Johnson - Audit Firm Tenure and The Quality of Financial Reports (For Madotory Rotation)

Audit-Firm Tenure and the Quality of
Financial Reports*
VAN E. JOHNSON, Georgia State University
INDER K. KHURANA, University of Missouri-Columbia
J. KENNETH REYNOLDS, Louisiana State University
Abstract
This study examines whether the length of the relationship between a company and an audit
firm (audit-firm tenure) is associated with financial-reporting quality. Using two proxies for
financial-reporting quality and a sample of Big 6 clients matched on industry and size, we
find that relative to medium audit-firm tenures of four to eight years, short audit-firm tenures
of two to three years are associated with lower-quality financial reports. In contrast, we find
no evidence of reduced financial-reporting quality for longer audit-firm tenures of nine or
more years. Overall, our results provide empirical evidence pertinent to the recurring debate
regarding mandatory audit-firm rotation — a debate that has, to date, relied on anecdotal
evidence and isolated cases.
Keywords Audit-firm tenure; Audit quality; Financial-reporting quality; Mandatory auditor

rotation
Condensé
Les auteurs étudient l’incidence de la durée du mandat des cabinets d’expertise comptable
sur la qualité de l’information financière. Les rapports financiers sont l’un des principaux
outils de communication de l’information financière aux tiers. Il existe une asymétrie de
l’information et des facteurs de motivation conflictuels entre les gestionnaires qui ont la res-
ponsabilité de préparer les rapports et les utilisateurs de ces rapports, et la vérification des
états financiers contribue à réduire l’incidence de ces conflits en améliorant la qualité de
l’information communiquée par les gestionnaires. Pourtant, la vérification n’améliore la
qualité de l’information que si elle réduit l’occurrence d’inexactitudes dans les états financiers.
La capacité de la vérification de réduire ces inexactitudes est généralement considérée
comme une fonction dépendant de deux variables : la capacité du vérificateur de déceler les
inexactitudes (soit la compétence du vérificateur) et son comportement subséquent, c’est-à-
dire le fait qu’il s’assure que les correctifs nécessaires soient apportés aux inexactitudes
décelées (soit le comportement de communication du vérificateur).
* Accepted by Dan Simunic. We thank Dick Dietrich, Mike Ettredge, Jere Francis, Tom Linsmeier,
Earl Wilson, two anonymous reviewers, participants at the 1999 American Accounting Associa-
tion midyear auditing section meeting, the central states accounting research workshop, and
workshop participants at University of Illinois, University of Kansas, University of Missouri, and
University of North Texas for their valuable comments on this paper.
Contemporary Accounting Research Vol. 19 No. 4 (Winter 2002) pp. 637–60 © CAAA
19113846, 2002, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1506/LLTH-JXQV-8CEW-8MXD by Korea University Library, Wiley Online Library on [21/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
638 Contemporary Accounting Research
L’on a souvent supposé que la durée des mandats des cabinets d’expertise comptable
avait des répercussions négatives sur la compétence des vérificateurs, leur comportement de
communication ou les deux. La durée du mandat est la période pendant laquelle est maintenue
la relation entre une société et ses vérificateurs. Ceux qui débattent de la question de la rotation
des vérificateurs invoquent des arguments antagonistes relatifs aux conséquences positives
ou négatives de relations prolongées entre une société et ses vérificateurs. En apparence, les
arguments des deux camps se défendent, bien que la majorité des preuves invoquées jusqu’à
maintenant dans les discussions tiennent essentiellement de l’anecdote et de la conjecture. Il
existe peu de données empiriques relatives à l’incidence de la durée des mandats des cabinets
d’expertise comptable sur l’information financière. La présente étude vise à combler ce vide
et à contribuer à faire avancer le débat sur la durée des mandats des cabinets d’expertise
comptable.
Dans leur analyse empirique, les auteurs classent la durée des mandats des cabinets
d’expertise comptable selon qu’elle est courte, moyenne ou longue. Suivant leur définition,
un mandat de courte durée est une relation vérificateur-client de deux à trois ans, un mandat
de durée moyenne, une relation de quatre à huit ans, et un mandat de longue durée, une relation
de neuf ans ou plus ; les tests effectués par les auteurs ne sont cependant pas sensibles aux
autres découpages possibles. Les auteurs formulent l’hypothèse selon laquelle les durées
courte et longue seraient toutes deux associées à une qualité d’information financière
inférieure à celle qui correspond à la durée moyenne.
Les auteurs s’attendent à ce que les mandats de courte durée aient une incidence négative
sur la qualité de l’information, principalement parce que la société faisant l’objet de la
vérification est peu familière aux vérificateurs. Bon nombre d’études ont invariablement
démontré que les vérificateurs ont tendance à fournir un travail légèrement moins efficace
au cours des deux premières années de leur mandat qu’au cours des années subséquentes.
Outre cette lacune, le vérificateur peut également faire face à différents facteurs de motivation
durant les premières années du mandat, comparativement aux années subséquentes. Il devra,
en effet, trouver un compromis entre le désir de tirer parti des possibilités commerciales
qu’offre le client et celui de préserver la réputation du cabinet et d’éviter les litiges onéreux.
Les facteurs de motivation initiaux du vérificateur peuvent être partiellement modifiés par
l’impératif du maintien de la relation avec le client pendant la période nécessaire au recouvre-
ment des coûts lorsqu’il y a sous-facturation systématique des services (ce que l’on appelle
la pratique du leurre-prix), au cours des premières années du mandat, dans le but d’attirer le
client.
Les auteurs s’attendent à ce que les mandats de longue durée aient une incidence négative
sur la qualité de l’information, principalement en raison de problèmes de motivation. Plus le
vérificateur accumule d’expérience en travaillant auprès du client, moins il est sensible au
risque de litige ou de préjudice à la réputation et plus ses motivations s’orientent vers la
fidélisation du client et ce qu’il peut rapporter. De plus, le vérificateur risque de devenir
moins vigilant dans la détection des inexactitudes en raison de la confiance qu’il a cultivée
au fil de ses relations prolongées avec le client.
Pour déterminer l’incidence de la durée des mandats d’un cabinet d’expertise comptable
sur la qualité de l’information financière, les auteurs évaluent cette qualité à l’aide de deux
mesures. Premièrement, ils utilisent la valeur absolue des ajustements imprévus à titre de sub-
stitut à l’ampleur des interventions de la direction dans les résultats déclarés. La comptabilité
CAR Vol. 19 No. 4 (Winter 2002)

Audit-Firm Tenure and the Quality of Financial Reports 639
d’exercice donne aux gestionnaires une grande latitude dans la détermination des résultats.
Les recherches précédentes ont fait ressortir que cette latitude peut être exploitée de façon
opportuniste, ce qui risque de réduire la qualité des rapports financiers. En l’absence
d’attentes précises a priori en ce qui a trait aux facteurs de motivation des gestionnaires, les
chercheurs ont généralement utilisé la valeur absolue des ajustements imprévus pour cerner
le comportement interventionniste de la direction. Deuxièmement, les auteurs étudient la
mesure dans laquelle les ajustements d’un exercice persistent dans les résultats de l’exercice
subséquent. L’utilisation opportuniste des ajustements aura tendance à augmenter la portion
transitoire des résultats de l’exercice, ce qui en diminue la persistance.
Les auteurs bornent leur analyse empirique aux clients des Six Grands cabinets
d’expertise comptable, étant donné que les recherches précédentes ont révélé d’importantes
différences entre les vérificateurs des Six Grands cabinets et les autres vérificateurs. À chacun
des cabinets appartenant au groupe de ceux dont la durée des mandats est courte (le plus
petit des trois groupes), les auteurs associent des cabinets de taille comparable et du même
secteur, appartenant aux groupes de ceux dont la durée des mandats est moyenne et longue,
afin d’éliminer les différences dans la qualité de l’information financière attribuables à des
facteurs autres que la durée des mandats. En utilisant les échantillons finals résultant
des combinaisons de 2 463 pour le test de la valeur absolue des ajustements et de 2 280 pour
le test de la persistance des ajustements, ils constatent uniformément que les mandats de
courte durée des cabinets d’expertise comptable sont associés à des rapports financiers
de qualité inférieure. En revanche, pour les deux tests, les auteurs ne relèvent rien qui permette
de conclure à une diminution de la qualité de l’information financière pour les mandats de
durée plus longue.
Les résultats obtenus par les auteurs viennent enrichir le fonds croissant des travaux sur
la qualité de la vérification. En utilisant un échantillon de sociétés clientes des services de
vérification des Six Grands cabinets d’expertise comptable, les auteurs démontrent que l’on
peut observer des variations de qualité de l’information financière — et, par déduction, des
variations de qualité de la vérification — dans les missions de vérification exécutées par les
Six Grands. Les résultats augmentent aussi la documentation relative à la qualité de l’infor-
mation financière. Dans l’optique selon laquelle les rapports financiers sont le résultat du
travail effectué de concert par la société et ses vérificateurs, les auteurs démontrent qu’il
pourrait être important de prendre en considération le vérificateur dans les prochaines analyses
de la qualité des rapports financiers. Enfin, les auteurs abordent la question de la rotation des
vérificateurs dans une perspective empirique. Leurs constatations viennent confirmer les
échos selon lesquels la qualité de l’information est inférieure pendant les premières années
de la relation vérificateur-client, mais elles ne permettent pas de conclure que, dans les
années subséquentes, les rapports sont de qualité inférieure, tout au moins sous le régime de
réglementation actuel.
1. Introduction
Congress and regulators periodically turn their attention to the role that auditors
play in the financial-reporting process (U.S. Governmental Accounting Office
(GAO) 1996). During these periods, which often follow high-profile frauds or
bankruptcies, the closeness of audit-firm–client relationships is frequently scrutinized

(cf. McClaren 1958; Winters 1976; United States Senate (Metcalf committee)
1976; Hoyle 1978; GAO 1991). Also during these periods, mandatory audit-firm
rotation is offered as a possible means of improving the quality of financial reporting.1
There is limited empirical evidence regarding the relationship between audit-
firm tenure (the length of the auditor–client relationship) and financial-reporting
quality. Thus, it is not clear whether the problem that mandatory rotation is
intended to solve is real or illusory. In this paper, we examine whether short and
long audit tenures are associated with the issuance of lower-quality financial
reports using two proxies for financial-reporting quality. First, we use the absolute
value of unexpected accruals to proxy for the extent of management interventions
in reported earnings numbers. Second, we investigate the extent to which current
accruals persist into earnings in the subsequent year as a proxy for the quality of
accruals reported in current period earnings.
Based on an industry- and size-matched sample of publicly traded companies
audited by Big 6 audit firms, our results suggest that short audit-firm–client rela-
tionships (two to three years) are associated with lower-quality financial reports
relative to medium audit-firm–client relationships (four to eight years). We observe
evidence of greater management intervention in reported earnings and lower-quality
accruals (current accruals that are less likely to persist in one-period-ahead earn-
ings) for short audit-firm tenures.
In contrast to the results for short audit-firm tenures, the same tests do not pro-
vide evidence that long audit-firm–client relationships (nine years or longer) are
associated with reduced financial-reporting quality relative to medium audit-firm–
client relationships. We are unable to detect statistically significant differences in
management interventions in reported earnings or accrual quality for long audit-
firm tenures.
Our results provide needed empirical evidence pertinent to the recurring
debate regarding mandatory audit-firm rotation — a debate that has, until recently,
relied on anecdotal evidence and isolated cases. This study, when combined with
other recent research, such as Dopuch, King, and Schwartz 2001 and Geiger and
Raghunandan 2002, can be used to move the debate forward. Our results document
lower financial-reporting quality for short audit tenures, but not for long audit ten-
ures. It is important to note that our results are based on the current regulatory
regime, and cannot be generalized to a regime in which audit-firm rotation is man-
datory. Under such a regime, where the horizon for the relationship between the
client and the firm is known, auditor incentives may differ significantly.
Our results also add to the growing body of literature on audit quality. Several
prior studies have investigated whether Big 6 auditors provide higher-quality
audits than non–Big 6 auditors (cf. DeAngelo 1981; Teoh and Wong 1993; Becker,
DeFond, Jiambalvo, and Subramanyam 1998; Francis, Maydew, and Sparks 1999).
We investigate audit tenure, a factor that varies within auditor size class rather than
across auditor size classes. Given the relative dominance of the Big 6 in the audits of
publicly listed companies, our study highlights quality differences within the Big 6
class. More importantly, we demonstrate that auditor tenure may be an additional
variable influencing financial-reporting outcomes. Our results are also consistent

with prior work (e.g., Knapp 1991; O’Keefe, King, and Gaver 1994) that has iden-
tified the importance of client-specific knowledge to audit quality.
The remainder of the paper is organized as follows. In section 2, we develop
our hypotheses. In section 3, we discuss specific research design issues associated
with the methodology used, and describe sample selection. Results are presented
in section 4, and conclusions are presented in the final section of the paper.
2. Hypotheses development
Financial reports are a principal means of communicating financial information to
those outside an entity. Given the existence of information asymmetries and the
potential for conflicts of interest between company management and outside users
of financial information, an audit of financial reports by a third party (or alternative
monitoring arrangements) can enhance the quality of the financial information
reported by management (Dopuch and Simunic 1982; Watts and Zimmerman
1986) and improve the quality of information that investors have about the value of
traded securities (Ronnen 1996).
Recognizing the importance of auditing in the financial-reporting process,
Antle and Nalebuff (1991) suggest that financial statements should be viewed as a
joint statement from the audit firm and company management. In general, the abil-
ity of the audit function to enhance financial-reporting quality is dependent on both
the likelihood that the audit will detect a material misstatement or omission
(henceforth auditor competence) and the auditor’s behavior subsequent to the
detection of a material misstatement (henceforth, auditor-reporting behavior). If
material misstatements are detected and corrected (or revealed), the quality of the
financial report is improved. Alternatively, a failure to detect material misstate-
ments or a failure to require that they be corrected before issuing a clean audit
report would not improve the quality of the financial report.
Prior work has documented that Big 5/6/8 audit firms are associated with
superior financial reporting outcomes (cf. Teoh and Wong 1993; Beasley and
Petroni 1996; Becker et al. 1998; Francis et al. 1999). The explanations offered for
the superiority of the larger firms generally focus on advantages in (1) perceived
competence (by virtue of their heavy spending on auditor-training facilities and pro-
grams), and (2) perceived independence in reporting (by virtue of their size and large
portfolio of clients, which presumably gives them the financial strength to stand up
to, or walk away from, a client if necessary). We investigate the potential effects of
audit tenure on financial-reporting quality by considering how changes in auditor
knowledge and incentives may impact financial-reporting quality as audit tenure
changes.
Short audit-firm tenure

Knowledge is a critical input to the auditor’s ability to detect material misstatements.
A great deal of knowledge necessary to the audit (such as knowledge of the client’s
accounting system and internal control structure) is client-specific.2 Although
auditors use other types of knowledge (such as general knowledge and industry
knowledge) to produce an audit, the importance of client-specific knowledge

creates a significant learning curve for new auditors (Knapp 1991) and results in
significant start-up costs (DeAngelo 1981). Less client-specific knowledge in the
early years of an engagement may result in a lower likelihood of detecting material
misstatements, thereby giving auditors a comparative advantage in detecting errors
over time as they obtain a deeper understanding of the client’s business (Beck,
Frecka, and Solomon 1988).3
An initial lack of client-specific knowledge on an engagement may not be
associated with lower financial-reporting quality if it is possible to overcome the
lack of knowledge by employing additional effort on new engagements.4 However,
knowledge and effort may not be perfect substitutes for one another. Arrunada and
Paz-Ares (1997) suggest that there may be a technological limit to the replacement
of client-specific assets; that many, if not most, of them cannot be replaced immedi-
ately. Thus, financial-reporting quality is expected to increase as client-specific
knowledge increases in the early years of an audit engagement.
Prior research also suggests that auditor incentives may differ in the early
years of an audit engagement. Generally, auditor incentives are traded off between
the auditor’s desire to maintain and profit from the auditor – client relationship
(Chow and Rice 1982; Citron and Taffler 1992) and the desire to protect the firm’s
reputation (brand name) and to avoid costly litigation (Balachandran and Nagara-
jan 1987; DeJong 1985; Melumad and Thoman 1990; Narayanan 1994; Nelson,
Ronen, and White 1988). DeAngelo (1981) noted that client-specific assets (such
as knowledge) along with transactions costs allow the incumbent auditor to earn
quasi rents from maintaining existing client relationships. Financial-reporting
quality could be reduced in early engagement years if the existence of quasi rents
skews the auditor’s incentives toward maintaining the client relationship. Geiger
and Raghunandan (2002) suggest that the competitive practice of low balling may
further skew the auditor’s incentives and cause the auditor to employ less effort or
be more accommodating to the client in the early years of an engagement in an
attempt to limit losses on the current engagement and to ensure a repeat engagement.
The above discussion leads to the following hypothesis (stated in the alternative
form).
HYPOTHESIS 1. The quality of financial reports issued by a company will be

lower when the audit firm’s tenure is short than when the audit firm’s tenure
is medium.
Long audit-firm tenure

Most arguments surrounding audit tenure and incentives have suggested that ceteris
paribus, over time, the auditor’s incentives shift toward maintaining and profiting
from the client and the auditor becomes less concerned with litigation relating to
the client. Accordingly, the auditor may become less objective and apply less effort
toward the detection of material misstatements (Hoyle 1978; Arrunada and Paz-
Ares 1997).5 Given the increased client-specific knowledge obtained from multiple
repeat engagements, some reduction in effort would be expected without threaten-
ing audit quality. However, arguments regarding the detrimental effects of long

audit-firm tenures implicitly assume that reductions in effort will exceed an opti-
mal level. Shockley (1981) characterizes this situation as the audit firm having a
“learned confidence” in the client as a result of the long relationship and suggests
that this learned confidence may result in the audit firm using less strenuous and
less innovative audit procedures.6
These arguments lead to the following hypothesis.
HYPOTHESIS 2. The quality of financial reports issued by a company will be

lower when the audit firm’s tenure is long than when the audit firm’s tenure
is medium.
The available empirical evidence regarding both hypotheses is limited. After

investigating 406 alleged cases of audit failures involving Securities and Exchange
Commission (SEC) clients, an AICPA quality control committee concluded that
audit failures are 3 times more likely in the first 2 years of an engagement than in
subsequent years (AICPA 1992). Two studies examining lawsuits involving audi-
tors found that audit failures are more common when audit tenure is 3 years or less
(St. Pierre and Anderson 1984; Stice 1991). Knapp (1991) varied audit tenure in an
experimental setting and found that experienced audit committee members per-
ceived that auditors with a 5-year tenure were more likely to detect errors than
auditors in the first year of an engagement or auditors with an audit tenure of 20
years. Most recently, Geiger and Raghunandan (2002) found that long-tenure audi-
tors were more likely than short-tenure auditors to issue going-concern opinions
for clients that subsequently declared bankruptcy. A series of studies examining
governmental audits consistently documented a negative correlation between audit
quality (measured as compliance with audit standards) and audit tenure (cf. Deis and
Giroux 1992; Giroux, Deis, and Bryan 1995; Deis and Giroux 1996; Copley and
Doucet 1993). To date, there is limited empirical evidence regarding the associa-
tion between short or long audit tenures and the quality of financial information
issued by publicly traded companies. This study is intended as a first step toward
providing such evidence.
3. Method
To investigate whether audit-firm tenure is associated with financial-reporting quality,
we use two empirical proxies for financial-reporting quality. These two proxies, the
measurement of audit-firm tenure, the control variables used in the investigation,
and the sample selection methods are discussed below.
Proxies for financial-reporting quality
Absolute value of unexpected accruals

Dechow (1994) highlights the potential timing and matching problems associated
with the use of cash flows as a short-term performance measure. Accrual-based
earnings under generally accepted accounting principles (GAAP) contain accruals
and deferrals to overcome the inherent limitations of cash flows. However, GAAP

also affords management substantial flexibility in reporting accrual-based earnings.

Prior research has noted that management may use this flexibility to opportunistically
manage earnings. Such opportunistic use of accruals by management will result in
lower-quality financial reports.
Several papers have used unexpected accruals as a proxy for management’s
active intervention in reporting earnings. The decision to use directional unex-
pected accruals or the absolute value of unexpected accruals is driven by the nature
of the study; specifically, whether there is an a priori expectation regarding manage-
ment’s incentives. For example, Jones (1991) expected that import relief investigations
would motivate managers to decrease earnings during the investigation period.
Accordingly, the tests were designed to detect significant income-decreasing unex-
pected accruals. Teoh, Welch, and Wong (1998) hypothesized that upcoming stock
issues would motivate managers to prefer higher income levels. They designed
tests to detect significant income-increasing unexpected accruals. When a
researcher has an a priori expectation of management’s incentives, directional
unexpected accruals represent a more powerful test.
Alternatively, several papers have examined the impact of factors that are not
associated with a clear directional management incentive. For example Warfield,
Wild, and Wild (1995) examine the association between managerial ownership and
earnings management. Klein (2002) examines the association between corporate
governance mechanisms and earnings management. In each study, the data is
pooled over a long time period. Because they do not have a priori directional
expectations regarding management’s motivation for particular firm-year observa-
tions, both Warfield et al. and Klein use the absolute value of unexpected accruals
to capture management’s intervention behavior, noting that the magnitude of dis-
cretionary accrual adjustments serves to measure the extent to which managers
intervene in reporting accounting earnings numbers. Similarly, Bartov, Gul, and
Tsui (2000) interpret a positive association between the absolute value of a firm’s
unexpected accruals and the likelihood of receiving a qualified audit report as evi-
dence that unexpected accruals capture the extent of earnings management behavior.
Following prior research, we use the absolute value of unexpected accruals as
a proxy for financial reporting quality.7 The magnitude of absolute value of unex-
pected accruals measures a company’s success in managing earnings either up or
down, as needed, depending on year-specific situations (Reynolds and Francis 2000).
In general, using unexpected accruals as a proxy for financial-reporting quality is a
joint test of earnings management and the model of expected accruals used.8
The persistence of the accrual components of earnings

As a second proxy for financial-reporting quality, we examine the relationship
between current-period accruals and future income. The time series properties of
earnings have been studied extensively in accounting (cf. Freeman, Ohlson, and
Penman 1984; Sloan 1996). One approach used in prior research is to estimate the
following regression equation
Earningst + 1 = α 0 + α1Earningst + ∈t + 1 (1).

The slope coefficient, α1, represents the extent to which earnings performance in
time t is expected to persist in subsequent-period earnings. A slope coefficient
equal to zero suggests that current-period earnings are purely transitory whereas a
slope coefficient equal to one suggests that earnings follow a random walk. Prior
research has generally found the slope coefficient to be between zero and one, sug-
gesting that earnings are mean reverting.
The extent to which current-period earnings persist or are purely transitory (as
reflected by the magnitude of the slope coefficient α1) is not independent of the
accounting choices of management. The opportunistic use of accruals tends to
reduce the slope coefficient, while the use of accruals to signal private information
tends to increase it. As an example, consider a manager with private information
that collection periods are increasing and several large customers are in financial
distress. This information suggests that future earnings may be negatively affected.
The manager may signal this information by increasing the size of the reserve for
doubtful accounts, or he or she may choose to postpone any adjustment to the net
realizable value of receivables. The two choices generally have quite different
effects on the extent to which current-period earnings persist in the future, with the
former choice resulting in a larger slope coefficient than the latter.
Measurement of audit-firm tenure

Audit-firm tenure is the length of the audit-firm–client relationship as of the fiscal
year-end covered by the audited financial statements.9 Auditor data are available
on the 1995 annual COMPUSTAT files beginning in 1976. Accordingly, the initia-
tion date of the auditor – client relationship is censored on the left in some cases
because we do not have auditor data before 1976. Firms that are added to the
COMPUSTAT data base over time are included in the sample subject to the con-
straint that audit tenure can be computed accurately (i.e., if the initiation date of the
auditor–client relationship is available, or if the company has had the same auditor
for nine or more years).
Following prior research (e.g., St. Pierre and Anderson 1984; Stice 1991), we
define audit tenure as short when the same auditor has audited the financial state-
ments of a company for two or three years (as discussed below, we omit the year of
the auditor change to control for potential confounds).10 We define audit tenure as
long when the same auditor has audited the financial statements of a company for
nine or more years.11 On the basis of the definition of short and long tenure, we
define audit tenure as medium when the same auditor has audited the financial
statements for four to eight years. However, given the data-availability constraint
on the initiation date of the auditor–client relationship and our definition of audit
tenure (short is two to three years, medium is four to eight years, and long is nine
or more years), we can accurately (without measurement error) classify observa-
tions as of 1986 (the beginning of our sample period).
Sample selection
The set of initial candidates for inclusion in the sample was all U.S. corporations
on the COMPUSTAT full coverage and research files that did not change fiscal

year-ends in the 10-year test period 1986–95. We then deleted individual firm-year
observations for the following reasons:
1. Data considerations: Observations were deleted if (a) financial data were not
available on COMPUSTAT; or (b) the auditor could not be determined from
the annual COMPUSTAT tapes.12
2. Auditor considerations: Observations were deleted if (a) the company was not
audited by a Big 6 auditor; (b) the company received a modified audit opinion
within two years before or after the fiscal year-end; or (c) the company
changed auditors within one year before or after the fiscal year-end.
These screens resulted in 11,148 firm-year observations.13 The screens relat-
ing to data considerations are obvious; however, the screens relating to auditor
considerations warrant some discussion. First, we restricted our investigation to
Big 6 auditors. Prior research suggests that audit quality and perceptions of audit
quality differ for companies audited by Big 6 auditors versus those audited by
non–Big 6 auditors (Becker et al. 1998; Francis et al. 1999; Teoh and Wong 1993).
As discussed previously, restricting our sample to Big 6 auditors avoids a potential
confound and allows us to study an observable auditor characteristic that varies
within a particular size class. The screens relating to opinion qualifications and
auditor changes are also used to avoid potentially confounding factors. Prior
research has documented that the market’s responsiveness to earnings announce-
ments is significantly lower in periods before and after the issuance of qualified
audit reports (Choi and Jeter 1992; Subramanyam and Wild 1996). Similarly, prior
research has found that auditor behavior may differ in the year before and after an
auditor change (DeFond and Subramanyam 1998).
In an attempt to enhance our ability to attribute differences in financial report-
ing quality to audit-firm tenure, we also match on industry and size. Matching on
industry can rule out alternative explanations because companies from the same
industry are more likely to face similar supply and demand uncertainties, to engage
in similar transactions, and to use similar accounting methods (Teoh and Wong
1993). Prior research has conjectured that the ability of companies to use discre-
tionary accruals in financial reporting (Francis et al. 1999) may vary across
industries. Matching on company size can rule out the alternative explanation of
auditor independence because larger clients generally represent a greater potential
loss of revenue to the auditor if the client switches to a different auditor. Prior
research (Haskins and Williams 1990; Krishnan 1994) has also documented a neg-
ative relation between client size and auditor switching.
We selected the industry- and size-matched sample in the following manner.
Companies audited by short-tenure audit firms were selected and matched with
companies audited by medium-tenure and long-tenure audit firms in the same fis-
cal year and four, three, and two-digit Standard Industrial Classification (SIC)
code, depending on data availability, and that are closest in size (defined in terms
of book value of total assets). The sample is further restricted to exclude observa-
tions falling in the extreme 1 percent of the entire distribution of variables used to
test Hypothesis 1 and Hypothesis 2. This procedure yielded a matched sample of

2,463 (2,280) firm-year observations for the unexpected accruals (persistence)

tests. The sample observations are evenly distributed across years.
Table 1 presents the descriptive statistics for the industry- and size-matched
sample companies, divided by tenure group, along with significance levels of
univariate tests comparing the mean values of the variables between medium- and
short-, and medium- and long-tenure groups.14 The table includes variables used in
prior research to proxy for economic environment. Observations in the short-tenure
group are not significantly different ( p-value > 0.10) than the observations in the
medium-tenure group for any of the variables reported in Table 1. Similarly, no
significant differences emerge between the long- and medium-tenure groups for
any of the variables except age (computed as the number of years since the listing
date). Although no matching procedure can rule out all alternative explanations,
the descriptive statistics in Table 1 suggest that any differences in financial-reporting
quality that are observed in the matched sample are unlikely to be due to differences
in growth opportunities, company size, risk, financial health, or profitability. How-
ever, because there was a significant age difference for the long-tenure group, we
explicitly control for age in subsequent analyses.
Control variables
Although our primary variable of interest in this investigation is audit-firm tenure,
other audit-firm characteristics and client characteristics can affect financial-reporting
quality. To the extent that another variable affecting financial-reporting quality is
correlated with audit-firm tenure, there is a potential that this other variable may be
driving any observed difference in financial-reporting quality.
In a broad sense, the client characteristics that affect financial-reporting qual-
ity can be grouped according to whether they are likely to affect the sophistication
(or accuracy) of the financial-reporting system or the incentives of management.
The sophistication (or accuracy) of the financial-reporting system is likely to differ
with the size and age of the company, with larger, more mature companies
expected to have more sophisticated financial-reporting systems. Management’s
incentives regarding financial reporting are a well-researched topic, and prior work
has identified factors such as the financial condition of the company and the tightness
of debt constraints. Companies in financial distress or under near-debt constraints
may be more motivated to manage earnings (DeFond and Jiambalvo 1994).
As previously noted, our primary tool for controlling for alternative explana-
tions is the use of an industry- and size-matched sample. As reported in Table 1,
the matching procedure was quite effective at minimizing differences in growth
opportunities, company size, risk, financial health, and profitability. Thus, even if
these factors are correlated with financial-reporting quality, they will not be viable
explanations for any observed differences in financial-reporting quality across ten-
ure categories. Still, including variables that may affect financial-reporting quality
(or our specific proxies) sharpens the analysis by increasing the explanatory power
of the model and enhancing our ability to detect differences that are associated
with the variable of interest.

TABLE 1
Descriptive statistics by audit tenure
Short tenure* Medium tenure† Long tenure‡

(n = 821) (n = 821) (n = 821)
Mean Median Mean Median Mean Median
MB 2.75 1.685 3.33 1.693 2.76 1.677
SIZE 545.8 109.1 540.6 106.8 508.0 123.2
LEV 0.53 0.53 0.54 0.54 0.53 0.52
FC 4.383 3.136 4.157 2.898 4.296 3.514
BETA 0.779 0.728 0.743 0.703 0.755 0.689
ROA 0.031 0.042 0.031 0.037 0.046 0.045
AGE 12.4 10 13.1 11 19.9§ 19
∆CFO 0.009 0.007 0.012 0.011 0.012 0.01
GROWTH 0.108 0.054 0.122 0.059 0.054 0.048
∆ACQN − 0.004 0 − 0.003 − 0.003 − 0.003 0
∆TIE 0.005 0.001 0.001 0.001 − 0.023 0.001
∆FIN − 0.011 − 0.007 − 0.012 − 0.001 − 0.007 − 0.002
SI 0.404 0 0.441 0 0.387 0
Notes:
* Two to three years of audit-client relationship.
† Four to eight years of audit-client relationship.
‡ Nine or more years of audit-client relationship.
§ Significantly different from the medium group ( p-value < 0.01).
MB is the ratio of the market-to-book value of common equity.
SIZE is the total assets in millions of dollars.
LEV is the ratio of total liabilities to total assets.
FC is the Altman-Z computed as 1.2*(working capital/total assets) + 1.4*(retained
earnings/total assets) + 3.3*(earnings before interest and taxes/total assets) +
0.6*(market value of equity/book value of total debt) + 1.0*(sales/total assets).
BETA is the value-weighted beta estimated over the fiscal year.
ROA is the ratio of operating income to total assets.
AGE is the number of years since listing date.
∆CFO is the change in operating cash flows scaled by assets.
GROWTH is the one-year growth in assets.
∆ACQN is the change in acquisition expenditures scaled by assets.
∆TIE is the change in times interest earned scaled by assets.
∆FIN is the change in new financing scaled by assets.
SI is one if a firm reported a special item, zero otherwise.

4. Empirical results
Level of unexpected accruals

Following prior research, the level of unexpected accruals for firm i is calculated as:
UAit = TAit /Ait − 1 − {φ1jt(1/Ait − 1) + φ 2jt [(∆REVit − ∆ARit )/Ait − 1]

+ φ 3jt(PPEit /Ait − 1)} (2),
where TA is total accruals, A is total assets, REV is total sales revenues, AR is

accounts receivables, and PPE is gross property, plant, and equipment. The coeffi-
cients φ1jt, φ 2jt , and φ 3jt are the parameters from the estimation of expected accruals
using all companies in the jth industry (defined by two-digit SIC codes) and fiscal
year t. Following DeFond and Jiambalvo (1994), and Subramanyam (1996),
expected accruals are estimated using a cross-sectional version of the modified
Jones 1991 model as shown below:
TAit /Ait − 1 = φ1jt (1/Ait − 1) + φ 2jt [(∆REVit − ∆ARit )/Ait − 1]

+ φ 3jt (PPEit /Ait − 1) + ε it (3),
where all variables are defined as above.

We perform a multivariate test that includes the control variables (size, age,
financial condition, and leverage) discussed previously. We also include operating
cash flows (OCF ) defined as cash flows from operations scaled by lagged total
assets because they have been shown to vary inversely with unexpected accruals
(Dechow, Sloan, and Sweeney 1995). Specifically, the following OLS regression is
used to test the hypothesized relationship between unexpected accruals and audit
tenure:15
|UAit| = β 0 + β1(OCFit /Ait − 1) + β2LAit + β 3LEVit + β4FCit + β 5AGEit

+ β6SHORTit + β 7LONGit + ε it (4),
where
|UAit | = absolute value of unexpected accruals for company i at time t,

OCFit = operating cash flows for company i at time t,
Ait − 1 = total assets for company i at time t − 1,
LAit = natural logarithm of total assets for company i at time t,
LEVit = ratio of total liabilities to total assets of company i at time t,
FCit = Altman Z-score for company i at time t,16
AGEit = number of years company i is listed in a stock exchange at time t,
SHORTit = one when the length of the auditor–client relationship is short (two or
three years); zero otherwise, and

LONGit = one when the length of the auditor–client relationship is long (nine
years or longer); zero otherwise.
Consistent with the theoretical formulation and prior research, the coefficients
β1, β 2, β4, and β5 (β 3) are predictably negative (positive). More importantly, on the
basis of the hypotheses developed above, we should observe higher levels of unex-
pected accruals for short and long audit-firm tenures relative to medium audit-firm
tenures. In other words, we expect the SHORT and LONG dummy variables to be
positively related to the absolute value of unexpected accruals (i.e., β 6 > 0 under
Hypothesis 1 and β 7 > 0 under Hypothesis 2).
Table 2 presents the results from estimating the cross-sectional OLS regres-
sion with the absolute value of unexpected accruals as the dependent variable. The
adjusted R 2 for (4) reported in Table 2 is 8.3 percent. All of the control variables
are significant in the expected direction except financial condition (FC). In terms
of Hypothesis 1, the parameter of interest in (4) is β 6. Ceteris paribus, β 6 captures
TABLE 2
Effect of auditor tenure on the absolute value of unexpected accruals (UA)
|UAit| = β 0 + β1(OCFit /Ait − 1) + β2LAit + β 3LEVit + β4FCit + β 5AGEit + β6SHORTit

+ β 7LONGit + ε it (4)
Predicted Parameter
Independent variable Parameter sign estimate t-statistic
Intercept β0 0.009 13.71*

OCFit /Ait − 1 β1 − −0.053 −4.01*
LAit β2 − −0.006 −5.62*
LEVit β3 + 0.026 3.68*
FCit β4 − −0.001 −0.01
AGEit β5 − −0.001 −2.74*
SHORTit β6 + 0.007 2.37*
LONGit β7 + 0.001 0.04
Adjusted R 2 0.083
F-statistic 14.9*
Notes:
* Significant at p-value < 0.01 (all one-tailed except for intercept).
OCFit /Ait − 1 is operating cash flows divided by total assets for company i at time t − 1.
LAit is the natural logarithm of total assets for company i at time t.
SHORTit is one when the length of the auditor–client relationship is short (two or three
years); zero otherwise.
LONGit is one when the length of the auditor–client relationship is long (nine years or
longer); zero otherwise.
All other variables are defined in the notes to Table 1.

the difference in the magnitude of unexpected accruals for short audit-firm tenure
observations relative to the medium audit-firm tenure observations. The coefficient
on the SHORT dummy variable is significantly positive (p < 0.01), which indicates
that the level of unexpected accruals reported by sample companies in the short-
tenure group is higher than that reported by sample companies in the medium-tenure
group.
In terms of Hypothesis 2, β 7 represents the difference in the magnitude of
unexpected accruals for long audit-firm tenure observations relative to the medium
audit-firm tenure observations. Hypothesis 2 predicts β 7 to be positive. The multi-
variate results for the long-tenure group fail to reject Hypothesis 2. The coefficient
on the LONG dummy variable is not statistically significant (p-value > 0.10). The
test provides no evidence of a statistically significant increase in the absolute value
of unexpected accruals for sample observations in the long-tenure group (relative
to sample observations in the medium-tenure group).
Sensitivity analysis
A further analysis is made to assure that the relation between tenure groups and the
absolute value of unexpected accruals is not confounded by firm size. We plotted
residuals from the estimated (4) against total assets, and the plot showed no sys-
tematic pattern that would suggest that the error terms are correlated with total
assets.17 We also reestimated (4) without operating cash flows as a control vari-
able; none of our inferences regarding SHORT and LONG coefficients were
affected.
To ensure that our results are not driven by spurious correlations, we also reesti-
mated (4) by focusing on variables identified in DeFond and Subramanyam 1998 and
available on COMPUSTAT. These variables include GROWTH (measured as one
year growth in assets), ∆ CFO (measured as change in operating cash flows scaled
by assets); ∆ FIN (measured as change in new financing scaled by assets); ∆ ACQN
(measured as change in acquisition expenditures scaled by assets and ∆ TIE (measured
as change in times interest earned scaled by assets). Unlike DeFond and Subramanyam
who use a 1/0 dummy variable for reported restructuring charges, we also include
SI (measured as 1/0 dummy variable if a firm was listed on COMPUSTAT as using
special items). Although the significance levels of our test variables (SHORT and
LONG) remained unchanged, none of the additional included variables except SI
were statistically significant (p-value > 0.10).
Finally, we test the sensitivity of our results to alternative cutoffs for short-,
medium-, and long-tenure groups. Choosing the wrong cutoff date weakens the
power of our statistical test and biases against finding the predicted differences if
they exist.18 We reran the models after changing the cutoff for long audit tenure to
8 and 10 years. In both cases the coefficient for long audit tenure remained insig-
nificant (p-value > 0.10). We also changed the cutoff for medium to 2 and 4 years.
Again, our primary result of lower financial-reporting quality for the short-tenure
group relative to the medium-tenure group was unaffected by the change in cutoff
dates to 4 years. However, the decline in financial-reporting quality of the short-
tenure group relative to the medium-tenure group was not significant when the

cutoff date used was 2 years. The lack of significance may be partially attributed to
the small sample size in the short-tenure group as a result of a change in the cutoff
date.
Persistence of accruals
To examine the impact of audit tenure on the persistence of the accrual component
of earnings using the matched sample, we modify the model used in Sloan 1996 to
permit the coefficients on operating cash flows and accruals to vary across tenure
groups. Specifically, we estimate the following cross-sectional model:
Eit + 1 /A it − 1 = γ 0 + γ 1(OCFit /A it − 1) + γ 2(TAit /A it − 1)

+ γ 3[(OCFit /A it − 1)*SHORTit] + γ 4[(OCFit /A it − 1)*LONGit]
+ γ 5[(TAit /A it − 1)*SHORTit] + γ 6[(TAit /A it − 1)*LONGit] + ε it + 1 (5),
where
Eit + 1 = earnings for company i at time t + 1, and
all other variables are as defined before.
The introduction of two interaction terms ((TAit /A it − 1)*SHORTit and (TAit /

A it − 1)*LONGit) in (5) allows the earnings persistence attributable to the accrual
component of earnings to differ between the short- and medium-tenure groups and
between the long- and medium-tenure groups. Based on Hypothesis 1 and Hypothesis
2, we predict a negative sign for γ5 and γ6, respectively.
Table 3 presents the results based on the matched sample. The adjusted R 2 of
(5) is 32.1 percent. The coefficients associated with current-period operating cash
flows and total accruals are positive and statistically significant (p-value < 0.01).
The earnings persistence attributable to the accrual component of earnings for the
medium-tenure group is 0.508, which is significant (p-value < 0.01). Comparable
estimates for the short- and long-tenure groups are 0.381 (γ 2 + γ 5) and 0.590 (γ 2
and γ6), respectively. Because both Hypothesis 1 and Hypothesis 2 predict lower
persistence of accruals for short- and long-tenure groups relative to the medium-
tenure groups, the statistical test of Hypothesis 1 (Hypothesis 2) is whether γ 5 (γ 6)
is negative. γ 5 is −0.127 and is significantly less than zero (p-value < 0.01). A sig-
nificantly negative coefficient on (TAit /A it − 1)*SHORTit interaction indicates that
the accrual component is less persistent for the clients of auditors with short tenure
than those for the clients of auditors with medium tenure. This result is consistent
with Hypothesis 1, which predicts lower financial-reporting quality associated
with short tenure. Alternatively, γ 6 is not statistically significant (p-value > 0.10),
thereby implying that accruals of clients with long-tenure auditors exhibit persis-
tence in earnings similar to those of clients with medium-tenure auditors. Like the
evidence on the unexpected accruals, this result is also inconsistent with the prediction
under Hypothesis 2. We are unable to find evidence of reduced financial-reporting
quality associated with long audit-firm tenures.

Sensitivity analysis
A further analysis is made to assure that the relation between tenure groups and the
persistence of accruals is not confounded by firm size. We plotted residuals from
(5) against total assets, and the plots showed no systematic pattern that would sug-
gest that the error terms are correlated with total assets.19 To test the sensitivity of
our results to differences in client age, we reestimated (5) after allowing the coeffi-
cients on cash flows and accrual to vary with age. The coefficients of the interaction
terms with age are not significant (p-value > 0.10), while the negative coefficient
on (TAit /A it − 1)*SHORTit remains significant (p-value > 0.05). To control for the
possibility that large accruals and losses may be less persistent by their very
TABLE 3
Effects of auditor tenure on the persistence of the accrual component of earnings
Eit + 1/A it − 1 = γ 0 + γ 1(OCFit /A it − 1) + γ 2(TAit /A it − 1) + γ 3[(OCFit /A it − 1)*SHORTit]

+ γ 4[(OCFit /A it − 1)*LONGit] + γ 5[(TAit /A it − 1)*SHORTit]
+ γ 6[(TAit /A it − 1)*LONGit] + ε it + 1 (5)
Predicted Parameter
Independent variable Parameter sign estimate t-statistic
Intercept γ0 −0.001 −0.02

OCFit /A it − 1 γ1 + 0.734 20.81*
TAit /A it − 1 γ2 + 0.508 12.3*
(OCFit /A it − 1)*SHORTit γ3 ? −0.061 −1.12
(OCFit /A it − 1)*LONGit γ4 ? 0.088 1.17
(TAit /A it − 1)*SHORTit γ5 − −0.127 −2.71*
(TAit /A it − 1)*LONGit γ6 − 0.082 1.16
Adjusted R 2 0.321
F-statistic 144.1*
Notes:
* Significant at p-value < 0.01 (all one-tailed except for intercept).
Eit + 1/A it − 1 is the operating income for company i at time t + 1 divided by total assets for
company i at time t − 1.
OCFit /Ait − 1 is operating cash flows for company i at time t divided by total assets for
company i at time t − 1.
TAit /A it − 1 is the total accruals for company i at time t divided by total assets for company i
at time t − 1.
SHORTit is one when the length of the auditor–client relationship is short (two or three
years); zero otherwise.
LONGit is one when the length of the auditor–client relationship is long (nine years or
longer); zero otherwise.

nature, we reestimated (5) by introducing two additional control variables, the

absolute value of total accruals and a dummy for the presence of losses.20 The nega-
tive coefficient on (TAit /A it − 1)*SHORTit continues to be significant ( p-value > 0.01).
Analogous to the sensitivity tests for unexpected accruals, we reran the models
after changing the cutoff for long audit tenure to 8 and 10 years. In both cases the
coefficient for long audit tenure remained insignificant (p-value > 0.10). We also
changed the cutoff for medium to 2 and 4 years. Again, our primary result of lower
financial reporting quality for the short-tenure group relative to the medium-tenure
group was unaffected by the change in cutoff dates to 4 years. However, the decline
in financial reporting quality of the short-tenure group relative to the medium-
tenure group was not significant when the cutoff date used was 2 years. The lack of
significance may be partially attributed to the small sample size in the short-tenure
group as a result of a change in the cutoff date.
5. Conclusions
The research question in this study is whether audit-firm tenure is related to financial-
reporting quality. An audit of financial reports by a third party can enhance the
quality of the financial information reported by management (Dopuch and Simunic
1982; Watts and Zimmerman 1986). However, either a learning curve with respect
to client-specific knowledge or incentives that stress retaining and profiting from
new clients could diminish the audit’s contribution to the quality of financial
reports. Similarly, a shift in incentives resulting from a sense of complacency or a
learned confidence in the client could gradually lessen the audit’s contribution to
the quality of financial reports.
To investigate this question, we examine the properties of reported accruals
for an industry- and size-matched sample of Big 6 clients that have been audited by
the same firm for two to three (short tenure), four to eight (medium tenure), or nine
or more (long tenure) years. Specifically, following prior research, we examine the
absolute value of the unexpected accruals and the extent to which reported accruals
persist in subsequent period’s reported income.21 Our results are consistent across
both measures. Short relationships between an audit firm and a client are associ-
ated with higher absolute levels of unexpected accruals and accruals that are less
persistent in subsequent earnings. Alternatively, we find no statistically significant
differences in the properties of the accruals reported by clients with medium or
long relationships with their auditors.
In addition to using a matched sample (which successfully created fairly
homogeneous samples across tenure categories), we also attempted to rule out the
most plausible alternative explanations for our results through multivariate analyses
and a variety of sensitivity tests. Although it is impossible to rule out all alternative
explanations for the short audit-tenure results, we are unable to identify a plausible
alternative explanation. The plausibility of an alternative explanation must be sub-
ject to the following three criteria. First, an alternative explanation should posit and
offer theoretical support for systematic differences in sample companies across
tenure categories. Second, the proposed difference must be related to the properties
of accruals used as proxies for financial-reporting quality in this study. Finally, the

proposed difference must not be adequately captured by the control variables

included in the multivariate analyses and sensitivity tests.
The results of our investigation may be pertinent to the debate surrounding the
issue of mandatory audit-firm rotation, which goes back at least 40 years (cf.
McClaren 1958) and seems to recur each decade (cf. Winters 1976; Metcalf com-
mittee 1976; Hoyle 1978; Dopuch et al. 2001). The tenor of the debate has changed
little over time. Those critical of the public accounting profession (usually in
response to one or more “audit failures”) argue that long relationships between
auditors and client management result in a decline in audit quality and are not in
the public’s interest. Mandatory auditor rotation is offered as a possible solution.
The profession in turn responds that rotation will increase audit costs and will not
improve audit quality (and in fact may reduce audit quality).
In the current regulatory regime, where audit-firm rotation is not mandatory,
we do not find that long audit-firm tenures are associated with a decline in financial-
reporting quality. Of course, it is possible that there is a decline in a dimension of
financial-reporting quality that is not captured by our empirical proxies, or that
more powerful tests might detect a difference. However, we did find a significant
decline in financial-reporting quality associated with short audit-firm tenures.
Whether the short tenure results generalize to a regulatory regime where audit-firm
rotation is mandatory depends on whether the reason for the observed decline in
financial-reporting quality is related to knowledge or incentives. To the extent that
the decline is due to an audit firm’s incentives to capture future revenue streams
from the client, mandatory rotation, by shortening the future revenue stream, could
mitigate the decline documented in this study. Alternatively, if the decline is due to
a lack of client-specific knowledge in the early years of an audit, the decline in
financial-reporting quality would also be likely to exist under mandatory rotation.
Endnotes
1. The debate is also not limited to the United States. Gietzmann and Sen (1997) review
the debate over mandatory rotation in European countries.
2. The use of analytical procedures provides an example of the importance of client-
specific knowledge. The effectiveness of analytical procedures is dependent on the
precision of the a priori expectation formed by the auditor. Kinney and McDaniel
(1996) note the importance of client data relationships in forming expectations. Client-
specific knowledge also affects the evaluation of explanations regarding observed
differences. A lack of client-specific knowledge may increase the likelihood of
accepting (incorrect) client-provided non-error explanations.
3. In testimony before the Metcalf subcommittee, then president of the American Institute
of Certified Public Accountants (AICPA), Wallace Olson, stated that the most effective
audits are generally performed by auditors who have a thorough knowledge of the
business entity gained over a considerable period of years (AICPA 1978).
4. Prior research suggests that audit firms spend more hours on initial engagements
(Palmrose 1989; Deis and Giroux 1996).
5. Interestingly, even the staunchest critics of long audit tenures have not suggested that
an auditor would be less likely to report (or require the correction of) a material

misstatement if one is found. However, the nature of the incentive trade-off problem
for the auditor is the same.
6. As mentioned, these arguments are all ceteris paribus. If other factors (such as
declining financial health) increase the salience of litigation risk, these factors are
likely to result in increased effort. Accordingly, the financial condition of the client (as
a proxy for the auditor’s perceived litigation risk) will be a critical control variable in
the analysis of the effects of audit-firm tenure.
7. An anonymous reviewer suggested that we also report results based on the signed level
of unexpected accruals. We do not (nor did we expect to) find any statistically
significant differences (p-value > 0.10) in signed unexpected accruals between short-
and medium-tenure groups, and between long- and medium-tenure groups.
8. Although we conduct and report several sensitivity checks to test the robustness of our
results, measurement error in the unexpected accruals proxy is a plausible alternative
explanation for tests using the absolute value of unexpected accruals.
9. We measure audit tenure as the length of the relationship between the audit firm and
the client. Although there may be turnover of individual auditors, including the audit
partner, the economic incentives of the firm with respect to the client do not change.
Similarly, audit approaches and working papers survive the turnover of individual
auditors.
10. During the sample period covered in this study, four of the Big 8 firms merged to form
the Big 6. In the event of a merger of auditors, the auditor–client relationship was
viewed as a continuing one with the new combined auditing firm. A switch of auditors
would only occur if the client then switched to a new auditor with no ties to either of
the premerger auditors.
11. The cutoff for medium versus long tenure is admittedly somewhat arbitrary. Arrunada
and Paz-Ares (1997) document that in European countries with mandatory auditor-
rotation rules rotation is required after a 9- to 12-year period.
12. For our sample period auditor information is not available on COMPUSTAT for banks,
life insurance, property, casualty companies, and public utilities. Accordingly, this
screen resulted in the elimination of most regulated industries.
13. An examination of the number of observations across tenure-years reveals that long
audit tenures are the norm. This finding is consistent with Beck et al. 1988, who
document that over an 11-year period, approximately 83 percent of the companies
examined had been with the same auditor for 9 or more years. They interpret this result
as being consistent with economic bonding between auditors and their clients.
14. Table 1 is based on a sample of 2,463 (821 in each tenure group) firm observations.
Descriptive statistics for the sample companies used to test Hypothesis 1 and
Hypothesis 2 on the other proxy of financial reporting quality (namely, the persistence
of accruals measure) are similar to those reported in Table 1 and are therefore not
reported.
15. We also estimated (4) using the log of the age variable (instead of the age variable) as a
control variable, with results essentially the same as those reported. Analysis was also
performed using the absolute value of operating cash flows rather than the signed cash
flows as a control variable and results were similar to those reported in the paper.

16. The elements of the Z-score with their associated weightings (in parentheses) are as
follows (Altman and McGough 1974): working capital/total assets (1.2), retained
earnings/total assets (1.4), earnings before interest and taxes/total assets (3.3), market
value of equity/book value of total debt (0.6), and sales/total assets (1.0). A lower Z-
score indicates greater financial distress.
17. We also reestimated (4) in Table 2 separately for the upper and lower half of client size
(split at the median values of total assets). The coefficient on SHORT remains positive
at p-value < 0.05 for both the smaller and upper halves of the sample. We also added
two additional terms to (4) in which the slope dummy variables SHORT and LONG are
interacted with an indicator variable for firm size (coded one if total assets are greater
than median total assets in the sample, zero otherwise). The interaction terms are not
significant (p-value > 0 .10), although SHORT continues to be positive (p-value < 0.05).
18. For example, assume that a decline in financial-reporting quality does occur for long
audit tenure, but that it does not occur until year 12. Choosing year 9 as the cutoff
includes years where there was no decline in financial-reporting quality in the long-
tenure group. The effect is to make the long- and medium-tenure groups more alike,
and to reduce the likelihood of detecting the actual difference.
19. We also reestimated (5) separately for the upper and lower half of client size (split at
the median values of total assets). The reestimation of (5) shows that the coefficient on
[(TAit /Ait − 1)*SHORTit ] remains negatively associated with one-year-ahead earnings
accruals at p-value < 0.05 for both the smaller and larger halves of the sample, and the
coefficient on [(TAit /Ait − 1)*LONGit] continues to be insignificant (p-value > 0.10) for
both halves of the sample.
20. We thank an anonymous reviewer for suggesting that we reestimate (5) with the two
additional control variables.
21. An anonymous reviewer suggested the possibility of examining an earnings smoothing
measure that focuses on the effect of unexpected accruals on the volatility of earnings.
However, computing a smoothing measure using a variability metric is problematic in
the context of our study because a variability metric requires annual earnings
observations over several years (and tenure is a temporal variable). An examination of
earnings smoothing across tenure categories is a potential avenue for future research.
References
Altman, E., and T. McGough. 1974. Evaluation of a company as a going concern. Journal of
Accountancy 138 (12): 50–7.
American Institute of Certified Public Accountants (AICPA). 1978. The commission on
auditors’ responsibilities: Report, conclusions, and recommendations. New York:
AICPA.
American Institute of Certified Public Accountants (AICPA) SEC Practice Section. 1992.
Statement of position regarding mandatory rotation of audit firms of publicly held
companies. New York: AICPA.
Antle, R., and B. Nalebuff. 1991. Conservatism and auditor-client negotiations. Journal of
Accounting Research 29 (Supplement): 31–54.

Arrunada, B., and C. Paz-Ares. 1997. Mandatory rotation of company auditors: A critical
examination. International Review of Law and Economics 17 (1): 31–61.
Balachandran, B., and N. Nagarajan. 1987. Imperfect information, insurance, and auditors
legal liability. Contemporary Accounting Research 3 (2): 281–301.
Bartov, E., F. Gul, and J. Tsui. 2000. Discretionary-accruals models and audit qualifications.
Journal of Accounting and Economics 30 (3): 421–52.
Beck, P., T. Frecka, and I. Solomon. 1988. A model of the market for MAS and audit
services: Knowledge spillovers and auditor-auditee bonding. Journal of Accounting
Literature 7: 50–64.
Becker, C., M. DeFond, J. Jiambalvo, and K. Subramanyam. 1998. The effect of audit
quality on earnings management. Contemporary Accounting Research 15 (1): 1–24.
Beasley, M., and K. Petroni. 1996. Errors in accounting estimates and their relation to
auditor firm type. Journal of Accounting Research 34 (Spring): 151–71.
Choi, S., and D. Jeter. 1992. The effects of qualified audit opinions on earnings response
coefficients. Journal of Accounting and Economics 15 (2): 229–47.
Chow, C., and S. Rice. 1982. Qualified audit opinions and auditor switching. Accounting
Review 57 (2): 326–36.
Citron, D., and R. Taffler. 1992. The audit report under going concern uncertainties: An
empirical analysis. Accounting and Business Research 22 (88): 337–46.
Copley, P., and M. Doucet. 1993. Auditor tenure, fixed fee contracts, and the supply of
substandard single audits. Public Budgeting and Finance 13 (3): 23–35.
DeAngelo, L. 1981. Auditor size and audit quality. Journal of Accounting and Economics
3 (3): 183–99.
Dechow, P. 1994. Accounting earnings and cash flows as measures of firm performance: The
role of accounting accruals. Journal of Accounting and Economics 18 (1): 3–42.
Dechow, P., R. Sloan, and A. Sweeney. 1995. Detecting earnings management. Accounting
Review 70 (2): 193–225.
DeFond, M., and J. Jiambalvo. 1994. Debt covenant violation and manipulation of accruals.
Journal of Accounting and Economics 17 (1–2): 145–76.
DeFond, M., and K. Subramanyam. 1998. Auditor changes and discretionary accruals.
Journal of Accounting and Economics 25 (1): 35–67.
Deis, D., and G. Giroux. 1992. Determinants of audit quality in the public sector.
Accounting Review 67 (3): 462–79.
Deis, D., and G. Giroux. 1996. The effect of auditor changes on audit fees, audit hours and
audit quality. Journal of Accounting and Public Policy 15 (1): 55–76.
DeJong, D. 1985. Class-action privileges and contingent legal fees: Investor and lawyer
incentives to litigate and the effect on audit quality. Journal of Accounting and Public
Policy 4 (3): 175–200.
Dopuch, N., and D. Simunic. 1982. The competition in auditing: An assessment. In Fourth
Symposium on Auditing Research, 401–50. Urbana: University of Illinois.
Dopuch, N., R. King, and R. Schwartz. 2001. An experimental investigation of retention and
rotation requirements. Journal of Accounting Research 38 (June): 93–117.
Francis, J., E. Maydew, and H. Sparks. 1999. The role of big-six auditors in the credible
reporting of accruals. Auditing: A Journal of Practice and Theory 18 (2): 17–34.

Freeman, R., J. Ohlson, and S. Penman. 1982. Book rate of return and prediction of earnings
changes. Journal of Accounting Research 20 (Autumn): 639–53.
Geiger, M., and K. Raghunandan. 2002. Auditor tenure and auditor reporting failures.
Auditing: A Journal of Practice and Theory 21 (1): 67–78.
Gietzmann, M., and P. Sen. 1997. Auditor rotation: an economic analysis of the European
Union’s proposed fifth directive. Working paper, London School of Business.
Giroux, G., D. Deis, and B. Bryan. 1995. The effect of peer review on audit economies.
Research in Accounting Regulation 9: 63–82.
Haskins, M., and D. Williams. 1990. A contingent model of intra-big eight auditor changes.
Auditing: A Journal of Practice and Theory 9 (3): 55–74.
Hoyle, J. 1978. Mandatory auditor rotation: The arguments and an alternative. Journal of
Accountancy 145 (5): 69–78.
Jones, J. 1991. Earnings management during import relief investigations. Journal of
Accounting Research 29 (Autumn): 193–228.
Kinney, W., and L. McDaniel. 1996. How to improve the effectiveness of substantive
analytical procedures. CPA Journal 66 (4): 52–4.
Klein, A. 2002. Audit committee, board of director characteristics, and earnings
management. Working paper, New York University.
Knapp, M. 1991. Factors that audit committees use as surrogates for audit quality. Auditing:
A Journal of Practice and Theory 10 (1): 35–52.
Krishnan, J. 1994. Auditor switching and conservatism. Accounting Review 69 (1): 200–15.
McClaren, L. 1958. Rotation of auditors. Journal of Accountancy 106 (4): 41–4.
Melumad, N., and L. Thoman. 1990. On auditors and the courts in an adverse selection
setting. Journal of Accounting Research 28 (Spring): 77–120.
Narayanan, V. 1994. An analysis of auditor liability rules. Journal of Accounting Research
32 (Supplement): 39–64.
Nelson, J., J. Ronen, and L. White. 1988. Legal liabilities and the market for auditing
services. Journal of Accounting, Auditing and Finance 3 (3): 255–95.
O’Keefe, T., R. King, and K. Gaver. 1994. Audit fees, industry specialization, and
compliance with GAAS reporting standards. Auditing: A Journal of Practice and
Theory 13 (2): 41–55.
Palmrose, Z. 1989. The relation of audit contract type to audit fees and hours. Accounting
Review 64 (3): 488–99.
Reynolds, K., and J. Francis. 2000. Does size matter? The influence of large clients on
office-level auditor reporting decisions. Journal of Accounting and Economics 30 (3):
375–400.
Ronnen, U. 1996. The effects of mandated versus voluntary auditing policy on the quality of
auditing. Journal of Accounting, Auditing and Finance 11 (3): 361–92.
Shockley, R. 1981. Perceptions of auditors independence: An empirical analysis.
Accounting Review 56 (4): 785–800.
Sloan, R. 1996. Do stock prices fully reflect information in accruals and cash flows about
future earnings? Accounting Review 71 (3): 289–315.
St. Pierre, K., and J. Anderson. 1984. An analysis of the factors associated with lawsuits
against public accountants. Accounting Review 59 (2): 242–63.

Stice, J. 1991. Using financial and market information to identify pre-engagement factors
associated with lawsuits against auditors. Accounting Review 66 (3): 516–33.
Subramanyam, K. 1996. The pricing of discretionary accruals. Journal of Accounting and
Economics 22 (1–3): 249–91.
Subramanyam, K., and J. Wild. 1996. Going-concern status, earnings persistence, and
informativeness of earnings. Contemporary Accounting Research 13 (1): 251–73.
Teoh, S., I. Welch, and T. Wong. 1998. Earnings management and the underperformance of
seasoned equity offerings. Journal of Financial Economics 50 (1): 63–99.
Teoh, S., and T. Wong. 1993. Perceived auditor quality and the earnings response
coefficient. Accounting Review 68 (2): 346–67.
U.S. Governmental Accounting Office (GAO). 1996. The accounting profession major
issues: Progress and concerns. Report to the ranking minority member, Committee on
Commerce, House of Representatives. Washington, DC: GAO.
U.S. Governmental Accounting Office (GAO). 1991. Failed banks: Accounting and auditing
reforms urgently needed. Washington, DC: GAO.
United States Senate (Metcalf committee). 1976. The accounting establishment: A staff
study prepared by the subcommittee on reports, accounting, and management of the
committee on government operations. Washington, DC: U.S. Government Printing
Office.
Warfield, T., J. Wild, and K. Wild. 1995. Managerial ownership, accounting choices, and
informativeness of earnings. Journal of Accounting and Economics 20 (1): 61–92.
Watts, R., and J. Zimmerman. 1986. Positive accounting theory. Englewood Cliffs, NJ:
Prentice Hall.
Winters, A. 1976. Looking at the auditor rotation issue. Management Accounting 57 (9):
29–30.

Johnson - Audit Firm Tenure and The Quality of Financial Reports (For Madotory Rotation)

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Johnson - Audit Firm Tenure and The Quality of Financial Reports (For Madotory Rotation)

Transféré par

Droits d'auteur :

Formats disponibles

Audit-Firm Tenure and the Quality of

VAN E. JOHNSON, Georgia State University

INDER K. KHURANA, University of Missouri-Columbia

J. KENNETH REYNOLDS, Louisiana State University

Keywords Audit-firm tenure; Audit quality; Financial-reporting quality; Mandatory auditor

CAR Vol. 19 No. 4 (Winter 2002)

CAR Vol. 19 No. 4 (Winter 2002)

CAR Vol. 19 No. 4 (Winter 2002)

Short audit-firm tenure

CAR Vol. 19 No. 4 (Winter 2002)

HYPOTHESIS 1. The quality of financial reports issued by a company will be

Long audit-firm tenure

CAR Vol. 19 No. 4 (Winter 2002)

HYPOTHESIS 2. The quality of financial reports issued by a company will be

The available empirical evidence regarding both hypotheses is limited. After

Proxies for financial-reporting quality

Absolute value of unexpected accruals

CAR Vol. 19 No. 4 (Winter 2002)

also affords management substantial flexibility in reporting accrual-based earnings.

The persistence of the accrual components of earnings

Earningst + 1 = α 0 + α1Earningst + ∈t + 1 (1).

CAR Vol. 19 No. 4 (Winter 2002)

Measurement of audit-firm tenure

CAR Vol. 19 No. 4 (Winter 2002)

CAR Vol. 19 No. 4 (Winter 2002)

2,463 (2,280) firm-year observations for the unexpected accruals (persistence)

CAR Vol. 19 No. 4 (Winter 2002)

Short tenure* Medium tenure† Long tenure‡

CAR Vol. 19 No. 4 (Winter 2002)

Level of unexpected accruals

UAit = TAit /Ait − 1 − {φ1jt(1/Ait − 1) + φ 2jt [(∆REVit − ∆ARit )/Ait − 1]

where TA is total accruals, A is total assets, REV is total sales revenues, AR is

TAit /Ait − 1 = φ1jt (1/Ait − 1) + φ 2jt [(∆REVit − ∆ARit )/Ait − 1]

where all variables are defined as above.

|UAit| = β 0 + β1(OCFit /Ait − 1) + β2LAit + β 3LEVit + β4FCit + β 5AGEit

|UAit | = absolute value of unexpected accruals for company i at time t,

CAR Vol. 19 No. 4 (Winter 2002)

|UAit| = β 0 + β1(OCFit /Ait − 1) + β2LAit + β 3LEVit + β4FCit + β 5AGEit + β6SHORTit

Intercept β0 0.009 13.71*

CAR Vol. 19 No. 4 (Winter 2002)

CAR Vol. 19 No. 4 (Winter 2002)

Eit + 1 /A it − 1 = γ 0 + γ 1(OCFit /A it − 1) + γ 2(TAit /A it − 1)

Eit + 1 = earnings for company i at time t + 1, and

all other variables are as defined before.

The introduction of two interaction terms ((TAit /A it − 1)*SHORTit and (TAit /

CAR Vol. 19 No. 4 (Winter 2002)

Eit + 1/A it − 1 = γ 0 + γ 1(OCFit /A it − 1) + γ 2(TAit /A it − 1) + γ 3[(OCFit /A it − 1)*SHORTit]

Intercept γ0 −0.001 −0.02

CAR Vol. 19 No. 4 (Winter 2002)

nature, we reestimated (5) by introducing two additional control variables, the

CAR Vol. 19 No. 4 (Winter 2002)

proposed difference must not be adequately captured by the control variables

CAR Vol. 19 No. 4 (Winter 2002)

CAR Vol. 19 No. 4 (Winter 2002)

CAR Vol. 19 No. 4 (Winter 2002)

CAR Vol. 19 No. 4 (Winter 2002)

CAR Vol. 19 No. 4 (Winter 2002)

CAR Vol. 19 No. 4 (Winter 2002)

Vous aimerez peut-être aussi