chi et al - 2009 - mandatory audit partner rotation, audit quality, and market perception - evidence from taiwan [mapr]

Contemporary Accounting Research

Vol. 26 No. 2 (Summer 2009) pp. 359–91 © CAAA
doi:10.1506/car.26.2.2

Mandatory Audit Partner Rotation, Audit Quality,
and Market Perception: Evidence from Taiwan*

WUCHUN CHI,

National Chengchi University

HUICHI HUANG,

Syracuse University

YICHUN LIAO,

National Taiwan University

HONG XIE,

University of Kentucky

1. Introduction

Mandatory audit partner rotation has existed in the United States since the 1970s,
when the American Institute of Certified Public Accountants (AICPA) required that
audit partners in charge of Securities and Exchange Commission (SEC) audits be
rotated at least once every seven years. The Sarbanes-Oxley Act of 2002 (SOX)

further strengthens this requirement by mandating a five-year rotation for the lead
and concurring partners. An implicit assumption in a policy of mandatory partner
rotation is that such rotation enhances audit quality. However, this assumption has
not been systematically tested in the literature due to the lack of partner informa-
tion in U.S. audit reports.
Unlike in the United States, audit reports in Taiwan contain both audit firm
and audit partner names. Exploiting this institutional feature, Chen, Lin, and Lin
(2008) and Chi and Huang (2005) examine the relation between earnings quality
and partner tenure. They find that earnings quality tends to increase in partner tenure,
consistent with findings in the United States based on audit firm tenure. However,
their sample periods are prior to 2003 when partner rotation in Taiwan was voluntary.
These studies, thus, do not directly investigate the effect of mandatory audit part-
ner rotation on earnings quality or audit quality.

1

In this paper, we use audit data in Taiwan, where a five-year audit partner rota-
tion became de facto mandatory in 2004, to examine the effectiveness of mandatory
audit partner rotation in promoting audit quality and perceived audit quality.
Inspired by SOX, two principal stock exchanges in Taiwan — Taiwan Stock

* Accepted by Michael Willenborg. An earlier version of this paper was presented at the 2005

Con-
temporary Accounting Research

Conference, generously supported by the

Canadian Institute of
Chartered Accountants

, the

Certified General Accountants of Ontario

, the

Certified Man-
agement Accountants of Ontario

, and the

Institute of Chartered Accountants of Ontario

. We
appreciate valuable comments from Linda Bamber (discussant), Rajib Doogar, Chan-Jane Lin,
James Myers, Dan Simunic, Ira Solomon, Theodore Sougiannis, Michael Willenborg (associate
editor), two anonymous reviewers, participants at the 2005

Contemporary Accounting Research

Conference, and workshop participants at National Chengchi University and National Taipei Uni-
versity. Professor Chi gratefully acknowledges the financial support from National Science Council
(NSC 93-2416-H-004-036).

360 Contemporary Accounting Research

CAR

Vol. 26 No. 2 (Summer 2009)

Exchange Corporation (TWSE) and GreTai Securities Market (GTSM) — adopted
a set of rules in April 2003 that, in effect, require a five-year mandatory partner
rotation.

2

These rules became fully effective in 2004 for both semi-annual and
annual reports, with 2003 as a transition period (more detail below). We use the
2004 semi-annual reports of Taiwanese companies listed in the

Taiwan Economic
Journal

(TEJ) database for this study. Semi-annual reports in Taiwan are audited
no differently from annual reports, and the 2004 semi-annual reports are the first set
of data that reflect the full force of the mandatory partner rotation rule in Taiwan.
Following prior studies (e.g., Myers et al. 2003), we examine the effect of
mandatory audit partner rotation on audit quality using absolute and signed
performance-matched abnormal accruals (Kothari, Leone, and Wasley 2005) as
proxies for audit quality. We identify a sample of companies in 2004 whose audit
partners were subject to mandatory rotation within the same audit firm (the manda-
tory rotation sample) and compare it with three benchmark samples.

3

First, we
compare the mandatory rotation sample with companies in 2004 whose audit part-
ners were not subject to mandatory rotation (the nonrotation sample). We find no
difference in audit quality between these two samples. Second, we compare the

mandatory rotation sample with itself one year earlier (2003) (the mandatory rotation
sample in the prior year). We find that the audit quality of companies in the manda-
tory rotation sample under new audit partners is lower than the audit quality of
these same companies one year earlier under old audit partners. Third, we compare
our mandatory rotation sample with companies in years before 2003 whose audit
partners were voluntarily rotated within the same audit firm (the voluntary rotation
sample). We again find no difference in audit quality between these two samples.
In sum, we find no support for the belief that mandatory audit partner rotation
enhances audit quality. Our findings are robust to various sensitivity checks.
Next, we examine the effect of mandatory audit partner rotation on investor
perceptions of audit quality, using the earnings response coefficient (ERC) as a
proxy for perceived audit quality (Teoh and Wong 1993; Ghosh and Moon 2005).
After controlling for common determinants of the ERC, we find that the ERC of
the mandatory rotation sample is not significantly different from that of the non-
rotation sample or that of the mandatory rotation sample in the prior year, but is
significantly larger than the ERC of the voluntary rotation sample. Overall, we find
no consistent support for the belief that mandatory audit partner rotation enhances
investor perceptions of audit quality.
This paper contributes to the literature on auditor tenure and audit quality. To
our knowledge, we are among the first to directly examine the effect of mandatory
partner rotation on audit quality. Our findings are inconsistent with the implicit
belief in a mandatory partner rotation policy that such rotation enhances audit
quality or perceptions of audit quality. Rather, our findings are consistent, in spirit,
with findings in the United States that mandatory audit firm rotation may not nec-
essarily improve audit quality (Johnson, Khurana, and Reynolds 2002; Myers et al.
2003; Ghosh and Moon 2005; Blouin, Grein, and Rountree 2007).
Our findings, however, must be interpreted with caution. Our inferences about
the effect of mandatory partner rotation on audit quality and perceptions of audit

Mandatory Audit Partner Rotation, Audit Quality, and Market Perception 361

CAR

Vol. 26 No. 2 (Summer 2009)

quality critically depend on the ability of our accrual-based proxies (abnormal
accruals) and market-based proxy (ERC) to capture audit quality and perceived
audit quality. Although widely used in the accounting literature, both types of
proxies are noisy, which weakens our inferences. In addition, our findings, based
on Taiwanese data, may not be generalizable to a post-SOX U.S. audit market or
other audit markets where mandatory audit partner rotation is adopted, due to insti-
tutional differences between Taiwan and those markets. We note, however, that
accounting and auditing standards in Taiwan are similar to those in the United
States and that important empirical regularities in the U.S. audit market (e.g.,
Myers et al. 2003) can also be found in Taiwanese audit data (Chen et al. 2008).
Thus, there are significant similarities between Taiwan and the United States, mak-
ing our findings relevant for the U.S. audit market.
The remainder of the paper is organized as follows. Section 2 describes Taiwan-
ese regulation of mandatory audit partner rotation. Section 3 reviews the literature
and develops hypotheses. We describe data and sample selection in section 4. We
present empirical models and findings in section 5 and conclude in section 6.

2. Mandatory audit partner rotation in Taiwan

Unlike in the United States, where audit reports of public companies show only
audit firm names, audit reports in Taiwan show both firm and partner names.

4

Again unlike in the United States, where partner rotation every seven years has

long been required, audit partner rotation in Taiwan was entirely voluntary until
2003.
In April 2003, after the passage of SOX in the United States, TWSE and
GTSM, two principal stock exchanges in Taiwan, promulgated two rules that, in
effect, require a five-year mandatory partner rotation. First, both stock exchanges
amended the procedures for auditing the financial statements of listed companies
and added a clause, stating that if the lead or concurring partner has performed
audit services for a listed company in five consecutive years (applied retroactively),
then that company’s financial statements are subject to the stock exchange’s
“substantive review” procedure.

5

Specifically, the stock exchange audits financial
statements of a company targeted for substantive review and takes appropriate
actions if significant irregularities are found (more below). Second, there was a
large percentage of audit firms with both partners auditing the same client in the
previous four or more years in Taiwan as of 2003. The Taiwanese Accountants
Union argued that it would be difficult for audit firms, especially small audit firms,
to rotate two partners in the same year. In response to this and other concerns, both
stock exchanges postponed the effective time for full implementation of the five-
year rule for both audit partners to 2004, with 2003 (annual audits) as a transition
period when audit firms were allowed to have one partner, but not both, auditing
the same client for five or more years.
After a stock exchange determines that a company’s financial statements are
subject to substantive review, it will request and review audit working papers from
the audit partners. If the exchange finds significant violations of accounting or
auditing standards, it will refer the case to relevant government agencies for

362 Contemporary Accounting Research

CAR

Vol. 26 No. 2 (Summer 2009)

administrative or punitive actions, which range from reprimand to suspension of
license or even criminal charges. Because the potential punishments are severe, the
two stock exchanges’ new rules that subject a company’s financial statements to
substantive review when audited by the same lead or concurring audit partner in the
recent five consecutive years, in effect, mandate a five-year rotation for both partners.

3. Literature review and hypothesis development

The separation of ownership and control in public companies creates conflicts of
interest between management and outside stakeholders. Due to the conflicts of inter-
est and asymmetric information, financial statements prepared by management are
audited by a third party (an auditor) to mitigate agency costs (Watts and Zimmer-
man 1986). The value of an audit, however, depends on audit quality, which, in
turn, depends on auditor competence and independence (DeAngelo 1981).

6

Auditor
competence and independence thus are critically important to the value or per-
ceived value of an audit.

7

Mandatory audit partner rotation and audit quality

Mandatory audit partner rotation is adopted or considered in many countries as a
mechanism to enhance auditor independence and audit quality. In the United
States, the AICPA has required partner rotation every seven years since the 1970s.
Moreover, SOX section 203 mandates a five-year rotation for the lead and review-
ing partners. Internationally, mandatory partner rotation is currently practiced in
Australia (Carey and Simnett 2006), Singapore, United Kingdom, France, Spain,
the Netherlands, Japan, and Germany, and is being considered in Canada (General
Accounting Office [GAO] 2003, Appendix V).
The arguments for and against mandatory partner rotation, to a certain extent,
are parallel to those for and against mandatory audit firm rotation and center on the
costs and benefits of the rotation. The costs of mandatory partner rotation include
(a) increased likelihood of audit failures due to new partners’ lack of client-specific
knowledge of risk, operations, and financial reporting practices in the initial years
(American Institute of Certified Public Accountants [AICPA] 1992; Pricewater-
houseCoopers 2002); and (b) direct increases in costs incurred by both audit firms
and client companies due to the need for the new partner(s) to become familiar
with the client practices. The benefits of mandatory partner rotation, on the other
hand, include a “fresh look” by the new partner(s) and enhanced auditor independ-
ence.

8

The adoption of mandatory partner rotation in the United States and else-
where suggests that the regulators believe that the benefits of rotation outweigh the
costs and thus a policy of mandatory partner rotation enhances audit quality. How-
ever, as we review below, the validity of such a belief has not been tested in the
accounting literature.

Literature review

Prior studies often use absolute and signed abnormal accruals as proxies for earn-
ings quality or audit quality.

9

Abnormal accruals have become an accepted proxy
for earnings management, and thus earnings quality, in the accounting literature

Mandatory Audit Partner Rotation, Audit Quality, and Market Perception 363

CAR

Vol. 26 No. 2 (Summer 2009)

(e.g., Healy and Wahlen 1999; Kothari 2001). A main justification for also using
abnormal accruals as a proxy for audit quality is that audited financial statements
should be viewed as a joint outcome from the audit firm and company management
(Antle and Nalebuff 1991) — that is, earnings quality as captured by abnormal
accruals is also affected by the audit firm and thus reflects audit quality.
Another justification for using abnormal accruals as a proxy for audit quality
is the large volume of studies documenting a link between abnormal accruals and
audit quality. For example, Heninger (2001) documents a positive relation between
auditor litigation and the level of income-increasing abnormal accruals. Menon
and Williams (2004) find that companies employing former audit partners as offic-
ers or directors (“the revolving door”) report larger signed and unsigned abnormal
accruals. Richardson, Tuna, and Wu (2002) show that earnings restatements are
positively related to various accrual measures. Finally, many studies use abnormal
accruals as a metric to gauge audit quality differential between, for example, Big 6
and non–Big 6 auditors (Becker et al. 1998) and between auditors with and with-
out industry expertise (Krishnan 2003).

Using accrual-based proxies for audit quality, recent studies have examined
the relation between audit firm tenure and audit quality. For example, Johnson et al.
(2002) document that short audit firm tenure of two to three years is associated
with lower-quality financial reporting relative to medium (four to eight years) or
long (nine or more years) tenure. Similarly, Myers et al. (2003) find a positive rela-
tion between audit quality and audit firm tenure.
Several recent studies examine the relation between audit partner tenure and
audit quality. For example, using data from Australia, where partner information is
publicly disclosed and when partner rotation was voluntary, Carey and Simnett
(2006) find a diminution in audit quality, as proxied by the propensity to issue
going-concern opinions and the incidence of just beating (missing) earnings
benchmarks, for long partner tenure. In contrast, using data from Taiwan when
partner rotation was voluntary, Chen et al. (2008) find that audit quality, as measured
by absolute abnormal accruals, increases with partner tenure after controlling for
audit firm tenure. On the other hand, Chi and Huang (2005) find that audit quality,
as proxied by signed abnormal accruals, initially increases but starts to decrease as
partner tenure exceeds five years when they examine the effect of partner tenure
alone on earnings quality.

10

After including audit firm tenure in regression analy-
ses, they find that audit quality initially increases in audit firm tenure but starts to
decrease as firm tenure exceeds five years but the coefficients on partner tenure and
partner tenure squared are both insignificant.
Recent studies also use market-based measures, such as the cost of debt and
the ERC, as proxies for investor perceptions of audit quality. For example, Mansi
et al. (2004) find a significantly negative relation between the cost of debt and audit
firm tenure, suggesting that audit firm tenure enhances audit quality.

11

Similarly,
Ghosh and Moon (2005) use the ERC as a proxy for investor perceptions of audit
quality and find a positive association between perceived audit quality and audit
firm tenure.

364 Contemporary Accounting Research

CAR

Vol. 26 No. 2 (Summer 2009)

Hypotheses

To summarize, recent studies on audit firm tenure suggest that audit quality
increases with firm tenure, consistent with a “learning curve” for new auditors and
the critical importance of client-specific knowledge and experience, which can
only be acquired over time with the client, for producing a high-quality audit.
Recent studies on audit partner tenure, however, produce conflicting evidence
regarding the relation between partner tenure and audit quality. Importantly, all
these studies examine the relation between auditor tenure (firm or partner) and
audit quality under voluntary rotation regimes and, thus, do not directly examine
the effect of mandatory partner rotation on audit quality. Because the incentives
and behavior of auditors may change significantly under a mandatory rotation
regime, relative to a voluntary rotation regime, whether prior findings under a vol-
untary audit firm or partner rotation regime can be generalized to a mandatory
audit firm or partner rotation regime is ultimately an empirical question (Johnson
et al. 2002, 640; Myers et al. 2003, 796; Ghosh and Moon 2005, 588; Carey and
Simnett 2006, 674). In brief, the extant literature has not examined the effect of

mandatory audit partner rotation on audit quality and has not tested the validity
of the implicit belief that mandatory audit partner rotation enhances audit quality.
We examine the effect of mandatory audit partner rotation on audit quality and
perceptions of audit quality using audit data from Taiwan under the mandatory
partner rotation regime. We formulate the following two hypotheses (stated in
alternative form) based on the implicit assumption in a mandatory partner rotation
policy:
H

YPOTHESIS

1.

The audit quality of companies whose audit partners are man-
datorily rotated is higher than the audit quality of companies whose
audit partners are not required to rotate.

H

YPOTHESIS

2.

Investor perceptions of audit quality of companies whose
audit partners are mandatorily rotated are higher than investor percep-
tions of audit quality of companies whose audit partners are not required
to rotate.

4. Sample selection and data

Data for this study are collected from the 2004 semi-annual TEJ database for com-
panies listed on TWSE or GTSM. We identify a sample of companies in 2004
whose audit partners (at least one of them) were subject to mandatory rotation
(MROTA sample) and another sample of companies in 2004 whose partners (both
of them) were not required to rotate (NROTA sample) using the following procedure.
First, we identify 1,022 companies in 2002 (annual data) from the TEJ database
after excluding three Taiwan depository receipts (TDR) because semi-annual finan-
cial statements of TDRs are only reviewed rather than audited.

12

We delete 21
companies with missing audit partner information and three companies with non-
calendar fiscal year-ends. We thus obtain a preliminary sample of 998 companies
in 2002.

Mandatory Audit Partner Rotation, Audit Quality, and Market Perception 365

CAR

Vol. 26 No. 2 (Summer 2009)

Second, we trace audit partners of these 998 companies in past years up to
2002 and find 832 companies with at least one partner who had performed audit
services for the same client for at least four consecutive years by the end of 2002
and 166 companies with both partners who had performed audit services for the
same client for less than four consecutive years by 2002. We classify the 832 com-
panies into our mandatory rotation sample (MROTA), because at least one audit
partner is subject to rotation either for the 2003 annual audits or 2004 semi-annul
audits.

13

On the other hand, we classify the 166 companies identified above into
the nonmandatory rotation sample (NROTA) because none of their audit partners is
subject to mandatory rotation for the 2004 semi-annual audits.
Third, we trace audit partners of companies in our MROTA and NROTA sam-
ples to years 2003–4 to determine whether they are rotated for 2004 semi-annual
reports. We lose additional companies for the following reasons in the MROTA
(NROTA) samples: (a) 30 (7) companies due to delisting; (b) 78 (not applicable)
companies due to their changing audit firms during 2003 and 2004;

14

(c) 109 (0)
companies because one audit partner was rotated off in 2003 but came back in
2004;

15

(d) 1 (0) company because both audit partners were rotated off in 2003 but
at least one came back in 2004; (e) 15 (0) companies for which one audit partner
should have been rotated in 2004 but was not rotated; (f) 28 (0) companies for
which both audit partners should have been rotated in 2004 but only one audit part-
ner was rotated; (g) 18 (23) companies in financial industries whose accruals are
difficult to interpret; (h) 54 (9) companies with missing data for tracing audit firm
tenure; and (i) 6 (2) companies due to our requirement of at least eight observa-
tions to estimate abnormal accruals for each industry-year combination using the
modified Jones 1991 model. The above process generates 493 (125) companies in
our MROTA and NROTA samples, respectively. Table 1, panel A summarizes the

sample selection process.
To test our hypotheses, we compare the MROTA sample with three bench-
marks. The first benchmark is the nonrotation sample (NROTA) described above.
The second benchmark is the mandatory rotation sample itself in the prior year
(MBEFR sample). MBEFR and MROTA samples thus contain exactly the same
companies, but semi-annual financial statements of MBEFR sample were audited
under the old audit partners in 2003, whereas those of the MROTA sample were
audited under new audit partners (at least one) in 2004. The third benchmark is the
voluntary rotation (VROTA) sample described below.
The sample selection process for our VROTA sample is summarized in Table 1,
panel B. Specifically, we identify companies before 2003 for which at least one
partner was voluntarily rotated within the same audit firm. We want to have
roughly the same number of observations in VROTA as in MROTA, and need only
go back to 1999 because we already identified 638 company-year observations by
1999. We delete 77 observations in financial institutions, 41 observations due to
missing audit firm tenure, and 7 observations due to fewer than eight companies in
their industry classifications in a year. The final VROTA sample consists of 513
company-year observations.

366 Contemporary Accounting Research

CAR

Vol. 26 No. 2 (Summer 2009)

5. Empirical models and findings

In this section, we examine the effect of mandatory audit partner rotation on audit
quality and investor perceptions of audit quality. We first present the empirical
model and findings using accrual-based proxies for audit quality, and then the

empirical model and findings using the market-based proxy for perceived audit
quality.

Accrual-based proxies for audit quality

Variable measurement and empirical model

Johnson et al. (2002) and Myers et al. (2003) use the Jones 1991 model-estimated
abnormal accruals as proxies for audit quality. We use the modified Jones model

TABLE 1

Sample selection

Panel A:

MROTA and NROTA sample selection
Companies on TWSE or GTSM in 2002 from the TEJ database
after deleting 3 TDRs 1,022
Less
Companies with missing audit partner information (21)
Companies with noncalendar year-end
(3)
Preliminary sample 998
Preliminary sample 832 166
Less
Companies delisted in 2003 or 2004 (30) (7)
Companies switching audit firms in 2003 or 2004 (78) (N/A)

*

Companies rotating one audit partner in 2003 but the audit
partner came back in 2004 (109) (0)
Companies rotating both audit partners in 2003 but one of
them came back in 2004 (1) (0)
Companies rotating both audit partners in 2003 but both
came back in 2004 (0) (0)
Companies should rotate one audit partner but rotated none (15) (0)
Companies should rotate both audit partners but rotated
only one (28) (0)
Companies should rotate both audit partners but rotated none (0) (0)
Companies in financial institutions (18) (23)
Companies with missing data for tracing audit firm tenure (54) (9)
Companies with less than eight observations in a
industry-year combination
(6) (2)

Final sample 493 125
(The table is continued on the next page.)
Sample label MROTA NROTA

Mandatory Audit Partner Rotation, Audit Quality, and Market Perception 367

CAR

Vol. 26 No. 2 (Summer 2009)

(Dechow, Sloan, and Sweeney 1995) to estimate abnormal accruals and then do
performance matching according to Kothari et al. (2005) because they demonstrate
that performance-matched abnormal accruals capture earnings management better

than do traditional Jones model-estimated abnormal accruals.
Specifically, we first estimate raw abnormal accruals (

MJAbnA

) as the residu-
als from the modified Jones model below (company subscript

i

is omitted except in
places where doing so causes confusion):

TAC

t

/

TA

t



Ϫ

1

ϭ



␣

t

(1/

TA

t



Ϫ

1

)

ϩ



␤

t

(

⌬

SALES

t

/

TA

t



Ϫ

1



Ϫ



⌬

AR

t

/

TA

t



Ϫ

1

)

ϩ



␥

t

(

PPE

t

/

TA

t



Ϫ

1

)

ϩ



␧

t

(1),
where

TAC

t

ϭ

total accruals in the first half of year

t

, calculated using the statement
of cash flow approach recommended by Hribar and Collins 2002

ϭ

income before discontinued operations and extraordinary items

Ϫ

(cash
from operations

Ϫ

discontinued operations and extraordinary items
from the statement of cash flows);

⌬

SALES

t

ϭ

change in sales revenue between the first half of year

t

and the first
half of year

t



Ϫ

1;

TABLE 1 (Continued)

Panel B:

VROTA sample selection
2002 157 157
2001 220 377
2000 143 520
1999 118 638
Less
Observations in financial institutions (77)
Observations with missing data for tracing
audit firm tenure (41)
Observations with fewer than eight companies in

an industry-year combination
(7)

Final sample 513

Note:

*

When a company in the preliminary NROTA sample switches its audit firm or rotates
audit partners within the same audit firm, that company is still included in the
NROTA sample because the switch of audit firm or rotation of audit partners is
not required by the mandatory partner rotation rule.
Number of
companies
Cumulative
company-year
observations

368 Contemporary Accounting Research

CAR

Vol. 26 No. 2 (Summer 2009)

⌬

AR

t

ϭ

change in accounts receivable between the first half of year

t

and the
first half of year

t



Ϫ

1;

PPE

t

ϭ

gross amount of property, plant, and equipment at the end of the first
half of year

t

; and

TA
t Ϫ 1
ϭ total assets at the end of year t Ϫ 1 (i.e., total assets at the beginning
of the first half of year t).
We estimate (1) in the cross-section in each year (from 1999 to 2004) for each TEJ
industry classification with at least eight observations using all companies with
required data in the TEJ database.
We then do performance matching based on current-period return on assets
(ROA
t
). Specifically, for each company i (i ϭ 1, 2, , n, and n Ն 8) in an industry-
year combination in year t, we find another company j, where j  i, among the
remaining companies (n Ϫ 1) in the same industry-year combination whose return on
assets (ROA
jt
) is closest to that of company i (ROA
it
). Our performance-matched,
modified Jones model-estimated abnormal accruals (PMMJAbnA) for company i in
year t are the difference in raw abnormal accruals between companies i and j.
After obtaining PMMJAbnA for all companies with required data in the TEJ
database during 1999–2004, we keep only company-year observations in our man-
datory rotation sample (MROTA) and three benchmark samples (NROTA,
MBEFR, and VROTA).
Following Myers et al. 2003, we examine the effectiveness of mandatory audit
partner rotation in promoting audit quality using the following regression model:
Acc ϭ
␣
ϩ
␤

1
BMK ϩ
␤
2
Age ϩ
␤
3
Size ϩ
␤
4
IndGrw ϩ
␤
5
CFO ϩ
␤
6
Big4
ϩ
␤
7
FTenure ϩ
␧
(2),
where
Acc ϭ performance-matched abnormal accruals (PMMJAbnA), measured in
absolute, positive, and negative values;
BMK ϭ a dummy variable equal to 1 if observations are from one of the three
benchmark samples (NROTA, MBEFR, or VROTA), and equal to 0
otherwise;
Age ϭ number of years since the company was listed;

Size ϭ natural logarithm of total assets at the end of the first half of year t;
IndGrw ϭ industry growth
ϭ
by the TEJ industry classification, and t and t Ϫ 1 refer to the first half
of years t and t Ϫ 1, respectively;
SALES
it
i 1ϭ
N
Α
SALES
it 1Ϫ
i 1ϭ
N
Α
ր
Mandatory Audit Partner Rotation, Audit Quality, and Market Perception 369
CAR Vol. 26 No. 2 (Summer 2009)
CFO ϭ cash from operations from the statement of cash flows for the first half
of year t, scaled by total assets at the end of year t Ϫ 1;
Big4 ϭ a dummy variable equal to 1 if the auditor is from a Big 4 or Big 5
audit firm, and equal to 0 otherwise;
16
and
FTenure ϭ audit firm tenure, measured as the number of consecutive years since
1988 that the company has retained the audit firm.
17
We estimate (2) in each of the three comparison samples: MROTA versus
NROTA, MROTA versus MBEFR, and MROTA versus VROTA. We first estimate
(2) over a comparison sample using the ordinary least squares (OLS) method when

the absolute value of accruals (͉PMMJAbnA͉) is the dependent variable. We then
estimate (2) over truncated subsamples, PMMJAbnA Ն 0 and PMMJAbnA Ͻ 0,
respectively, using the maximum likelihood method when the truncated signed
value of accruals is the dependent variable where OLS would bias coefficient esti-
mates toward zero (see Myers et al. 2003).
Our variable of primary interest is BMK. Hypothesis 1 predicts a positive
coefficient on BMK when using ͉PMMJAbnA͉ as the dependent variable and a
positive (negative) coefficient on BMK for the positive (negative) PMMJAbnA sub-
sample in a truncated regression. In other words, our first hypothesis predicts that
accruals are more extreme (and thus audit quality is lower) for the benchmark sample
(i.e., BMK ϭ 1), relative to the mandatory rotation sample (i.e., BMK ϭ 0).
We include several control variables for other known determinants of accruals
in (2), based on Myers et al. 2003 and other prior studies. First, we include Age to
control for changes in accruals over a company’s life cycle (Anthony and Ramesh
1992) and expect accruals to become less extreme as a company’s age increases.
Second, we include Size to control for a size effect and expect accruals for larger
companies to be less extreme (Watts and Zimmerman 1986). Third, we include
IndGrw to control for a potentially positive effect of industry growth on a com-
pany’s accruals. However, Myers et al. show a mixed relation between accruals and
IndGrw. Therefore, we do not predict the sign for IndGrw. Fourth, we include CFO
to control for a negative relation between accruals and cash from operations
(Dechow 1994). On the basis of findings in Myers et al., we expect a negative coef-
ficient on CFO. Fifth, we include Big4 to control for prior findings that Big 4 or
Big 5 audit firms tend to be more conservative and tend to limit their clients’
extreme accruals. However, Myers et al. find mixed results for Big4 and, thus, we
do not predict the sign for Big4. Finally, we include FTenure to control for the
effect of audit firm tenure on accruals and expect accruals to be less extreme for
companies with longer firm tenure (Johnson et al. 2002; Myers et al. 2003).
Empirical findings based on performance-matched abnormal accruals
To mitigate the potential undue influences of extreme values, we winsorize

PMMJAbnA, Age, Size, IngGrw, and CFO at the top and bottom 1 percent of their
respective distributions.
18
Descriptive statistics for variables in (2) are reported in
Table 2. First, we compare the MROTA sample with the NTOTA sample. The
370 Contemporary Accounting Research
CAR Vol. 26 No. 2 (Summer 2009)
TABLE 2
Descriptive statistics
Panel A: Mandatory rotation sample (MROTA, n ϭ 493)
͉PMMJAbnA͉ 0.063 0.062 0.000 0.018 0.044 0.088 0.292
Age 9.063 7.288 2.000 4.000 7.000 11.000 40.000
Size 22.251 1.204 19.963 21.467 22.064 22.871 25.772
IndGrw 1.041 0.076 0.846 1.024 1.024 1.072 1.305
CFO 0.007 0.065 Ϫ0.246 Ϫ0.021 0.010 0.042 0.169
Big4 0.826 0.380 0.000 1.000 1.000 1.000 1.000
FTenure 8.099 4.262 2.000 5.000 7.000 9.000 17.000
Panel B: Nonrotation sample (NROTA, n ϭ 125)
͉PMMJAbnA͉ 0.068 0.070 0.001 0.017 0.055 0.079 0.393
Age 9.800 9.214 2.000 4.000 6.000 11.000 42.000
Size 22.202 1.236 20.155 21.432 21.980 22.843 26.009
IndGrw 1.041 0.088 0.846 1.024 1.024 1.072 1.305
CFO Ϫ0.003 0.084 Ϫ0.366 Ϫ0.026 Ϫ0.001 0.041 0.162
Big4 0.744
‡
0.438 0.000 0.000 1.000 1.000 1.000
FTenure 4.864
*
3.859 1.000 2.000 4.000
*

5.000 17.000
Panel C: Mandatory rotation sample in prior year (MBEFR, n ϭ 493)
͉PMMJAbnA͉ 0.055
†
0.052 0.000 0.016 0.038 0.082 0.246
Age 8.063
†
7.288 1.000 3.000 6.000
*
10.000 39.000
Size 22.140 1.191 19.943 21.300 21.957 22.832 25.467
IndGrw 1.004
*
0.060 0.826 1.016 1.016
*
1.041 1.085
CFO 0.017
†
0.059 Ϫ0.192 Ϫ0.011 0.017
†
0.045 0.191
Big4 0.826 0.380 0.000 1.000 1.000 1.000 1.000
FTenure 7.099
*
4.262 1.000 4.000 6.000
*
8.000 16.000
Panel D: Voluntary rotation sample (VROTA, n ϭ 513)
͉PMMJAbnA͉ 0.064 0.059 0.000 0.021 0.045 0.086 0.284
Age 7.817

†
8.698 1.000 2.000 5.000
*
10.000 39.000
Size 22.227 1.158 20.155 21.414 22.115 22.891 25.580
IndGrw 0.994
*
0.108 0.791 0.929 1.003
*
1.085 1.135
CFO 0.018
*
0.062 Ϫ0.213 Ϫ0.008 0.020
*
0.051 0.184
Big4 0.809 0.393 0.000 1.000 1.000 1.000 1.000
FTenure 5.858
*
3.831 2.000 3.000 5.000
*
7.000 15.000
(The table is continued on the next page.)
Variable Mean s.d. Min. Q1 Median Q3 Max.
Variable Mean s.d. Min. Q1 Median Q3 Max.
Variable Mean s.d. Min. Q1 Median Q3 Max.
Variable Mean s.d. Min. Q1 Median Q3 Max.
Mandatory Audit Partner Rotation, Audit Quality, and Market Perception 371
CAR Vol. 26 No. 2 (Summer 2009)
mean ͉PMMJAbnA͉ for the MROTA sample is 0.063, whereas that for the NROTA
sample is 0.068. A two-tailed t-test suggests that the difference of Ϫ0.005 is insignif-

icantly different from zero. In addition, a two-tailed nonparametric Wilcoxon z-test
suggests that the median ͉PMMJAbnA͉ for the MROTA sample is insignificantly
different from that for the NROTA sample. Thus, univariate comparisons of the
mean and median ͉PMMJAbnA͉ suggest that audit quality of the mandatory rota-
tion sample is indifferent from that of the nonrotation sample, failing to support
Hypothesis 1. Turning to other variables, the differences in means and medians
between MROTA and NROTA samples are all insignificant except for Big4 and
FTenure. The mean Big4 for the MROTA sample is 0.826, whereas that for the
NROTA sample is 0.744. A two-tailed t-statistic suggests that the difference of
0.082 is significant at the 0.1 level. We indicate this significance by placing
a ‡ sign on the mean Big4 for the NROTA sample without reporting the specific
t-statistic. As for FTenure, the mean and median for the MROTA sample are both
significantly larger than their counterparts for NROTA.
19
Second, we compare the MROTA sample with itself one year earlier (MBEFR).
We find that the mean, but not the median, ͉PMMJAbnA͉ for the MROTA sample is
TABLE 2 (Continued)
Notes:
Variables are defined as follows:
͉PMMJAbnA͉ ϭ absolute performance-matched abnormal accruals;
Age ϭ number of years since the company was listed on stock exchanges;
Size ϭ natural logarithm of total assets at the end of the first half of year t;
IndGrw ϭ industry growth
ϭ
by the TEJ industry classification, and t and t Ϫ 1 refer to the first half of
years t and t Ϫ 1, respectively;
CFO ϭ cash from operations from the statement of cash flows for the first half of
year t, scaled by total assets at the end of year t Ϫ 1;
Big4 ϭ a dummy variable equal to 1 if the auditor is from a Big 4 or Big 5 audit
firm, and equal to 0 otherwise; and

FTenure ϭ number of consecutive years since 1988 that the company has retained the
audit firm.
*
Difference in mean (median) between the mandatory rotation sample and the benchmark
sample significant at the 0.01 level using a two-tailed t-test (Wilcoxon z-test).
†
Difference in mean (median) between the mandatory rotation sample and the benchmark
sample significant at the 0.05 level using a two-tailed t-test (Wilcoxon z-test).
‡
Difference in mean (median) between the mandatory rotation sample and the benchmark
sample significant at the 0.10 level using a two-tailed t-test (Wilcoxon z-test).
SALES
it
i 1ϭ
N
Α
SALES
it 1Ϫ
i 1ϭ
N
Α
ր
372 Contemporary Accounting Research
CAR Vol. 26 No. 2 (Summer 2009)
significantly larger (i.e., the audit quality is lower) than that for the MBEFR sam-
ple.
20
Third, we find no significant differences in the mean and median ͉PMMJAbnA͉
between the MROTA and VROTA samples. Both findings are inconsistent with
Hypothesis 1.

Next, we examine the effect of mandatory audit partner rotation on audit qual-
ity in a multivariate setting using (2). Table 3, panel A reports our findings using
absolute abnormal accruals (͉PMMJAbnA͉) as the dependent variable. First, we
find that the coefficient on BMK is insignificant (0.001, t ϭ 0.166) in the “MROTA
versus NROTA” column. This suggests that ͉PMMJAbnA͉ (i.e., audit quality) is
indistinguishable between our mandatory rotation sample and the nonrotation
sample, after controlling for common determinants of abnormal accruals in (2).
21
Second, the coefficient on BMK is significantly negative (Ϫ0.010, t ϭ Ϫ2.750) in
the “MROTA versus MBEFR” column. This suggests that the audit quality of com-
panies subject to mandatory rotation in 2004 under new audit partners is lower than
the audit quality of these same companies one year earlier under old partners.
Third, the coefficient on BMK is insignificant in the “MROTA versus VROTA” col-
umn (Ϫ0.002, t ϭ Ϫ0.521), suggesting that the audit quality of the mandatory
rotation sample is not different from that of the voluntary rotation sample.
TABLE 3
Performance-matched abnormal accruals and mandatory audit partner rotation
Panel A: Absolute performance-matched abnormal accruals (͉PMMJAbnA͉) results
Intercept ? Ϫ0.028 0.077 Ϫ0.048
(Ϫ0.470) (1.588) (Ϫ1.049)
BMK ϩ 0.001 Ϫ0.010
*
Ϫ0.002
(0.166) (Ϫ2.750) (Ϫ0.521)
Age ϪϪ0.001
*
Ϫ0.001
*
Ϫ0.001
*

(Ϫ3.558) (Ϫ3.128) (Ϫ3.920)
Size Ϫ 0.009
*
0.004
†
0.006
*
(4.328) (2.439) (3.203)
IndGrw ? Ϫ0.090
*
Ϫ0.078
†
0.000
(Ϫ2.600) (Ϫ2.521) (0.013)
CFO ϪϪ0.303
*
Ϫ0.188
*
Ϫ0.194
*
(Ϫ5.153) (Ϫ4.031) (Ϫ3.990)
Big4 ? Ϫ0.002 Ϫ0.006 0.007
(Ϫ0.310) (Ϫ1.275) (1.488)
FTenure ϪϪ0.001 Ϫ0.001
†
Ϫ0.002
*
(Ϫ1.056) (Ϫ2.234) (Ϫ3.592)
Adj. R
2

0.143 0.077 0.079
n 618 986 1,006
(The table is continued on the next page.)
Variable Exp. sign
MROTA versus
NROTA
MROTA versus
MBEFR
MROTA versus
VROTA
Mandatory Audit Partner Rotation, Audit Quality, and Market Perception 373
CAR Vol. 26 No. 2 (Summer 2009)
Following Myers et al. 2003, we also estimate (2) for positive and negative
abnormal accruals separately.
22
We report our findings from the truncated regres-
sions in Table 3, panels B and C. First, for income-increasing accruals (panel B),
the coefficient on BMK is significantly negative (Ϫ0.022, t ϭ Ϫ1.756) for the
“MROTA versus MBEFR” column, suggesting that new audit partners in the MROTA
sample constrain extremely positive accruals to a smaller extent than old audit
partners in the MBEFR sample. Second, for income-decreasing accruals (panel C),
the coefficient on BMK is significantly positive (0.066, t ϭ 1.998) for the “MROTA
versus MBEFR” column, again suggesting that new audit partners in the MROTA
sample constrain extremely negative accruals to a smaller extent than do old audit
partners in the MBEFR sample. Both findings suggest that the audit quality of
companies subject to mandatory rotation under new audit partners is lower than the
audit quality of these same companies one year earlier under old audit partners.
Third, the audit quality of the mandatory rotation sample is indistinguishable from
that of the nonrotation sample (the “MROTA versus NROTA” column) and the vol-
untary rotation sample (the “MROTA versus VROTA” column), as indicated by

insignificant coefficients on BMK in Table 3, panels B and C.
TABLE 3 (Continued)
Panel B: Positive performance-matched abnormal accruals (PMMJAbnA Ն 0) results
Intercept ? Ϫ0.531
*
Ϫ0.244 Ϫ0.446
*
[Ϫ3.000] [Ϫ1.510] [Ϫ3.270]
BMK ϩϪ0.025 Ϫ0.022
‡
0.002
[Ϫ1.311] [Ϫ1.756] [0.189]
Age ϪϪ0.003
†
Ϫ0.001 Ϫ0.003
†
[Ϫ1.985] [Ϫ1.046] [Ϫ2.037]
Size Ϫ 0.030
*
0.014
†
0.020
*
[4.246] [2.226] [3.617]
IndGrw ? Ϫ0.096 Ϫ0.037 0.022
[Ϫ1.183] [Ϫ0.472] [0.376]
CFO ϪϪ1.100
*
Ϫ1.255
*

Ϫ1.088
*
[Ϫ10.442] [Ϫ10.101] [Ϫ12.644]
Big4 ? Ϫ0.020 Ϫ0.021 0.006
[Ϫ1.183] [Ϫ1.449] [0.415]
FTenure ϪϪ0.002 Ϫ0.003 Ϫ0.005
†
[Ϫ0.778] [Ϫ1.449] [Ϫ2.200]
Adj. R
2
0.533 0.419 0.419
n 321 521 522
(The table is continued on the next page.)
Variable Exp. sign
MROTA versus
NROTA
MROTA versus
MBEFR
MROTA versus
VROTA
374 Contemporary Accounting Research
CAR Vol. 26 No. 2 (Summer 2009)
In sum, we have two major findings in Table 3. First, the audit quality of the
mandatory rotation sample is indistinguishable from the audit quality of the non-
rotation sample and the voluntary rotation sample. Second, the audit quality of
companies in the mandatory rotation sample under new audit partners is lower than
the audit quality of these same companies one year ago under old audit partners.
Chen et al. (2008) document that audit quality is positively related to audit
partner tenure under the voluntary rotation regime in Taiwan. The essence of Chen
et al. 2008, Johnson et al. 2002, and Myers et al. 2003 is that client-specific knowl-

edge and experience, as captured by the length of audit partner or audit firm tenure,
TABLE 3 (Continued)
Panel C: Negative performance-matched abnormal accruals (PMMJAbnA Ͻ 0) results
Intercept ? Ϫ0.212 Ϫ0.215 0.186
[Ϫ0.278] [Ϫ0.649] [0.454]
BMK ϪϪ0.047 0.066
†
0.051
[Ϫ0.510] [1.998] [1.347]
Age ϩ 0.006 0.004 0.003
[0.815] [1.012] [0.914]
Size ϩϪ0.011 Ϫ0.004 Ϫ0.001
[Ϫ0.377] [Ϫ0.324] [Ϫ0.088]
IndGrw ? 0.930 0.441
‡
0.104
[1.283] [1.887] [0.542]
CFO ϪϪ2.074 Ϫ1.747
*
Ϫ2.340
*
[Ϫ1.541] [Ϫ3.574] [Ϫ3.873]
Big4 ? Ϫ0.005 0.040 Ϫ0.023
[Ϫ0.048] [1.001] [Ϫ0.505]
FTenure ϩϪ0.003 0.001 0.005
[Ϫ0.236] [0.148] [0.774]
Adj. R
2
0.046 0.130 0.162
n 297 465 484

Notes:
Variable definitions: BMK is a dummy variable equal to 1 if observations are from one of the
three benchmark samples (NROTA, MBEFR, or VROTA), and equal to 0 otherwise.
Other variables are as defined in Table 2.
*
Significant at the 0.01 level based on a two-tailed t-statistic (in parentheses) in
panel A and a two-tailed z-statistic (in brackets) in panels B and C.
†
Significant at the 0.05 level based on a two-tailed t-statistic (in parentheses) in
panel A and a two-tailed z-statistic (in brackets) in panels B and C.
‡
Significant at the 0.10 level based on a two-tailed t-statistic (in parentheses) in
panel A and a two-tailed z-statistic (in brackets) in panels B and C.
Variable Exp. sign
MROTA versus
NROTA
MROTA versus
MBEFR
MROTA versus
VROTA
Mandatory Audit Partner Rotation, Audit Quality, and Market Perception 375
CAR Vol. 26 No. 2 (Summer 2009)
are essential for auditors to produce a high-quality audit. The mean average partner
tenure for the MROTA, NROTA, MBEFR, and VROTA samples are 1.506, 2.332,
4.825, and 2.724 years, respectively (untabulated).
23
The fact that partner tenure
for MROTA is the shortest and that for MBEFR is the longest is due to the nature
of these two samples and by construction. The MROTA sample consists of com-
panies with new partner(s), whereas the MBEFR sample consists of the same

companies as MROTA but in the previous year with old partners and thus long ten-
ure. Our finding that the audit quality of companies in MROTA under new audit
partners is lower than the audit quality of these same companies one year earlier
under old audit partners (MBEFR) is consistent with the Chen et al. 2008 finding
that audit quality increases in audit partner tenure.
We now turn to discussing the control variables in Table 3. Results on Age are
generally consistent with our prediction. Specifically, abnormal accruals are nega-
tively related to Age in panel A and panel B, as expected, although they are not pos-
itively related to Age in panel C. Our findings on Size, for the most part, are
opposite to our expectation based on prior studies. Our findings of a negative coef-
ficient on CFO in panels A, B, and C are consistent with Chen et al. 2008 and
Myers et al. 2003. Finally, we find some evidence that longer audit firm tenure
(FTenure) reduces extreme accruals in panel A and panel B, consistent with Chen
et al. 2008, Johnson et al. 2002, and Myers et al. 2003.
Additional tests
Alternative measures of accruals
We test the robustness of our findings to alternative measures of accruals. Specifically,
we calculate abnormal accruals using three alternative approaches: (a) performance-
matched abnormal accruals estimated using the Jones 1991 model (PMJAbnA);
24
(b) raw abnormal accruals estimated using the modified Jones model without per-
formance matching (MJAbnA); and (c) raw abnormal accruals estimated using the
Jones model without performance matching (JAbnA).
With abnormal accruals estimated using these three alternative approaches, we
reestimate (2). For brevity, we only report the coefficient on BMK in Table 4.
25
To
facilitate comparison, we reproduce the coefficients on BMK in Table 3 at the top
of each panel in Table 4. Overall, results based on these three alternative models
are qualitatively similar to our main results, with the following key exceptions.

First, the coefficient on BMK in the “MROTA versus NROTA” column in Table 4,
panel B is significantly negative for the modified Jones model-estimated accruals
(Ϫ0.029, t ϭ Ϫ1.871), suggesting that the audit quality of the mandatory rotation
sample is lower than the audit quality of the nonrotation sample. Second, the coef-
ficient on BMK in the “MROTA versus NROTA” column in panel C is significantly
negative for the modified Jones model-estimated accruals (Ϫ0.060, t ϭ Ϫ2.703)
and for the Jones model-estimated accruals (Ϫ0.079, t ϭ Ϫ2.630), suggesting that
the audit quality of the mandatory rotation sample is higher than the audit quality
of the nonrotation sample. Because these two results are contradictory, we con-
clude that there is no consistent evidence regarding whether the audit quality of the
376 Contemporary Accounting Research
CAR Vol. 26 No. 2 (Summer 2009)
MROTA sample is higher or lower than that of the NROTA sample. We thus main-
tain our previous conclusion that the audit quality of the mandatory rotation sample
is not significantly different from that of the nonrotation sample.
Prior studies suggest that current accruals are more susceptible to earnings
management than total accruals. Consequently, measures of current accruals or
working capital accruals might capture earnings management or audit quality
TABLE 4
Alternative measures of accruals and mandatory audit partner rotation
Panel A: Absolute accruals (͉Acc͉) results
Performance-matched modified Jones model ϩ 0.001 Ϫ0.010
*
Ϫ0.002
(PMMJAbnA) (0.166) (Ϫ2.750) (Ϫ0.521)
Performance-matched Jones model ϩ 0.003 Ϫ0.011
*
Ϫ0.003
(PMJAbnA) (0.460) (Ϫ2.836) (Ϫ0.694)
Modified Jones model ϩ 0.005 Ϫ0.007

†
Ϫ0.002
(MJAbnA) (0.884) (Ϫ2.572) (Ϫ0.547)
Jones model ϩ 0.007 Ϫ0.007
†
Ϫ0.002
(JAbnA) (1.235) (Ϫ2.474) (Ϫ0.770)
Current accruals ϩ 0.005 Ϫ0.010
*
0.000
(CurA) (0.786) (Ϫ3.169) (0.084)
Abnormal working capital accruals ϩ 0.017 Ϫ0.014
*
Ϫ0.008
(AbnWCA) (1.167) (Ϫ2.854) (Ϫ1.461)
Panel B: Positive accruals (Acc Ͼ 0) results
Performance-matched modified Jones model ϩϪ0.025 Ϫ0.022
‡
0.002
(PMMJAbnA)[Ϫ1.311] [Ϫ1.756] [0.189]
Performance-matched Jones model ϩϪ0.033 Ϫ0.027
‡
Ϫ0.005
(PMJAbnA)[Ϫ1.431] [Ϫ1.811] [Ϫ0.345]
Modified Jones model ϩϪ0.029
‡
Ϫ0.004 Ϫ0.008
(MJAbnA)[Ϫ1.871] [Ϫ0.439] [Ϫ1.063]
Jones model ϩϪ0.023 Ϫ0.000 Ϫ0.010
(JAbnA)[Ϫ1.574] [Ϫ0.054] [Ϫ1.516]

Current accruals ϩϪ0.048 Ϫ0.189 0.005
(CurA)[Ϫ0.550] [Ϫ1.482] [0.687]
Abnormal working capital accruals ϩ 0.011 Ϫ0.121 Ϫ0.114
(AbnWCA) [0.081] [Ϫ0.789] [Ϫ0.291]
(The table is continued on the next page.)
Variable
Exp.
sign
MROTA
versus
NROTA
MROTA
versus
MBEFR
MROTA
versus
VROTA
Variable
Exp.
sign
MROTA
versus
NROTA
MROTA
versus
MBEFR
MROTA
versus
VROTA
Mandatory Audit Partner Rotation, Audit Quality, and Market Perception 377

CAR Vol. 26 No. 2 (Summer 2009)
better than measures based on total accruals. We thus adopt current accruals
(CurAt) and abnormal working capital accruals (AbnWCAt) as two additional meas-
ures of audit quality.
26
We reestimate (2) using these two alternative measures of audit quality and
report the coefficients on BMK in Table 4. Again, the tenor of our main findings in
Table 3 is unchanged. The only key exception is a significantly negative coefficient
on BMK in Table 4, panel C for abnormal working capital accruals in the “MROTA
versus NROTA” comparison sample (Ϫ0.541, t ϭ Ϫ1.676), which provides some
weak evidence that the audit quality of the mandatory rotation sample is higher
than that of the nonrotation sample.
27
To summarize, our findings in Table 3 are robust to five alternative accrual-
based measures of audit quality. Taken together (Tables 3 and 4), we find that
(a) the audit quality of the MROTA sample is largely indistinguishable from that of
the NROTA sample and that of the VROTA sample, and (b) the audit quality of the
MROTA sample is lower than the audit quality of MBEFR. Therefore, we find no
support for a belief that mandatory audit partner rotation enhances audit quality.
TABLE 4 (Continued)
Panel C: Negative accruals (Acc Ͻ 0) results
Performance-matched modified Jones model ϪϪ0.047 0.066
†
0.051
(PMMJAbnA)[Ϫ0.510] [1.998] [1.347]
Performance-matched Jones model ϪϪ0.073 0.070
‡
0.033
(PMJAbnA)[Ϫ0.941] [1.932] [1.026]
Modified Jones model ϪϪ0.060

*
0.031
*
0.001
(MJAbnA)[Ϫ2.703] [2.969] [0.096]
Jones model ϪϪ0.079
*
0.032
*
Ϫ0.007
(JAbnA)[Ϫ2.630] [2.931] [Ϫ0.600]
Current accruals ϪϪ0.065 0.031 0.028
(CurA)[Ϫ1.536] [1.419] [1.639]
Abnormal working capital accruals ϪϪ0.541
‡
0.382 0.145
(AbnWCA)[Ϫ1.676] [1.100] [0.600]
Notes:
Accrual measures are estimated or calculated using the procedure listed in the “Accrual
measure” column.
*
Significant at the 0.01 level based on a two-tailed t-statistic (in parentheses) in
panel A and a two-tailed z-statistic (in brackets) in panels B and C.
†
Significant at the 0.05 level based on a two-tailed t-statistic (in parentheses) in
panel A and a two-tailed z-statistic (in brackets) in panels B and C.
‡
Significant at the 0.10 level based on a two-tailed t-statistic (in parentheses) in
panel A and a two-tailed z-statistic (in brackets) in panels B and C.
Variable

Exp.
sign
MROTA
versus
NROTA
MROTA
versus
MBEFR
MROTA
versus
VROTA
378 Contemporary Accounting Research
CAR Vol. 26 No. 2 (Summer 2009)
Expanded sample period
The results in Tables 3 and 4 are based on year 2004 semi-annual financial state-
ments, which are the first set of data that reflect the full force of the mandatory
partner rotation rule in Taiwan. An interesting question is whether our findings are
robust when we expand the sample period.
28
To address this concern, we collected
additional semi-annual data for 2005 and 2006 from the TEJ database. We identify
132 companies whose audit partners are mandatorily rotated during 2005 and 2006
(MROTA0506) and 668 companies whose audit partners are not required to rotate
during 2005 and 2006 (NROTA0506). We then combine our original MROTA
(NROTA) with MROTA0506 (NROTA0506) and obtain MROTA0406
(NROTA0406). We still use VROTA as the voluntary rotation benchmark sample.
We repeat the analyses in Table 3 and Table 4 using our expanded comparison
samples: MROTA0406 versus NROTA0406, MROTA0406 versus MBEFR0406,
and MROTA0406 versus VROTA. Our findings are qualitatively identical to
those in Table 3 and Table 4 except that the evidence that the audit quality of

MROTA0406 is lower than the audit quality of MBEFR0406 becomes much
weaker (results untabulated). Additional analyses suggest that this is due to the fact
that the difference in partner tenure between MBEFR0406 and MROTA0406
becomes smaller than its counterpart between MBEFR and MROTA after adding
new observations in the years 2005 and 2006. The mean average partner tenure for
MBEFR0506 is only 2.981 years (its counterpart for MBEFR is 4.825 years)
because the MBEFR0506 sample covers years 2004–5 when the mandatory rota-
tion rule had become fully effective and thus there were no long-tenure partners
(i.e., tenure longer than five years) in the MBEFR0506 sample anymore.
One-partner rotation versus two-partner rotation
Our MROTA sample contains both one-partner rotation (MROTA1), where only
one of the two partners was rotated, and two-partner rotation (MROTA2), where
both partners were rotated. Conceivably, these two subsamples could differ in their
effect on audit quality. We conduct a battery of tests to address this concern.
First, we compare one-partner rotation sample (MROTA1) with three bench-
marks: NROTA, MBEFR1 (one-partner mandatory rotation sample in the prior year),
and VROTA1 (one-partner voluntary rotation sample, which is a part of VROTA).
We repeat all analyses in Tables 3 and 4 for MROTA1. Our findings are broadly
consistent with those in Tables 3 and 4, except that the evidence that the audit quality
of MROTA1 is lower than that of MBEFR1 is weaker (results untabulated).
Second, we compare the two-partner rotation sample (MROTA2) with three
benchmarks: NROTA, MBEFR2 (two-partner mandatory rotation sample in the
prior year), and VROTA2 (two-partner voluntary rotation sample, which is the other
part of VROTA). We repeat all analyses in Tables 3 and 4 for MROTA2. Our find-
ings are again broadly consistent with those in Tables 3 and 4, except that the evi-
dence that the audit quality of MROTA2 is lower than that of MBEFR2 is weaker
(results untabulated).
Finally, we directly compare MROTA2 with MROTA1 within the MROTA
sample. Specifically, we define a dummy variable, DM2, where DM2 is equal to 1
Mandatory Audit Partner Rotation, Audit Quality, and Market Perception 379

CAR Vol. 26 No. 2 (Summer 2009)
if two partners were both rotated, and 0 if only one partner was rotated. We then
estimate (2) after BMK is replaced by DM2 for the “MROTA2 versus MROTA1”
comparison for each of our six measures of accruals, with each accrual measured at
the absolute, positive, and negative values, respectively. We thus obtain 18 coeffi-
cients on DM2, each of which captures the incremental effect of MROTA2 relative
to MROTA1. We only find one coefficient that is significantly negative at the 0.1
level; the other 17 coefficients are all insignificant. Taken together, we conclude
that there is no significant incremental effect of MROTA2 relative to MROTA1.
Market-based proxy for investor perceptions of audit quality
In this section, we use the earnings response coefficient (ERC) estimated in concur-
rent returns–earnings regressions as a market-based proxy for investor perceptions
of audit quality, following Teoh and Wong 1993 and Ghosh and Moon 2005. We
first present the empirical model and then report our findings.
Empirical model and variable measurement
Following Ghosh and Moon 2005, we use the following model to examine whether
investors perceive mandatory audit partner rotation as enhancing audit quality:
CAR ϭ
␣
ϩ
␤
1
E ϩ
␤
2
⌬E ϩ
␤
3
BMK ϩ
␤

4
E ϫ BMK ϩ
␤
5
⌬E ϫ BMK
ϩ E ϫ CV
j
ϩ ⌬E ϫ CV
j
ϩ CV
j
ϩ
␧
(3),
where
CAR ϭ cumulative value-weighted market-adjusted abnormal returns over eight
months during January–August;
29
E ϭ income from continuing operations for the first half of year t, scaled by
the market value of equity at the beginning of January of year t;
⌬E ϭ change in income from continuing operations between the first half of
year t and the first half of year t Ϫ 1, scaled by the market value of equity
at the beginning of January of year t;
BMK ϭ a dummy variable equal to 1 if observations are from one of the three
benchmark samples (NROTA, MBEFR, or VROTA), and equal to 0 other-
wise; and
CV
j
ϭ one of the nine control variables discussed below, j ϭ 1, 2, … , 9.
The reason we measure cumulative abnormal returns (CAR) over an eight-

month period (January– August), instead of a six-month contemporaneous CAR
␤
62ϩ j 1Ϫ()
j
1ϭ
9
Α
␤
72ϩ j 1Ϫ()
j
1ϭ
9
Α
␤
23 jϩ
j
1ϭ
9
Α
380 Contemporary Accounting Research
CAR Vol. 26 No. 2 (Summer 2009)
(March– August), is that Ghosh and Moon (2005) argue that the ERC estimated
from (3) using a six-month contemporaneous CAR might be biased downward due
to prices leading earnings (see also Kothari 1992). Ghosh and Moon recommend
the use of a longer-period CAR to mitigate such a bias (see also Collins and Kothari
1989).
We estimate (3) over three comparison samples, MROTA versus NROTA,
MROTA versus MBEFR, and MROTA versus VROTA. Our primary interest is to
compare the ERC for the mandatory rotation sample (BMK ϭ 0) with the ERC for
a benchmark sample (BMK ϭ 1). The sum of the coefficients on E ϫ BMK and

⌬E ϫ BMK (
␤
4
ϩ
␤
5
) is the incremental ERC for a benchmark sample relative to
the MROTA sample after controlling for common determinants of the ERC and,
thus, is our coefficient of interest. If investors perceive mandatory audit partner
rotation as enhancing audit quality, then the ERC for the mandatory rotation sample
will be larger than the ERC for a benchmark sample. Hypothesis 2 thus predicts a
negative incremental ERC for the benchmark sample (i.e.,
␤
4
ϩ
␤
5
Ͻ 0).
We include nine control variables and their respective interactions with earn-
ings levels (E) and earnings changes (⌬E) in (3) following Ghosh and Moon 2005.
These control variables are: (a) audit firm tenure, FTenure, defined above; (b) com-
pany age, Age, defined above; (c) auditor type, Big4, defined above; (d) growth
potential, Growth, calculated as the sum of market value of equity and book value
of total debt, divided by book value of total assets at the end of the first half of year t;
(e) earnings persistence, Persist, calculated as the first-order autocorrelation of
income from continuing operations per share for the past 16 quarters; (f) earnings
volatility, Volaty, calculated as the standard deviation of income from continuing
operations per share for the past 16 quarters; (g) systematic risk, Beta, calculated
using the past 60 monthly returns with at least 45 nonmissing returns; (h) size,
MVE, measured as the natural logarithm of market value of equity at the end of the

first half of year t; and (i) financial leverage, LEV, measured as the ratio of total
debt to total assets at the end of the first half of year t.
Empirical findings based on earnings response coefficients
Due to the need for additional variables, our three comparison samples, MROTA
versus NROTA, MROTA versus MBEFR, and MROTA versus VROTA, are all
reduced in size. We lose some observations due to missing cumulative abnormal
returns (CAR). The biggest source for data loss is the need to have 60 past monthly
returns to calculate Beta. The final sample size for the above three comparison
samples is 422, 632, and 580 observations, respectively.
To mitigate potential undue influence of extreme values, we winsorize CAR,
E, ⌬E, Age, Growth, Persist, Volaty, Beta, MVE, and LEV at the top and bottom
1 percent of their distributions. We briefly discuss descriptive statistics for key
variables in each of the three comparison samples (untabulated). The mean CAR is
positive (0.003), whereas the median is negative (Ϫ0.014) for the MROTA versus
NROTA comparison sample. The mean and median CAR are both negative (Ϫ0.046
and Ϫ0.072) for the MROTA versus MBEFR comparison sample, and are again
both negative (Ϫ0.018 and Ϫ0.062) for the MROTA versus VROTA comparison
Mandatory Audit Partner Rotation, Audit Quality, and Market Perception 381
CAR Vol. 26 No. 2 (Summer 2009)
sample. The mean CAR is larger than median CAR for all three samples, suggesting
that returns are right-skewed. The mean E is smaller than median E for all three
samples, similar to the patterns in the United States (see, e.g., Ghosh and Moon
2005). The mean and median ⌬E are both positive, also similar to the U.S. data
reported in Ghosh and Moon 2005.
Our findings from estimating (3) are reported in Table 5. We find that the sum
of the coefficients on earnings levels and earnings changes (
␤
1
ϩ
␤

2
) is not sig-
nificant in each of our three comparison samples.
30
Our primary interest is the
incremental ERC (
␤
4
ϩ
␤
5
) for the benchmark sample. We find that the incremental
ERC is insignificantly different from zero for the MROTA versus NROTA compar-
ison sample (
␤
4
ϩ
␤
5
ϭ Ϫ0.021, F-stat. ϭ 0.00) and for the MROTA versus
MBEFR comparison sample (
␤
4
ϩ
␤
5
ϭ Ϫ0.279, F-stat. ϭ 0.42). These results
suggest that the ERC for the MROTA sample is not significantly different from the
ERC for the NROTA sample or the ERC for the MBEFR sample, failing to support
Hypothesis 2. On the other hand, the incremental ERC is significantly negative for

the MROTA versus VROTA comparison, suggesting that the ERC for the MROTA
sample is significantly larger than that for the VROTA sample, consistent with
Hypothesis 2. Out of three comparison samples, we only find support for Hypothe-
sis 2 in one comparison sample. We thus conclude that there is no consistent evi-
dence to suggest that investors perceive mandatory partner rotation as enhancing
audit quality.
We now briefly discuss control variables in Table 5. As in Ghosh and Moon
2005, we do not report the individual coefficients on the interactions between each
control variable and earnings levels (E) and earnings changes (⌬E). Instead, we
report the sum of the two interaction coefficients. First, we do not find that the ERC
increases with audit firm tenure after controlling for other determinants of the ERC
because
␤
6
ϩ
␤
7
is not significant in all three comparison samples, inconsistent
with Ghosh and Moon 2005. Second, we find some evidence that the ERC decreases
with firm age in the MROTA versus NROTA comparison sample (
␤
8
ϩ
␤
9
Ͻ 0),
whereas Ghosh and Moon (2005) do not find any relation between ERC and firm
age. Third, our ERC is larger for companies audited by Big4 auditors in the
MROTA versus MBEFR comparison sample, consistent with Teoh and Wong
1993. Finally, consistent with Ghosh and Moon 2005 and prior studies, we find that

the ERC increases in growth potential (Growth) and earnings persistence (Persist),
and decreases in systematic risk (Beta) and financial leverage (LEV).
6. Conclusion
In this paper, we examine the effect of mandatory audit partner rotation on audit
quality and perceptions of audit quality using audit data from Taiwan, where a five-
year partner rotation became de facto mandatory in 2004. Audit reports in Taiwan
contain both audit firm and partner names so that researchers can identify years in
which audit partners are rotated either voluntarily or mandatorily.
We first examine the effect of mandatory partner rotation on audit quality,
using both absolute and signed abnormal accruals as proxies for audit quality
(Johnson et al. 2002; Myers et al. 2003). We find that the audit quality of companies
382 Contemporary Accounting Research
CAR Vol. 26 No. 2 (Summer 2009)
TABLE 5
Earnings response coefficients and investor perceptions of mandatory audit partner rotation
Intercept
␣
0.270 (1.08) Ϫ0.661
*
(Ϫ2.82) Ϫ0.151 (Ϫ0.49)
E
␤
1
Ϫ11.098
*
(Ϫ2.91) 0.938 (0.22) 4.784 (1.58)
⌬E
␤
2
6.765

‡
(1.91) Ϫ2.418 (Ϫ0.50) Ϫ2.859 (Ϫ0.85)
␤
1
ϩ
␤
2
Ϫ4.333 [1.71] Ϫ1.480 [0.10] 1.924 [0.44]
BMK
␤
3
Ϫ0.033 (Ϫ0.92) Ϫ0.068
*
(Ϫ2.98) Ϫ0.020 (Ϫ0.57)
E ϫ BMK
␤
4
Ϫ0.801
‡
(Ϫ1.73) Ϫ0.245 (Ϫ0.80) Ϫ1.710
*
(Ϫ3.76)
⌬E ϫ BMK
␤
5
0.780 (1.50) Ϫ0.035 (Ϫ0.11) Ϫ0.105 (Ϫ0.28)
␤
4
ϩ
␤

5
Ϫ0.021 [0.00] Ϫ0.279 [0.42] Ϫ1.815
*
[16.98]
Control variables
E ϫ FTenure (
␤
6
)/⌬E ϫ FTenure (
␤
7
)
␤
6
ϩ
␤
7
0.007 [0.01] 0.048 [0.49] Ϫ0.043 [0.69]
E ϫ Age (
␤
8
)/⌬E ϫ Age (
␤
9
)
␤
8
ϩ
␤
9

Ϫ0.036
‡
[3.55] Ϫ0.017 [0.30] 0.016 [0.80]
E ϫ Big4 (
␤
10
)/⌬E ϫ Big4 (
␤
11
)
␤
10
ϩ
␤
11
0.577 [1.95] 0.927
‡
[2.81] Ϫ0.603 [1.79]
E ϫ Growth (
␤
12
)/⌬E ϫ Growth (
␤
13
)
␤
12
ϩ
␤
13

2.787
†
[5.60] 3.595
*
[7.68] Ϫ0.081 [0.02]
E ϫ Persist (
␤
14
)/⌬E ϫ Persist (
␤
15
)
␤
14
ϩ
␤
15
1.619
*
[6.88] 0.831 [1.32] Ϫ0.790 [1.36]
E ϫ Volaty (
␤
16
)/⌬E ϫ Volaty (
␤
17
)
␤
16
ϩ

␤
17
0.144 [0.17] 0.528 [0.60] Ϫ0.001 [0.00]
E ϫ Beta (
␤
18
)/⌬E ϫ Beta (
␤
19
)
␤
18
ϩ
␤
19
Ϫ1.842
*
[18.53] Ϫ1.381
*
[7.60] Ϫ1.012
†
[4.39]
E ϫ MVE (
␤
20
)/⌬E ϫ MVE (
␤
21
)
␤

20
ϩ
␤
21
0.287 [2.47] 0.080 [0.11] 0.133 [0.79]
E ϫ LEV (
␤
22
)/⌬E ϫ LEV (
␤
23
)
␤
22
ϩ
␤
23
Ϫ1.099 [0.66] Ϫ3.266
†
[3.86] Ϫ0.992 [0.95]
FTenure
␤
24
0.007
‡
(1.90) 0.000 (0.10) 0.005 (1.12)
Age
␤
25
0.003 (1.46) 0.000 (0.16) 0.002 (0.72)

(The table is continued on the next page.)
MROTA versus NROTA MROTA versus MBEFR MROTA versus VROTA
Variable Coefficient Coeff. est. Test-stat. Coeff. est. Test-stat. Coeff. est. Test-stat.
Mandatory Audit Partner Rotation, Audit Quality, and Market Perception 383
CAR Vol. 26 No. 2 (Summer 2009)
TABLE 5 (Continued)
Control variables (cont.)
Big4
␤
26
Ϫ0.010 (Ϫ0.30) Ϫ0.033 (Ϫ1.11) 0.033 (0.86)
Growth
␤
27
0.046 (0.94) Ϫ0.014 (Ϫ0.32) 0.140
*
(3.62)
Persist
␤
28
Ϫ0.020 (Ϫ0.52) 0.031 (0.81) 0.041 (0.82)
Volaty
␤
29
Ϫ0.013 (Ϫ0.27) 0.015 (0.36) 0.037 (0.74)
Beta
␤
30
Ϫ0.099
*

(Ϫ2.97) Ϫ0.065
†
(Ϫ2.18) Ϫ0.074
‡
(Ϫ1.69)
MVE
␤
31
Ϫ0.019 (Ϫ1.44) 0.030
†
(2.41) Ϫ0.008 (Ϫ0.47)
LEV
␤
32
0.139 (1.63) 0.129 (1.62) 0.245
†
(2.26)
Adj. R
2
0.364 0.219 0.148
n 422 632 580
(The table is continued on the next page.)
MROTA versus NROTA MROTA versus MBEFR MROTA versus VROTA
Variable Coefficient Coeff. est. Test-stat. Coeff. est. Test-stat. Coeff. est. Test-stat.