Note. This table reports the results of probit regressions investigating the association between
nancial statement fraud and auditor size for regulated versus nonregulated industries.
Regulated industries are dened as energy, iron and steel, oil renery and petrochemicals,
communications and heavy machinery industries. The dependent variable is Fraud, which
equals 1 if the rm committed nancial statement fraud in year t and 0 otherwise. See
Table 4 for all other variable descriptions. t-Statistics clustered by rm are presented in
parentheses. ***, ** and * denote signicance at the 1%, 5% and 10% level, respectively.
1192
Table 7
Probit regression of fraud type on auditor size.
Revenue-related fraud
Asset-related fraud
(1)
(2)
(3)
0.3532**
(2.29)
Top 10
0.1665***
(2.74)
0.0600**
(2.54)
0.1025
(1.16)
0.3655**
(2.12)
0.1010
(0.89)
0.0097
(0.12)
0.1011
(0.29)
1.9150***
(3.73)
0.0121
(0.04)
0.1850*
(1.73)
2.6796
(1.30)
Included
Included
7328
0.0964
Ln(Auditor Size)
0.0620***
(2.62)
0.1169
(1.33)
0.3665**
(2.11)
0.1035
(0.89)
0.0009
(0.01)
0.0970
(0.28)
1.8827***
(3.59)
0.0080
(0.03)
0.1836*
(1.75)
0.9321
(0.56)
Included
Included
7328
0.0926
Auditor Tenure
ROE Dummy
State-Owned
Ln(Company Age)
Ln(Company Size)
Leverage
ROA
Book-to-Market
Losst 1
Intercept
Year xed effects
Industry xed effects
N
Pseudo R2
(4)
0.2381*
(1.86)
0.0222
(1.01)
0.0660
(0.66)
0.3234**
(2.45)
0.1419
(1.21)
0.0467
(0.64)
0.3279
(1.07)
1.9500***
(3.75)
0.2805
(0.94)
0.0850
(0.84)
2.1004
(1.42)
Included
Included
7214
0.0716
0.0031
(0.07)
0.0244
(1.13)
0.0605
(0.61)
0.3216**
(2.44)
0.1360
(1.18)
0.0150
(0.21)
0.2874
(0.95)
1.9663***
(3.75)
0.1980
(0.67)
0.0840
(0.82)
1.4226
(0.86)
Included
Included
7214
0.0666
Note. This table reports the results of probit regressions investigating the association between revenue-related fraud and auditor size and between asset-related fraud and auditor size. The
dependent variable in regressions (1) and (2) is revenue-related fraud, which equals 1 if the fraud is related to revenues in year t and 0 otherwise. The dependent variable in regressions
(3) and (4) is asset-related fraud, which equals 1 if the fraud is related to assets in year t and 0 otherwise. See Table 4 for all other variable descriptions. t-Statistics clustered by rm are
presented in parentheses. ***, ** and * denote signicance at the 1%, 5% and 10% level, respectively.
Table 8
Probit regression of the incidence of fraud on auditor size after excluding rms listing in
other countries or listing B-shares.
Fraud
Top 10
Ln(Auditor Size)
Auditor Tenure
ROE Dummy
State-Owned
Ln(Company Age)
Ln(Company Size)
Leverage
ROA
Book-to-Market
Losst 1
Intercept
Year xed effects
Industry xed effects
N
Pseudo R2
Fraud
Coefcient
t-Statistic
0.4226***
(2.85)
0.0480**
0.0777
0.3100**
0.1612
0.0094
0.2885
2.2039***
0.1859
0.0535
0.4528
Included
Included
7042
0.09
(2.36)
(1.00)
(2.35)
(1.63)
(0.14)
(1.05)
(4.76)
(0.74)
(0.58)
(0.33)
Coefcient
t-Statistic
0.1212**
0.0404**
0.0761
0.3198**
0.1564
0.0043
0.2714
2.2001***
0.1903
0.0559
2.2137
Included
Included
7042
0.09
(2.57)
(2.04)
(0.98)
(2.45)
(1.62)
(0.06)
(0.99)
(4.80)
(0.76)
(0.61)
(1.33)
Note. This table reports the results of probit regressions investigating the association between
nancial statement fraud and auditor size after excluding rms listing in other jurisdictions
or listing B-shares. The dependent variable is Fraud, which equals 1 if the rm committed
nancial statement fraud in year t and 0 otherwise. See Table 4 for all other variable
descriptions. t-Statistics clustered by rm are presented in parentheses. ***, ** and * denote
signicance at the 1%, 5% and 10% level, respectively.
1193
Table 9
Ordered probit regression of fraud severity on auditor size.
Fraud
Top 10
Ln(Auditor Size)
Auditor Tenure
ROE Dummy
State-Owned
Ln(Company Age)
Ln(Company Size)
Leverage
ROA
Book-to-Market
Losst 1
Intercept 1
Intercept 2
Year xed effects
Industry xed effects
N
Pseudo R2
Fraud
Coefcient
t-Statistic
0.2461*
(1.88)
0.0504***
0.0894
0.4171***
0.1295
0.0062
0.2101
2.2376***
0.0689
0.0582
0.6522
0.8426
Included
Included
7809
0.0782
(2.72)
(1.20)
(3.40)
(1.43)
(0.10)
(0.81)
(5.30)
(0.29)
(0.74)
(0.54)
(0.70)
Coefcient
t-Statistic
0.1058**
0.0441**
0.0872
0.4217***
0.1286
0.0151
0.2100
2.2425***
0.0876
0.0597
1.5949
1.4042
Included
Included
7809
0.0763
(2.37)
(2.43)
(1.17)
(3.46)
(1.43)
(0.25)
(0.81)
(5.36)
(0.37)
(0.75)
(1.04)
(0.91)
Note. This table reports the results of ordered probit regressions investigating the association
between nancial statement fraud and auditor size. The dependent variable is Fraud, dened
as 0 when there is no fraud, 1 when there is fraud and punishment is public criticism (but not
monetary nes), and 2 when there is fraud and punishment includes monetary nes. See
Table 4 for all other variable descriptions. t-Statistics clustered by rm are presented in
parentheses. ***, ** and * denote signicance at the 1%, 5% and 10% level, respectively.
1194
Table 10
Probit regression of the incidence of fraud (as identied by high abnormal accruals) on auditor size.
Top 10
(1)
(3)
0.1944**
(2.02)
Ln(Auditor Size)
Auditor Tenure
ROE Dummy
State-Owned
Ln(Company Age)
Ln(Company Size)
Leverage
ROA
Book-to-Market
Losst 1
Intercept
Year xed effects
Industry xed effects
N
Pseudo R2
(2)
0.0107
(0.81)
0.2088***
(2.61)
0.1180
(1.39)
0.3588***
(4.62)
0.0040
(0.08)
1.5165***
(6.91)
8.0448***
(8.71)
0.7717***
(3.20)
0.1212
(1.13)
1.7667*
(1.71)
Included
Included
7299
0.23
(4)
0.1315**
(2.35)
0.0787**
(2.06)
0.0042
(0.32)
0.1870**
(2.34)
0.1299
(1.53)
0.3600***
(4.63)
0.0069
(0.13)
1.5338***
(6.91)
8.3854***
(9.38)
0.7500***
(3.15)
0.0728
(0.66)
0.2330
(0.20)
Included
Included
7299
0.23
0.0131
(1.54)
0.0116
(0.24)
0.1267**
(2.33)
0.2248***
(4.50)
0.0108
(0.35)
0.4267***
(3.27)
2.5361***
(7.93)
0.8680***
(7.21)
0.0225
(0.36)
0.2566
(0.42)
Included
Included
7299
0.10
0.0423*
(1.77)
0.0103
(1.14)
0.0371
(0.75)
0.1611***
(2.86)
0.2722***
(5.31)
0.0263
(0.78)
0.5409***
(3.80)
3.1145***
(8.99)
0.9408***
(7.12)
0.0018
(0.03)
1.6567**
(2.21)
Included
Included
7299
0.13
Note. This table reports the results of probit regressions investigating the association between nancial statement fraud (as identied by high abnormal accruals measured using the
performance-adjusted modied Jones model) and auditor size. The dependent variable is Fraud, which equals 1 if the rm committed nancial statement fraud in year t and 0 otherwise.
See Table 4 for all other variable descriptions. t-Statistics clustered by rm are presented in parentheses. ***, ** and * denote signicance at the 1%, 5% and 10% level, respectively.
References
Aharony, J., Lee, C.W.J., & Wong, T.J. (2000). Financial packaging of IPO rms in China.
Journal of Accounting Research, 38(1), 103126.
Allen, F., Qian, J., & Qian, M. (2005). Law, nance, and economic growth in China. Journal
of Financial Economics, 77(1), 57116.
Allen, F., Qian, J., & Zhang, C. (2011). An alternative view on law, institutions, nance and
growth. Working paper.
Burns, N., & Kedia, S. (2006). The impact of performance-based compensation on
misreporting. Journal of Financial Economics, 79(1), 3567.
Carcello, J.V., & Nagy, A.L. (2004). Audit rm tenure and fraudulent nancial reporting.
Auditing: A Journal of Practice and Theory, 23(2), 5569.
Chen, H., Chen, J.Z., Lobo, G.J., & Wang, Y. (2011). Effects of audit quality on earnings management and cost of equity capital: Evidence from China. Contemporary Accounting
Research, 28(3), 892925.
Chen, G., Firth, M., Gao, D.N., & Rui, O.M. (2006). Ownership structure, corporate governance, and fraud: Evidence from China. Journal of Corporate Finance, 12(3), 424448.
Chen, C.J.P., Su, X., & Zhao, R. (2000). An emerging market's reaction to initial modied audit
opinions: Evidence from the Shanghai stock exchanges. Contemporary Accounting
Research, 17(3), 429455.
Chen, S., Sun, S.Y.J., & Wu, D. (2010). Client importance, institutional improvements, and
audit quality in China: An ofce and individual auditor level analysis. The Accounting
Review, 85(1), 127158.
Chen, K.C.W., & Yuan, H. (2004). Earnings management and capital resource allocation:
Evidence from China's accounting-based regulation of rights issues. The Accounting
Review, 79(3), 645665.
Davis, L.R., Soo, B.S., & Trometer, G.M. (2009). Auditor tenure and the ability to meet or
beat earnings forecasts. Contemporary Accounting Research, 26(2), 517548.
DeAngelo, L.E. (1981). Auditor size and audit quality. Journal of Accounting and Economics,
3(3), 183199.
DeFond, M.L., Wong, T.J., & Li, S. (1999). The impact of improved auditor independence on
audit market concentration in China. Journal of Accounting and Economics, 28(3),
269305.
Ding, X. (2013). Financial frauds analysis on agriculture companies. National Business,
2013-07, 5859 (In Chinese).
Ding, Y., Zhang, H., & Zhang, J. (2007). Private versus state ownership and earnings
management: Evidence from Chinese listed companies. Corporate Governance: An
International Review, 15(2), 223238.
Erickson, M., Hanlon, M., & Maydew, E.L. (2006). Is there a link between executive
equity incentives and accounting fraud? Journal of Accounting Research, 44(1),
113143.
Ettredge, M.L., Sun, L., Lee, P., & Anandarajan, A.A. (2008). Is earnings fraud associated
with high deferred tax and/or book minus tax levels? Auditing: A Journal of Practice
& Theory, 27(1), 133.
Fan, J.P.H., & Wong, T.J. (2005). Do external auditors perform a corporate governance role in
emerging markets? Evidence from East Asia. Journal of Accounting Research, 43(1),
3572.
Firth, M., Mo, P.L.L., & Wong, R.M.K. (2005). Financial statement frauds and auditor sanctions: An analysis of enforcement actions in China. Journal of Business Ethics, 62(4),
367381.
Francis, J.R., & Krishnan, J. (1999). Accounting accruals and auditor reporting conservatism. Contemporary Accounting Research, 16(1), 135165.
Frankel, R., Johnson, M., & Nelson, K. (2002). The relation between auditors' fees for
nonaudit services and earnings management. The Accounting Review, 71105
(Supplement).
Geiger, M.A., & Raghunandan, K. (2002). Auditor tenure and audit reporting failures.
Auditing: A Journal of Practice and Theory, 21(1), 6778.
Ghosh, A., & Moon, D. (2005). Auditor tenure and perceptions of audit quality. The
Accounting Review, 80(2), 585612.
Glaeser, E., Johnson, S., & Shleifer, A. (2001). Coase versus the Coasians. Quarterly Journal
of Economics, 116(3), 853899.
1195
Liu, F., Xie, B., & Huang, Y. (2009). Audit size and audit quality. The Audit Research, 2009,
4554 (In Chinese).
Liu, F., & Zhou, F. (2007). Do Big 4 means high-quality audit services? Accounting Research,
2007, 7994 (In Chinese).
Loebbecke, J.K., Eining, M.M., & Willingham, J.J. (1989). Auditors' experience with material
irregularities: Frequency, nature, and detectability. Auditing: A Journal of Practice and
Theory, 9(1), 128.
Myers, J.N., Myers, L.A., & Omer, T.C. (2003). Exploring the term of the auditorclient
relationship and the quality of earnings: A case for mandatory auditor rotation. The
Accounting Review, 78(3), 779799.
Si, Zhanfeng (2012). Research on the recognition, measurement and correlation with
operation performance of intangible assets in China's publishing and media companies.
Working paper (In Chinese).
Teoh, S.H., & Wong, T.J. (1993). Perceived auditor quality and the earnings response
coefcient. The Accounting Review, 68(2), 346366.
Tian, L., & Estrin, S. (2008). Retained state shareholding in Chinese PLCs: Does government ownership always reduce corporate value? Journal of Comparative Economics,
36(1), 7489.
Wei, Z., Xie, F., & Zhang, S. (2005). Ownership structure and rm value in China's
privatized rms: 19912001. Journal of Financial and Quantitative Analysis, 40(1),
87108.
Wu, X. (2007). Regulatory propensity toward auditor liability in audit failures: An
empirical analysis. Accounting Research in China, 2007-7, 5361 (In Chinese).
Zheng, G., & Xu, Y. (2011). The consequences of capital market regulation: Evidence from
CSRC auditing enforcement. Working Paper (In Chinese).