35
1 Mandatory Audit Partner Rotation and Audit Quality: Effect of Personal Relationships between Audit Partners Abstract Whether auditor rotation can improve audit quality is a focus of accounting and auditing research. This study considers the effect of audit partners’ personal relationships. Using a sample of Chinese firms undergoing auditor partner rotation, we find that personal relationships between previous and new audit partners are more likely to exist if the previous audit partner rotate back again after the cooling-off period. After rotation, if no personal relationship exists between the previous and new partners, the audit quality improves; however, if the previous and new partners have a personal relationship, the audit quality improvement is significantly reduced and in some cases diminishes completely. This study shows that previous audit partners may use new transitive audit partners to evade mandatory audit partner rotation, causing the desired objective to fail. Keywords: mandatory audit partner rotation; audit quality; personal relationships between audit partners.

Mandatory Audit Partner Rotation

Embed Size (px)

Citation preview

Page 1: Mandatory Audit Partner Rotation

1

Mandatory Audit Partner Rotation and Audit Quality: Effect of Personal

Relationships between Audit Partners

Abstract

Whether auditor rotation can improve audit quality is a focus of accounting and auditing research.

This study considers the effect of audit partners’ personal relationships. Using a sample of

Chinese firms undergoing auditor partner rotation, we find that personal relationships between

previous and new audit partners are more likely to exist if the previous audit partner rotate back

again after the cooling-off period. After rotation, if no personal relationship exists between the

previous and new partners, the audit quality improves; however, if the previous and new partners

have a personal relationship, the audit quality improvement is significantly reduced and in some

cases diminishes completely. This study shows that previous audit partners may use new

transitive audit partners to evade mandatory audit partner rotation, causing the desired objective

to fail.

Keywords: mandatory audit partner rotation; audit quality; personal relationships between audit

partners.

Page 2: Mandatory Audit Partner Rotation

2

1. Introduction

We investigate the effect of mandatory audit partner rotation on audit quality by considering

the effect of audit partners’ personal relationships. Audit quality determinants are a focus of the

accounting and auditing literature. Researchers have debated the effect of auditor tenure on audit

quality over the last half-century. One school suggests an “extended auditor-client relationships”

story, which argues that extended auditor-client relationships could have a negative effect on

audit quality due to the loss of auditor independence (e.g., see Mautz and Sharaf, 1961;Catanach

and Walker,1999; Raghunanthan et al.,1994; Dopuch et al.,2001). This story predicts that longer

auditor tenure is associated with lower audit quality and that mandatory audit partner rotation

should improve audit quality. Another school suggests a “new auditors learning costs” story,

which argues that audit are more likely to fail in the first years of a new auditor-client

relationship, as new auditors initially have to incur learning costs. This story predicts that longer

auditor tenure is associated with higher audit quality and that mandatory audit partner rotation

should not improve audit quality (e.g., see Berton, 1991; Palmrose, 1986; Palmrose, 1991;Petty

and Cuganesan, 1996;Geiger andRaghunandan,2002).

The empirical findings on this important debate are mixed. Consistent with the “extended

auditor-client relationships” explanation, Chi and Huang (2005) find that while audit quality

initially increases, it begins to decrease as a partner or audit firm’s tenure exceeds five years.

Carey and Simnett (2006) find a lower propensity to issue a going-concern opinion and just

beating (missing) earnings benchmarks following a long partner tenure. Consistent with the “new

auditors learning costs” explanation, Myers et al. (2003) investigate a U.S. sample to find that

Page 3: Mandatory Audit Partner Rotation

3

higher earnings quality (audit quality) is associated with longer auditor tenure, and conclude that

setting a longer auditor tenure results in auditors placing greater constraints on extreme

management decisions related to financial performance reporting on average. Further, Carcello

and Nagy (2004) find that fraudulent financial reporting is more likely to occur in the first three

years of an audit firm’s tenure.

Recent research examines the effects of audit partner rotation to shed light on this debate

directly. Nevertheless, the findings are mixed. The “extended auditor-client relationships”

explanation predicts that audit quality improves after auditor partner rotation. For example,

Hamilton et al. (2005) examine Australian data and find that as of 2002, new partners of Big 5

firms have potentially led to greater earnings conservatism. Firth et al. (2012b) examine Chinese

data and find that firms with mandatory audit partner rotations are associated with a significantly

higher likelihood of a modified audit opinion than no-rotation firms. The “new auditors learning

costs” explanation predicts that audit quality may not increase after partner rotation. For instance,

Chi et al. (2009) analyze Taiwan firms’ experienced mandatory auditor partner rotation and find

no support for the belief that it increases audit quality.

In this study, we propose reconciling prior findings by considering the personal relationships

between audit partners when examining the effects of audit partner rotation. We argue that an

audit partner rotation may not achieve its objective when the existing and incoming audit

partners have a close relationship. In essence, audit quality may not improve after such a false

rotation. We examine data from Chinese companies and find that false rotations do exist in the

country’s audit firms. We also consider interviews conducted with audit partners that revealed

extreme cases in which audit practices continued to be conducted by previous audit partners

Page 4: Mandatory Audit Partner Rotation

4

while the new audit partner signed the report afterwards.1We conjecture that this type of personal

relationship between audit partners can significantly alter the effect of audit partner rotation on

audit quality. We define a personal relationship as whether the previous and new audit partners

have a pre-existing working relationship (i.e., specifically whether the previous and new audit

partners already cooperated on auditing any listed company before mandatory rotation).If both

serve as CPAs signing an audit report, we identify them as having a personal relationship.

Researchers widely use working relationships as proxies for the close social net working

relationships between previous and new audit partners. This method is generally accepted by

academics such as Liu et al.(2011). Following the literature(Chenet al., 2009; Bartov et al., 2000;

Myers et al.,2003),we measure audit quality proxies as the absolute values of non-recurring

profit and loss (NRI), and use discretionary accrual(|DA|)to conduct robustness checks.

Using a sample of Chinese firms undergoing auditor partner rotations, we hypothesize and

find that personal relationships between previous and new audit partners are more likely to exist

if the previous audit partner rotate back again after the cooling-off period. We also hypothesize

and find that if no personal relationship exists between the previous and new partners, the audit

quality increases. However, if a personal relationship exists between the previous and new

partners, the auditor quality improvement is significantly reduced and in some cases diminishes

completely after controlling for other variables.

This study contributes to the literature in the following three ways. First, by considering the

effect of personal relationships, we reconcile the different findings on the effects of audit partner

1 For example, in 2010, Shanghai Aijian Corporation (stock code: 600643) was exposed as being involved in a

financial fraud of RMB1.7 billion Yuan during 1998-2002. At the same time, Aijian was audited by Shanghai’s

Lixin Audit firm. The auditing firm always issued unqualified opinions (standard audit opinions).However,

according to later investigation, the two CPAs, Dai Dingyi and Zhou Qi, did not participate in the audit process but

only signed the reports. Source: http://finance.cb.com.cn/13531828/20100305/176801_2.html.

Page 5: Mandatory Audit Partner Rotation

5

rotation. In particular, a mandatory rotation may not be effective if there is no arm’s-length

distance between the previous and new audit partners.

Second, we provide new evidence in the debate on the effectiveness of mandatory audit

rotation. Using an audit rotation setting, we demonstrate that drawing inferences on the effects of

audit tenure on audit quality is more complicated than it appears. Considering factors such as

personal relationships could change the conclusion. We further demonstrate that while authentic

rotation improves audit quality, a false rotation may not.

Last but not least, this study has policy implications for regulators. We show that even

mandatory auditor rotation cannot necessarily fix the negative effect of “extended auditor-client

relationships.”Regulators should be aware of such loopholes in their requirements and amend

them accordingly.

The rest of the paper is organized as follows. Section 2 reviews the relevant literature,

introduces the institutional knowledge in our research setting and develops our hypotheses.

Section 3 presents the sample, data and regression models. Section 4 shows the empirical

analysis, and Section 5 concludes the paper.

2. Literature review, institutional background and hypothesis development

2.1. Literature review

The effects of auditor tenure and audit partner rotation on audit quality are widely debated.

One school of researchers suggests an “extended auditor-client relationships” story, pointing out

that auditor-client relationships can get increasingly closer along with audit tenure extensions.

When “extended auditor-client relationships” are established, auditors are more inclined to give

favorable opinions because their relationships are close. For instance, auditors are more likely to

Page 6: Mandatory Audit Partner Rotation

6

trust their clients and thus reduce necessary audit processes, handle investigations carelessly and

accept written or oral evidence. These practices can consequently lead to decreases in audit

quality(Mautz and Sharaf, 1961; International Federation of Accountants,2010). The argument

that audit quality is negatively correlated with audit tenure is demonstrated in some empirical

studies. Chi and Huang (2005) find that while audit quality increases initially, it begins to

decrease as a partner or audit firm’s tenure exceeds 5years.Carey and Simnett (2006) find a

lower propensity to issue a going-concern opinion and just beating (missing) earnings

benchmarks following a long partner tenure. According to the “extended auditor-client

relationships” view, a longer audit tenure can easily lead to the loss of audit independence and

the erosion of audit quality, and thus supports regulators in enforcing mandatory auditor rotation

policies.

The other school of researchers suggests a “new auditors learning costs” story. They argue that

longer auditor tenures indicate higher audit quality. Their underlying reasoning is that an

auditor’s tenure extension can help the auditor gain more special audit experience and

professional competence and increase the level of his or her knowledge of specific risks, thereby

improving audit quality (Petty and Cuganesan, 1996; Geiger and Raghunandan,2002; Myers et

al.,2003). Many empirical findings support this view. Johnson et al. (2002) find that whereas

short audit firm tenures are associated with larger absolute discretionary accruals, long audit firm

tenures are not. Carcello and Nagy (2004) find that fraudulent financial reporting is more likely

to occur in the first 3 years of an audit firm’s tenure. Chen et al. (2008) find that absolute

discretionary accruals decrease significantly with both audit partner and audit firm tenures.

Myers et al. (2003) find that accruals decline with longer audit firm tenures. Ghosh and

Moon(2005)find a positive association between earnings response coefficients and audit firm

Page 7: Mandatory Audit Partner Rotation

7

tenure. Mansi et al. (2004) find a significantly negative relation between audit firm tenure and

the cost of corporate bonds. According to the “new auditor learning costs” view, restrictions on

auditor tenure can limit the function of the “learning effect,” thus weakening the auditor’s

professional competence. Therefore, it does not support regulators implementing the mandatory

auditor rotation requirement.

Studies offer indirect evidence on the effects of audit firm or auditor tenures on audit quality.

We can directly examine how mandatory auditor rotation affects audit quality. More and more

studies have considered this question in recent years. Many use experimental research methods,

as researchers in most countries (including the U.S. and Canada) cannot obtain information about

auditor partners. Regulators do not require detailed disclosures on this kind of

information(Bamber and Bamber, 2009).The literature’s findings are also mixed on whether

auditor rotation can improve audit quality. Some studies find that rotation cannot improve audit

quality. Chi et al.(2009)arrive at this conclusion using data from Taiwan. Ruiz-Barbadillo et al.

(2009) achieve the same result using Spanish archival data. However, other studies draw

different conclusions. Hamilton et al. (2005) examine Australian data and find that as of 2002,

new partners of Big 5 firms have potentially led to greater earnings conservatism. Firth et al.

(2012b) find that firms with mandatory audit partner rotations are associated with a significantly

higher likelihood of modified audit opinion than no-rotation firms.

The literature that uses experimental research methods to examine the effect of mandatory

auditor rotation produces various findings. For instance, Emby and Favere-Marchesi (2005) find

that new concurring partners are more likely than continuing partners to conclude that goodwill

is impaired, indicating that partner switches can improve audit quality. Bedard and Johnstone

(2010) find that while planned engagement efforts can increase after a partner rotation, clients do

Page 8: Mandatory Audit Partner Rotation

8

not compensate for this. In their experimental study, Dopuch et al. (2001) find that the rotation

requirements in the third and fourth regimes decrease auditor-subjects’ willingness to issue

biased reports. Some of the literature finds that other factors can influence the effect of

mandatory auditor rotation. For example, Fargher et al. (2008) find that abnormal accruals

decrease in the early years following a partner switch from the same audit firm, and increase if

the new partner is from a different audit firm.

In summary, the literature has not yet reached consistent conclusions about the relationship

between auditor tenure and audit quality or the effectiveness of mandatory auditor rotation.

Taking advantage of China’s unique institutional background on audit data, we investigate the

effects that personal relationships between previous and new auditors have on audit quality after

auditor rotation.

2.2.Institutional background on China’s mandatory auditor rotation policy

The differences between mandatory auditor rotation policies among various countries are huge.

Mandatory auditor rotation has not appeared in countries such as Italy and Brazil in decades.

Other countries such as Spain adopted mandatory rotation only to later abolish it due to intense

opposition. Some jurisdictions such as China and Taiwan did not carry out mandatory auditor

rotation until recently, when the disclosure of financial fraud at Enron and other firms caught the

authorities’ attention.

For the moment, China’s regular auditor rotation policy is aimed at certified public

accountants (CPAs), whose names are signed on audit reports. In addition, regulators stipulate a

series of rules. As early as June 2002, the Chinese Institute of Certified Public Accountants

(CICPA) issued the Directive Suggestions on CPA’s Professional Ethics, Article 15 of which

Page 9: Mandatory Audit Partner Rotation

9

requires audit firms to regularly rotate auditors in charge of projects and the CPAs signed on to

the audit reports to improve auditor independence.

In October 2003, the CICPA and Ministry of Finance (MOF) of China jointly issued the

Regulations on the Regular Rotation of the CPA Engaging on the Auditing of Securities and

Futures, the major articles of which areas follows. One CPA cannot spend more than 5 years

continuously auditing a firm. If a CPA changes to a new audit firm after working at a previous

firm, the time he or she spends auditing a company in the two different audit firms should be

aggregated. A CPA who reaches the limit of 5 years and is rotated cannot resume audit services

for the former firm within 2 years; this is defined as the “cooling-off period.” If a firm completes

its initial public offering(IPO), a CPA cannot spend longer than 2 full financial years

continuously auditing the firm. Above all, the regulations were required to be enforced on

January 1, 2004. With the implementation of these policies, China began to adopt mandatory

auditor rotation in practice.

On January 17, 2004, the MOF published the Regulations on Improving and Enhancing the

Quality of Corporate Financial Reports Audit, which focuses on the auditor rotations of all kinds

of non-financial companies owned or partly owned by the state. Article 14 of the regulations

requires a company to change auditors after being audited by the same auditor for 5 consecutive

years.

In summary, Chinese mandatory auditor rotation has the following basic features: the previous

auditor cannot audit a firm for more than 5 consecutive years (or 2 years for an IPO company).

After a 2-year cooling-off period in which the previous auditor cannot resume his or her audit

service, her or she can rotate back to audit the firm. At the same time, according to the relevant

regulations, both of the CPAs must sign and stamp the audit reports. One auditor must be either a

Page 10: Mandatory Audit Partner Rotation

10

partner of the audit firm or the chief CPA, and the other must be responsible for the fieldwork.

The CRSC also requires that every listed company report the names of the two incumbent

auditors, which audit firm(s) they come from, and other details such as the names of the previous

and new auditors and the time of rotation. This unique disclosure institution is rarely seen in

other countries, offering us a special research setting to investigate how the personal

relationships between previous and new auditors affect the mandatory auditor rotation policy.

2.3.Hypothesis development

According to China’s current auditor rotation policy, if an auditor has audited a firm for 5

years, he or she must be replaced by another auditor from the same or a different audit firm. This

means the previous auditor must transfer his or her own client resources to other auditors. It

reflects a loss for the rotated auditor, especially in a country like China. Because the number of

listed companies is relatively small and the auditing market has become intensively competitive,

losing a listed company client means a large revenue loss for an auditor. From a self-interest

perspective, the previous auditor may attempt to rotate back to continue providing audit services

to the client after the cooling-off period. The action of rotating back is common. According to

the research sample used by Firth et al. (2012a), as many as 46.4% of previous auditors rotate

back to former clients after the cooling-off period. If the client has a close relationship with the

previous auditor, it is also inclined to continue the relationship after the cooling-off period. For

example, one interesting finding derived from the literature is that many clients follow their

auditors to new firms (Blouin et al., 2007; Chen et al., 2009).

If an auditor is motivated to rotate back, he or she may prefer that the new auditor be someone

with whom he has a personal relationship. A close relationship between a previous and new

auditor increases the chance that the previous auditor will rotate back to serve the original client

Page 11: Mandatory Audit Partner Rotation

11

after the cooling-off period. In a relationship-based society like China’s, people are inclined to

build “inner circles” in which intimate connections can encourage mutual help and construct

“win-win” scenarios (Bedford, 2011; Zhang and Li, 2003). Both previous and new auditors who

have close relationships prefer to help each other and introduce the other side as the new auditor.

Further, if a new auditor has a close relationship with a previous auditor, the latter is more likely

to interfere in the former’s auditing process to cater to the needs of clients and avoid losing them

to other audit firms. For instance, much of the literature finds that clients prefer to change audit

firms if they receive modified audit opinions (Chow and Rice, 1982).

In fact, previous auditors may influence the choice of new auditor candidates. As the CPA

partners, previous auditors are senior employees that make many important decisions. Choices

related to designating a new auditor are less important operational decisions that audit partners

can influence. At the same time, partners have absolute administrative and decision-making

authority over the resources they control. Because this mandate comes from the ways in which

audit resources embedded in the partner responsibility system adopted by almost all countries are

distributed, partners may choose their successors based on their own opinions.

In general, previous auditors are motivated to retain clients for as long as possible and prefer

to have close relationships with new auditors. Meanwhile, successors are inclined to compromise

with clients over the course of an audit, which may decrease its quality. This practice greatly

influences the effect of mandatory auditor rotation and suggests that the policy could fail. In

contrast, if a previous auditor has no motive to rotate back to a client, he or she has no need to

manipulate the new auditor choice. As a result, successors could be more independent of

predecessors, and audit quality may increase after the rotation, therefore optimizing the effect of

mandatory audit rotation. All of these discussions lead to the following two related hypotheses.

Page 12: Mandatory Audit Partner Rotation

12

H1: For a company whose audit is conducted by an auditor who rotates back, the possibility of

a personal relationship between the previous and new auditor is higher than that for companies

whose audit is conducted by an auditor who is not rotating back.

H2: If the previous and new auditors have a personal relationship, the audit quality after

mandatory auditor rotation does not improve. If the previous and new auditors have no personal

relationship, the audit quality after mandatory auditor rotation improves.

3. Date, sample and regression models

3.1. Data and sample

While we obtained most of the data from the China Stock Market Accounting Research

(CSMAR) database, the database lacked certain required details related to CPAs. We consulted

the Shanghai Securities Exchange(http://www.sse.com.cn/)and Shenzhen Securities Exchange

(http://www.szse.cn/)to supplement the relevant data. Because the Chinese regular mandatory

auditor rotation policy was put into force on January 1, 2004, only the post-2003financial reports

could be influenced. Thus, to analyze the information from the 5 prior years, our sample period

covers1998-2010. We treated the data according to the following measures. First, we excluded

observations of the financial industry, as its applied accounting standard is unique. Second, we

excluded the observations with missing data on the type of audit report. Third, we excluded the

observations missing the data required to calculate the DA. Fourth, we excluded the observations

that could not meet the criteria of mandatory audit partner rotations within audit firms.2 Finally,

we identified 3,755 firm-year observations from 454 listed companies. Panel A of Table 1

reports the sample selection process.

2Certain observations were excluded, such as 1) the observations of audit firms with tenures under 5 years (2years)and 2) the

observations of the auditor voluntary rotation within auditing firms in cases where the auditors’ tenures had not yet reached the 5- or2-year limits or where the auditors were not replaced despite reaching their tenure limits (violating the regulations).

Page 13: Mandatory Audit Partner Rotation

13

We partition the mandatory rotation timeline into three periods (shown in Figure 1) as follows.

The first period is the pre-mandatory rotation period, i.e., PreMR, which includes the 5 years

prior to the previous audit rotation(2 years in IPO companies). The second period is the post-

mandatory rotation period, i.e., PostMR, in which the new auditors provide audit services for

clients. As for the previous auditors, this is a cooling-off period. The final period is the rotating-

back or not-rotating-back period, i.e., RB or NotRB, in which the previous auditors either rotate

or do not rotate back to service their original clients.

Figure 1. The mandatory partner rotation periods

According to whether the previous auditors rotate back after the cooling-off period, we divide

the sample into two parts: the rotating-back sample (RB sample) and not-rotating-back sample

(NRB sample). Panel B of Table 1 shows the distributions of both samples in each year. Among

the 454 listed companies whose audit partners were mandatorily rotated during 1998-2010, 210

are included in the RB sample and 244 are included in NRB sample. The numbers of both

samples are roughly identical.

According to whether the previous and new auditors have a personal relationship, we divide

the sample into two additional parts: the personal-relationship sample (PR sample)and the no-

personal-relationship sample(NPR sample). Panel C of Table 1 shows the distributions of the PR

and NPR samples in the RB and NRB samples, respectively. Among the 210RB sample

observations, 168 observations belong to the PR sample for a weight of 80%. Among the 244

PreMR(5or2)…

*

PreMR(1)

(5/2)

(5/2)

2.PostMR 3.RB or NotRB

PostMR(1)… PostMR(n)

1.PreMR 1.PreMR

Page 14: Mandatory Audit Partner Rotation

14

NRB sample observations, 150 observations belong to the PR sample for a weight of 61%. If a

previous auditor rotates back after the cooling-off period, the odds that he or she has a personal

relationship with the new auditor is 19%higher than for a previous auditor who doesn’t rotate

back. This finding is consistent with H1.

[Insert Table 1]

3.2. Regression models

We use the following logistic regression model to test H1:

𝑃𝑅 = 𝛼0 + 𝛼1𝑅𝐵 + 𝛼2𝐺𝐸𝑁𝐷𝐸𝑅 + 𝛼3𝐴𝐺𝐸 + 𝛼4𝐸𝐷𝑈 + 𝛼5𝐸𝑋𝑃 + 𝛼6𝐵𝐼𝐺4

+𝛼7𝐺𝑅 + 𝛼8𝐿𝑇𝐴 + 𝛼9𝑂𝑃𝑅𝑂𝐴 + 𝛼10𝐿𝐸𝑉 + 𝛼11𝐿𝐼𝑄 + 𝛼12𝑅𝐸𝐶𝑉 + μ (1)

The explained variable in the model is PR, a dummy variable indicating whether the previous

and new audit partner have a personal relationship. We measure personal relationships as

whether the previous and new audit partners have a working relationship. Previous and new audit

partners are considered to have a personal relationship when they jointly perform audit services

for a listed company as the signing auditors. Many papers use the working relationships between

previous and new audit partners to measure their personal relationships, and academia generally

accredits this approach (e.g., Liu et al., 2011).

We are interested in the variable RB, a dummy variable that equals1 if the previous audit

partner rotates back and continues to perform audit services for his or her former clients after the

cooling-off period and0 otherwise. According to H1, the regression coefficient of RB should be

significantly positive.

Because the selection of a new audit partner maybe influenced by the characteristics of both

the previous audit partner and the client, we control for these two factors. We choose the

Page 15: Mandatory Audit Partner Rotation

15

previous partner’s gender, age, educational level and practice time and whether he or she works

at the Big Four as the controlling variables. GENDER is a dummy variable that equals 1 when

the previous audit partner is male and 0 otherwise. AGE represents his or her age; EDU is a

dummy variable that controls for his or her education level, andequals1 for a university degree or

above and 0 otherwise.EXP is used to control the audit partner’s practice time in years.BIG4

controls for the independent and professional competence of the previous audit partner,

andequals1 if he or she comes from the Big Four and 0 otherwise.

Following Wong et al. (2008), we control for the client level variables as follows. GROWTH

controls for the client’s growth, and is equal to the sales revenue growth rate. LTA controls for

the firm size, and is equal to the natural logarithm of the total assets at the end of the year.

OPROA is used to control for the performance, and is equal to the net profit divided by the year’s

total assets.LEV is the financial leverage of the client, and is equal to the total liabilities at the

end of the year divided by the total assets at the end of the year.LIQ is the current ratio, is equal

to the current assets divided by the current liabilities and indicates the short-term credit capacity.

RECV controls for the client’s accounts receivable level, which is equal to the accounts

receivable divided by the year’s total assets.

We apply the following OLS regression model to test H2:

𝑁𝑅𝐼 = 𝛽0 + 𝛽1𝑃𝑅 + 𝛽2𝑃𝑂𝑆𝑇 + 𝛽3𝑃𝑅 ∗ 𝑃𝑂𝑆𝑇 + 𝛽4𝐿𝑇𝐴 + 𝛽5𝐿𝐸𝑉 + 𝛽6𝑂𝑃𝑅𝑂𝐴

+𝛽7𝐿𝑂𝑆𝑆 + 𝛽8𝑅𝐸𝐶𝑉 + 𝛽9𝐼𝑁𝑉 + 𝛽10𝐶𝐴𝑆𝐻𝐹𝐿𝑂𝑊 + 𝛽11𝐵𝐼𝐺4 + 𝛽12𝐿𝐼𝑆𝑇𝐴𝐺𝐸

+𝛽13𝑌𝐸𝐴𝑅 + 𝛽14 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 + σ (2)

NRI is the explained variable in the model and refers to audit quality. A great deal of the

literature shows that the extent of earnings management is a valid audit quality proxy (Heninger,

2001; Geiger and Raghunandan, 2002; Bartov et al., 2000; Myers et al., 2003). Because listed

companies in China’s capital market generally prefer to manage earnings according to non-

Page 16: Mandatory Audit Partner Rotation

16

recurring items, much of the literature uses them to measure companies’ earnings management

levels. Whereas Chen and Wang (2004) use non-recurring items to measure the extent of

earnings management, Chen and Yuan (2004) consider industry-median-adjusted after-tax non-

operating gains and losses a more reasonable measurement (ENOI=industry-median-adjusted

after-tax non-operating profit/owner’s equity). We adopt the measurement used by Chen et al.

(2009), who measure audit quality according to industry-median-adjusted NRI, calculated as (net

profit–operating profit+profit from other operations)/total assets.

We are interested in the interaction term of PR and POST(PR*POST). POST is a dummy

variable that equals 1 when the study period occurs after mandatory audit rotation and0

otherwise. According to H2, the PR*POST regression coefficient should be significantly positive.

Following Blouin et al. (2007) and Chen et al. (2009), we control for the following variables

in the model. LTA controls for the corporate size, LEV for the financial risk and OPROA for the

profitability. LOSS is a dummy variable that equals 1 when the client incurs a loss and 0

otherwise. RECV and INV control for the receivables and inventories, respectively, which are

projects usually used for earnings management. They are also the key points of the audit, and

both influence the audit quality.BIG4controls for the effect of the audit firm type. CASHFLOW

controls for the client’s cash flow status, and is equal to the ratio of the net operating cash inflow

to the total assets. LISTAGE refers to the number of years the client has been listed. Studies find

it to have a significant effect on the extent of earnings management. YEAR and INDUSTRY

control for the differences in years and industries, respectively. LTA, LEV, OPROA, RECV, INV

and BIG4 are measured consistently.

4. Empirical results

Page 17: Mandatory Audit Partner Rotation

17

4.1. Descriptive statistics

The descriptive statistics of the variables are reported in Table 2. We divide the sample into

two groups and compare the variable differences. In the PR sample, the previous and new audit

partners have personal relationships. In the NPR sample, the partners do not have personal

relationships. The mean and median values of OPROA in the PR sample are 0.198 and 0.170,

respectively, significantly less than the 0.213 and 0.181 values in the NPR sample. The mean and

median values of INV for the PR sample(0.163 and 0.126, respectively) are significantly larger

than their counterparts for the NPR sample (0.143 and 0.118, respectively).The mean and median

values of LISTAGE in the PR sample are 5.679 and 5.000, respectively, both significantly larger

than the 4.777 and 4.000 values in the NPR sample. In the PR sample, there are 32 clients whose

net profits are less than 0 (1.67% of the total), and the corresponding number of clients in the

NPR sample is 11 (1.49% of the total). However, the differences between the two samples are

not significant, meaning that the proportions of clients in poor financial condition in the two

groups are indistinguishable. Fifty-three clients hire firms from the Big Four as their audit firms,

accounting for 2.77% of the total within the PR sample. However, a significantly larger

percentage (7.71%, 57) of clients hire from the Big Four within the NPR sample, showing that

the probability of personal relationships between previous and new audit partners is lower if they

come from the Big Four. We find no significant differences between the remaining variables in

the two groups.

[Insert Table 2]

4.2. Regression results

Model 1’s regression results are presented in Table 3. We begin by excluding all of the control

variables, and then add the control variables from the audit partner and client levels, respectively.

Page 18: Mandatory Audit Partner Rotation

18

Finally, we add all of the control variables. The results show that RB’s regression coefficients

are significantly positive in each regression, which is consistent with H1. For companies whose

audits are conducted by rotating-back auditors, the possibilities of a personal relationship

between the previous and new auditors is higher than for other companies whose audits are

conducted by not-rotating-back auditors. This finding suggests that previous audit partners affect

the selection of a new audit partner to make it easier to rotate back after the cooling-off period

and allow their own people to continue as their successors.

The control variable coefficient estimations are also reasonable. The GENDER coefficients are

significantly negative, meaning that female audit partners prefer to choose those they have

personal relationships with as new audit partners, possibly because they worry moreabout losing

clients. The BIG4coefficients are also significantly negative, suggesting that previous and new

audit partners are less likely to have a personal relationship if the previous audit partner is from

the Big Four.

[Insert Table 3]

We report our Model 2 regression results in Table 4.We begin by directly investigating the

effectiveness of mandatory audit partner rotation in Panel A without adding PR and PR*POST.

We then investigate the effectiveness under the condition that the previous and new audit

partners have a personal relationship in Panel B. In both of the regression results listed in Panels

A and B, we compare the NRI of all of the years before rotation with the NRI of the 2 years after

rotation [PreMRvs. PostMR(1-2)], and then respectively compare the NRI of all of the years

before rotation with the NRI of the first[PreMRvs. PostMR(1)]and second [PreMRvs.

PostMR(2)]years after rotation.

Page 19: Mandatory Audit Partner Rotation

19

Panel A shows that the POST coefficients are negative, although not significantly, in all three

regressions. This shows that regardless of the personal relationships of the previous and the new

audit partners, audit quality is not significantly improved under mandatory audit partner rotation,

which is consistent with Chi and Huang’s (2009) results.

As for Panel B, the POST coefficients are significantly negative in all of the regressions,

showing an increase in audit quality after mandatory audit partner rotation if no personal

relationship between the previous and new audit partners exists. The PR*POST coefficients are

significantly positive in the first two regressions and positive and marginally significant in the

final regression. This indicates that compared with the non-personal-relationship condition, the

increase in audit quality under mandatory audit partner rotation is significantly weaker under the

personal-relationship condition. The PR*POST+POST coefficients are insignificantly negative,

meaning there is no evidence that mandatory audit partner rotation is positively related to audit

quality if the previous and new audit partners have a personal relationship. Therefore, H2 is

supported.

The LAT, LEV, OPROA and LISTAGE coefficients are significantly negative, indicating a

lower extent of earnings management under a larger corporation size, a higher asset-liability ratio,

a better return on assets and a longer listing period. The LOSS coefficients are significantly

negative, showing that clients with negative net profits tend to limit their earnings management.

The CASHFLOW and BIG4coefficients are significantly positive, indicating that clients with

sufficient operating cash flows and Big Four audit firms may have higher earnings management

levels.

[Insert Table 4]

4.3. Robustness tests

Page 20: Mandatory Audit Partner Rotation

20

Following the literature, we test the robustness of our findings by using the absolute value of

discretionary accruals (|DA|) as an alternative audit quality measurement(Heninger, 2001; Geiger

and Raghunandan, 2002; Bartov et al., 2000; Myers et al., 2003).|DA| is calculated according to

the modified Jones Model (Dechow et al., 1996):

𝑇𝐴𝐶𝐶𝑖𝑡

𝑇𝐴𝑖𝑡−1= 𝛾1(1/𝑇𝐴𝑖𝑡−1) + 𝛾2[(∆𝑅𝐸𝑉𝑖𝑡 − ∆𝑅𝐸𝐶𝑖𝑡 )/𝑇𝐴𝑖𝑡−1] + 𝛾3 𝑃𝑃𝐸𝑖𝑡 / 𝑇𝐴𝑖𝑡−1 + 𝜈𝑖𝑡

(a)

TACCit refers to the total accruals and is equal to the change of the current assets of year t–the

change of the cash and cash equivalents of year t–the change of the current liabilities of year

t+the change of the short-term borrowing in current liabilities of year t–the change of the

depreciation and amortization expense of year t.TAit−1is equal to the total assets at the end of

year t-1. ∆REVit is the change in sales revenue between years t and t-1.RECit is equal to the

difference between the net receivables of years t and t-1.PPEit is equal to the book value of the

fixed assets.

We do the regression analysis by industry and year to estimateγ1, γ2 , γ3.;We then substitute the

parameters into equation B to calculate DA, and take its absolute value:

𝐷𝐴 = TACCit /TAit−1 − [𝛾1(1/𝑇𝐴𝑖𝑡−1) + 𝛾2 ∆REVit − ∆RECit

TAit−1 + 𝛾3(𝑃𝑃𝐸𝑖𝑡 /𝑇𝐴𝑖𝑡−1)]

(b)

The regression results of Model 1 and Model 2 with |DA| as the dependent variable are

reported in Table 5. Consistent with the previous analysis, in Panel A we directly investigate the

effectiveness of mandatory audit partner rotation without adding PR and PR*POST. In Panel B

we reinvestigate that effectiveness in the case of the predecessor and successor having a personal

relationship. In both of the regression results listed in Panels A and B, we compare the |DA| of all

Page 21: Mandatory Audit Partner Rotation

21

of the years before rotation with the |DA| of 2 years after rotation [PreMRvs. PostMR(1-2)], and

then respectively compare the |DA| of all of the years before rotation with the |DA| of the first

[PreMRvs. PostMR(1)]and second [PreMRvs. PostMR(2)]years after rotation.

The results in Panel A show that the POST coefficients are significantly negative in all three

regressions, indicating that audit quality is significantly improved after mandatory audit partner

rotation, even when personal relationships between the previous and new audit partners are not

considered. This differs from using NRI as an audit quality measurement. This finding is

consistent with studies on the influence of mandatory audit partner rotation or partner tenure on

audit quality, meaning different results may be achieved under alternative audit quality

measurements. For example, using a sample of U.S. companies from 1981-1998,Davis et al.

(2002)measure audit quality according to discretionary accrual variables and find that audit

quality decreases with the extension of a partner’s tenure.

Panel B shows that, consistent with the previous results, the POST coefficients are

significantly negative in all of the regressions, indicating an audit quality increase after

mandatory audit partner rotation if no personal relationship between the previous and new audit

partners exists. The PR*POST coefficients significantly and positively indicate that the

improvement in audit quality under mandatory audit partner rotation is significantly weaker

under the personal-relationship condition compared with the no-personal-relationship condition.

The PR*POST+POST coefficients are significantly negative in the first two regressions and

insignificant in the third. This indicates that although the audit quality increase is significantly

weakened, there is a certain degree of improvement when the previous and new audit partner

have a personal relationship.

Page 22: Mandatory Audit Partner Rotation

22

To summarize, while the |DA| and |NRI| regression results are not completely consistent, we

are interested in whether personal relationships have a negative influence on the effectiveness of

mandatory audit partner rotation. The results are consistent when two different audit quality

indicators are used, thus providing preferable support for H2.

[Insert Table 5]

5. Conclusion

Researchers have not yet reached a consensus on the effectiveness of mandatory audit partner

rotation and whether this rotation should be implemented. We conjecture that the mixed findings

may result from the omission of a correlated variable, i.e., personal relationships between audit

partners. Taking advantage of China’s unique practice of audit partner information disclosure,

we find that previous audit partners have an effect on the selection of a new audit partner, and

that those who rotate back prefer to choose partners with whom they have close personal

relationships. While this leads to no or limited audit quality improvement, if no personal

relationship between the previous and new audit partners exists, audit quality after rotation is

significantly enhanced. The results of this study show that in practice, a previous audit partner

may use a new, transitive audit partner to evade mandatory audit partner rotation, causing the

desired objective to fail.

Our findings enrich the literature on mandatory audit partner rotation, and make an important

inference on its perfection and supervision. While the requirement can be effective, in the

implementation process, a former audit partner’s rotation back after the cooling-off period may

need to be restricted, and his or her influence on the selection of a new audit partner may need to

be limited.

Page 23: Mandatory Audit Partner Rotation

23

References

Bamber, E. M., and Bamber, L. S., 2009, Discussion of discretionary accruals, audit-firm tenure and

audit-partner tenure: Empirical evidence from Taiwan, Contemporary Accounting Research,26(2):

393-402.

Bartov, E., Gul, F. A., and Tsui, J. S., 2000, Discretionary-accruals models and audit qualifications,

Journal of Accounting and Economics, 30: 421-452.

Bedford, O., 2011, Guanxi-building in the workplace: A dynamic process model of working and

backdoor guanxi, Journal of Business Ethics, 104: 149-158.

Bedard, J. C., Johnstone, K. M., 2010, Audit partner tenure and audit planning and pricing, Auditing: A

Journal of Practice & Theory, 29(2): 45-70.

Berton, L., 1991, GAO weighs auditing plan for big banks, Wall Street Journal (March).

Blouin, J., Grein, B., and Roundtree, B., 2007, An analysis of forced auditor change: The case of former Arthur Andersen clients, The Accounting Review, 82(3): 621-650.

Carcello, J. V., and Nagy, A. L., 2004, Audit firm tenure and fraudulent financial reporting, Auditing: A Journal of Practice and Theory, 23(2): 55-69.

Carey, P., and Simnett, R., 2006, Audit partner tenure and audit quality, The Accounting Review, 81: 653-676.

Catanach, A. H., and Walker, P. L., 1999, The international debate over mandatory auditor rotation: a conceptual research framework, International Accounting, Auditing & Taxation, 8(1): 43-66.

Chen, C.Y., Lin, C. J., and Lin, Y. C., 2008, Audit partner tenure, audit firm tenure, and discretionary

accruals: does long auditor tenure impair earnings quality? Contemporary Accounting Research, 25: 415-445.

Chen, C. J. P., Su, X., and Wu, X., 2009, Forced audit firm change, continued partner-client relationship, and financial reporting quality, Auditing: A Journal of Practice & Theory, 28(2): 227-246.

Chen, K. C. W., and Yuan, H., 2004, Earnings management and capital resource allocation: Evidence from China’s accounting-based regulation of rights issues, The Accounting Review, 79(3): 645–665.

Chen, S. M., and Wang, Y. T., 2004, Evidence from China on the value relevance of operating incomes. below-the-line items, The International Journal of Accounting,39(4): 339-364.

Chi, W., and Huang, H., 2005, Discretionary accruals, audit-firm tenure and audit-partner tenure:

Empirical evidence from Taiwan, Journal of Contemporary Accounting and Economics, 1(1): 65-92.

Chi, W., Huang, H., Liao, Y., and Xie, H., 2009, Mandatory audit partner rotation, audit quality, and

market perception: Evidence from Taiwan, Contemporary Accounting Research, 26(2): 359-391.

Chow, C.W., and Rice, S. J., 1982, Qualified audit opinions and auditor switching, The Accounting

Review, 57 (2): 326–335.

Dopuch, N., King, R. R., and Schwartz, R., 2001, An experimental investigation of retention and rotation

requirements, Journal of Accounting Research, 39(6): 93-117.

Page 24: Mandatory Audit Partner Rotation

24

Davis, L. R., Soo, B., and Trompeter, G., 2002, Auditor Tenure, Auditor Independence and Earnings

Management, Working paper, Boston College, Chestnut Hill, MA.

Dechow, P. M., Sloan, R. G., and Sweeny, A. P., 1995, Detecting earnings management, The Accounting

Review, 70(4): 193-225.

Emby, C., and Favere-Marchesi, M., 2005, The impact of continuity on concurring partner reviews: an

exploratory study, Accounting Horizons, 19(1): 1-10.

Fargher, N., Lee, H-Y., and Mande, V., 2008, The effect of audit partner tenure on client managers’

accounting discretion, Managerial Auditing Journal, 23:161-186.

Firth, M., Rui, O. M., and Wu, X., 2012a, Rotate back or not after mandatory audit partner rotation?

Journal of Accounting and Public Policy, 31(4): 356-373.

Firth, M., Rui, O. M., and Wu, X., 2012b, How do various forms of auditor rotation affect audit quality?

Evidence from China, The International Journal of Accounting, 47(3): 109-138.

Geiger, M. A., and Raghunandan, K., 2002, Auditor tenure and audit reporting failures, Auditing: A

Journal of Practice and Theory, 21(3): 67-78.

Ghosh, A., and Moon, D., 2005, Auditor tenure and perceptions of audit quality, The Accounting Review, 80(2): 585-612.

Heninger, W. G., 2001, The association between auditor litigation and abnormal accruals, The Accounting Review, 76(1): 111-26.

Hamilton, J., Ruddock, C. M. S., Stokes., D.J., and Taylor, S. L.,2005, Audit partner rotation, earnings quality and earnings conservatism, working paper, available at SSRN:

http://ssrn.com/abstract=740846.

International Federation of Accountants, 2010, Handbook of the code of ethics for professional

accountants.

Johnson, V. E., Khurana, I. K., and Reynpreviouss., J. K., 2002, Audit-firm tenure and the quality of financial reports, Contemporary Accounting Research, 19(4): 637-660.

Liu, J., Wang, Y., and Wu, L., 2011, The effect of Guanxi on audit quality in China, Journal of Business Ethics, 103 (4):621-638.

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., and Sharaf, H.A., 1961, The Philosophy of Auditing, American Accounting Association: 312-356.

Myers, J. N., Myers, L. A., and Omer, T.C., 2003, Exploring the term of the auditor-client relationship

and the quality of earnings: A case for mandatory auditor rotation? The Accounting Review, 78 (3): 779-799.

Palmrose, Z., 1986, The effect of non-audit service on the pricing of audit services: Further evidence, Journal of Accounting Research, 24: 405-411.

Page 25: Mandatory Audit Partner Rotation

25

Palmrose, Z., 1991, Trials of legal disputes involving independent auditors: Some empirical evidence,

Journal of Accounting Research, supplement: 149-185.

Petty, R., and Cuganesan, S., 1996, Auditor rotation: Framing the debate, Australian Accountant, 5: 40-41

Raghunanthan, B., Lewis, B. L. and Evans, J. H., 1994, An empirical investigation of problem Audits,

Research in Accounting Regulation, 8: 33-58.

Ruiz-Barbadillo, E., Gomez-Aguilar, N., and Carrera, N., 2009, Does mandatory audit firm

rotation enhance auditor independence? Evidence from Spain, Auditing: A Journal of

Practice and Theory, 28(1): 113-135.

Zhang, X., and Li, G., 2003, Does guanxi matter to nonfarm employment? Journal of Comparative Economics, 31: 315-331.

Page 26: Mandatory Audit Partner Rotation

26

TABLE 1

Descriptive Information on Sample Selection and Sample Distribution

Panel A: Sample selection

Total firm year observations available on CSMAR from 1998-2010 19,604

Less:

Observations of firms in the financial industry

(324)

Observations without audit opinions

(3)

Observations with insufficient data to calculate discretional accruals (882)

Observations that do not meet the criteria of mandatory audit partner rotation

within audit firms (14,640)

Final sample 3,755

Panel B: Sample (firms) composition by mandatory rotation year

Year RB Percent(%) NRB Percent(%)

2003

39

19

32

13

2004

42

20

32

13

2005

37

18

22

9

2006

47

22

52

21

2007

17

8

25

10

2008

22

10

31

13

2009

6

3

50

21

2010

0

0

0

0

Total 210 100 244 100

RB: after PostMR, the previous audit partner rotated back and performed audit services for the original client; NRB:

after PostMR, the previous audit partner did not rotate back to perform audit services for the original client; PR: the

previous and new audit partners have a personal relationship; NPR: the previous and new audit partners have no

personal relationship.

Page 27: Mandatory Audit Partner Rotation

27

TABLE 2

Descriptive Statistics for Dependent and Explanatory Variables (PR vs. NPR)

PR

NPR

Test of differences

Continuous

variables

(PR=1)

(PR=0)

(p-value)

Mean

Median

Mean

Median

Mean

Median

NRI

0.013

0.001

0.012

0.003

0.914

0.760

LTA

21.191

21.071

21.170

20.951

0.643

0.106

LEV

0.525

0.523

0.513

0.499

0.254

0.201

OPROA

0.198

0.170

0.213

0.181

0.016

0.012

RECV

0.134

0.108

0.132

0.110

0.615

0.690

INV

0.163

0.126

0.143

0.118

0.000

0.048

CASHFLOW 0.051

0.049

0.055

0.050

0.260

0.589

LISTAGE

5.679

5.000

4.777

4.000

0.000

0.000

Categorical

Variables

No. of

Obs.

Percent

No. of

Obs.

Percent

Mean

Median

LOSS

32

1.67%

11

1.49%

0.739

0.745

BIG4 53 2.77% 57 7.71% 0.020 0.000

PR: the previous and new audit partners have a personal relationship; NPR: the previous and new audit partners have

no personal relationship.

NRI: industry median adjusted NRI=(net profit–operating profit+profit from other operations)/total assets at the end

of the year; LTA: the natural logarithm of total assets at the end of the year=ln (total assets at the end of the

year);LEV: the asset-liability ratio=the total assets at the end of the year/the total liabilities at the end of the year;

OPROA: the return on assets=net profit/the year’s total assets; RECV: the receivable level=net accounts receivable/the year’s total assets; INV: the inventory level=net inventories/the year’s total assets; CASHFLOW: the

cash flow level=net operating cash inflow/the year’s total assets; LISTAGE: the number of years the client has been

listed; LOSS: a dummy variable equal to 1 when the net profit of the client is negative and 0 otherwise;BIG4: a

dummy variable equal to1 when the audit firm is from the Big Four and0 otherwise.

Page 28: Mandatory Audit Partner Rotation

28

TABLE 3

Model 1 Regression Results

Dependent Variable: PR

Intercept

0.4673

0.1662

-0.9836

-3.8862

(12.62)***

(0.05)

(0.24)

(2.53)

RB

0.9190

0.9119

0.9478

0.9435

(17.94)***

(15.99)***

(18.51)***

(16.66)***

GENDER

-0.4549

-0.4640

(3.39)*

(3.40)*

AGE

0.0071

0.0078

(0.14)

(0.16)

EDUC

0.1321

0.1762

(0.33)

(0.57)

EXPER

0.0316

0.0277

(0.52)

(0.39)

BIG4

-1.0028

-1.3614

(5.13)***

(6.82)***

GROWTH

0.2067

0.2228

(0.61)

(0.62)

LTA

0.0490

0.1705

(0.27)

(2.40)

OPROA

-1.0255

-0.7097

(0.59)

(0.26)

LEV

0.6752

0.6026

(1.58)

(1.30)

LIQ

0.0366

0.0230

(0.73)

(0.36)

RECV

0.5606

0.9758

(0.31)

(0.85)

Pseudo R2

0.3650

0.6490

0.6340

0.6670

χ2 statistic

18.87***

27.25***

23.49***

34.02***

n 454 436 454 436

*Significant at the 10% level. **Significant at the 5% level. ***Significant at the 1% level.

There are 18 auditors whose education level data are not available. When the control variables for the previous audit

partner’s personal characters are added to Model 1 for regression, the sample size is reduced to 436.

Regression model: PR = α0 + α1RB + α2GENDER + α3AGE + α4EDU + α5EXP + α6BIG4 + α7GROWT

+α8LTA+ α9OPROA + α10 LEV + α11 LIQ + α12 RECV + μ PR: a dummy variable equal to1 when the previous and new audit partners have a personal relationship and 0

otherwise; RB: a dummy variable equal to1 when the previous audit partner rotated back after PostMR and

performed audit services for the original client and 0 otherwise.

All of the control variables are computed based on the data in the year before the rotation year. GENDER: a dummy

variable equal to 1 when the previous audit partner is male and 0 otherwise; AGE: the previous audit partner’s age before mandatory audit partner rotation; EDU: a dummy variable controlling for the previous audit partner’s

Page 29: Mandatory Audit Partner Rotation

29

education level, equal to 1 for a university degree and above and 0 otherwise; EXP: the number of years the previous

partner practiced before mandatory audit partner rotation; BIG4: a dummy variable equal to 1 when the previous

audit partner is from the Big Four and 0 otherwise; GROWTH: the rate of sales revenue growth before mandatory

audit partner rotation, equal to(sales revenue of this year–sales revenue of last year)/sales revenue last year; LTA: the

natural logarithm of the total assets at the end of the year before mandatory audit partner rotation, equal to ln (total

assets at the end of the year);OPROA: the return on assets before mandatory audit partner rotation, equal to the net profit/the year’s total assets; LEV: the asset-liability ratio before mandatory audit partner rotation, equal to the total

assets at the end of the year/the total liabilities at the end of the year; LIQ: the current ratio before mandatory audit

partner rotation, equal to the current assets/the current liabilities; RECV: the receivable level before mandatory audit

partner rotation, equal to the net accounts receivable/the year’s total assets.

Page 30: Mandatory Audit Partner Rotation

30

TABLE 4

Model 2 Regression Results

Dependent Variable: NRI

Panel A

PreMRvs.PostMR(1-2)

PreMRvs.PostMR(1)

PreMRvs.PostMR(2)

Intercept

1.5838

1.6163

1.6135

(53.48)***

(51.74)***

(50.07)***

POST

-0.0038

-0.0037

-0.0034

(-1.22)

(-1.02)

(-0.80)

LTA

-0.0700

-0.0718

-0.0716

(-52.93)***

(-50.67)***

(-49.4)***

LEV

-0.0362

-0.0349

-0.0287

(-6.40)***

(-5.72)***

(-4.68)***

OPROA

-0.4932

-0.4937

-0.4869

(-53.67)***

(-49.36)***

(-47.64)***

LOSS

-0.0224

-0.0201

-0.0181

(-2.46)**

(-1.97)**

(-1.83)*

RECV

-0.0100

-0.0059

-0.0150

(-0.85)

(-0.48)

(-1.17)

INV

0.0259

0.0235

0.0228

(2.54)**

(2.17)**

(2.03)**

CASHFLOW

0.1415

0.1387

0.1439

(9.31)***

(8.51)***

(8.58)***

BIG4

0.0518

0.0492

0.0530

(8.16)***

(7.25)***

(7.43)***

LISTAGE

-0.0011

-0.0014

-0.0014

(-3.26)***

(-3.64)***

(-3.58)***

YEAR

Control

Control

Control

INDUSTRY

Control

Control

Control

Adj. R2

0.7125

0.7202

0.7115

F value

200.09

182.02

165.22

<.0001(p)

<.0001(p)

<.0001(p)

n 2,652 2,252 2,198

Page 31: Mandatory Audit Partner Rotation

31

Panel B

PreMRvs.PostMR(1-2) PreMRvs.PostMR(2) PreMRvs.PostMR(2)

Intercept

1.5887

1.6205

1.6169

(53.54)***

(51.77)***

(50.07)***

PR

0.0008

0.0008

0.0011

(0.24)

(0.25)

(0.35)

PR*POST

0.0111

0.0122

0.0110

(2.08)**

(1.84)*

(1.54)

POST

-0.0119

-0.0124

-0.0114

(-2.40)**

(-2.09)**

(-1.70)*

LTA

-0.0702

-0.0719

-0.0717

(-53.02)***

(-50.71)***

(-49.42)***

LEV

-0.0359

-0.0348

-0.0284

(-6.36)***

(-5.71)***

(-4.63)***

OPROA

-0.4930

-0.4936

-0.4866

(-53.70)***

(-49.39)***

(-47.63)***

LOSS

-0.0226

-0.0206

-0.0179

(-2.48)**

(-2.02)**

(-1.80)*

RECV

-0.0101

-0.0061

-0.015

(-0.86)

(-0.50)

(-1.17)

INV

0.0253

0.0230

0.0222

(2.48)**

(2.13)**

(1.98)**

CASHFLOW

0.1412

0.1381

0.1442

(9.30)***

(8.48)***

(8.60)***

BIG4

0.0532

0.0503

0.0539

(8.34)***

(7.36)***

(7.51)***

LISTAGE

-0.0012

-0.0014

-0.0014

(-3.41)***

(-3.72)***

(-3.65)***

YEAR

Control

Control

Control

INDUSTRY

Control

Control

Control

Adj. R2

0.7131

0.7205

0.7117

F value

189.24

171.68

155.99

<.0001(p)

<.0001(p)

<.0001(p)

n

2,652

2,252

2,198

F statistic for

POST+PR*POST=0

-0.0007

-0.0003

-0.0004

(p-value) (0.8350) (0.9449) (0.9274)

*Significant at the 10% level. **Significant at the 5% level. ***Significant at the 1% level.

PreMR:5 years before mandatory audit partner rotation (2 years in IPO companies); PostMR(1-2): the first and

second year after mandatory audit partner rotation; PostMR(1): the first year after mandatory audit partner rotation;

PostMR(2): the second year after mandatory audit partner rotation.

Regression model: 𝑁𝑅𝐼 = β0 + β1PR + β2POST + β3PR ∗ POST + β4LTA+ β5LEV + β6OPROA + β7LOSS + β8RECV + β9INV

Page 32: Mandatory Audit Partner Rotation

32

+β10 CASHFLOW + β11 BIG4 + β12 LISTAGE+ β13 YEAR + β14 INDUSTRY+ σ NRI: industry median adjusted NRI=(net profit–operating profit+profit from other operations)/total assets at the end

of the year; PR: a dummy variable equal to 1 when the previous and new audit partners have a personal relationship

and 0 otherwise; POST: a dummy variable equal to 1 when the study period is after mandatory audit partner rotation

and 0 otherwise; LTA: the natural logarithm of total assets at the end of the year=ln (total assets at the end of the year); LEV: the asset-liability ratio=the total assets at the end of the year/the total liabilities at the end of the year;

OPROA: the return on assets=net profit/the year’s total assets; LOSS: a dummy variable equal to 1 when the net

profit of the client is negative and 0 otherwise; RECV: the receivable level=net accounts receivable/the year’s total

assets; INV: the inventory level=net inventories/the year’s total assets; CASHFLOW: the cash flow level=net

operating cash inflow/the year’s total assets; BIG4: a dummy variable equal to 1 when the audit firm is from the Big

Four and 0 otherwise; LISTAGE: the number of years the client has been listed; YEAR: a year control variable;

INDUSTRY: an industry control variable.

Page 33: Mandatory Audit Partner Rotation

33

TABLE 5

Regression Results – Robustness Check

Dependent variable: |DA|

Panel A

PreMRvs.PostMR(1-2)

PreMRvs.PostMR(1)

PreMRvs.PostMR(2)

Intercept

0.3059

0.3467

0.3453

(5.46)***

(5.82)***

(5.52)***

POST

-0.0265

-0.0239

-0.0183

(-4.52)***

(-3.55)***

(-2.25)**

LTA

-0.0075

-0.0099

-0.0091

(-2.97)***

(-3.66)***

(-3.22)***

LEV

0.0522

0.0531

0.0504

(4.36)***

(4.10)***

(3.77)***

OPROA

0.0263

0.0344

0.0338

(1.49)

(1.78)*

(1.68)*

LOSS

0.0017

0.0099

0.0057

(0.10)

(0.52)

(0.30)

RECV

-0.0044

0.0075

0.0060

(-0.19)

(0.32)

(0.24)

INV

-0.0022

-0.0054

-0.0115

(-0.12)

(-0.26)

(-0.53)

CASHFLOW

-0.1033

-0.1158

-0.0795

(-3.63)***

(-3.76)***

(-2.46)**

BIG4

0.0228

0.0248

0.0228

(1.92)*

(1.93)*

(1.66)*

LISTAGE

-0.0086

-0.0109

-0.0107

(-13.25)***

(-15.22)***

(-14.34)***

YEAR

Control

Control

Control

INDUSTRY

Control

Control

Control

Adj. R2

0.1715

0.2153

0.1893

F value

17.63

20.31

16.54

<.0001(p)

<.0001(p)

<.0001(p)

n 2,652 2,252 2,198

Page 34: Mandatory Audit Partner Rotation

34

Panel B

PreMRvs.PostMR(1-2) PreMRvs.PostMR(1) PreMRvs.PostMR(2)

Intercept

0.3160

0.3540

0.3542

(5.62)***

(5.93)***

(5.65)***

PR

-0.0047

-0.0038

-0.0032

(-0.79)

(-0.65)

(-0.53)

PR*POST

0.0225

0.0208

0.0264

(2.25)**

(1.67)*

(1.93)*

POST

-0.0427

-0.0389

-0.0375

(-4.60)***

(-3.46)***

(-2.92)***

LTA

-0.0076

-0.0100

-0.0092

(-3.03)***

(-3.67)***

(-3.25)***

LEV

0.0527

0.0531

0.051

(4.40)***

(4.10)***

(3.81)***

OPROA

0.0265

0.0345

0.0343

(1.51)

(1.78)***

(1.71)*

LOSS

0.0016

0.0095

0.0065

(0.09)

(0.50)

(0.34)

RECV

-0.0045

0.0073

0.0060

(-0.20)

(0.31)

(0.24)

INV

-0.0025

-0.0054

-0.0118

(-0.13)

(-0.26)

(-0.55)

CASHFLOW

-0.1038

-0.1166

-0.079

(-3.64)***

(-3.79)***

(-2.45)**

BIG4

0.0239

0.0252

0.0234

(2.00)**

(1.95)*

(1.70)*

LISTAGE

-0.0086

-0.0109

-0.0107

(-13.29)***

(-15.20)***

(-14.34)***

YEAR

Control

Control

Control

INDUSTRY

Control

Control

Control

Adj. R2

0.1726

0.2156

0.1900

F value

16.80

19.20

15.72

<.0001(p)

<.0001(p)

<.0001(p)

n

2,652

2,252

2,198

F statistic for POST+PR*POST=0

-0.0202

-0.0181

-0.0111

(p-value) (0.0019) (0.0172) (0.2156)

*Significant at the 10% level. **Significant at the 5% level. ***Significant at the 1% level.

PreMR: 5 years before mandatory audit partner rotation (2 years in IPO companies); PostMR(1-2): the first and

second year after mandatory audit partner rotation; PostMR(1): the first year after mandatory audit partner rotation; PostMR(2): the second year after mandatory audit partner rotation.

Regression model:

Page 35: Mandatory Audit Partner Rotation

35

𝐷𝐴 = β0 + β1PR + β2POST + β3PR ∗ POST + β4LTA+ β5LEV + β6OPROA + β7LOSS + β8RECV + β9INV

+β10 CASHFLOW + β11 BIG4 + β12 LISTAGE+ β13 YEAR + β14 INDUSTRY+ σ |DA|: the absolute value of the discretionary accruals, calculated by the modified Jones Model; PR: a dummy variable

equal to 1 when the previous and new audit partners have a personal relationship and 0 otherwise; POST: a dummy

variable equal to 1 when the study period is after mandatory audit partner rotation and 0 otherwise; LTA: the natural logarithm of total assets at the end of the year=ln (total assets at the end of the year); LEV: the asset-liability ratio=the

total assets at the end of the year/the total liabilities at the end of the year; OPROA: the return on assets=net profit/the

year’s total assets; LOSS: a dummy variable equal to 1 when the net profit of the client is negative and 0 otherwise;

RECV: the receivable level=net accounts receivable/the year’s total assets; INV: the inventory level=net

inventories/the year’s total assets; CASHFLOW: the cash flow level=net operating cash inflow/the year’s total assets;

BIG4: a dummy variable equal to 1 when the audit firm is from the Big Four and 0 otherwise; LISTAGE: the number

of years the client has been listed; YEAR: a year control variable; INDUSTRY: an industry control variable.