Advances in Accounting, incorporating Advances in International Accounting 30 (2014) 18–31

Contents lists available at ScienceDirect

Advances in Accounting, incorporating Advances in

International Accounting
journal homepage: www.elsevier.com/locate/adiac

Mandatory audit partner rotation, audit market concentration, and audit

quality: Evidence from China☆
Sati P. Bandyopadhyay a,1, Changling Chen a,⁎, Yingmin Yu b,2
a
School of Accounting and Finance, University of Waterloo, 200 University Avenue West, Waterloo, Ontario N2L 3G1, Canada
b
School of Accountancy, Central University of Finance and Economics, 39 College South Road, Haidian District, Beijing 100081, China

a r t i c l e i n f o a b s t r a c t

Available online 29 December 2013 This research examines the audit quality consequences of China's mandatory audit partner rotation (MPR) reg-
ulation, which became effective in 2004. The rule requires firms to rotate signing audit partners of audit reports
JEL classification: every five years. We find that audit quality improves in the three years immediately following a client firm's MPR
M41 during the 2004–2011 period for a sample of 273 Chinese publicly listed firms. Specifically, we find that the im-
M42 provement is most pronounced in those Chinese provinces with both low levels of audit market concentration
and low levels of legal development. However, MPR does not improve audit quality in jurisdictions where
Keywords:
Mandatory audit partner rotation legal conventions are more developed and/or where audit markets are highly concentrated with a handful of
Audit market concentration large audit firms dominating the market.
Audit quality © 2013 Elsevier Ltd. All rights reserved.

1. Introduction reviewing partner in order to improve audit independence and thus

In this study we examine the effects of mandatory audit partner ro-
tation (MPR) on audit quality. Specifically, we look at the effects of MPR
under varying audit market concentration (AMC) conditions in the Chi-
nese audit market, where, starting in 2004, regulators required client
firms to rotate audit partners every five years. We find that MPR im-
proves audit quality in provinces with low levels of AMC but not in
provinces with high levels of AMC. Our results suggest that the effec-
tiveness of MPR policy on audit quality depends on the structure of
the audit market. Our research contributes to audit literature by relating
two long-standing issues, namely, the consequence of MPR on audit
quality, and the impact of AMC on audit quality. Both these issues
have recently re-entered the public debate in the USA (Public
Company Accounting Oversight Board, 2011a) and the European
Commission (EU) (2011).
In the post-Enron era, several countries have mandated periodic ro-
tation of the lead audit engagement partner and the concurring
☆ We thank the editor Philip Reckers and two anonymous referees for their insightful
comments. We also thank Xi Wu, Junsheng Zhang, and other workshop participants at
the Central University of Finance and Economics. This study is supported by the research
grants from the “Project 211” Fund of the Central University of Finance and Economics,
China and grants from the “2011 Synergetic Innovation” Key Project on “Development of
Public Accounting Profession” of the Central University of Finance and Economics, China.
We also gratefully acknowledge the financial support from the Research Fellowship
Program at the School of Accounting and Finance of the University of Waterloo.
⁎ Corresponding author. Tel.: +1 519 888 4567x35731; fax: +1 519 888 7562.
E-mail addresses: bandy@uwaterloo.ca (S.P. Bandyopadhyay), clchen@uwaterloo.ca
(C. Chen), yuym168@gmail.com (Y. Yu).
1
Tel.: +1 519 888 4567x32533; fax: +1 519 888 7562.
2
Tel.: +86 10 62156441; fax: +86 10 62288114.
audit quality. For example, the Sarbanes–Oxley Act (henceforth SOX,
2002) mandates US audit partners to rotate their audit clients every
five years, and the European Union requires audit firms to replace
audit partners in charge of their clients that are Public Interest Entities
every seven years. Similar MPR requirements are also in vogue in
Australia, China, Taiwan, and many other jurisdictions.3 The conse-
quence of mandatory auditor rotation (at firm or partner level) on
audit quality depends on the tradeoff of improvement in audit indepen-
dence versus loss in client-specific audit experience (Kinney &
McDaniel, 1996; Knapp, 1991; Mautz & Sharaf, 1961).4 On the one
hand, a fresh look into the audit engagement by the rotated-in audit
partner improves audit independence and thus the quality of the
audit. On the other hand, the rotated-in partner does not possess the
client-specific expertise of the rotated-out partner, and this lack of ex-
perience could reduce audit quality. The final effect of MPR on audit
quality is an empirical issue determined by the tradeoff. Empirical evi-
dence on the relation between MPR and audit quality is mixed. Studies
based on the Taiwanese audit market indicate either no effect or a neg-
ative effect of MPR on audit quality (Chen, Lin, & Lin, 2008; Chi & Huang,
3
Other countries adopting MPR include Singapore, Japan, United Kingdom, France,
Spain, the Netherlands, and Germany (General Accounting Office 2003, Appendix V; Chi
et al., 2009).
4
These authors use the generic “auditor” term in their papers and do not specify wheth-
er their arguments apply to audit firm or audit partner level. These arguments apply equal-
ly well to both firm and partner level (Chen et al., 2008). DeAngelo (1981) defines audit
quality as the joint probability that an auditor detects a breach of accounting standards
and the probability that the auditor reports the breach. MPR will likely decrease the prob-
ability of detecting a breach because of lost audit knowledge but increase the probability of
reporting the breach.

0882-6110/$ – see front matter © 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.adiac.2013.12.001
 S.P. Bandyopadhyay et al. / Advances in Accounting, incorporating Advances in International Accounting 30 (2014) 18–31
2005; Chi, Huang, Liao, & Hong, 2009). In contrast, research based on the
Australian audit market (Carey & Simnett, 2006) provides some evi-
dence that MPR tends to enhance audit quality. Experimental evidence
(Dopuch, King, & Schwartz, 2001; Tan, 1995) also suggests that MPR im-
proves audit quality. Note that while the identity of audit partners is
public information in Taiwan and Australia, and thus the effect of MPR
on audit quality can be evaluated directly, this is not the case in many
jurisdictions, including the USA.
Recently, the European Commission (EU) (2011) has expressed the
view that the practice of MPR does not improve audit independence
(and hence audit quality). The argument is that MPR does not remove
the familiarity threat that might cloud audit judgment and reduce pro-
fessional audit skepticism of a new (rotated-in) audit partner, who
would not have incentives to take decisions that might cause the audit
firm to lose a long-standing client firm. Hence, new (replaced) audit
partners “likely feel obliged to live with the decisions and agreements
made by the former (rotated-out audit) partner; he/she may have little
flexibility to reopen them” (European Commission (EU), 2011, page 17).
On the basis of these arguments, European Commission (EU) (2011)
proposes mandatory audit firm rotation (MFR) to replace MPR. Consis-
tent with the view of the European Commission (EU) (2011), the US
Public Company Accounting Oversight Board (henceforth PCAOB) is-
sued a concept release in August 2011 (PCAOB, 2011a) that also sug-
gests MFR for US firms.
The accounting community, however, has generally opposed the
proposal to replace MPR with MFR. For example, a summary of re-
sponses to the MFR study by the General Accounting Office (GAO,
2004, Question 73) shows that about two-thirds of the respondents ap-
pear to believe that relative to potentially more costly MFR, MPR suffi-
ciently achieves the intended benefits of taking a fresh look at the
audit engagement by the rotated-in partner. In their response letter to
the PCAOB (PCAOB, 2011b), the International Federation of Accountants
(henceforth IFAC) argues that “these changes [MPR] are still relatively
new, and have not been in place sufficiently long enough to objectively
assess their impact.” This IFAC response implies the need for further ex-
amination of the consequences of MPR.
In summary, while many in the professional accounting and audit
community believe that MPR enhances audit quality, international reg-
ulators do not seem to share that view. Moreover, as discussed above,
academic research provides mixed evidence on the impact of MPR on
audit quality.
Responding to the call of IFAC for further research on the effects of
MPR, we examine the relation between MPR and audit quality in the
Chinese audit market to provide some insight into this contentious de-
bate. Also, in contrast to extant research, we consider the impact of re-
gional variation in AMC, whose role has recently been highlighted in
the foregoing MPR versus MFR deliberations. For example, the Center
for Audit Quality (henceforth CAQ) argues in its written statement
(PCAOB, 2011c) to PCAOB's (2011a) concept release that many small
audit firms will find the costs of periodic tendering, documentation
and staffing associated with MFR too onerous and will be forced to
“abandon their public practice and focus instead on private company
audits” (PCAOB, 2011c, page 12). This suggests that abandonment of
MPR for MFR could lead to a higher level of concentration of the Big 4
audit firms in many public audit markets that are already considered
highly concentrated by policymakers.
For example, European Commission (EU) (2011) has expressed con-
cerns that a potential demise of any of the existing Big 4 audit firms in
highly concentrated audit markets might de-stabilize the financial sys-
tem. There is also concern that “concentration among a few firms en-
abled the largest accounting firms to exercise greater influence over
the audit standard setting process and regulatory requirements.”
(GAO, 2003) Audit quality could also be compromised through moral
hazard issues if large audit firms believe they are “too few to fail”
(GAO, 2003). The lack of choice in audit markets dominated by Big 4
audit firms is another concern. Client firms might not want to be audited
19
by the same audit firm that audits its competitors (European Commission
(EU), 2011).
Regardless of such comments, which reflect serious policy concerns
about the consequences of heightened AMC, recent research often finds
a positive relation between AMC and audit quality (Francis, Michas, &
Seavey, 2013; Kallapur, Sandaraguruswamy, & Zang, 2010). In our
paper we examine whether MPR can improve audit quality in low
AMC jurisdictions where extant research tends to report lower audit
quality than that in high AMC jurisdictions. It might be noted that, to
date, audit research has either examined the relation between audit
quality and MPR without controlling for AMC (e.g. Carey & Simnett,
2006; Chi et al., 2009) or the relation between AMC and audit quality
without controlling for MPR (Kallapur et al., 2010). The unequal pace
of audit market development across Chinese provinces makes it an
ideal setting for our analysis. For example, our Chinese MPR sample ex-
hibits a wide range of variation in AMC at the provincial level. During
the 2004 to 2011 sample period, the provincial Herfindahl index
based on audit fees and client locations varies between a minimum of
0.071 and a maximum of 0.982, with the first quartile of 0.125, a median
of 0.155, and the third quartile of 0.217.5
We identify 273 unique Chinese publicly listed client companies
countrywide that are subject to the MPR rule and compare their audit
quality both pre- and post-MPR.6 We require our sample companies ro-
tate out their signing audit partners when they have met the maximum
five-year tenure requirement. We also impose the condition that audit
partner rotation not be accompanied by audit firm rotation in the
periods before and after MPR to avoid a potentially confounding
effect of audit firm rotation on audit quality. Following prior research
(e.g. Chen et al., 2008; Francis et al., 2013), we use abnormal (discretion-
ary) accruals as our measure of audit quality.
We find that on average MPR has a positive effect on audit quality in
the post-rotation years, especially in the second and third years after
MPR. When we partition our sample by the provincial AMC levels, we
find that the incremental benefit of MPR on audit quality is observed
only in low, but not high, AMC provinces in China. We then extend
Firth, Rui, and Wu (2012a) to examine whether the interaction of the
level of legal development with AMC has an effect on how MPR en-
hances audit quality. Firth, Rui, and Wu (2012a) show that MPR has a
positive effect on audit quality in China only in regions with low levels
of legal development. We find that the beneficial effect of MPR is ob-
served in those provinces that not only have low levels of legal develop-
ment but also low levels of AMC. Our results are robust to alternative
audit quality measures such as the discretionary working capital ac-
cruals (Carey & Simnett, 2006), different AMC measures, and various
pre-MPR periods (one year before MPR and a three-year period before
MPR), after controlling for audit firm tenure, client company size, cash
flows, industry growth, age, state ownership, and leverage.
Our paper contributes to the literature by bringing together two dif-
ferent streams of extant audit research. One stream examines the effects
of MPR on audit quality, but the results are indeterminate. The other
stream of literature examines the effect of AMC on audit quality in papers
such as Kallapur et al. (2010) and Francis et al. (2013). We contribute to
the literature by investigating the effects of MPR on audit quality under
different AMC conditions and different levels of legal development. We
show that MPR is able to enhance audit quality in Chinese provinces
where both of these features are relatively underdeveloped. However,
5
Kallapur et al. (2010) report US MSA (metropolitan statistical area) Herfindahl index-
es of 0.230, 0.252, and 0.293 respectively for the first quartile, median, and the third quar-
tile distributions in the 2000–2006 period. However, these authors study the relation
between AMC and audit quality but not in the MPR setting.
6
Relative to a cross-sectional comparison between MPR client firms and a control sam-
ple (e.g. voluntary audit partner rotation or no-rotation sample, as in Chi et al., 2009) our
pre- versus post-MPR comparison for MPR firms has the advantage of highlighting the
consequence of MPR by eliminating the potentially confounding effects of covariates relat-
ed to client firms' characteristics.
20 S.P. Bandyopadhyay et al. / Advances in Accounting, incorporating Advances in International Accounting 30 (2014) 18–31
the improvement in audit quality does not seem to be large enough to
catch up with the high levels of audit quality enjoyed by client firms in
provinces with high AMC and a high level of legal development. Our find-
ings suggest that the effect of MPR on audit quality needs to be examined
on a case-by-case basis and cannot be subject to generalized conclusions
for different audit market structures. For example, the contradictory MPR
results in Taiwan versus Australia might arise from not considering inter-
nal variations in audit market conditions in these countries.
Our results have some potential policy implications. As previously
mentioned, while regulators are concerned about the potential adverse
effects of high AMC, audit quality is also demonstrated to be high in
these jurisdictions. Our study shows that MPR is a policy tool that
could potentially improve audit quality in low AMC areas and thus
avoid the attendant policy problems of high AMC. However, given that
our sample firms (273) constitute less than 20% of Chinese publicly
listed companies, the generalizability of our results needs to be treated
with caution.
The remainder of the paper is organized as follows. In Section 2,
we describe the institutional background and summarize the litera-
ture. We develop our hypotheses in Section 3. In Section 4, we dis-
cuss the sample selection procedure, variable measurement, and
research methods. Section 5 summarizes our results. Section 6
concludes.
2. Institutional background and literature review
Since the 1990s, Chinese accounting and auditing regulators, in-
cluding the Chinese Institute of Certified Public Accountants
(CICPA), the Ministry of Finance (MOF), and the China Securities
Regulatory Commission (henceforth CSRC), have undertaken a num-
ber of steps to enhance audit independence. In 1996, these national
regulators required all Chinese audit firms that had previously had
government affiliations to break off their government ties (Gul,
Fung, & Jaggi, 2009). Regulators also adopted a new set of auditing
standards in 1995, which ultimately resulted in their convergence
with the International Standards on Auditing (ISA) promulgated by
the IFAC (Firth et al., 2012a; Lin & Chan, 2000; Xiao, Zhang, & Xie,
2000).
Moreover, in the wake of a series of accounting scandals that took
place in the late 1990s and early 2000s, which resulted in the bankrupt-
cy of Yinguangxia Company and other listed companies, the auditing li-
cense of the then largest audit firm in China, Zhongtianqin
(Yinguangxia's auditors),7 was suspended. The Chinese government
then implemented a series of measures to restore public confidence in
the financial reporting process (Chen, Sun, & Wu, 2010). In October
2003, the CSRC and MOF jointly issued a mandatory audit partner rota-
tion rule (Chinese Securities Regulatory Commission (CSRC Regulation)
& China Ministry of Finance (MOF), 2003 No. 13) to improve audit inde-
pendence and thus audit quality. Under this rule, which became effec-
tive January 1, 2004, all Chinese-listed companies are required to
rotate out their audit partners who have signed the company's audit re-
ports for five consecutive years.8 As in Taiwan, Chinese audit reports
have two signatories: a lead audit partner responsible for fieldwork
7
Chen et al. (2010), Appendix 1) provides a list of scandals in the Chinese stock market
in the early 2000s.
8
Chen et al. (2008) raise the concern that MPR may be superficial if a partner rotates
back to the client after a very short “cooling-off” period. They find that more than half of
the partners in their Taiwanese sample, who rotated off in 2003 or 2004, rotated back after
one year. Note that there is no minimum cooling-off period for a rotation-off audit partner
in Taiwan. In contrast, Chinese regulations require a minimum two-year “cooling-off” pe-
riod. In our sample, 53 client companies are found to have re-appointed the same audit
partners to their prior audit clients in Year 8, the year right after the required two-year
cooling-off period. We conduct sensitivity tests by eliminating these client companies
and our results hold. Firth, Rui, and Wu (2012b) findings are consistent with reduced audit
quality associated with “rotated back” partners.
and a reviewing partner who must be at least a deputy executive of
the audit firm.9 These two signing auditors are required to assume the
same legal liability unless proved to the contrary (Firth et al., 2012a).10
To be consistent with prior literature, we refer to signing auditors as
audit partners.
The usefulness of mandatory auditor rotation (at the firm or partner
level) has been a matter of debate both in the financial press and in the
academic auditing literature for a number of years. Apart from a few
jurisdictions, like Italy and Brazil, that have mandatory audit firm
rotation, individual audit partner rotation has become a require-
ment in several jurisdictions including Australia (Carey & Simnett,
2006), Taiwan (Chi et al., 2009), China (Firth et al., 2012a), USA
(SOX, 2002), Singapore, United Kingdom, France, Spain, Netherlands,
Japan, and Germany (GAO, 2003, Appendix V). It is also currently
being considered in Canada. The adoption of MPR rules in these audit
markets reflects regulators' concerns that lengthy audit partner ten-
ure reduces audit quality on account of its adverse effect on audit
independence.
Researchers have argued both in favor of and against mandatory ro-
tation at the audit firm level.11 Some of these arguments apply to man-
datory rotation at the audit partner level as well, because individual
partners have incentives to maintain their relationships with a client
in order to retain the client (Chen et al., 2008, page 420). The argument
in favor of mandatory rotation is that a long association with an audit
client clouds the auditor's judgment and leads to impairment of audit
independence (Mautz & Sharaf, 1961). A new audit partner who rotates
in to periodically replace an incumbent audit partner is expected to take
a “fresh look” into different aspects of the engagement and improve
audit quality. Healey and Kim (2003) argue that mandatory rotation
will restore “badly shaken investor confidence in the financial account-
ing system.” In an experimental study, Dopuch et al. (2001) provide ev-
idence that MPR increases auditor independence. In another
experimental study, Tan (1995, page 115) concludes that “staff rotation
and review awareness can improve the quality of the audit decision pro-
cess.” Raghunathan, Lewis, and Evans (1994) provide some evidence
consistent with greater frequency of US Security Exchange Committee
(SEC) actions against longer tenure audit firms, suggesting a potential
beneficial role of MPR practice.
An important argument against mandatory rotation is that it will de-
prive the client of the firm-specific knowledge acquired by the incum-
bent auditor through a steep learning curve over a period of time
(Knapp, 1991). This knowledge is not available to the replaced (rotat-
ed-in) auditor. This view is largely supported in the extant US-based
studies which, with few exceptions (e.g. Davis, Soo, & Trompeter,
2009; Deis & Giroux, 1992) find that audit quality is poor in the early
9
In the review process, a review partner examines audit plan, audit risk (inherent risk
and control risk), audit evidence (e.g. audit sampling), auditing adjustments, and audit re-
port draft. This is consistent with ISA regulations, under which engagement control review
partners are required to review the risk of material misstatement (inherent risk and con-
trol risk) and audit sampling among others. In our sample, we are unable to identify which
of the two signing auditors is the lead or reviewing partner. The authors' interviews with
practitioners in China reveal that the order of two signatures is random for many auditing
firms. We find that our results are not sensitive to whether we center our MPR analyses on
audit tenure of the first versus the second signing auditor. In our sample, 108 of the 273
client companies of our sample rotated out their first signing audit partners, 83 rotated
out the second signing audit partners, and 82 rotated out both partners.
10
Chinese Independent Auditing Standards, adopted in 1995, are highly convergent
with the ISA of the IFAC (Firth et al., 2012a; Lin & Chan, 2000; Xiao et al., 2000). According
to the Chinese auditing standards, reviewing partners must hold the position of at least
deputy executive in an audit firm. In the review process, a review partner shall examine
the audit plan, audit risk (inherent risk and control risk), audit evidence (from audit sam-
pling), auditing adjustments, and audit report draft. By ISA, the engagement control re-
view partner shall also review the risk of material misstatement (inherent risk and
control risk) and audit sampling among others. The Chinese review audit partners' tasks,
hence, share similarities with ISA's engagement quality control on audit review.
11
In our paper, mandatory rotation refers to mandatory audit partner rotation, unless in-
dicated otherwise.
 S.P. Bandyopadhyay et al. / Advances in Accounting, incorporating Advances in International Accounting 30 (2014) 18–31
years of an audit firm's tenure with a specific client.12 This is consistent
with the notion that audit quality improves as an audit partner acquires
client-specific knowledge over time. As a corollary, it could be argued
that the loss of client-specific expertise when an incumbent audit part-
ner is rotated out under MPR could hurt audit quality. However, none of
the foregoing US-based papers provide direct evidence on the conse-
quences of MPR on measures of audit quality because they mostly
study audit firm tenure but not individual audit partner tenure.
Direct empirical tests of the consequences of MPR on audit quality
are mostly based on the audit and financial data obtained from
Taiwan and Australia where13, unlike in the US, identities of the audit
partners who perform the audit are disclosed. In China, as in these coun-
tries, the names of the two audit partners engaged in auditing a client
are public information.
Carey and Simnett (2006) examine the association between audit
quality and long audit partner tenure using data from Australia for a pe-
riod when audit partner rotation was not mandatory. They find that
audit quality declines with audit partner tenure when audit quality is
measured as (1) the propensity to issue a going concern opinion for dis-
tressed firms, or (2) the probability of exceeding earnings benchmarks.
Also using Australian data, Hamilton, Ruddock, Stokes, and Taylor
(2005) find less income-increasing discretionary accruals following
MPR. In contrast, using Taiwanese audit partner data, Chen et al.
(2008) find that audit quality increases with audit partner tenure and
audit firm tenure, thus concluding that MPR (and also MFR) is likely
to impair audit quality. Similarly, Chi et al. (2009) conclude that MPR
does not help audit quality in Taiwan. These authors find that the level
of discretionary accruals of client companies in the year of MPR is no
lower than the pre-rotation year level.
The above mentioned mixed results show that the effect of MPR on
audit quality in Taiwan versus Australia is different, probably due to the
specific audit market in which the relationship is examined. In our
study, we first examine the average effect of MPR on audit quality
using Chinese audit data over all Chinese provinces. This analysis pro-
vides fresh insights into the benefits or otherwise of MPR in a different
audit market. We then analyze how this relation changes with the pro-
vincial variation in AMC, which is an audit market characteristic not
studied in previous Australian/Taiwanese research. It is important to
note that AMC has an independent effect on audit quality (e.g. Francis
et al., 2013; Kallapur et al., 2010) and might confound empirical tests
of the relation between MPR and audit quality if its effect not controlled
for. Furthermore, international regulators are concerned about the ad-
verse effects on audit quality of dominance of a handful of audit firms
in international audit markets.
China exhibits wide variation in provincial AMC. For example, the
Herfindahl index based on audit fees and client locations is around
0.07 in the populous Guangdong province through our sample period,
whereas the figure is around 0.55 for the equally populous Zhejiang
province over the same period. In the entire country, 64 audit firms
audited 1570 listed clients in China (Chinese Institute of Certified
12
Gul, Jaggi, and Krishnan (2007) demonstrate greater earnings management in the ear-
ly years of an audit firm's tenure. Arel, Brody, and Pany (2005) show that audit failures are
common in the initial years of an audit firm's audit engagement. Geiger and Raghunandan
(2002) measure audit failure as the inability of the auditor to issue a modified audit opin-
ion before its client goes bankrupt and find more instances of audit failures in the early
years of the audit firm's tenure. Carcello and Nagy (2004) come to similar conclusions
about the timing of financial statement frauds. Stanley and DeZoort (2007) also find that
audit failures tend to take place in the early years of an audit firm's tenure. Several studies
find a positive relation between financial reporting quality and audit firm tenure (e.g.
Johnson et al., 2002; Mansi, Maxwell, & Miller, 2004; Myers et al., 2003). This suggests that
financial statement quality improves with audit firm tenure, which is in contrast to the
non-linear relation between earnings management and audit firm tenure demonstrated
by Davis et al. (2009), who find that audit quality of firms with short (two to three years)
or very long (13–15 years or more) tenure tends to be low. Ghosh and Moon (2005) find a
positive relation between earnings response coefficients and audit firm tenure.
13
Recently, PCAOB has proposed that US audit firms disclose the name of the audit part-
ner in charge of a client's audit (Rapoport, 2013).
21
Public Accountants (CICPA), 2007 Bulletin No. 15), averaging at less
than 25 clients per audit firm. The dominance of Big 4 audit firms is
much less evident in China (auditing 17% of all listed Chinese companies
during 1999–2007 period) relative to many other jurisdictions (Francis
et al., 2013) including the USA (61%), Australia (71%), and Taiwan (74%).
Chinese regulators are wary of its audit markets being dominated by Big
4 audit firms and is keen to encourage the growth of domestic firms to
compete with large international audit firms that were allowed to oper-
ate directly in China after the country's accession to the World Trade Or-
ganization (henceforth WTO) in the post-2001 period (Chan & Wu,
2011).14 Variations in AMC across different Chinese provinces provide
us with the opportunity to examine whether the beneficial effects of
MPR on audit quality, if any, vary with AMC levels.
Finally, we examine if the relation between MPR and audit quality is
affected by the level of legal development of the province in which the
audit client is located, especially when viewed in conjunction with the
AMC level of the province. Chinese provinces do not have a uniform
level of legal development, and they exhibit large differences in this re-
spect. Fan, Wang, and Zhu (2004) measure the provincial level of legal
development in terms of the number of lawyers as a percentage of the
population, the efficiency of the local courts, and the protection of prop-
erty rights. Wang, Wong, and Xia (2008) show that there are great geo-
graphical disparities in legal development in China in terms of
protection of property rights and efficiency of law courts among other
legal elements. Firth et al. (2012a) examine the audit quality conse-
quence of MPR in China in strong versus weak legal environments
using the Wang et al. (2008) criteria. These authors find that client com-
panies subject to MPR exhibit better audit quality than non-MPR com-
panies in 2004, the first year that MPR was adopted in China.
However, this positive audit quality effect is restricted to client compa-
nies located in Chinese provinces suffering from low levels of legal de-
velopment; the positive audit quality effect of MPR does not apply to
more legally developed provinces.
However, Firth et al. (2012a) do not examine, as we do, the change
in audit quality pre- and post-MPR for both the mandatory rotation
year and up to two years subsequently. More importantly, they do not
examine the audit quality effect of MPR conditional on varying provin-
cial AMC levels. In contrast, in our paper we examine how MPR affects
audit quality for different provinces that exhibit different levels of
legal development and AMC.
3. Hypotheses development
As stated earlier, the effect of MPR on audit quality is the result of
a tradeoff between having an independent and fresh look at the en-
gagement by a new (rotated-in) audit partner versus losing the
audit expertise of the departing (rotated-out) auditor. Note that
audit independence is defined in the literature as the joint probability
that a given auditor will both (a) “discover a breach in the client's
accounting system, and (b) report the breach” (DeAngelo, 1981). The
rotated-in partner will probably have less hesitation in reporting a breach,
if a breach is discovered, than the outgoing partner, because he or she
would not have had enough time to build up a relationship with the
client. This is likely to enhance audit independence and thus audit quality.
However, the effects of MPR on the probability of discovering a
breach is not very clear. On the one hand, the new partner, by taking a
“fresh look” (Mautz & Sharaf, 1961), might find it easier to uncover a
14
The motivation for Chinese regulators to encourage the growth of domestic audit
firms is to improve the supply of quality audit services to meet the increased demand
resulting from the transformation of state-owned enterprises (SOE) into public-listed
companies (Chan & Wu, 2011). Note that before SOEs were privatized, they were audited
by the National Audit Office of the People's Republic of China, not independent audit firms.
In addition, the demand for large domestic audit firms arose from sovereignty consider-
ations, rather than audit market considerations, especially after the Chinese accession to
WTO. In other words, the Chinese government felt it was more desirable to have large local
audit firms dominate the Chinese audit market than large foreign Big 4 audit firms.
22 S.P. Bandyopadhyay et al. / Advances in Accounting, incorporating Advances in International Accounting 30 (2014) 18–31
breach, compared to the outgoing partner. On the other hand, the new
partner probably will not possess the firm-specific expertise enjoyed
by the outgoing partner, thereby making it harder for him or her to de-
tect a breach, if one exists. The outcome of the tradeoff of the “fresh
look” versus reduction of client-specific expertise on the probability of
discovering a breach is an empirical issue.15 Thus, because it is deter-
mined by the joint probability of discovering a breach and disclosing a
breach, overall improvement of audit quality due to MPR is hard to
predict.
Extant evidence on the effect of MPR on audit quality, which relies
on empirical research using Australian (Carey & Simnett, 2006;
Hamilton et al., 2005) and Taiwanese (Chen et al., 2008; Chi et al.,
2009) audit data, is mixed. Experimental studies, on the other hand,
show positive effects of MPR on audit independence (Dopuch et al.,
2001) and audit quality (Tan, 1995). Given the theoretical and empirical
evidence, our first hypothesis about average audit quality in Chinese
audit markets across all provinces is non-directional.
H1. Average audit quality across Chinese provinces does not change
after mandatory audit partner rotation as compared to the pre-
rotation period.
As stated earlier, AMC varies greatly across different provinces in
China. In this paper we also examine the effect of MPR on audit quality
under varying AMC conditions at the provincial level. To the best of our
knowledge, there has not been any theoretical analysis of this issue.
However, the literature does provide a number of conflicting predic-
tions about the potential effects of AMC on audit quality. For example,
in a theoretical paper, Chaney, Jeter, and Shaw (2003) argue that the
cost of losing a single client when an auditor tells the truth about a
breach is quite small under high competition (low AMC) because of
the low profit margins in these markets and a large client pool.
These authors state that the low cost of “telling the truth” would en-
courage disclosure of a breach and improve audit quality in high
competition audit markets. It is unclear, however, how MPR will
change this cost of telling the truth in highly competitive (low
AMC) audit markets.
In contrast, DeAngelo (1981) argues that large audit firms with a
large client base have strong incentives to remain independent because
of large quasi rents they earn from their clients. Large quasi rents exist
because the potential reputation cost arising from an audit failure is
likely to be high for large audit firms. This implies that audit markets
that are dominated by a small number of large audit firms (or high
AMC) will exhibit a high level of audit independence and therefore
high audit quality. If this argument is true, MPR will likely not affect
audit quality in high AMC markets incrementally either, given that
audit quality is probably high to begin with.16
In addition, Watts and Zimmerman (1981) argue that larger audit
firms, which dominate concentrated audit markets, are better at moni-
toring their auditors in comparison with small audit firms. Since the cost
of audit failure is high for large audit firms, resulting in costly decline in
reputation and probable shareholder lawsuits, these audit firms moni-
tor their audit partners very closely. Therefore, the incremental effect
of MPR, with its fresh look into the engagement by a rotated-in partner,
is likely to be minimal. This implies that any enhancement to audit qual-
ity from MPR is more likely to be significant in less concentrated (highly
competitive) audit markets where audit firm, on average, are likely to
be smaller, and the level of monitoring of audit partners by the audit
firm is probably less intense.
The forgoing discussions indicate that the predicted effects of MPR
on audit quality under differing AMC conditions are uncertain. Empirical
15
In any case, we control for audit expertise effects by including incoming partner's in-
dustry experience in our empirical tests.
16
“Theory suggests that auditor independence and audit quality are inextricably linked,
with auditor independence being an integral component of audit quality.” (GAO, 2003).
evidence tends to show that audit quality is better in high AMC US city
markets (Kallapur et al., 2010) and international country-level markets
(Francis et al., 2013). Despite this, some influential commentators hold a
different view. For example, Honorable Richard Breeden (see PCAOB,
2012, page 5), an ex-chairman of the SEC, has stated that the present
value of future revenues from a few large audit engagements of any
Big N audit firm, assuming “continued incumbency,” could exceed sev-
eral billion dollars, which “helps to explain why the issue of auditors try-
ing to please the largest clients continues to arise.” This view is
consistent with audit quality being low in highly concentrated markets,
where MPR might have a beneficial effect.
Since the relation between MPR and audit quality under different
audit market conditions is ambiguous, our second hypothesis, also
non-directional, is as follows:
H2. The improvement in audit quality after mandatory audit
partner rotation in provinces with low audit market concentration
is not different from that in provinces with high audit market
concentration.
Finally, we examine the effect of market concentration in interaction
with legal development on the change in post-MPR audit quality. As
stated earlier, Firth et al. (2012a) find that the level of legal develop-
ment in Chinese provinces affects the relation between MPR and audit
quality using the provincial legal environment index of Wang et al.
(2008).17 Arguably, market discipline and more effective monitoring
in the high AMC and high legal development provinces allow audit
firms to maintain a high level of audit quality regardless of MPR. The
code of auditor ethics and legal environment likely provide an effective
monitoring mechanism for maintaining audit independence in jurisdic-
tions with high levels of AMC and a strong legal enforcement frame-
work. Thus, incremental benefits from MPR, if any, might be less
significant in provinces with high AMC and a strong legal environment
relative to provinces of low AMC and low legal development. However,
we do not have a directional hypothesis on this interaction effect be-
cause it is also possible that audit quality in low AMC and low legal de-
velopment provinces does not improve after MPR due to potentially
weak market discipline and government regulation. Our third hypothe-
sis also has no directional predictions:
H3. The improvement in audit quality after MPR in provinces with low
audit market concentration and low legal development is not different
from that in provinces with high audit market concentration and high
legal development.
4. Sample selection and research design
4.1. Sample selection
We obtain our audit and financial data from a database compiled by
Sinofin Technology Limited Co. and China Center for Economic Research
(CCER) of Beijing University. The names of the signing auditors and
audit firm data of Chinese publicly listed companies are available in
the CCER database starting from 2000. We manually collect 1999 audit
data from publicly available CICPA archives to compile data over a
five-year period from 1999 to 2003, as illustrated in Appendix 1, for
identifying clients firms' MPR in 2004, the first year of adoption of the
MPR rule.18
Our sample covers the 1999–2011 period. We require signing audi-
tors' identity data for consecutive six-year rolling windows starting
from 1999 in order to identify mandatory audit partner changes starting
in 2004, the first year of MPR. Any effect of MPR on audit quality
17
We provide details on the legal environment index measure in Section 5.
18
Our sample does not include those companies that adopted the MPR rule early in
2003. All Chinese-listed companies are required to have a fiscal year end of December 31.
 S.P. Bandyopadhyay et al. / Advances in Accounting, incorporating Advances in International Accounting 30 (2014) 18–31
could take longer than one year because rotated-in audit partners
might require some time to familiarize themselves with the new
client's business. Therefore, we compare client firms' audit quality
in the pre-MPR period versus the MPR year, and two subsequent
years as well, in order to examine longer-term effects of MPR, if
any. Our last MPR year is 2009 because 2011 is the last year of our
sample period.
Appendix 1 describes the method we follow to classify an audit part-
ner change as mandatory rotation. We refer to the first year that we
start counting audit partner tenure as Year 1, the last year prior to
MPR as Year 5, the MPR year with new rotated-in audit partner as
Year 6, and the two subsequent years as Years 7 and 8, respectively.
We classify an audit partner change as a mandatory rotation if the pre-
vious incumbent audit partner “A” had audited a client from Years 1 to 5
but was replaced by audit partner “B” in Year 6. In our main tests, we
compare audit quality of Year 5 versus Years 6, 7, and 8. For example,
for client companies subject to MPR in 2004 (Year 6), we use 2003
data to evaluate Year 5 audit quality, and 2006 and 2007 data to evalu-
ate Years 7 and 8, respectively.
As shown in Table 1, we begin with the CCER sample of 13,287
firm–year observations for the 2004–2011 period. After eliminating
observations with missing audit partner names in Year 6 and the
five preceding years, we obtain 7094 firm–year observations,
representing an average of 887 client companies per year. Out of
this sample, we select 455 unique client companies who rotate
their audit partners after they finish their five-year tenure for
the first time during 2004–2009. We remove another 169 client
Table 1
Sample selection and composition.
Panel A: Sample selection
China Center for Economic Research (CCER) Database
Total number of firm–year observations during 2004–2011
Less: Number of firm–year observations that had missing
signing audit partners' names in the current year
(2004–2011) and the five preceding years (1999 audit partner
and audit firm data were hand-collected)
Preliminary sample:
Number of firm–year observations (2004–2011)
Average number of client firms per year
Mandatory partner rotation sample selection:
Number of companies that had auditor rotations after a five-year
tenure for the first time at Year 6 (2004–2009)
Less: Number of companies with audit firm change in Year 6 or
the five preceding years
Less: Number of companies that should rotate both partners but
only rotate one after mandated tenure (one-year extension
allowed when both audit partners' tenure reached five years
in the same year)
Less: Number of companies that have missing abnormal accruals
in Year 6 and Year 5
Remaining number of companies:
Number of firm–year observations in the period 2004–2011
(279 times 4 years, namely, Year 5, Year 6, Year 7, and Year 8)
Less: Observations with audit firm change in Years 7 or 8
Less: Observations with missing abnormal accrual values in
Years 7 or 8
Less: Observations with missing values in control variables
Remaining number of firm–year observations (273 for Year 5,
273 for Year 6, 197 for Year 7, and 144 for Year 8):
Panel B: Distribution of mandatory rotation years:
2004
2005
2006
2007
2008
2009
Total
23
companies that change audit firms during the Year 1 to Year 6 peri-
od.19 We then exclude 7 companies which rotated only one audit
partner when both partners should have been replaced after the
mandated maximum five-year tenure, as well as 6 companies with
missing accrual data during Years 5–6. We are left with 273 remain-
ing unique client companies (or 1092 firm–year observations) after
completing these data screening procedures.
As described above, our sample selection procedure requires client
companies to have the same audit firms in the five-year period (Years
1 to 5) preceding the identified mandatory rotation year (Year 6). In ad-
dition, in the two-year period subsequent to the MPR year, if there is an
audit firm change in Year 7, we eliminate the corresponding Year 7 and
Year 8 observations. Alternatively, if there is no audit firm change in
Year 7 but there is one change in Year 8, we keep the Year 7 observation
and eliminate only the Year 8 observation. In this way, we attempt to re-
move any confounding effect of audit firm change in our empirical anal-
yses; we also try to restrict loss of power of tests arising from small
sample size by not requiring the inclusion of all three post-MPR years
(Years 6, 7, and 8) in the sample selection procedure.20 Our final sample
has 887 firm–year observations, including 273 respectively for each of
Year 5 and Year 6, 197 for Year 7, and 144 for Year 8 (see Panel B of
Table 1).
Note that we identify Year 6 as the first year of a client company's
first-time mandatory audit partner rotation after the adoption of the
MPR rule. As in Australia and Taiwan, Chinese audit reports have two
signing audit partners, namely, a lead engagement partner and a
reviewing partner. The rule allows a one-year postponement for one
of the two signing auditors if both signing auditors reach their five-
year tenure at the same time (Chinese Securities Regulatory
Commission (CSRC Regulation) & China Ministry of Finance (MOF),
2003 No. 13). For example, consider a client with two signing audit part-
ners, one of them having audited the client for five years, but the other
for only four years. According to the mandatory rotation rule, the first
auditor is required to rotate out in Year 6, but the second auditor is
No. of Obs. allowed to audit the client for one more year. Therefore, for this case,
13,287 the second mandatory rotation happens in Year 7. We identify the
(6193) first year of the client firm's MPR as Year 6; we treat the second MPR
year as Year 7. It is also possible that audit partner changes in Years 7
and 8 that take place before the expiry of the MPR imposed maximum
five-year tenure are voluntary. In order to address any confounding ef-
7094 fects arising from multiple audit partner rotations in post-MPR years,
887
we eliminate those observations that are associated with audit partner
455 changes (71 voluntary and 47 mandatory) in Years 7 and 8 in our sensi-
tivity checks. Our results hold.
(169)
4.2. Audit quality measures
(7)
Following the literature, we use an earnings quality measure, name-
ly, discretionary (abnormal) accruals, as a proxy of audit quality. Prior
(6) research shows a positive relation between measures of high quality
audit (such as audit firm size or industry expertise) and high quality fi-
273
1092
nancial reporting (Balsam, Krishnan, & Yang, 2003; Ghosh & Moon,
2005; Johnson, Khurana, & Reynolds, 2002; Krishnan, 2003; Myers,
(202) Myers, & Omer, 2003). The underlying argument is that high quality au-
(2) ditors are capable of detecting questionable accounting practices and
misrepresentations as reflected in discretionary accruals. If managers
(1)
887 are unwilling to address the auditor's concerns about their discretion
in financial reporting, high quality auditors are more likely than other
54 19
We eliminate client companies with audit firm changes to isolate the effect of manda-
72
tory audit partner rotation. Companies can change their audit firms for many reasons. For
63
example, Johnson and Lys (1990) show that audit firm changes are related to client char-
39
acteristics and audit firm cost structures. Schwartz and Menon (1985) report that client
33
firms with poor performance may switch their audit firms on account of disputes regard-
12
ing audit opinions, reporting issues, management changes, and audit fees.
273 20
We thank our anonymous reviewer for this suggestion.
24 S.P. Bandyopadhyay et al. / Advances in Accounting, incorporating Advances in International Accounting 30 (2014) 18–31
auditors to issue qualified audit reports (see DeAngelo, 1981). Myers
et al. (2003), page 783) argue that “high quality audits mitigate
more extreme management reporting decisions, and suggest that
accruals can be used to identify these extreme reporting decisions.”
Empirical studies with Chinese data provide evidence that discre-
tionary accruals measures are capable of capturing earnings manip-
ulation behavior in China (Ting, Yen, & Huang, 2009; Yu, Du, & Sun,
2006).
Consistent with prior research on MPR and audit quality (Chen et al.,
2008; Chi et al., 2009), we measure discretionary accruals using the
modified Jones model (Dechow, Sloan, & Sweeney, 1995) matched by
performance (Kothari, Leone, & Wasley, 2005). For simplicity, firm sub-
script is omitted in Eq. (1) and later equations.
TotalAccuralst =Assetst−1 ¼ α0 þ α1 ð1=Assetst−1 Þ þ β1 ðΔSt−ΔARt Þ=Assetst−1
β2 ðPPEt−1 =Assetst−1 Þ þ β3 ROAt−1 þ εt ð1Þ
where total accruals (TotalAccruals) equal net income before extraordi-
nary items minus operating cash flows, ΔS is change in sales, ΔAR is
change in account receivables, PPE is net property, plant and equipment,
ROA is return on assets, and t is the year subscript. All variables are
scaled by lagged assets (Assetst − 1).
We estimate Eq. (1) using the full CCER sample with non-missing
values of the regression variables by industry–year. We require at
least 15 valid observations for each industry–year regression.
The industry category follows the 13-industry codes announced
by CSRC. The mean coefficients of industry–year regressions are
used to estimate the fitted values of non-discretionary accruals.
Discretionary accrual (DA) is computed as the total accruals minus
the fitted values. Following Carey and Simnett (2006), we also use
discretionary working capital accruals (DWCA) in our sensitivity
test.
Extant literature uses a number of different measures of audit qual-
ity, for example, absolute values (unsigned) of discretionary accruals,
signed values, or positive/negative values of discretionary accruals. In
this paper, we focus on the signed value of discretionary accruals for
two reasons. The first reason follows from the results of Hribar and
Nichols (2007) indicating that firm characteristics such as operating
volatility are related to the error variance in discretionary accruals.
These authors argue that this correlation could weaken the statistical
power of detecting low quality earnings using the unsigned (or abso-
lute values of) discretionary accruals measure. Besides Hribar and
Nichols (2007), Carey and Simnett (2006) and Francis et al. (2013)
also focus on the signed discretionary accruals in their analysis
of audit quality. The second reason is regulatory; Chinese-listed
companies have disincentives to recognizing extreme income-
decreasing accruals arising from CSRC-imposed regulations for eligi-
bility for rights offerings. The most rigid regulation is the minimum
ROE requirement.21 This regulation is widely applicable because al-
most all listed firms tend to seek permission to undertake rights of-
ferings (Chen & Yuan, 2004).
21
The requirement for rights offerings was changed a number of times, reflecting the
CSRC's efforts to protect shareholders against management expropriation. For example,
rights offerings required two consecutive years of profits after December 1993, and a
three-year average ROE not below 10% after September 1994. The rules changed in Janu-
ary 1996, which required that for each of previous three years prior to a rights offering, the
incumbent firms' three-year average ROE must not fall below 10%. This rule was changed
again in March 1999, which required the maintenance of a minimum ROE level of not be-
low 6% for each of previous three years before rights offerings. After May 2006, these firms
needed to maintain a minimum three-year weighted average ROE of 6%. For regulations
on rights offerings, see the CSRC's regulation released on May 8, 2006, available on the of-
ficial webpage http://www.csrc.gov.cn.
4.3. Research models
We estimate the following equation to compare the audit quality of
Year 5, the last year prior to MPR, versus the three post-mandatory rota-
tion Years: 6, 7, and 8.
DA ¼ α0 þ α1 POST þ β1 IBIG4 þ β2 CIBIG10 þ β3 FT þ β4 SIZE þ β5 CFO
þβ6 GROW þ β7 AGE þ β8 SOE þ β9 LEV þ β10 TTYPE þ β11 EXP
þβ12 CLIENTP þ δ Year þ γIndustry þ ε ð2Þ
where DA is discretionary accruals. POST is an indicator variable that
equals one for Years 6, 7, and 8 and zero for Year 5.
The constant term (α0) represents audit quality of Year 5, after con-
trolling for covariates. A negative coefficient (α1) on the POST variable is
consistent with audit quality improvement in the post-rotation period
relative to the benchmark last pre-rotation year (Year 5), after control-
ling for covariates. A positive coefficient of POST indicates lower audit
quality after MPR.
Following prior research (Carey & Simnett, 2006; Chen et al., 2008;
Chi et al., 2009; DeFond, Raghunandan, & Subramanyam, 2002), we in-
clude several control variables in Eq. (2). Audit firm size variables
IBIG4 and CBIG10 equal one respectively for Big 4 international audit
firms and Top 10 Chinese audit firms22 and zero otherwise. FT is audit
firm tenure measured as the number of years that a client firm is audited
by the same audit firm.23 SIZE is client company size computed as the
natural logarithm of total assets. Audit firm size and audit firm tenure
control for the audit quality effects of audits by large audit firms and
by audit firms with a long tenure with its incumbent client. We include
cash flow variable CFO (operating cash flows scaled by lagged assets) be-
cause prior research shows that it is negatively associated with accruals
(Chi et al., 2009). GROW is growth in sales, measured as the total sales of
the current year scaled by the industry total sales of the last year. AGE is
the number of years that the client firm is listed in a Chinese stock ex-
change, which controls for the propensity of relatively younger compa-
nies in high-growth industries to recognize extreme accruals.
In addition, we include the SOE variable, which equals one, for state-
owned enterprises, and zero otherwise. Chen, Chen, Lobo, and Wang
(2011) argue that SOE companies have less incentive to manage earn-
ings relative to non-SOE companies because the differences in the na-
ture of the ownership, agency relations, and bankruptcy risks.
However, we do not predict the sign of the SOE variable. Aharony, Lee,
and Wong (2000) argue that while SOE managers do not have the
same incentives as US managers to manage earnings since they do not
own any shares of the firm, they might still earn high prestige and
other non-pecuniary benefits from reporting higher earnings by under-
taking earnings management activities. We also include a leverage var-
iable (LEV) to capture the effects of risk associated with high levels of
debt (Carey & Simnett, 2006). LEV equals total liability scaled by total
assets. TTYPE is an indicator variable for special treatment stocks to cap-
ture earnings incentives arising from potential delisting risks.24 We do
22
The top 10 Chinese audit firms are selected based on the ranking prepared by CICPA in
2009. Chen, Chen, Lobo, and Wang (2011) use Top 8 to categorize big audit firms in their
study.
23
Audit data in CCER become available in 2000. Therefore, we count the number of years
that an audit firm serves a client (audit firm tenure) starting in 2000. For observations with
MPR year in 2004, we manually check 1999 audit firm data to ensure that audit firms are
the same in 1999.
24
ST/PT/ ∗ ST firms are firms that report consecutive losses and face substantial delisting
risk. According to the rules introduced by the CSRC in 1999, a firm is designated as a special
treatment (ST) firm if it incurs losses for two consecutive years and a particular treatment
(PT) firm if it continues to report a loss for another year. A PT firm is delisted if it fails to
become profitable in the following year. PT stocks cannot be traded except on Fridays
and are limited to a maximum 5% price increase over the last Friday's close (no downside
limit). The PT designation has been discontinued by the CSRC in 2002. Since 2003, the
CSRC introduced a new designation called “*ST”, which is similar to ST, to further advise
the market of the risk that a company will be delisted if its loss continues the following
year. If a firm incurs losses for three consecutive years, it is de-listed.
 S.P. Bandyopadhyay et al. / Advances in Accounting, incorporating Advances in International Accounting 30 (2014) 18–31 25

Table 2
Variable distributions and correlation matrix.

Panel A: variable distributions

Variable Mean Std. dev. Min. 25% Median 75% Max.

DA 0.001 0.091 −0.387 −0.043 0.006 0.042 0.706

DWCA −0.140 5.594 −0.584 −0.063 0.004 0.075 0.679
IBIG4 0.054 0.226 0 0 0 0 1
CBIG10 0.157 0.364 0 0 0 0 1
HERF 0.202 0.152 0.071 0.125 0.155 0.217 0.982
LEGAL 5.951 1.230 2.620 5.050 5.630 6.980 7.970
FT 6.515 1.387 4 5 6 8 9
SIZE 21.515 0.990 17.967 20.831 21.472 22.147 25.585
CFO 0.059 0.099 −0.645 0.014 0.054 0.105 0.601
GROW 1.266 0.205 0.787 1.161 1.212 1.357 3.284
AGE 9.079 3.036 0 7 9 11 18
SOE 0.661 0.474 0 0 1 1 1
LEV 0.540 0.366 0.021 0.389 0.544 0.657 7.788
TTYPE 0.159 0.366 0 0 0 0 1
EXP 0.020 0.037 0 0.002 0.006 0.018 0.297
CLIENTP 0.045 0.066 0 0.009 0.022 0.056 0.598

Panel B: Pearson correlation matrix

Variable DA DWCA HERF LEGAL FT SIZE CFO GROW AGE LEV EXP

DWCA 0.164
HERF −0.010 0.014
LEGAL −0.065 0.020 0.194
FT −0.041 0.020 0.129 0.026
SIZE 0.087 −0.011 0.053 0.175 0.197
CFO −0.639 −0.017 −0.042 −0.073 0.020 0.096
GROW −0.007 0.021 −0.004 0.015 −0.071 0.081 −0.008
AGE −0.027 0.013 −0.007 0.171 0.397 −0.003 0.003 −0.033
LEV −0.160 −0.097 0.049 −0.053 0.043 −0.032 −0.076 −0.066 0.082
EXP −0.002 0.007 0.121 0.134 −0.103 0.000 −0.045 −0.085 −0.071 0.145
CLIENTP 0.021 0.059 −0.120 −0.105 −0.156 −0.008 0.049 −0.034 −0.067 0.026 −0.033

See Appendix 2 for variable definitions. Bold numbers in Panel B are correlations significant with p b 0.05 (including b0.01). EXP (CLIENTP) variable values in Panel B are the residuals from
the first-stage regression EXP (CLIENTP) on SIZE (see Section 4.3 for the details of the two-stage method).

not have predictions for LEV and TTYPE. While financially distressed
firms with high debt and delisting risk may be aggressive in accruals
recognition, these firms may have to reduce instances of earnings man-
agement because of potentially restrictive debt covenants for high le-
verage firms and/or the tight levels of regulations on TTYPE companies.
The EXP variable measures the two incumbent audit partners' com-
bined experience in auditing clients in the incumbent client's industry.
An audit partner's industry experience is measured as the percentage
of his/her audited client companies' assets in the same industry. Note
that the EXP score associated with the rotated-in partner(s) compen-
sates partially for the audit expertise that is lost with the departing part-
ner(s). As mentioned earlier, the effects of MPR on audit quality reflect a
tradeoff of enhanced audit independence versus loss of client-specific
audit expertise. After controlling for EXP (and effects of covariates),
the indicator variable POST captures the change in post-MPR audit qual-
ity arising from potentially improved audit independence.
The CLIENTP variable measures client power, which equals the per-
centage of a client's total assets as a percentage of its audit firm's total
audited assets countrywide. This variable and EXP are both highly corre-
lated with the SIZE variable, with correlation coefficients of around 0.50
(un-tabulated). In an attempt to reduce the potential effects of
multicollinearity, we use a two-stage approach. We first run two regres-
sions, namely EXP on SIZE, and CLIENTP on SIZE. We derive the residuals
of each regression by taking the actual EXP (CLIENTP) value minus its
fitted value. Then we use the residual values, which are orthogonal to
SIZE values, to estimate Eq. (2). We control for Industry (Industry)
and year (Year) fixed effects in all regressions. Appendix 2 provides de-
tails on the measurement of all variables.
To assess the effect of MPR on audit quality in each of the individual
post-MPR years, we revise Eq. (2) by replacing the POST variable with
three indicator variables Y6, Y7, and Y8, that equals one, respectively,
for Years 6, 7, and 8, and zero for pre-rotation period (Year 5).
DA ¼ α0 þ α1 Y6 þ α2 Y7 þ α3 Y8 þ β1 IBIG4 þ β2 CBIG10 þ β3 FT
þβ4 SIZE þ β5 CFO þ β6 GROW þ β7 AGE þ β8 SOE þ β9 LEV ð3Þ
þβ10 TTYPE þ β11 EXP þ β12 CLIENT þ δ YEAR þ γIndustry þ ε
5. Data descriptive statistics and research findings
5.1. Data descriptive statistics
Table 2 Panel A reports the descriptive statistics of the selected sam-
ple. The means of different indicator variables show that 5.4% of the
sample observations are audited by Big 4 international audit firms
(IBIG4), and 15.7% by the Top 10 largest Chinese audit firms (CBIG10);
66.1% reflect state ownership (SOE) of client firms; and finally, 15.9%
were listed as special treatment stocks or particular transfer stocks
(TTYPE) at some point by the CSRC in Year 5 and/or Year 6. The average
firm tenure (FT) is approximately 6.5 years. The average leverage ratio
(LEV) is 54.0%, with a median of 54.4%. The minimum and maximum
values of discretionary accruals (DA), discretionary working capital ac-
cruals (DWCA), and other variables do not exhibit obvious outliers.25
Table 2 Panel B reports the Pearson correlation matrix of our variables.
25
To address the effects of potentially influential observations, we winsorize variable
values at the levels of 1% and 99%. We also use r-student screening (Belsley, Kuh, & Welsch,
1980) in all our regressions by eliminating observations with r-student absolute values
greater than 4 in the regressions.
26 S.P. Bandyopadhyay et al. / Advances in Accounting, incorporating Advances in International Accounting 30 (2014) 18–31

Table 3
Audit Quality Pre versus Post-MPR.

Pred. sign DA DA (DA N 0) DA (DA b 0) DA

Coef. (t-value) Coef. (t-stat) Coef. (t-stat) Coef. (t-stat)

POST +/− −0.012*** (−2.54) −0.059** (−1.97) −0.001 (−0.12)

Y6 +/− −0.008 (−0.99)
Y7 +/− −0.014*** (−2.66)
Y8 +/− −0.017*** (−2.63)
IBIG4 − −0.016* (−1.77) −0.093** (−2.37) 0.001 (0.08) −0.016* (−1.78)
CBIG10 − −0.006 (−1.04) −0.041 (−1.43) −0.001 (−0.06) −0.006 (−0.99)
FT − −0.001 (−0.20) 0.007 (0.53) −0.001 (−0.20) −0.001 (−0.31)
SIZE + 0.014*** (5.80) 0.039*** (2.98) 0.002 (0.34) 0.014*** (5.52)
CFO − −0.637*** (−18.85) −0.898*** (−6.96) −0.582*** (−13.22) −0.607*** (−18.05)
GROW + 0.001 (0.06) −0.114 (−1.33) 0.029 (0.83) −0.005 (−0.26)
AGE − 0.001 (1.52) 0.005 (1.12) 0.001 (0.37) 0.011* (1.64)
SOE +/− 0.009*** (2.15) 0.041* (1.78) 0.000 (−0.03) 0.008* (1.75)
LEV ? −0.080*** (−6.09) −0.384*** (−5.24) −0.015 (−0.73) −0.080*** (−6.00)
TTYPE ? 0.014** (2.04) 0.041 (1.35) 0.030*** (2.62) 0.014** (1.97)
EXP − −0.035 (−0.44) 0.102 (0.28) −0.151 (−0.92) −0.073 (−0.89)
CLIENTP + 0.063 (1.56) −0.091 (−0.41) 0.109 (1.58) 0.058 (1.42)
Constant ? −0.240*** (−4.24) −0.435 (−1.48) −0.033 (−0.31) −0.237*** (−3.99)
Obs. no. 879 479 408 881
Adj. R2 (%) 54.01 39.87 46.87 53.29

See Appendix 2 for variable definitions. All regressions control for industry and year fixed effects. Truncated regressions were used in the positive and negative abnormal accrual
regressions. *, **, *** significant respectively at 10%, 5%, and 1% levels with two-tailed tests.

DA and DWCA are positively correlated. DA is negatively correlated with and Francis et al. (2013). Our positive coefficient on SOE is consistent
HERF, LEGAL, CFO, and LEV. with SOE companies having incentives to use income-increasing ac-
cruals to boost bottom line net incomes (Aharony et al., 2000). We
5.2. Results of H1 find some evidence of high audit quality associated with Big 4 interna-
tional firms' audits. We also find that high leverage clients report
Table 3 reports the regression results of estimating Eqs. 2 and 3. Note more income-decreasing accruals and that there is some evidence that
that all our tests are two-tailed t-tests because our hypotheses are non- TTYPE clients are more aggressive in recognizing income-increasing
directional. Standard errors are clustered at individual client firm level accruals.
since discretionary accruals are firm-specific.26 Using signed discretion-
ary accruals (DA) as the dependent variable in Eq. (2), we find
5.3. Results of H2 and H3
that the indicator variable, POST, has a negative coefficient of −0.012
(t-stat = −2.54) significant at less than 5% level. This result is consis-
Hypothesis H2 examines whether the MPR's effect on audit qual-
tent with the notion that audit quality in the post-MPR years is superior
ity in provinces with less concentrated audit markets is different
to that in the last year prior to the start of MPR.
from that in other provinces. To test H2, we compute the Herfindahl
We also estimate truncated regressions respectively for positive
index for each client-province as the sum of squared audit firm fee
DA and negative DA observations. We find that a positive DA subsample
shares of all audit firms in a province. We estimate Eq. 2 for two sub-
has less income-increasing abnormal accruals in the POST period
samples grouped by the observations that reflect clients located in
(α3 = −0.059, t-stat = −1.97, p b .10), but a negative DA subsample
provinces with higher than the median provincial Herfindahl index
exhibits no significant change in audit quality in the post-MPR period.
versus those with lower than the median provincial Herfindahl
This is consistent with sample firms not undertaking “big baths” (poten-
index.28
tial reasons for firms recognizing negative accruals) in the post-MPR pe-
Table 4 summarizes the regression estimates of Eq. (2) for clients lo-
riod. For Eq. 3, while the coefficient on the first rotation year (Y6)
cated in provinces that exhibit Herfindahl indices less than the median
is insignificant, both Year 7 (Y7) and Year 8 (Y8) coefficients are
level (low AMC) versus more than the median level (high AMC). The re-
negative and significant (α2 = −0.014, t-stat = −2.66; α3 = −0.017,
sults show that the POST variable is insignificant for the high AMC group
t-stat = −2.63) at less than the 1% level. Collectively, these results indi-
but significant at less than the 5% level for the low AMC group using
cate that client companies report less extreme income-increasing abnor-
two-tailed tests (coefficient = − 0.017, t-stat = − 2.30). The control
mal accruals after MPR.27
variables have similar coefficients across the two regressions. SIZE and
The control variables in Table 3 are generally consistent with our
LEV are both significant at less than 1% level.
predictions. Specifically, our findings of positive coefficients on SIZE
We also examine the stand-alone effect of legal environment on the
and negative coefficient on CFO are consistent with Chi et al. (2009)
relation between MPR and audit quality (Firth et al., 2012a). The provin-
cial legal environment index scores are taken from Wang et al. (2008),
Table A1). These authors report the levels of legal development index
26
We use the SAS programming codes provided by Noah Stoffman, Kelley School of
Business, Indiana University (http://kelley.iu.edu/nstoffma/fe.html) to run regressions
with clustered standard errors and fixed effects models. While the degree of freedom is
28
much less after controlling for (client) firm effects in the fixed effects model, our results We also estimate Eq. (3) using the same sample decompositions by median of the me-
(un-tabulated) hold. dian of provincial Herfindahl index, and the median of legal development index. Consis-
27
The truncated regressions do not report the adjusted R-Squared. When using OLS re- tent with Eq. (2) results reported in Table 4, our results (un-tabulated) suggest audit
gressions, the adjusted R-Squared is respectively 39.87% and 46.87% for the income- quality improvement in the post-MPR period, especially in Years 7 and 8, for the low
increasing and income-decreasing accrual regressions. Herfindahl index and low legal development index groups, respectively.
 S.P. Bandyopadhyay et al. / Advances in Accounting, incorporating Advances in International Accounting 30 (2014) 18–31 27

Table 4
Audit quality pre versus post-MPR: conditioning on AMC or legal environment.

Pred. DA DA DA DA

Sign Low AMC High AMC Low Legal High Legal

Coef. (t-value) Coef. (t-stat) Coef. (t-stat) Coef. (t-stat)

POST +/− −0.017** (−2.30) −0.009 (−1.46) −0.017*** (−2.47) −0.002 (−0.25)
IBIG4 − −0.040** (−2.23) 0.004 (0.46) −0.019 (−1.49) −0.021 (−1.50)
CBIG10 − −0.017 (−1.23) 0.002 (0.38) −0.008 (−0.87) −0.003 (−0.36)
FT − 0.002 (0.47) −0.002 (−0.66) 0.002 (0.53) −0.007* (−1.69)
SIZE + 0.014*** (4.17) 0.013*** (3.61) 0.015*** (4.46) 0.014*** (3.92)
CFO − −0.584*** (−9.75) −0.652*** (−18.83) −0.600*** (−13.33) −0.730*** (−12.93)
GROW + −0.042 (−1.24) 0.055** (2.06) 0.024 (0.69) −0.012 (−0.45)
AGE − 0.001 (0.98) 0.001 (0.83) 0.001 (0.85) 0.001 (1.50)
SOE +/− 0.013** (2.14) 0.005 (0.78) 0.012* (1.83) 0.003 (0.49)
LEV ? −0.080*** (−4.34) −0.080*** (−4.06) −0.098*** (−4.97) −0.074*** (−4.21)
TTYPE ? 0.025*** (2.60) 0.003 (0.37) 0.013 (1.31) 0.028** (2.37)
EXP − −0.117 (−1.12) −0.005 (−0.04) −0.044 (−0.27) 0.112 (1.06)
CLIENTP + 0.160*** (2.99) 0.005 (0.08) −0.001 (−0.02) 0.169*** (2.81)
Constant ? −0.214*** (−2.65) −0.265*** (−3.25) −0.273*** (−3.26) −0.223*** (−2.69)
Obs. no. 448 433 454 426
Adj. R2 (%) 49.05 59.73 45.22 67.03

See Appendix 2 for variable definitions. Low and high AMC regressions are respectively for the two subsamples of below or above median audit market concentration levels measured by
Herfindahl index (HERF). Low and high legal regressions are respectively for the two subsamples of below or above median legal environment index (LEGAL). All regressions control for
industry and year fixed effects. *, **, *** significant respectively at 10%, 5%, and 1% levels with two-tailed tests.

of 30 Chinese provinces' comprising of the following sub-indices, name-
ly, (1) the number of lawyers as a percentage of the provincial popula-
tion, (2) the efficiency of local courts (percentage of lawsuits
pursued by the courts), and (3) protection of property rights.29 As
Table 4 summarizes, POST is significantly negative (coefficient =
− 0.017, t-stat = − 2.47) only for the low legal development
provinces (i.e., those with legal indices less than median index
for the country) but not for the high legal development group. The re-
sults, consistent with Firth et al. (2012a), suggest that MPR is less like-
ly to improve audit quality in high AMC or high levels of legal
development audit markets, but is more effective in provinces low
levels of AMC or low levels of legal environment.
In the foregoing regressions, we use the median split to estimate
separate regressions for high versus low AMC subsamples and also
high versus low legal development subsamples. The reason for estimat-
ing separate regressions is that the relation between audit quality and
control variables (e.g. IBIG4) likely varies with high versus low AMC
and legal development. For example, in Table 4, the coefficients of
IBIG4 and CLIENTP are significant in the low AMC group but not in the
high AMC group. Consequently, estimating one single regression
would force equality of coefficients on the control variables in this re-
gression that could potentially bias the coefficient of POST, the post-
MPR year indicator variable.
To test hypothesis H3, we estimate the effect of MPR on audit quality
with provincial AMC in interaction with the provincial legal develop-
ment in empirical tests. Table 5 presents the regression results of the
two groups with low AMC and low legal development (low–low) versus
those with high AMC and high legal development (high–high) based on
medians of these variables. To save space, we only report the results of
Eq. (3) with the POST variable to capture the overall change in audit
quality after MPR. The regression results are similar when we substitute
Y6, Y7 and Y8 indicator variables for POST in regression models (results
un-tabulated). For the low–low group, the POST variable is significantly
negative at less than 1% level (coefficient = −0.027, t-stat = −2.74).
The results from the high–high group show insignificant change in
audit quality in the post-MPR period. The results (un-tabulated) are
also insignificant for the two middle groups, namely, the low AMC and
high legal development and the high AMC and low legal development
provinces.
We also estimate a single regression with the full sample where
POST interacts respectively with the AMC variable (LHERF) and the
legal environment variable (LLEGAL). The regression includes a
two-way interaction of LHERF ∗ LLEGAL, and a three-way interaction
of POST ∗ LHERF ∗ LLEGAL. In order to simplify the interpretation of
the coefficient signs, we reverse decile rank both the provincial
Herfindahl indices and the provincial legal development indices.
For example, the variable LHERF is set to one for provincial
Herfindahl indices that are included in the lowest decile of observa-
tions. Conversely, LHERF is set to zero for Herfindahl indices falling in
the highest decile rank. LHERF takes values between zero and one for
the intervening Herfindahl indices.30 We repeat this procedure for
the legal development index variable, LLEGAL. Decile ranking pro-
vides more information in the variable measures relative to binary
indicator measures.
Consistent with our decile ranking approach, in the following dis-
cussions, LL provinces are those that have the lowest AMC decile
rank and the lowest legal development index decile rank as well.
For these provinces, both LHERF and LLEGAL take the value of one.
Similarly, the HH provinces reflect the highest AMC decile and
highest legal development index decile, with LHERF and LLEGAL
both taking the value of zero.31
Table 5 shows that audit quality improves in the LL (LHERF = 1 and
LLEGAL = 1) provinces in the post-MPR period, but not in the HH
provinces (LHERF = 0 and LLEGAL = 0). Note that the change in the
(pre- versus post-MPR) audit quality in LL provinces, after controlling
for covariates, equals the sum of coefficients of POST + POST ∗ LHERF +
POST ∗ LLEGAL + POST ∗ LHERF ∗ LLEGAL. Since the coefficient on POST
is insignificantly different from zero, and all the other foregoing

29
The index is obtained from the National Economic Research Institute (NERI) Index of
Marketization of China's provinces in 2000 to measure the quality of market-supporting
institutions at the provincial level. The NERI Index project was sponsored by the National
Economic Research Institute and the China Reform Foundation and conducted by Fan,
Wang, and Zhu (2004).
30
Setting the lowest value to zero and the highest value to one is convenient for the in-
terpretation of coefficients. The intervening eight values are set to 1/9, 2/9, …, up to 8/9
consistent with the Proc Rank procedure under SAS.
31
Note that in Table 4, we use median-based ranks to define low–low/high–high groups;
in Table 5, we used decile-based ranks to define LL/HH groups.
28 S.P. Bandyopadhyay et al. / Advances in Accounting, incorporating Advances in International Accounting 30 (2014) 18–31

Table 5
Audit quality pre versus post-MPR: conditioning on AMC/legal environment.

D.V. = DA Pred. Sign Low–Low High–High Full Sample

Coef. (t-stat) Coef. (t-stat) Coef. (t-stat)

POST +/− −0.027*** (−2.74) 0.003 (0.32) 0.000 (0.03)

POST ∗ LHERF +/− −0.022* (−1.80)
POST∗ LLEGAL − −0.014** (−2.31)
POST∗ LHERF∗ LLEGAL +/− −0.011*** (−2.60)
LLEGAL + 0.016** (2.37)
LHERF + 0.021*** (2.91)
LHERF∗ LLEGAL + 0.019*** (2.51)
IBIG4 − −0.051** (−2.16) −0.037** (−2.34) −0.018* (−1.86)
CBIG10 − −0.019 (−1.41) 0.002 (0.26) −0.002 (−0.41)
FT − 0.003 (0.54) −0.008 (−1.47) 0.000 (−0.05)
SIZE + 0.021*** (3.06) 0.018*** (3.82) 0.015*** (6.40)
CFO − −0.623*** (−8.96) −0.810*** (−17.34) −0.665*** (−22.80)
GROW + −0.086* (−1.83) 0.041* (1.74) 0.001 (0.03)
AGE − 0.002 (1.36) 0.002* (1.72) 0.001* (1.67)
SOE +/− 0.021* (1.90) 0.003 (0.36) 0.009** (2.13)
LEV ? −0.127*** (−4.45) −0.087*** (−3.44) −0.084*** (−6.49)
TTYPE ? 0.034** (2.45) 0.028 (1.52) 0.012* (1.80)
EXP − 0.266 (0.72) 0.192 (1.13) −0.018 (−0.23)
CLIENTP + 0.035 (0.33) 0.042 (0.53) 0.053 (1.31)
Constant ? −0.303* (−1.79) −0.451*** (−4.08) −0.292*** (−5.08)
Obs. no. 232 211 876
Adj. R2 (%) 47.68 79.2 56.32
Chi-Square
POST + POST∗ LHERF + POST∗ LLEGAL + POST∗ LHERF∗ LLEGAL −0.047*** 7.69
LHERF + LLEGAL + LHERF∗ LLEGAL 0.056* 2.85
POST + POST∗ LHERF + POST∗ LLEGAL + POST∗ LHERF∗ LLEGAL 0.009** 3.58
+LHERF + LLEGAL + LHERF∗ LLEGAL

See Appendix 2 for variable definitions. Low–Low and High–High regressions are respectively for the two subsamples with both audit market concentration (HERF) and legal environment
index (LEGAL) levels below or above their median values. Variable LHERF is reversely ranked HERF deciles scaled between zero and one, with zero as the highest decile and one as the
lowest decile HERF values. Similarly, LLEGAL is reversely ranked LEGAL values scaled between zero and one, with zero as the highest decile and one as the lowest decile LEGAL values.
All regressions control for industry and year fixed effects. *, **, *** significant respectively at 10%, 5%, and 1% levels with two-tailed tests.

coefficients are negative and significant, these results are consistent with
smaller income-increasing discretionary accruals (or improved accrual
quality) for the LL provinces after MPR. The sum of these coefficients is
− 0.047 and significantly different from zero (Chi-Square = 7.69,
p b 0.001). The difference in (pre- versus post-MPR) audit quality
in HH provinces is reflected by the coefficient on the POST variable,
which is insignificantly different from zero, indicating no change in
the audit quality of these provinces in the post-MPR period, after
controlling for covariates.
Moreover, Table 5 also shows that the audit quality in HH
provinces is superior to that in LL provinces, both before and after
MPR is introduced. Specifically, the sum of the coefficients of
LLEGAL + LHERF + LHERF ∗ LLEGAL reflects the difference in the
audit quality (DA) of LL provinces relative to that of the HH prov-
inces in Year 5 (the last year prior to MPR), after controlling for co-
variates. All these coefficients are positive and significant,
and their sum is 0.056 and significantly different from zero (Chi-
Square = 2.85, p b 0.10). This indicates that LL provinces exhibit lower

Table 6
Audit quality pre versus post-MPR: sensitivity tests.

Low–Low High–High

Coef. (t-stat) Coef. (t-stat)

Alternative audit quality measure

D.V. = DWCA −0.060* (−1.84) −0.273 (−0.90)
Obs. no. 230 203

Alternative sample
Add Years 3 and 4 data −0.025*** (−2.92) 0.009 (1.12)
Obs. No. 387 309
Remove Years 7 and 8 audit partner change obs. −0.029*** (−2.96) 0.002 (0.30)
Obs. no. 205 176

Alternative Herfindahl indices

Client sales revenue by province–year −0.016* (−1.72) −0.002 (−0.22)
Obs. no. 202 193
Audit fees within Big 4 international accounting firms by province–year −0.020** (−2.40) 0.005 (0.44)
Obs. no. 213 195
Audit fees within Big 10 domestic accounting firms by province–year −0.036*** (−4.30) 0.002 (0.25)
Obs. no. 211 179

See Appendix 2 for variable definitions. Low–Low and High–High regressions are respectively for the two subsamples with both audit market concentration (HERF) and legal environment
index (LEGAL) levels below or above their median values. To save space, Table 6 summarizes the coefficients of only POST variables in alternative sensitivity check regressions. All regres-
sions control for industry and year fixed effects. *, **, *** significant respectively at 10%, 5%, and 1% levels with two-tailed tests.
 S.P. Bandyopadhyay et al. / Advances in Accounting, incorporating Advances in International Accounting 30 (2014) 18–31 29

audit quality (greater income-increasing DA values) relative to HH prov- 6. Conclusions

inces in Year 5. In the post-MPR period, the difference in audit quality
between LL versus HH provinces, after controlling for covariates, Our research examines audit quality in China in the years immedi-
is the sum of coefficients of LLEGAL + LHERF + POST ∗ LHERF + ately following mandatory audit partner rotations. Using a sample of
POST ∗ LLEGAL + LLEGAL ∗ LHERF + POST ∗ LLEGAL ∗ LHERF. The sum 273 Chinese firms that are subject to mandatory audit partner rotations,
of these coefficients is 0.009 and significantly different from zero (Chi- we provide evidence of the consequences of mandatory audit partner
Square = 3.58, p b 0.001). This indicates that MPR helps narrow, but rotation on discretionary accruals, a proxy of earnings (audit) quality.
does not eliminate, the difference in audit quality between the LL and We find that while audit quality in the first post-rotation year (Year 6)
HH provinces. Note that while the difference in DA values between the is not significantly different from that in the final pre-rotation year
LL and HH provinces decreases from 0.056 in Year 5 to 0.009 after MPR, (Year 5), audit quality does improve significantly in the two subsequent
the audit quality of LL provinces is still lower. years (Years 7 and 8). This is a new result because prior research on
In summary, Table 5 results suggest that in the post-MPR period, audit partner rotation focuses on changes to audit quality only in the
audit quality improves the most for client firms in low AMC and low first year after mandatory rotation and largely finds no evidence of im-
legal development provinces. Firth et al. (2012a) document the effective- provement in audit quality. We extend the focus of our study to three
ness of MPR in low legal development provinces in China. In contrast, we years post-rotation, on the ground that new rotated-in audit partners
show that AMC and legal development both incrementally moderate the likely face a learning curve in a new audit engagement after taking
effect of MPR on audit quality. Our results imply that the positive effect of over clients from incumbent audit partners. We document an improve-
MPR on audit quality occurs to clients in legally less developed provinces ment in audit quality following mandatory audit partner rotation, par-
that have low AMC, where audit quality is low to begin with. ticularly in Years 7 and 8.
We condition our tests on two factors that exhibit significant varia-
tion across different Chinese provinces. We find that audit quality in
5.4. Sensitivity analyses the post-rotation period improves in provinces that exhibit not only
low audit concentration (high audit competition) but also low levels
We check the robustness of our results with an alternative earnings of legal development. In contrast, mandatory audit partner rotation
quality measure used by Carey and Simnett (2006), namely, discretionary does not have any significant effect on clients located in provinces that
working capital accruals (DWCA). The value of DWAC in year t equals cur- exhibit high audit market concentration levels and high legal develop-
rent realized working capital (WCt) minus the expected level of working ment levels, where audit quality is higher even in the pre-rotation peri-
capital needed to support the current level of sales (WCt − 1/St − 1) ∗ St). od. Mandatory audit partner rotation helps narrow the difference in
We estimate Eq. (3) after replacing the dependent variable DA with audit quality of low concentration/low legal development provinces
DWAC. As Table 6 summarizes, the POST variable is significantly nega- versus high concentration/high legal development provinces, but it
tive (coefficient = − 0.060, t-stat = − 1.84) for the low–low prov- does not eliminate the difference completely.
inces, but insignificant for the high–high provinces grouped by the Our results are robust to our alternative audit quality measure and
median values of AMC and legal development. Therefore, our results alternative samples. The overall positive audit quality effect of mandato-
hold when we use DWAC as a measure for audit quality. ry audit partner rotation in the less legally developed provinces with
In the foregoing tables, we compare the pre-MPR Year 5 earnings' low levels of audit market concentration suggests that the rule is likely
quality with post-MPR Years 6, 7, and 8. In our sensitivity analysis, we to enhance audit quality in jurisdictions that have low audit quality to
compare the pre-MPR Years 3, 4, and 5 to post-MPR Years 6, 7, and 8 begin with and thus need it the most. Our results also indicate that the
with symmetric three-year periods both before and after MPR. This effects of MPR on audit quality cannot be generalized and must be eval-
approach addresses potential unknown bias in the last pre-MPR year uated on a case-by-case basis in different audit markets.
precisely because it is the last year for the audit partner who is
rotated off.31 Table 6 shows that our results continue to hold with a Appendix 1. Categorization of mandatory partner rotation
significantly negative coefficient on POST at less than 1% level
(coefficient = −0.025, t-stat = −2.92).
As we mentioned before, some client firms in our sample change audit
Mandatory audit partner rotation year
partners in Years 7 or 8. In our sensitivity checks, we eliminate these Years
7 and 8 observations that are associated with audit partner changes. The 2004 2005 2006 2007 2008 2009
two post-MPR years' partner rotations could be mandatory (the second 1999 A (Year 1)
audit partner also finishes five-year tenure after Year 6; 47 observations) 2000 A (Year 2) A (Year 1)
or voluntary (before finishing the five-year tenure; 71 observations). Our 2001 A (Year 3) A (Year 2) A (Year 1)
2002 A (Year 4) A (Year 3) A (Year 2) A (Year 1)
results do not change after removing these 118 observations. Table 6 2003 A (Year 5) A (Year 4) A (Year 3) A (Year 2) A (Year 1)
shows that the coefficient of the POST variable is significantly negative, 2004 B (Year 6) A (Year 5) A (Year 4) A (Year 3) A (Year 2) A (Year 1)
with p b 0.01 (coefficient = −0.029, t-stat = −2.96) for the low–low 2005 B (Year 7) B (Year 6) A (Year 5) A (Year 4) A (Year 3) A (Year 2)
provinces, but insignificant for the high–high provinces. 2006 Year 8 Year 7 B (Year 6) A (Year 5) A (Year 4) A (Year 3)
2007 Year 8 Year 7 B (Year 6) A (Year 5) A (Year 4)
Table 6 also reports the regression results using alternative AMC
2008 Year 8 Year 7 B (Year 6) A (Year 5)
Herfindahl indices. Our results hold for all three alternative 2009 Year 8 Year 7 B (Year 6)
Herfindahl index measures at the province–year level calculated 2010 Year 8 Year 7
by (1) client companies' sales revenues for all audit firms, 2011 Year 8
(2) audit fees within Big 4 international audit firms, and (3) audit
fees within the top 10 Chinese audit firms. 32 The coefficients of
POST are all significantly negative for the low–low provinces but Appendix 1 describes the method to classify an audit partner change
not the high–high. as mandatory audit partner rotation. For example, if an audit partner “A”
had audited a client in the past five consecutive years from 1999 to
31
We thank our anonymous reviewer for this suggestion. 2003, and was replaced by audit partner “B” in the sixth year (2004),
32
We also do a robustness check using the Herfindahl index computed by client compa-
nies' assets. The results (un-tabulated) hold. We derive these alternative AMC measures
we categorize this audit partner change as mandatory audit partner ro-
based on prior research, such as Wolk, Michelson, and Wootton (2001) and Francis et al. tation. The first MPR year is 2004; the last MPR year is 2009. Our post-
(2013). MPR period includes Year 6, Year 7, and Year 8.
30 S.P. Bandyopadhyay et al. / Advances in Accounting, incorporating Advances in International Accounting 30 (2014) 18–31

Appendix 2. Variable definitions

Variable Definitions

DA Discretionary accruals derived from modified Jones model (Dechow et al., 1995) matched by performance (Kothari et al., 2005), estimated with the entire CCER
TotalAccrualst =Assetst−1 ¼ α0 þ α1 ð1=Assetst−1 Þ þ β1 ðΔSt −ΔARt Þ=Assetst−1
database sample from the corresponding industry–year regression:
þβ2 ðPPEt−1 =Assetst−1 Þ þ β3 ROAt−1 þ εt
where Total Accruals equals net income before extraordinary items minus operating cash flows. DA equals the deviation of total accruals from the model predicted
values. The industry category follows the 13-industry codes announced by CSRC. Each industry–year regression requires a minimum of 15 observations.
DWCA Discretionary working capital accruals. It equals current realized working capital (WCt) minus the expected level of working capital needed to support the current
level of sales (WCt − 1/St − 1)*St.
Y6; Y7; Y8 Y6 is an indicator variable that equals one for the first post-mandatory rotation year (Year 6) and zero otherwise; Y7 equals one for Year 7 and zero otherwise; Y8
equals one for Year 8 and zero otherwise.
POST Post-mandatory rotation period indicator variable; equals one for the first three years subsequent to the mandatory rotation (Years 6, 7, and 8), and zero otherwise.
HERF Audit market concentration measured by Herfindahl index of audit firms' audit fee income at the province level. It equals the sum of squared audit fee shares of all
audit firms in a province.
LHERF Reversed decile rank in the provincial Herfindahl index (HERF). We scale the reversed rank between zero and one. LHERF equals one for provincial Herfindhal indices
that are included in the lowest decile of observations; zero for the highest decile rank.
LEGAL Legal environment index at the province level, from Table A1 of Wang et al. (2008).
LLEGAL Reversed decile rank in the provincial legal environment index (LEGAL). We scale the reversed rank between zero and one. LLEGAL equals one for provincial legal
environment indices that are included in the lowest decile of observations; zero for the highest decile rank.
SOE An indicator variable that equals one if a company is a state-owned enterprise and 0 otherwise.
IBIG4 An indicator variable of big audit firms that equals one for big 4 international audit firms and zero otherwise.
CBIG10 An indicator variable of big audit firms that equals one for top 10 Chinese audit firms and zero otherwise.
SIZE Client company size variable, which equals the natural log of total assets.
CFO Client cash flows from operation (scaled by lagged assets).
GROW The growth in sales, measured as the industry total sales of the auditor rotation year t scaled by the industry total sales of year t − 1.
FT Audit firm tenure counting from 2004, the first year that the audit firm data became available in CCER database.
AGE No. of years since a client company was first listed in Chinese stock exchanges.
TTYPE And indicator variable that equals one if a client company was listed as ST/*ST company by the Chinese Stock Regulatory Committee in the period of Year 1 to Year 6,
and zero otherwise.
LEV Leverage ratio that equals total liability scaled by total assets.
EXP Two incumbent audit partners' experience in auditing clients in the same industry as their current client, measured as the sum of each audit partner's percentage of
their audited same-industry assets out of the total industry assets. The industry category follows the 13-industry codes announced by China Securities Regulatory
Commission (CSRC).
CLIENTP Client power measured as the percentage of a client's total assets out of an audit firm's total audited assets.

References
Aharony, J., Lee, C. J., & Wong, T. J. (2000). Financial packaging of IPO firms in China.
Journal of Accounting Research, 38(1), 103–126.
Arel, B., Brody, R., & Pany, K. (2005). Audit firm rotation and audit quality. CPA Journal,
36–39 (January).
Balsam, S., Krishnan, J., & Yang, J. S. (2003). Auditor industry specialization and earnings
quality. Auditing: A Journal of Practice and Theory, 22(2), 71–97.
Belsley, D. A., Kuh, E., & Welsch, R. E. (1980). Regression diagnostics: Identifying influential
data and sources of collinearity. New York: John Wiley and Sons.
Carcello, J. V., & Nagy, A. L. (2004). Audit firm tenure and fraudulent financial reporting.
Auditing: A Journal of Practice and Theory, 23(2), 55–69.
Carey, P., & Simnett, R. (2006). Audit partner tenure and audit quality. The Accounting
Review, 81(3), 653–676.
Chan, K. H., & Wu, D. (2011). Aggregate quasi rents and auditor independence: Evidence
from audit firm mergers in China. Contemporary Accounting Research, 28(1), 175–213.
Chaney, P., Jeter, D., & Shaw, P. (2003). The impact on the market for audit services of aggres-
sive competition by auditors. Journal of Accounting and Public Policy, 22(6), 487–516.
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), 892–925.
Chen, C., Lin, C., & Lin, Y. (2008). Audit partner tenure, audit firm tenure, and discretionary
accruals: Does long auditor tenure impair earnings quality? Contemporary Accounting
Research, 25(2), 415–445.
Chen, S., Sun, S. Y. J., & Wu, D. (2010). Client importance, institutional improvements, and
audit quality in China: An office and individual auditor level analysis. The Accounting
Review, 85(1), 127-15.
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, 74(3), 645–665.
Chi, W., & Huang, H. (2005). Discretionary accruals, audit–firm tenure and auditor tenure:
An empirical case in Taiwan. Journal of Contemporary Accounting & Economics, 1(1),
65–92.
Chi, W., Huang, H., Liao, Y., & Hong, X. (2009). Mandatory audit partner rotation, audit
quality, and market perception: Evidence from Taiwan. Contemporary Accounting
Research, 26(2), 359–391.
Chinese Institute of Certified Public Accountants (CICPA) (2007). The No.15 bulletin on
2007 annual financial reports auditing updates. (www.cicpa.org.cn (in Chinese))
Chinese Securities Regulatory Commission (CSRC Regulation), & China Ministry of Finance
(MOF) (2003). The regulation regarding the mandatory rotation of auditors issuing
auditing reports for companies issuing and trading stocks and options No.13. (in Chinese).
European Commission (EU) (2011). Proposal for a directive of the European Parliament
and the Council amending directive 2006/43/EC on statutory audits of annual
accounts and consolidated accounts and a proposal for a regulation of the European
Parliament and of the Council on specific requirements regarding statutory audit of
public-interest entities. Commission Staff Working Paper. Impact assessment accompa-
nying the document. Brussels, 30.11.2011.
Davis, L., Soo, B., & Trompeter, G. (2009). Auditor tenure and the ability to meet or beat
earnings forecasts. Contemporary Accounting Research, 26(2), 517–548.
DeAngelo, L. (1981). Auditor size and audit quality. Journal of Accounting and Economics,
3(3), 183–199.
Dechow, P.M., Sloan, R. G., & Sweeney, A. P. (1995). Detecting earnings management. The
Accounting Review, 70(2), 193–225.
DeFond, M. L., Raghunandan, K., & Subramanyam, K. R. (2002). Do nonaudit service fees
impair auditor independence? Evidence from going-concern audit opinions. Journal
of Accounting Research, 40(4), 1247–1274.
Deis, D. R., & Giroux, G. A. (1992). Determinants of audit quality in the public sector. The
Accounting Review, 67(3), 462–479.
Dopuch, N., King, R. R., & Schwartz, R. (2001). An experimental investigation of re-
tention and rotation requirements. Journal of Accounting Research, 39(1),
93–117.
Fan, G., Wang, X., & Zhu, H. (2004). The report on the relative process of marketization of
each region in China. The Economic Science Press, Beijing, Chapter 1 (in Chinese).
Firth, M., Rui, O. M., & Wu, X. (2012a). How do various forms of auditor rotation affect
audit quality? Evidence from China. The International Journal of Accounting, 47,
109–138.
Firth, M., Rui, O. M., & Wu, X. (2012b). Rotate back or not after mandatory audit partner
rotation? Journal of Accounting and Public Policy, 31(4), 356–373.
Francis, J. R., Michas, P. N., & Seavey, S. E. (2013). Does audit market concentration harm
the quality of audited earnings? Evidence from Audit Markets in 42 Countries.
Contemporary Accounting Research, 30(1), 325–355.
GAO (2003). Public accounting firms: Required study on the potential effects of mandato-
ry audit firm rotation. United States General Accounting Office. Report to the Senate
Committee on Banking, Housing, and Urban Affairs and the House Committee on Finan-
cial Services: November 2003.
GAO (2004). Mandatory audit firm rotation study: Study questionnaires, responses and
summary of respondents' comments. United States General Accounting Office. Report
to the Senate Committee on Banking, Housing, and Urban Affairs and the House Commit-
tee on Financial Services: February 2004.
Geiger, M., & Raghunandan, K. (2002). Auditor tenure and auditor reporting failures.
Auditing: A Journal of Practice & Theory, 21(1), 67–78.
Ghosh, A., & Moon, D. (2005). Auditor tenure and perceptions of audit quality. The
Accounting Review, 80(2), 585–612.
Gul, F. A., Fung, S. Y. K., & Jaggi, B. (2009). Earnings quality: Some evidence on the role of
auditor tenure and auditors' industry expertise. Journal of Accounting and Economics,
47(3), 265–287.
 S.P. Bandyopadhyay et al. / Advances in Accounting, incorporating Advances in International Accounting 30 (2014) 18–31
Gul, F., Jaggi, B., & Krishnan, G. V. (2007). Auditor independence: Evidence on the joint ef-
fects of auditor tenure and non-audit fees. Auditing: A Journal of Practice & Theory,
26(2), 117–142.
Hamilton, J. M., Ruddock, C. M. S., Stokes, D. J., & Taylor, S. L. (2005). Audit partner rota-
tion, earnings quality and earnings conservatism. Working Paper. La Trobe University,
University of New South Wales, Monash University.
Healey, T., & Kim, Y. (2003). The benefits of mandatory auditor rotation. Regulation, 26(3),
10–11.
Hribar, P., & Nichols, D. R. (2007). The use of unsigned earnings quality measures in tests
of earnings management. Journal of Accounting Research, 45(5), 1017–1053.
Johnson, V., Khurana, I., & Reynolds, J. (2002). Audit–firm tenure and the quality of finan-
cial reports. Contemporary Accounting Research, 19(4), 637–660.
Johnson, W. B., & Lys, T. (1990). The market for audit services: Evidence from voluntary
auditor changes. Journal of Accounting and Economics, 12(1–3), 281–309.
Kallapur, S., Sandaraguruswamy, S., & Zang, Y. (2010). Audit market concentration and
audit quality. Working paper. Indian School of Business, National University of
Singapore, and Singapore Management University.
Kinney, W., & McDaniel, L. (1996). How to improve the effectiveness of substantive ana-
lytical procedures. CPA Journal, 66(4), 52–54.
Knapp, M. (1991). Factors that audit committees use as surrogates for audit quality.
Auditing: A Journal of Practice & Theory, 10(1), 35–52.
Kothari, S. P., Leone, A. J., & Wasley, C. E. (2005). Performance matched discretionary ac-
crual measures. Journal of Accounting and Economics, 39(1), 163–197.
Krishnan, G. V. (2003). Does big 6 auditor industry expertise constrain earnings manage-
ment? Accounting Horizons, 17, 1–16.
Lin, K. Z., & Chan, K. H. (2000). Auditing standards in China: A comparative analysis
with relevant international standards and guidelines. The International Journal of
Accounting, 35(4), 559–577.
Mansi, S. A., Maxwell, W. F., & Miller, D. P. (2004). Does auditor quality and tenure matter to
investors? Evidence from the bond market. Journal of Accounting Research, 42(4), 755–793.
Mautz, R. K., & Sharaf, H. A. (1961). The philosophy of auditing. Monograph, 6. Sarasota, FL:
American Accounting Association.
Myers, J. N., Myers, L. A., & Omer, T. C. (2003). Exploring the term of the auditor–client re-
lationship and the quality of earnings: A case for mandatory auditor rotation? The
Accounting Review, 78(3), 779–799.
Public Company Accounting Oversight Board (PCAOB) (2011a). Concept release on auditor
independence and audit firm rotation: Notice of roundtable. PCAOB release No. 2011–006
August 16, 2011. PCAOB rulemaking docket No. 37. PCAOB 2011.
Public Company Accounting Oversight Board (PCAOB) (2011b). Comments by Internation-
al Federation of Accountants (IFAC). (http://pcaobus.org/Rules/Rulemaking/
Docket037/367_IFAC.pdf December 14, 2011. PCAOB 2011.)
31
Public Company Accounting Oversight Board (PCAOB) (2011c). Comments by Center for
Audit Quality. (http://pcaobus.org/Rules/Rulemaking/Docket037/519_CAQ.pdf De-
cember 14, 2011. PCAOB 2011.)
Public Company Accounting Oversight Board (PCAOB) (2012). Statement of Hon. Richard
C. Breeden, Former Chairman, U.S. Securities and Exchange Commission and Chairman,
Breeden Capital Management, to the Public Company Oversight Board regarding auditor
independence and audit firm rotation. (http://pcaobus.org/Rules/Rulemaking/
Docket037/ps_Breeden.pdf. March 21, 2012. PCAOB 2012.)
Raghunathan, B., Lewis, B.L., & Evans, J. H., III (1994). An empirical investigation of prob-
lem audits. Research in Accounting Regulation, 8, 33–58.
Rapoport, M. (2013, Dec 04). PCAOB proposes rule to identify partners in charge of audits;
audit regulator proposes accounting firms identify partner in charge of each client's
audit. Wall Street Journal (Online). (Retrieved from http://search.proquest.com.
proxy.lib.uwaterloo.ca/docview/1464719554?accountid=14906)
Sarbanes–Oxley Act of 2002 (2002). An Act to protect investors by improving the accura-
cy and reliability of corporate disclosures made pursuant to the securities laws and
for other purposes. Public Law (July 30, 2002. 107th. Congress.).
Schwartz, K. B., & Menon, K. (1985). Auditor switches by failing firms. The Accounting
Review, 60(2), 248–262.
Stanley, J., & DeZoort, T. (2007). Audit firm tenure and financial restatements: An analysis
of industry specialization and fee effects. Journal of Accounting and Public Policy, 26(2),
131–159.
Tan, H. -T. (1995). Effects of expectations, prior involvement, and review awareness on
memory for audit evidence and judgment. Journal of Accounting Research, 33(1),
113–135.
Ting, W., Yen, S., & Huang, S. (2009). Top management compensation, earnings manage-
ment and default risk: Insights from the Chinese stock market (2009). The
International Journal of Business and Finance Research, 3(1), 31–46.
Wang, Q., Wong, T. J., & Xia, L. (2008). State ownership, the institutional environment, and
auditor choice: Evidence from China. Journal of Accounting and Economics, 46(1),
112–134.
Watts, R., & Zimmerman, J. (1981). The markets for independence and independent
auditors. Graduate School of Management, University of Rochester unpublished
manuscript (Series No. GPB 80-10.).
Wolk, C. M., Michelson, S. E., & Wootton, C. W. (2001). Auditor concentration and market
shares in the US: 1988–1999. A descriptive note. The British Accounting Review, 33(2),
157–174.
Xiao, J. Z., Zhang, Y., & Xie, Z. (2000). The making of independent auditing standards in
China. Accounting Horizons, 14(1), 69–89.
Yu, Q., Du, B., & Sun, Q. (2006). Earnings management at rights issues thresholds—
Evidence from China. Journal of Banking & Finance, 30, 3453–3468.