gold et al - 2012 -the effect of engagement and review partner tenure and rotation on audit quality - evidence from germany [mapr]

The Effect of Engagement and Review Partner Tenure and
Rotation on Audit Quality: Evidence from Germany
Anna Gold*
VU University Amsterdam

Friederike Molls
Westfälische Wilhelms-Universität Münster

Christiane Pott
Westfälische Wilhelms-Universität Münster

Christoph Watrin
Westfälische Wilhelms-Universität Münster


*Corresponding author: Department of Accountancy (PGO); Faculty of
Economics and Business Administration; VU University Amsterdam; De
Boelelaan 1105; NL-1081 HV Amsterdam; The Netherlands; Tel: +31 20 59
82592

We thank the participants of the 2009 European Accounting Association Annual
Meeting, the 2009 Meeting of the Accounting Section of the German Academic
Association for Business Research (AS-VHB) in collaboration with the
International Association for Accounting Education and Research (IAAER), the
2009 Annual meeting of the German Academic Association for Business
Research (VHB), the 2009 EARNet Symposium, and the 2010 AAA American
Accounting Association Annual Meeting for many helpful suggestions. We
greatly appreciate the opportunity to use The Annual-Report-database available
from the Chair of Business Administration and Controlling at the Westfälische
Wilhelms-Universität Münster. All remaining errors are our own.

Electronic copy available at: />

The Effect of Engagement and Review Partner Tenure and
Rotation on Audit Quality: Evidence from Germany
Abstract:
This study contributes to the recent debate over the effect of audit partner tenure
and rotation on auditor independence, expertise and, ultimately, audit quality. We
investigate the effect of partner tenure and rotation on audit quality, using unique
1995-2010 data from all German listed companies for which we identify not only
the audit engagement but also the quality review partner. Using multiple absolute
and signed values earnings management measures as proxies for audit quality, we
find evidence of less income-reducing accounting with an increase in review
partner tenure (but not engagement partner tenure). Further, rotation of the review
partner (but not the engagement partner) is associated with more income-reducing
accounting. These findings underline the importance of distinguishing between
the two partner roles and support the expertise hypothesis with respect to the
review partner.
Keywords:

Audit partner rotation; Audit partner tenure; Audit quality;
Review partner

JEL descriptors:

G34, G38, M41, M42, M48

1

Electronic copy available at: />

The Effect of Engagement and Review Partner Tenure and
Rotation on Audit Quality: Evidence from Germany
1. Introduction
The harsh criticism of the auditing profession in the wake of corporate scandals
had a significant impact on regulators‘ activities worldwide. One issue early on
identified as potentially problematic (e.g., AICPA 1977; European Commission
1996) is the possibility of an excessive tenure period for audit partners, the
consequence of which might be that the auditor accepts questionable accounting
and reporting practices because of familiarity threats and/or in order to retain the
client. This reasoning is often referred to as the independence hypothesis (e.g.,
Mautz and Sharaf 1961; AICPA 1978; DeAngelo 1981a; SEC 1994; IFAC 2008).
As part of regulatory initiatives world-wide, the Sarbanes-Oxley Act (SOX 2002)
now requires rotation of both the audit engagement and review partner every five
(previously seven) years. Similarly, the 8th EU directive 2006/43/EC (European
Commission 2006) requires all 27 EU member states to implement mandatory
rotation of both engagement and review partners for audit reports dated after June
2008.1 The major expected benefit is that rotation will improve audit quality
because inappropriate accounting practices are more likely identified when a new,
and hence more independent, engagement partner assumes responsibility for the
audit (Carey and Simnett 2006). Meanwhile, opponents to the rotation

1

Member states are given discretion with regard to the exact rotation period.

2


requirement suggest a number of potential pitfalls that may counter the benefits.
Aside from the additional costs incurred to audit firms, a new partner may lack

the client-specific knowledge of risk, operations, and financial reporting practices,
which could potentially lead to lower audit quality—an argument known as the
expertise hypothesis (e.g., Solomon et al. 1999; Johnson et al. 2002). Therefore,
despite the widely adopted requirement of audit partner rotation, the standard‘s
consequences remain unclear and could be either beneficial or harmful for the
quality of the audit.
Recently, a few archival studies in Australia, Taiwan, and the United
States have examined tenure and rotation at the within-firm partner level by
observing signature changes on audit reports. With few exceptions (i.e., Chen et
al. 2008), results of these studies suggest that (excessive) tenure adversely affects
audit quality (Chi and Huang 2005; Carey and Simnett 2006; Manry et al. 2008)
and that audit quality is higher for firms under a mandatory partner rotation
regime than for firms not subject to rotation (Chi et al. 2009). Hence, these studies
largely support the independence hypothesis.
While prior studies have focused exclusively on tenure and/or rotation of
the engagement partner, relevant standards clearly require rotation of all key
partners, hence including the quality review partner (hereafter called ‗review
partner‘) alongside the engagement partner (e.g., SOX 2002; European
Commission 2006). The purpose of the engagement quality review is to provide

3


quality control for audit engagements and to serve as an evaluation of the
performance of the audit engagement partner and team (Epps and Messier 2007).
As such, the review partner plays a vital role in the audit review process and for
the attainment of a high quality audit; however, no prior research has examined
either tenure or rotation effects with respect to the review partner separately. In
Germany, both partners‘ signatures must be visible and identifiable in the audit
report and both partners are formally responsible for the audit (§ 322 HGB of the

German Commercial Code). This practice is a characteristic of the dual control
principle common in German business law. As such, German audit report data
offers the unique opportunity to examine tenure and rotation of the review partner
separately from the engagement partner by means of the location of their
respective signatures in the audit report. While (excessive) tenure of the
engagement partner has largely been shown to aggravate audit quality and there is
some evidence that engagement partner rotation provides a viable solution,
dynamics might be significantly different when considering the role of the review
partner. More specifically the review partner might be relatively less susceptible
to independence threats than the engagement partner, because involvement with
the client is less extensive and frequent due to the review partner‘s primary role of
quality control. Further, it is possible that the review partner would primarily
benefit from tenure and that rotation might reduce audit quality, because the
accumulated experience with the client is not always transferred to a new review

4


partner. While refraining from formulating directional hypotheses in this regard,
this paper contributes to the literature by examining whether explanations of
independence or expertise differ for tenure/rotation effects of the engagement and
review partner, respectively.
Consistent with prior studies (e.g., Chi et al. 2009), we examine these
effects using absolute and signed performance-matched abnormal accruals as
proxies for audit quality. We identify a sample of 2,636 firm-year observations for
the period 1995-2010 in Germany, hence including a period where partner
rotation was legally mandated from 2002 onwards.2 Descriptive results reveal that
in the German mandatory rotation setting, rotation occurs even more frequently
than the standard requires, such that the mean of partner tenure is less than three
years, and the great majority of German firms implement rotation of the review

partner (98.56%) and audit engagement partner (99.81%) before the seventh year
of tenure. Our regression results suggest that tenure of the review partner is
associated with an increase in audit quality (reflected by income-decreasing
accounting procedures), while rotation of the review partner leads to a significant
decrease in audit quality. Interestingly, audit quality is not significantly associated
with either engagement partner tenure or engagement partner rotation. Overall,
our paper provides evidence in favor of the expertise hypothesis, but only for the
review partner.
2

The standard was introduced retro-actively; meaning that engagement and review partners
starting in 1995 or earlier were forced to rotate ultimately by 2002.

5


Our study contributes to the debate on mandatory auditor rotation in at
least two respects. First, in an extension of Chi and Huang (2005), Carey and
Simnett (2006), Chen et al. (2008), and Manry et al. (2008), our analysis provides
new empirical evidence by using financial reports on partner (and firm) level
rotation and tenure data in a highly developed audit environment, i.e., Germany.
Second and more importantly, our unique distinction between engagement partner
and review partner signatures allows us to differentiate between tenure and
rotation of these two partner roles, respectively, and to further disentangle
competing hypotheses of expertise and independence. Although our findings are
in part based on a time period preceding the introduction of mandatory rotation,
we believe study results are interesting for policy makers and regulators in
countries where empirical investigation of engagement and review partner tenure
and rotation is difficult or impossible (e.g., the United States).
The remainder of the paper is organized as follows. Section 2 describes

recent key policy initiatives in the areas of audit partner rotation and the role of
the review partner in major jurisdictions (including Germany), reviews related
empirical literature, and develops study hypotheses. Section 3 discusses the
research design, sections 4 and 5 report the empirical results, and section 6
concludes the paper.

6


2. Recent key policy initiatives, literature review and hypotheses
Audit partner tenure and rotation
Audit partner rotation is required in many countries, but the maximum tenure
allowed varies considerably. In the 1970s, the United States was the first
legislation to require rotation of the audit partner after seven years of tenure on
audits of SEC-registered clients. With the implementation of SOX in 2002, the
rotation period was further reduced to five years for public company engagements
and now includes the rotation of both the engagement partner and the review
partner. Similarly, the Department of Trade and Industry and the U.K. Treasury
implemented a regulation that requires rotation of the lead engagement partner
after five years and of other key engagement partners after seven years. By the
end of June 2008, all member states of the European Union were required to enact
the revised 8th Directive‘s requirements into national law (European Commission
2006). One important detail of the Directive is rotation of the key audit partner,3
but member states are given discretion regarding the maximum tenure allowed.
Even before the EU Directive was issued, Germany had revised its
Commercial Code by means of the ‘Gesetz zur Kontrolle und Transparenz im
Unternehmensbereich’ implemented in March 1998, with the rotation requirement
3

The key audit partner is typically the partner that signs the audit report, but may also denote

either (a) the statutory auditor(s) designated by an audit firm for a particular audit engagement as
being primarily responsible for carrying out the statutory audit on behalf of the audit firm; or (b) in
the case of a group audit, at least the statutory auditor(s) designated by an audit firm as being
primarily responsible for carrying out the statutory audit at the level of the group and the statutory
auditor(s) designated as being primarily responsible at the level of material subsidiaries.

7


becoming effective in 2002. As a result, an audit partner was forced to rotate from
the engagement if he/she had signed the audit report six times during a ten-year
auditor-client relationship. The ‘Bilanzrechtsreformgesetz’ in October 2004
modified the existing requirement to a situation where rotation was mandated
after a seven-year engagement. This law is considered a direct response to SOX
(2002) and the European Commission Recommendation (European Commission
2002) (Veltins 2004).4 Next, we discuss prior research and develop hypotheses
regarding the potential effects of engagement partner tenure and rotation on audit
quality.
Engagement partner tenure, rotation, and audit quality
Audit quality is defined as the joint probability that an auditor will both detect and
report material misstatements (DeAngelo 1981b). According to this definition,
audit quality is a function of the auditor‘s ability to detect material misstatements
(auditor expertise) and to report those detected misstatements (auditor
independence). Prior literature and theory explain how the length of the
relationship between the auditor and the client (i.e., tenure) may affect audit
quality. Interestingly, there is an inherent conflict regarding such effects, as
4

As part of this major reform, § 319a HGB was issued for public interest companies. This
legislation defines circumstances under which an auditor must immediately be excluded from the

audit. Beyond the reasons stated in § 319 II and III HGB, an auditor is also excluded from the
audit of a client using an organized market, as defined in § 2 V Securities Trading Act4
(Wertpapierhandelsgesetz), if the key audit partner(s) has issued (i.e. signed) the audit opinion
(Bestätigungsvermerk–§ 322 HGB) for seven or more consecutive years (§ 319a I no. 4). This
requirement is waived if there has been a cooling-off period of three or more years since the last
audit.

8


reflected by the hypotheses of independence and expertise (e.g., Mautz and Sharaf
1961; Shockley 1981; Iyer and Rama 2004).
On one hand, audit quality may be compromised as auditor tenure
increases, an argument often referred to as the independence hypothesis. More
specifically, an increase in audit partner tenure might introduce a familiarity threat
to auditor independence (IFAC 2008). The IFAC Code of Ethics states that a
―familiarity threat occurs when, by virtue of a close relationship with an assurance
client, its directors, officers or employees, a firm or a member of the assurance
team becomes too sympathetic to the client‘s interests‖ (IFAC 2008, p. 18). This
threat could result in a decline in the audit partner‘s ability to accurately judge the
company‘s performance and the client‘s reporting decisions. A related
explanation is the threat of routine. While an audit partner will likely apply
creative approaches to audit-testing in the early years of tenure (AICPA 1978;
Hoyle 1978; McLaren 1958), deeper knowledge over time about the client‘s
systems and control procedures may result in more routine audit programs (Hoyle
1978). Shockley (1981, p. 789) asserts that ―complacency, lack of innovation, less
rigorous audit procedures and a developed confidence in the client may arise after
a long association with the client.‖ The major concern is that prior year audits
may cause the partner to anticipate current results instead of carefully and
objectively examining the current year‘s data for material misstatements

(Arrunada and Paz-Ares 1997). In conclusion, the independence hypothesis

9


predicts that auditor tenure is negatively associated with audit quality, while
auditor rotation would have positive outcomes.
In contrast, the expertise hypothesis (e.g., Solomon et al. 1999; Johnson et
al. 2002) contends that audit quality may increase with auditor tenure because the
auditor gains more knowledge of the client and the client‘s industry over time.
This explanation is based on information asymmetry between the client and the
auditor, which decreases over time, as the auditor acquires client- and industryspecific knowledge. Because increased client-specific knowledge provides a
comparative advantage in detecting material misstatements in financial reports,
lack of this knowledge in the early years of an audit engagement may result in a
lower quality audit, while tenure should have a favorable effect on audit quality
(Beck et al. 1988; Hoyle 1978; Knapp 1991; Solomon et al. 1999; Geiger and
Raghunandan 2002; Myers et al. 2003). Consequently, auditor rotation would
have adverse effects on audit quality according to the expertise hypothesis.
Given the limited availability of empirical data on the association between
auditor tenure/rotation and audit quality at the audit partner level, most prior
research has examined this relationship at the audit firm level. Archival evidence
on audit firm tenure shows mixed results as to which theory dominates with
respect to audit quality – the independence hypothesis (e.g., Kealey et al. 2007;
Mansi et al. 2004; Stanley and DeZoort 2007) or the expertise hypothesis (e.g.,
Carcello and Nagy 2004; Davis et al. 2009; DeFond and Subramanyam 1998;

10


Ghosh and Moon 2005; Johnson et al. 2002). However, Bamber and Bamber

(2009) caution that findings based on audit firm rotation may not be generalizable
to audit partner rotation, because of different cost and benefits effects.
Recent archival research has considered the issue of rotation and tenure at
the engagement partner level and provides mixed results. The mandatory
inclusion of partner signatures on audit reports in some countries, such as
Australia and Taiwan, has enabled such endeavors. Carey and Simnett (2006)
conducted one of the first studies in this area by examining Australian listed
companies. The authors reach varying conclusions, depending on the measure of
audit quality used. The non-Big 6 (but not Big 6) auditors‘ propensity to issue
going-concern audit opinions for distressed companies was less likely as audit
partner tenure increased. However, the authors found no evidence of an
association between audit partner tenure and either the signed or absolute amount
of abnormal working capital accruals. Using Taiwanese data, Chi and Huang
(2005) found that engagement partner (and audit firm) tenure initially led to
higher earnings quality (particularly for Big 5 auditors), but excessive familiarity
with the client (i.e., more than five years) negatively affected earnings quality.
Again using a sample of Taiwanese companies, Chen et al. (2008) found that the
absolute and positive values of discretionary accruals decreased significantly with
audit partner tenure, hence leading to an increase in audit quality. Chi et al. (2009)
investigated the effects on audit quality of the mandatory audit partner rotation

11


implemented in Taiwan in 2004. Using absolute and signed abnormal accruals
and abnormal working capital accruals as proxies for audit quality, they found
that in 2004, audit quality was higher for companies subject to the mandatory
rotation regime than for firms not subject to rotation. Finally, the only United
States study on audit partner tenure to date (Manry et al. 2008) found that partner
tenure was positively associated with audit quality, using estimated discretionary

accruals as a proxy. Excessive tenure (i.e., more than seven years) again had
negative effects on audit quality, but only for small clients.
In experimental settings, partner rotation decreases auditors‘ willingness to
issue biased audit reports (Dopuch et al. 2001) and strengthens independence in
appearance of non-professional investors (Kaplan and Mauldin 2008), suggesting
that partner rotation may have favorable effects on audit quality. However, while
audit firm rotation evoked higher confidence in financial statements by MBA and
law students, partner rotation did not have this effect (Gates et al. 2007).
In summary, archival and experimental results are inconsistent with
respect to the association between engagement partner tenure/rotation and audit
quality. Some research supports the independence hypothesis; other results
confirm the expertise hypothesis. Some studies even fail to find an effect of
tenure/rotation on audit quality, which could indicate that the two explanations
compensate for each other. Because no solid theoretical support is available to
predict directional hypotheses, we posit two null hypotheses to empirically test

12


these relationships in the German setting, which uniquely allows us to isolate
engagement partners from engagement quality review partners.
HYPOTHESIS 1a: Engagement partner tenure is not associated with audit
quality.
HYPOTHESIS 1b: Engagement partner rotation is not associated with audit
quality.
Next, we consider how tenure and rotation of the review partner may affect audit
quality.
Quality review partner tenure, rotation, and audit quality
The general purpose of the engagement quality review is to provide
quality control for audit engagements and to serve as an evaluation of the

performance of the engagement partner and the engagement team (Epps and
Messier 2007). The potential strengths of the review partner are manifold. First,
while the engagement partner is susceptible to familiarity threats (leading to
possibly favorable attitudes toward the client and potential reluctance in the final
review stage to challenge or question decisions made in earlier stages), the review
partner has significantly less interaction with the client, which may reduce such
familiarity threats in the first place (Schneider et al. 2003). Second, review
partners typically hold little accountability toward the client but have considerable
knowledge about the client‘s industry; therefore, they offer an opportunity for
objective and high-level judgments and decisions (Matsumura and Tucker 1995;

13


Rich et al. 1997). According to German audit standards, the review partner‘s
responsibilities include obtaining information about the fundamental content of
the audit opinion, the structure of the audit, and its material aspects. Furthermore,
adequate information procurement includes a critical review of the audit report.
Consistent with other legislations, a review of the audit report by a separate
review partner who is not involved in planning and conducting the audit is
mandatory for German financial statement audits (§ 24d II BS WP/vBP–
Berufssatzung für Wirtschaftsprüfer/vereidigte Buchprüfer). The review process
consists of evaluating whether the audit procedures and audit results are
consistent (§ 24d I BS WP/vBP).
Germany provides a unique setting for researchers, because both the
engagement partner and the review partner are formally responsible for the audit
and the reasonableness of the audit opinion, and both partners‘ signatures must be
visible and identifiable in the audit reports (§ 322 HGB of the German
Commercial Code).5 By providing signatures in the audit report, both partners
accept the role as key audit partners with the consequence of having to rotate after

seven years of tenure with the audit client. Relevant to our study, it is common
dual control practice in Germany that the review partner‘s signature is located on
5

Although the 8th EU directive requires only the signature of one audit partner responsible for the
audit, it is German audit practice to provide the signature of the review partner in addition.
Providing both signatures in the audit report is a characteristic of the dual control principle
common in German business law. For instance, legislation mandates company representation by at
least two directors (§ 709 Abs. 1 BGB und § 35 Abs. 2 Satz 1 GmbHG, § 77 Abs. 1 Satz 1 AktG)
and dual quality control of audit firms (§ 24d Abs. 1 Satz 1 BS WP/vBP).

14


the lower left-hand side of the audit report, next to the right-hand side signature of
the engagement partner (§ 27a BS WP/vBP).6 Both engagement partner and
review partner have reached partner level and typically work for the same audit
firm. 7 The review partner‘s signature implies that the signer is sufficiently
familiar with the audit to be able to take co-responsibility for the audit opinion.
Although recently established regulation in Germany and elsewhere
underlines the importance of the engagement quality review as an effective
technique to uncover potential failures and detect deficiencies in financial
statements (Luehlfing et al. 1995), only a handful of (experimental and analytical)
studies to date have examined audit review partner effectiveness with respect to
audit quality. In an experiment, Kraut and Davidson (1998) found that review
partners‘ audit opinions (vis-à-vis engagement partners) were influenced by
neither client importance nor the importance of the client‘s industry, thus
confirming their relative strength in independence. Matsumura and Tucker
investigated analytically and experimentally how engagement partners respond to
6

To verify this signature practice for our study, we elicited survey responses of more than half of
the audit firms in our sample. The results clearly support that the signature on the left-hand side of
the audit report always identifies the person who conducted the quality review. Furthermore, to
support the survey results, we received the internal directives on quality assurance detailing the inhouse practice of three of the Big 4 audit firms. The directive explicitly states that the partner
providing the audit quality review signs the left-hand side of the audit report and the audit
engagement partner signs the right-hand side of the report. Thus, we can safely conclude that the
location of the signature in the audit report informs us of the role that an individual served during
the audit process (Gelhausen 2007).
7
As explained in more detail in our results section, we observe 24 joint audits in our sample
suggesting that two firms jointly conducted the audit and therefore a total of four partners (two
engagement partners and two review partners) sign the audit report (i.e., four signatures).
Controlling for joint audit cases in our regression models does not change our results.

15


the presence of a review partner and found that engagement partners planned
higher levels of audit testing (Matsumura and Tucker 1995) and exhibited less
reporting bias (Tucker and Matsumura 1997) in the presence of a review partner.
Without second-partner reviews, engagement partners showed significantly more
strategic behavior in their reporting. Similarly, Ayers and Kaplan (2003) found
that engagement partners performed better audit risk assessments during client
acceptance in the presence of a review partner.
To our knowledge, no prior research has investigated the effect of auditor
tenure or rotation on review partners‘ reporting decisions, with two notable
exceptions employing experimental methods. First, Favere-Marchesi and Emby‘s
(2005) results revealed that continuing review partners (i.e., those with prior
involvement in the previous year‘s engagement) were less likely than new review

partners (i.e., no prior involvement) to reach conclusions that differed from the
engagement partner when the evidence refuted the engagement partner‘s view.
This finding may suggest that not even review partners are immune to potential
independence threats. However, Schneider et al. (2003) manipulated prior review
partner involvement in audit planning and found that the degree of a review
partner‘s agreement with an engagement team‘s conclusion was unaffected by
prior involvement; hence, this finding suggests that review partners have the
ability to remain objective.

16


On one hand, we suggest that the likelihood of independence threats due to
tenure could be significantly lower for review partners than for engagement
partners. Review partners remain at a greater distance from the client and are less
involved in daily audit practices. Given such circumstances, they are in a unique
position to enjoy the learning benefits without risking the accompanying and
potentially detrimental independence threats. Hence, the expertise hypothesis
could more likely be at stake for review partners as compared to the independence
hypothesis. However, as suggested by the experimental results of Favere-Marchei
and Emby (2005), even review partners appear to be susceptible to bias in their
reporting decisions. Given limited and inconclusive prior research evidence, we
posit two null hypotheses to empirically test the association between review
partner tenure and audit quality, and review partner rotation and audit quality,
respectively:
HYPOTHESIS 2A: Review partner tenure is not associated with audit
quality.
HYPOTHESIS 2B: Review partner rotation is not associated with audit
quality.
3. Research design

Sample
Our data were obtained from two different sources. First, the accruals data were
obtained from the Compustat Global Industrial database. As shown in Table 1, the

17


initial sample consisted of 9,597 firm-year observations of German listed
companies, excluding financial institutions and insurance companies and covering
the period 1995-2010. We removed 4,026 observations with missing accrual
items.
[Insert Table 1 about here]
The data on audit firm, audit engagement, and review partner signatures
was manually collected either from an Annual Reports database containing
consolidated financial statements or from the respective company‘s website. We
collected data from all German-listed companies for which annual financial
statements including the auditor‘s report were available. Our final sample consists
of 2,636 firm-year observations including data on the audit firm, audit
engagement and audit review partner.
Empirical model
Audit quality can be measured in multiple ways. Experimental and survey
research commonly assesses perceptions of audit quality (for a review, see Quick
2004). Proxies for audit quality in archival research include materiality judgments
(e.g., Kemp et al. 1983), propensity to issue unqualified audit opinions (e.g.,
Vanstraelen 2000) and audit firm litigations (e.g., Latham et al. 1998). More
recent studies have used a firm‘s accrual accounting behavior as a proxy for audit
quality (e.g., Myers et al. 2003, Chi et al. 2005, Carey and Simnett 2006; Francis
and Wang 2008; Cameran et al. 2008). Lower accrual levels are associated with

18

greater auditor conservatism and are interpreted as indicating higher quality
audits, mitigating extreme management reporting decisions. Johnson et al. (2002)
and Myers et al. (2003) use the Jones 1991 model-estimated abnormal accruals as
proxies for audit quality. Consistent with Chi et al. (2009), we use the modified
Jones model (Dechow et al. 1995) to estimate abnormal accruals.
Assuming that the total accruals of a firm can be divided into discretionary
and non-discretionary accruals and that the apportionment cannot be observed,
our measure estimates discretionary accruals. Following Dechow et al. (1995), we
calculate total accruals

as the change in current accruals

minus the change in current liabilities other than financial liabilities
and less depreciation

:8

(1)
Where
=

Total accruals of firm i in year t

=

Non-discretionary accruals of firm i in year t

=

Discretionary accruals of firm i in year t

=

Change in current assets of firm i from year t to year t-1

8

We are aware of the limitations of using balance sheet data to partition net income into accruals
and operating flows (Collins and Hribar 2002). However, in accordance with prior research using
international data (e.g., Leuz, Nanda and Wysocki 2003) we use the indirect approach to compute
total accruals, due to the lack of cash flow data for international firms in the commonly used
databases.

19


=

Change in cash of firm i from year t to year t-1

=

Change in current liabilities of firm i from year t to year t-1

=

Change in short-term debt of firm i from year t to year t-1

=

Depreciation of firm i in year t

Our measure is a modification of the original Jones model (Jones 1991), that is,
we use the cash flow Jones model (Kasznik 1999). These models are both based
on the assumption that the model indicates non-discretionary accruals. They
assume non-discretionary accruals as a component of total accruals for an
estimation period, which is not (or at least less) affected by earnings management.
The coefficients resulting from this regression are used to estimate the total
accruals of the firms in the actual sample period. Comparing the results from this
regression to the expected total accruals calculated by the model (equation (1))
results in differences that are considered to be discretionary accruals. To control
for industry-specific effects on the accruals, the model is calibrated separately for
each industry sector. The original Jones model (Jones 1991) is based on the idea
that the non-discretionary part of the accruals is dependent on economic
conditions of each firm. Therefore, property, plant and equipment
change in revenues

and

are included in the model to control for changes in

non-discretionary accruals caused by changing conditions.
The original Jones model is modified in multiple ways. The model implies
that changes in revenues are non-discretionary. Dechow et al. (1995) modified the

20

Jones model by reducing the revenues by the receivables as receivables could
increase current assets if they include bad debt. Kasznik (1999) included a change
in the cash flow. This model assumes that the accruals of a current period are
dependent on the cash flows of the previous period. Therefore the coefficients of
the cash flow Jones model (Kasznik 1999) are estimated for each industry sector
based on the following equation9:

(2)
Where
=

Last year‘s total assets of firm i in industry sector p

=

Change in revenues of firm i in industry sector p

=

Change in receivables of firm i in industry sector p

=

Property, plant and equipment of firm i in industry sector p

=

Change in cash flow of firm i in industry sector p

All variables are scaled by total assets to control for size effects and to reduce the

possibility of heteroscedasticity. To test hypotheses H1a and H2a, we examine the
impact of audit engagement tenure (EPTENURE) and review partner tenure
(RPTENURE) on earnings management using the following regression model:
9

In non-tabulated tests we compare the explanatory power of five different measures to estimate abnormal
accruals: the Jones 1991 model, the modified Jones model, the cash flow model, the performance-matched
Jones and modified Jones model. We selected the measure of abnormal accruals with the most consistent Rsquare overall for our different industry sectors, leaving us with the cash flow model.

21


(3)
Where
=
=

=
=
=
=
=
=
=

discretionary accruals by cash flow Jones
measured in absolute, positive and negative values
years of tenure of the engagement partner
conducting the audit (i.e., first year = 1; second year
= 2; etc.)

tenure of the engaged review partner
tenure of the engaged audit firm
dummy variable equal to ―1‖ if the auditor is from a
Big 4 audit firm, and equal to ―0‖ otherwise
sales growth rate calculated as the sales in year t
minus sales in t-1 and scaled by sales in year t-1
natural logarithm of total sales
cash flow in year t scaled by total assets in year t-1
ratio of retained earnings to total common equity

To test hypotheses H1b and H2b, we examine the impact of audit engagement
partner switches (EPS) and review partner switches (RPS) on earnings
management using the regression model below.

(4)
Where
=

discretionary accruals by Jones

measured

22


=

=

=

=
=
=
=
=

in absolute, positive and negative values
dummy variable equal to ―1‖ if the audit
engagement partner was switched from t-1 to t, and
equal to ―0‖ otherwise
dummy variable equal to ―1‖ if the audit review
partner was switched from t-1 to t, and equal to ―0‖
otherwise
tenure of the engaged audit firm
dummy variable equal to ―1‖ if the auditor is a Big 4
audit firm, and equal to ―0‖ otherwise
sales growth rate calculated as the sales in year t
minus sales in t-1 and scaled by sales in year t-1
natural logarithm of total sales
cash flow in year t scaled by total assets in year t-1
ratio of retained earnings to total common equity

Consistent with prior studies (Cameran et al. 2008; Chi et al. 2009), we use the
exact value, the absolute value, and the signed values of accruals as dependent
variables for our measures. To control for industry-specific effects, we use onedigit codes from the North American Industry Classification system (NAICS). We
further include fixed effects for year.
We include several control variables for other known determinants of
accruals in our model (l) (e.g., Chen et al. 2008; Chi et al. 2009; Myers et al.
2003). First, we include audit firm tenure in our models to control for any effect
on audit quality from the number of years that a given audit firm has been

performing the audit. Moreover, since accrual models are often unable to
completely capture the effects of companies‘ performance on nondiscretionary
accruals (Dechow et al. 1995; Kothari et al. 2005), we include cash flow to
control for the nondiscretionary component of accruals, which is not extracted by

23


our accrual model. We include lifecycle to control for changes in accruals over a
company‘s life cycle (Anthony and Ramesh 1992; Dechow et al. 2001).
Furthermore, the size of a company determines its earnings management behavior
in the sense that larger companies face higher political costs (Watts and
Zimmerman 1983; 1986; 1990) and higher litigation risk (Lang and Lundholm
1993). In addition, larger companies tend to report larger and more stable accruals
(Dechow and Dichev 2002). The growth of a company‘s sales is likely to increase
the absolute value of accruals. Finally, we control for Big4 audit firms, which tend
to be more conservative and to limit their clients‘ extreme accruals more than
non-Big 4 firms. However, findings related to this argument are mixed (Myers et
al. 2003).10
4. Results
Descriptive results
Panel A of Table 2 shows the distribution of audit engagement and review partner
switches over the entire sample period and other descriptive details on the joint
and individual rotation of audit partners as well as audit firms.
[Insert Table 2 about here]
Out of a total of 2.636 audits, 758 (687) engagement partner (review
partner) switches occurred during our sample period, compared with 1.878
10

Results remain unchanged when we include leverage calculated as total liabilities/equity in all

our models. However, since our sample size decreases to 2,367 we refrain from the inclusion of
leverage in our main tests.

24